{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 7813,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00012799180852425445,
      "grad_norm": 11.5625,
      "learning_rate": 8.510638297872341e-06,
      "loss": 15.304,
      "step": 1
    },
    {
      "epoch": 0.0002559836170485089,
      "grad_norm": 12.875,
      "learning_rate": 1.7021276595744682e-05,
      "loss": 15.0659,
      "step": 2
    },
    {
      "epoch": 0.00038397542557276333,
      "grad_norm": 13.3125,
      "learning_rate": 2.5531914893617022e-05,
      "loss": 15.3122,
      "step": 3
    },
    {
      "epoch": 0.0005119672340970178,
      "grad_norm": 10.1875,
      "learning_rate": 3.4042553191489365e-05,
      "loss": 15.3051,
      "step": 4
    },
    {
      "epoch": 0.0006399590426212722,
      "grad_norm": 11.0625,
      "learning_rate": 4.2553191489361704e-05,
      "loss": 15.0721,
      "step": 5
    },
    {
      "epoch": 0.0007679508511455267,
      "grad_norm": 11.5,
      "learning_rate": 5.1063829787234044e-05,
      "loss": 15.0955,
      "step": 6
    },
    {
      "epoch": 0.0008959426596697811,
      "grad_norm": 8.9375,
      "learning_rate": 5.957446808510639e-05,
      "loss": 14.8674,
      "step": 7
    },
    {
      "epoch": 0.0010239344681940356,
      "grad_norm": 7.75,
      "learning_rate": 6.808510638297873e-05,
      "loss": 14.8543,
      "step": 8
    },
    {
      "epoch": 0.00115192627671829,
      "grad_norm": 9.0,
      "learning_rate": 7.659574468085106e-05,
      "loss": 14.8684,
      "step": 9
    },
    {
      "epoch": 0.0012799180852425445,
      "grad_norm": 8.25,
      "learning_rate": 8.510638297872341e-05,
      "loss": 14.8875,
      "step": 10
    },
    {
      "epoch": 0.001407909893766799,
      "grad_norm": 8.125,
      "learning_rate": 9.361702127659574e-05,
      "loss": 14.8473,
      "step": 11
    },
    {
      "epoch": 0.0015359017022910533,
      "grad_norm": 8.0,
      "learning_rate": 0.00010212765957446809,
      "loss": 14.4045,
      "step": 12
    },
    {
      "epoch": 0.0016638935108153079,
      "grad_norm": 7.15625,
      "learning_rate": 0.00011063829787234042,
      "loss": 14.5803,
      "step": 13
    },
    {
      "epoch": 0.0017918853193395622,
      "grad_norm": 7.9375,
      "learning_rate": 0.00011914893617021278,
      "loss": 14.0811,
      "step": 14
    },
    {
      "epoch": 0.0019198771278638167,
      "grad_norm": 7.875,
      "learning_rate": 0.0001276595744680851,
      "loss": 14.1291,
      "step": 15
    },
    {
      "epoch": 0.0020478689363880712,
      "grad_norm": 9.125,
      "learning_rate": 0.00013617021276595746,
      "loss": 13.6935,
      "step": 16
    },
    {
      "epoch": 0.0021758607449123258,
      "grad_norm": 11.75,
      "learning_rate": 0.00014468085106382977,
      "loss": 13.5914,
      "step": 17
    },
    {
      "epoch": 0.00230385255343658,
      "grad_norm": 10.9375,
      "learning_rate": 0.00015319148936170213,
      "loss": 13.6709,
      "step": 18
    },
    {
      "epoch": 0.0024318443619608344,
      "grad_norm": 13.6875,
      "learning_rate": 0.00016170212765957449,
      "loss": 13.1205,
      "step": 19
    },
    {
      "epoch": 0.002559836170485089,
      "grad_norm": 12.9375,
      "learning_rate": 0.00017021276595744682,
      "loss": 13.4647,
      "step": 20
    },
    {
      "epoch": 0.0026878279790093435,
      "grad_norm": 11.0625,
      "learning_rate": 0.00017872340425531915,
      "loss": 12.7896,
      "step": 21
    },
    {
      "epoch": 0.002815819787533598,
      "grad_norm": 12.3125,
      "learning_rate": 0.00018723404255319148,
      "loss": 12.3087,
      "step": 22
    },
    {
      "epoch": 0.002943811596057852,
      "grad_norm": 8.0,
      "learning_rate": 0.00019574468085106384,
      "loss": 12.8961,
      "step": 23
    },
    {
      "epoch": 0.0030718034045821067,
      "grad_norm": 9.5,
      "learning_rate": 0.00020425531914893618,
      "loss": 11.93,
      "step": 24
    },
    {
      "epoch": 0.003199795213106361,
      "grad_norm": 7.21875,
      "learning_rate": 0.0002127659574468085,
      "loss": 12.0264,
      "step": 25
    },
    {
      "epoch": 0.0033277870216306157,
      "grad_norm": 7.46875,
      "learning_rate": 0.00022127659574468084,
      "loss": 11.706,
      "step": 26
    },
    {
      "epoch": 0.0034557788301548703,
      "grad_norm": 7.8125,
      "learning_rate": 0.0002297872340425532,
      "loss": 11.9963,
      "step": 27
    },
    {
      "epoch": 0.0035837706386791244,
      "grad_norm": 7.4375,
      "learning_rate": 0.00023829787234042556,
      "loss": 11.552,
      "step": 28
    },
    {
      "epoch": 0.003711762447203379,
      "grad_norm": 8.75,
      "learning_rate": 0.00024680851063829787,
      "loss": 11.6734,
      "step": 29
    },
    {
      "epoch": 0.0038397542557276334,
      "grad_norm": 6.875,
      "learning_rate": 0.0002553191489361702,
      "loss": 11.8145,
      "step": 30
    },
    {
      "epoch": 0.003967746064251888,
      "grad_norm": 17.75,
      "learning_rate": 0.00026382978723404253,
      "loss": 12.1279,
      "step": 31
    },
    {
      "epoch": 0.0040957378727761425,
      "grad_norm": 7.71875,
      "learning_rate": 0.0002723404255319149,
      "loss": 11.6992,
      "step": 32
    },
    {
      "epoch": 0.004223729681300397,
      "grad_norm": 6.0625,
      "learning_rate": 0.00028085106382978725,
      "loss": 11.4785,
      "step": 33
    },
    {
      "epoch": 0.0043517214898246516,
      "grad_norm": 10.125,
      "learning_rate": 0.00028936170212765953,
      "loss": 11.3125,
      "step": 34
    },
    {
      "epoch": 0.004479713298348906,
      "grad_norm": 8.6875,
      "learning_rate": 0.0002978723404255319,
      "loss": 11.2372,
      "step": 35
    },
    {
      "epoch": 0.00460770510687316,
      "grad_norm": 5.65625,
      "learning_rate": 0.00030638297872340425,
      "loss": 11.3262,
      "step": 36
    },
    {
      "epoch": 0.004735696915397414,
      "grad_norm": 5.59375,
      "learning_rate": 0.00031489361702127664,
      "loss": 11.1482,
      "step": 37
    },
    {
      "epoch": 0.004863688723921669,
      "grad_norm": 5.90625,
      "learning_rate": 0.00032340425531914897,
      "loss": 10.9973,
      "step": 38
    },
    {
      "epoch": 0.004991680532445923,
      "grad_norm": 7.46875,
      "learning_rate": 0.00033191489361702125,
      "loss": 11.0743,
      "step": 39
    },
    {
      "epoch": 0.005119672340970178,
      "grad_norm": 5.375,
      "learning_rate": 0.00034042553191489364,
      "loss": 11.0801,
      "step": 40
    },
    {
      "epoch": 0.005247664149494432,
      "grad_norm": 6.59375,
      "learning_rate": 0.00034893617021276597,
      "loss": 10.7041,
      "step": 41
    },
    {
      "epoch": 0.005375655958018687,
      "grad_norm": 5.625,
      "learning_rate": 0.0003574468085106383,
      "loss": 10.452,
      "step": 42
    },
    {
      "epoch": 0.0055036477665429415,
      "grad_norm": 4.6875,
      "learning_rate": 0.00036595744680851063,
      "loss": 10.4645,
      "step": 43
    },
    {
      "epoch": 0.005631639575067196,
      "grad_norm": 5.25,
      "learning_rate": 0.00037446808510638297,
      "loss": 10.4276,
      "step": 44
    },
    {
      "epoch": 0.005759631383591451,
      "grad_norm": 6.34375,
      "learning_rate": 0.0003829787234042553,
      "loss": 11.0687,
      "step": 45
    },
    {
      "epoch": 0.005887623192115704,
      "grad_norm": 6.75,
      "learning_rate": 0.0003914893617021277,
      "loss": 10.4207,
      "step": 46
    },
    {
      "epoch": 0.006015615000639959,
      "grad_norm": 6.09375,
      "learning_rate": 0.0004,
      "loss": 10.0271,
      "step": 47
    },
    {
      "epoch": 0.006143606809164213,
      "grad_norm": 5.5,
      "learning_rate": 0.00040851063829787235,
      "loss": 10.7434,
      "step": 48
    },
    {
      "epoch": 0.006271598617688468,
      "grad_norm": 4.90625,
      "learning_rate": 0.0004170212765957447,
      "loss": 10.2806,
      "step": 49
    },
    {
      "epoch": 0.006399590426212722,
      "grad_norm": 5.8125,
      "learning_rate": 0.000425531914893617,
      "loss": 10.2049,
      "step": 50
    },
    {
      "epoch": 0.006527582234736977,
      "grad_norm": 5.21875,
      "learning_rate": 0.0004340425531914894,
      "loss": 10.5081,
      "step": 51
    },
    {
      "epoch": 0.0066555740432612314,
      "grad_norm": 4.46875,
      "learning_rate": 0.0004425531914893617,
      "loss": 10.194,
      "step": 52
    },
    {
      "epoch": 0.006783565851785486,
      "grad_norm": 4.4375,
      "learning_rate": 0.000451063829787234,
      "loss": 9.9446,
      "step": 53
    },
    {
      "epoch": 0.0069115576603097405,
      "grad_norm": 3.765625,
      "learning_rate": 0.0004595744680851064,
      "loss": 9.823,
      "step": 54
    },
    {
      "epoch": 0.007039549468833995,
      "grad_norm": 5.0,
      "learning_rate": 0.00046808510638297874,
      "loss": 9.4676,
      "step": 55
    },
    {
      "epoch": 0.007167541277358249,
      "grad_norm": 4.875,
      "learning_rate": 0.0004765957446808511,
      "loss": 9.8968,
      "step": 56
    },
    {
      "epoch": 0.007295533085882503,
      "grad_norm": 5.53125,
      "learning_rate": 0.0004851063829787234,
      "loss": 9.6318,
      "step": 57
    },
    {
      "epoch": 0.007423524894406758,
      "grad_norm": 4.6875,
      "learning_rate": 0.0004936170212765957,
      "loss": 9.6559,
      "step": 58
    },
    {
      "epoch": 0.007551516702931012,
      "grad_norm": 6.15625,
      "learning_rate": 0.0005021276595744681,
      "loss": 9.6202,
      "step": 59
    },
    {
      "epoch": 0.007679508511455267,
      "grad_norm": 3.765625,
      "learning_rate": 0.0005106382978723404,
      "loss": 9.545,
      "step": 60
    },
    {
      "epoch": 0.007807500319979521,
      "grad_norm": 5.46875,
      "learning_rate": 0.0005191489361702127,
      "loss": 9.8627,
      "step": 61
    },
    {
      "epoch": 0.007935492128503776,
      "grad_norm": 4.34375,
      "learning_rate": 0.0005276595744680851,
      "loss": 9.7528,
      "step": 62
    },
    {
      "epoch": 0.00806348393702803,
      "grad_norm": 3.21875,
      "learning_rate": 0.0005361702127659575,
      "loss": 9.4818,
      "step": 63
    },
    {
      "epoch": 0.008191475745552285,
      "grad_norm": 4.59375,
      "learning_rate": 0.0005446808510638298,
      "loss": 9.5754,
      "step": 64
    },
    {
      "epoch": 0.008319467554076539,
      "grad_norm": 3.40625,
      "learning_rate": 0.0005531914893617022,
      "loss": 9.704,
      "step": 65
    },
    {
      "epoch": 0.008447459362600794,
      "grad_norm": 2.96875,
      "learning_rate": 0.0005617021276595745,
      "loss": 9.126,
      "step": 66
    },
    {
      "epoch": 0.008575451171125048,
      "grad_norm": 11.375,
      "learning_rate": 0.0005702127659574468,
      "loss": 9.9577,
      "step": 67
    },
    {
      "epoch": 0.008703442979649303,
      "grad_norm": 6.3125,
      "learning_rate": 0.0005787234042553191,
      "loss": 9.7423,
      "step": 68
    },
    {
      "epoch": 0.008831434788173557,
      "grad_norm": 6.09375,
      "learning_rate": 0.0005872340425531915,
      "loss": 9.7837,
      "step": 69
    },
    {
      "epoch": 0.008959426596697812,
      "grad_norm": 5.21875,
      "learning_rate": 0.0005957446808510638,
      "loss": 9.4718,
      "step": 70
    },
    {
      "epoch": 0.009087418405222066,
      "grad_norm": 2.953125,
      "learning_rate": 0.0006042553191489362,
      "loss": 9.6516,
      "step": 71
    },
    {
      "epoch": 0.00921541021374632,
      "grad_norm": 3.703125,
      "learning_rate": 0.0006127659574468085,
      "loss": 9.6938,
      "step": 72
    },
    {
      "epoch": 0.009343402022270575,
      "grad_norm": 4.03125,
      "learning_rate": 0.0006212765957446808,
      "loss": 9.1787,
      "step": 73
    },
    {
      "epoch": 0.009471393830794829,
      "grad_norm": 2.796875,
      "learning_rate": 0.0006297872340425533,
      "loss": 9.5102,
      "step": 74
    },
    {
      "epoch": 0.009599385639319084,
      "grad_norm": 2.78125,
      "learning_rate": 0.0006382978723404256,
      "loss": 9.433,
      "step": 75
    },
    {
      "epoch": 0.009727377447843338,
      "grad_norm": 4.25,
      "learning_rate": 0.0006468085106382979,
      "loss": 9.8887,
      "step": 76
    },
    {
      "epoch": 0.009855369256367593,
      "grad_norm": 2.578125,
      "learning_rate": 0.0006553191489361702,
      "loss": 9.3536,
      "step": 77
    },
    {
      "epoch": 0.009983361064891847,
      "grad_norm": 2.875,
      "learning_rate": 0.0006638297872340425,
      "loss": 9.356,
      "step": 78
    },
    {
      "epoch": 0.010111352873416102,
      "grad_norm": 3.734375,
      "learning_rate": 0.0006723404255319148,
      "loss": 9.0571,
      "step": 79
    },
    {
      "epoch": 0.010239344681940356,
      "grad_norm": 2.390625,
      "learning_rate": 0.0006808510638297873,
      "loss": 9.1523,
      "step": 80
    },
    {
      "epoch": 0.01036733649046461,
      "grad_norm": 2.546875,
      "learning_rate": 0.0006893617021276596,
      "loss": 9.4769,
      "step": 81
    },
    {
      "epoch": 0.010495328298988865,
      "grad_norm": 2.8125,
      "learning_rate": 0.0006978723404255319,
      "loss": 9.2867,
      "step": 82
    },
    {
      "epoch": 0.010623320107513119,
      "grad_norm": 3.125,
      "learning_rate": 0.0007063829787234043,
      "loss": 9.328,
      "step": 83
    },
    {
      "epoch": 0.010751311916037374,
      "grad_norm": 3.1875,
      "learning_rate": 0.0007148936170212766,
      "loss": 9.1725,
      "step": 84
    },
    {
      "epoch": 0.010879303724561628,
      "grad_norm": 2.421875,
      "learning_rate": 0.000723404255319149,
      "loss": 9.0622,
      "step": 85
    },
    {
      "epoch": 0.011007295533085883,
      "grad_norm": 5.4375,
      "learning_rate": 0.0007319148936170213,
      "loss": 9.0586,
      "step": 86
    },
    {
      "epoch": 0.011135287341610137,
      "grad_norm": 2.421875,
      "learning_rate": 0.0007404255319148936,
      "loss": 9.1943,
      "step": 87
    },
    {
      "epoch": 0.011263279150134392,
      "grad_norm": 2.703125,
      "learning_rate": 0.0007489361702127659,
      "loss": 8.9902,
      "step": 88
    },
    {
      "epoch": 0.011391270958658646,
      "grad_norm": 2.59375,
      "learning_rate": 0.0007574468085106383,
      "loss": 9.2359,
      "step": 89
    },
    {
      "epoch": 0.011519262767182901,
      "grad_norm": 3.375,
      "learning_rate": 0.0007659574468085106,
      "loss": 8.8138,
      "step": 90
    },
    {
      "epoch": 0.011647254575707155,
      "grad_norm": 2.8125,
      "learning_rate": 0.000774468085106383,
      "loss": 9.4292,
      "step": 91
    },
    {
      "epoch": 0.011775246384231408,
      "grad_norm": 2.625,
      "learning_rate": 0.0007829787234042554,
      "loss": 9.4025,
      "step": 92
    },
    {
      "epoch": 0.011903238192755664,
      "grad_norm": 3.90625,
      "learning_rate": 0.0007914893617021277,
      "loss": 9.2509,
      "step": 93
    },
    {
      "epoch": 0.012031230001279918,
      "grad_norm": 2.328125,
      "learning_rate": 0.0008,
      "loss": 9.0982,
      "step": 94
    },
    {
      "epoch": 0.012159221809804173,
      "grad_norm": 3.4375,
      "learning_rate": 0.0008085106382978723,
      "loss": 9.3691,
      "step": 95
    },
    {
      "epoch": 0.012287213618328427,
      "grad_norm": 2.25,
      "learning_rate": 0.0008170212765957447,
      "loss": 8.8108,
      "step": 96
    },
    {
      "epoch": 0.012415205426852682,
      "grad_norm": 2.09375,
      "learning_rate": 0.000825531914893617,
      "loss": 8.7706,
      "step": 97
    },
    {
      "epoch": 0.012543197235376936,
      "grad_norm": 2.8125,
      "learning_rate": 0.0008340425531914894,
      "loss": 9.2015,
      "step": 98
    },
    {
      "epoch": 0.012671189043901191,
      "grad_norm": 2.296875,
      "learning_rate": 0.0008425531914893617,
      "loss": 9.0042,
      "step": 99
    },
    {
      "epoch": 0.012799180852425445,
      "grad_norm": 2.171875,
      "learning_rate": 0.000851063829787234,
      "loss": 8.9005,
      "step": 100
    },
    {
      "epoch": 0.012927172660949698,
      "grad_norm": 4.1875,
      "learning_rate": 0.0008595744680851064,
      "loss": 8.7074,
      "step": 101
    },
    {
      "epoch": 0.013055164469473954,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0008680851063829788,
      "loss": 9.098,
      "step": 102
    },
    {
      "epoch": 0.013183156277998207,
      "grad_norm": 2.390625,
      "learning_rate": 0.0008765957446808511,
      "loss": 9.0511,
      "step": 103
    },
    {
      "epoch": 0.013311148086522463,
      "grad_norm": 2.921875,
      "learning_rate": 0.0008851063829787234,
      "loss": 9.0597,
      "step": 104
    },
    {
      "epoch": 0.013439139895046717,
      "grad_norm": 2.96875,
      "learning_rate": 0.0008936170212765957,
      "loss": 9.014,
      "step": 105
    },
    {
      "epoch": 0.013567131703570972,
      "grad_norm": 2.25,
      "learning_rate": 0.000902127659574468,
      "loss": 8.687,
      "step": 106
    },
    {
      "epoch": 0.013695123512095226,
      "grad_norm": 2.359375,
      "learning_rate": 0.0009106382978723405,
      "loss": 9.2103,
      "step": 107
    },
    {
      "epoch": 0.013823115320619481,
      "grad_norm": 2.609375,
      "learning_rate": 0.0009191489361702128,
      "loss": 9.077,
      "step": 108
    },
    {
      "epoch": 0.013951107129143735,
      "grad_norm": 2.984375,
      "learning_rate": 0.0009276595744680851,
      "loss": 9.2118,
      "step": 109
    },
    {
      "epoch": 0.01407909893766799,
      "grad_norm": 5.53125,
      "learning_rate": 0.0009361702127659575,
      "loss": 9.2105,
      "step": 110
    },
    {
      "epoch": 0.014207090746192244,
      "grad_norm": 2.703125,
      "learning_rate": 0.0009446808510638298,
      "loss": 8.9497,
      "step": 111
    },
    {
      "epoch": 0.014335082554716497,
      "grad_norm": 6.09375,
      "learning_rate": 0.0009531914893617022,
      "loss": 9.3107,
      "step": 112
    },
    {
      "epoch": 0.014463074363240753,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009617021276595745,
      "loss": 9.1432,
      "step": 113
    },
    {
      "epoch": 0.014591066171765006,
      "grad_norm": 4.0625,
      "learning_rate": 0.0009702127659574468,
      "loss": 9.3258,
      "step": 114
    },
    {
      "epoch": 0.014719057980289262,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0009787234042553192,
      "loss": 9.4424,
      "step": 115
    },
    {
      "epoch": 0.014847049788813516,
      "grad_norm": 2.234375,
      "learning_rate": 0.0009872340425531915,
      "loss": 8.9795,
      "step": 116
    },
    {
      "epoch": 0.014975041597337771,
      "grad_norm": 3.359375,
      "learning_rate": 0.000995744680851064,
      "loss": 8.8052,
      "step": 117
    },
    {
      "epoch": 0.015103033405862025,
      "grad_norm": 1.875,
      "learning_rate": 0.0010042553191489361,
      "loss": 9.0904,
      "step": 118
    },
    {
      "epoch": 0.01523102521438628,
      "grad_norm": 3.71875,
      "learning_rate": 0.0010127659574468084,
      "loss": 8.3967,
      "step": 119
    },
    {
      "epoch": 0.015359017022910534,
      "grad_norm": 2.171875,
      "learning_rate": 0.0010212765957446808,
      "loss": 9.2066,
      "step": 120
    },
    {
      "epoch": 0.015487008831434787,
      "grad_norm": 4.65625,
      "learning_rate": 0.0010297872340425532,
      "loss": 9.24,
      "step": 121
    },
    {
      "epoch": 0.015615000639959043,
      "grad_norm": 1.734375,
      "learning_rate": 0.0010382978723404255,
      "loss": 8.9721,
      "step": 122
    },
    {
      "epoch": 0.015742992448483296,
      "grad_norm": 1.9375,
      "learning_rate": 0.001046808510638298,
      "loss": 9.3257,
      "step": 123
    },
    {
      "epoch": 0.015870984257007552,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0010553191489361701,
      "loss": 8.7488,
      "step": 124
    },
    {
      "epoch": 0.015998976065531807,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0010638297872340426,
      "loss": 9.3335,
      "step": 125
    },
    {
      "epoch": 0.01612696787405606,
      "grad_norm": 1.8359375,
      "learning_rate": 0.001072340425531915,
      "loss": 9.3622,
      "step": 126
    },
    {
      "epoch": 0.016254959682580315,
      "grad_norm": 1.765625,
      "learning_rate": 0.0010808510638297872,
      "loss": 9.0444,
      "step": 127
    },
    {
      "epoch": 0.01638295149110457,
      "grad_norm": 1.65625,
      "learning_rate": 0.0010893617021276597,
      "loss": 8.8469,
      "step": 128
    },
    {
      "epoch": 0.016510943299628825,
      "grad_norm": 4.5,
      "learning_rate": 0.001097872340425532,
      "loss": 9.2005,
      "step": 129
    },
    {
      "epoch": 0.016638935108153077,
      "grad_norm": 2.0,
      "learning_rate": 0.0011063829787234043,
      "loss": 8.6386,
      "step": 130
    },
    {
      "epoch": 0.016766926916677333,
      "grad_norm": 3.65625,
      "learning_rate": 0.0011148936170212766,
      "loss": 9.0343,
      "step": 131
    },
    {
      "epoch": 0.016894918725201588,
      "grad_norm": 1.7890625,
      "learning_rate": 0.001123404255319149,
      "loss": 9.0456,
      "step": 132
    },
    {
      "epoch": 0.01702291053372584,
      "grad_norm": 3.34375,
      "learning_rate": 0.0011319148936170214,
      "loss": 8.9293,
      "step": 133
    },
    {
      "epoch": 0.017150902342250095,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0011404255319148937,
      "loss": 9.1034,
      "step": 134
    },
    {
      "epoch": 0.01727889415077435,
      "grad_norm": 2.125,
      "learning_rate": 0.0011489361702127661,
      "loss": 8.9307,
      "step": 135
    },
    {
      "epoch": 0.017406885959298606,
      "grad_norm": 1.828125,
      "learning_rate": 0.0011574468085106381,
      "loss": 8.9665,
      "step": 136
    },
    {
      "epoch": 0.017534877767822858,
      "grad_norm": 3.015625,
      "learning_rate": 0.0011659574468085106,
      "loss": 8.7356,
      "step": 137
    },
    {
      "epoch": 0.017662869576347114,
      "grad_norm": 1.9609375,
      "learning_rate": 0.001174468085106383,
      "loss": 8.8461,
      "step": 138
    },
    {
      "epoch": 0.01779086138487137,
      "grad_norm": 3.0,
      "learning_rate": 0.0011829787234042552,
      "loss": 8.7686,
      "step": 139
    },
    {
      "epoch": 0.017918853193395624,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0011914893617021277,
      "loss": 8.7767,
      "step": 140
    },
    {
      "epoch": 0.018046845001919876,
      "grad_norm": 2.0,
      "learning_rate": 0.0012,
      "loss": 8.629,
      "step": 141
    },
    {
      "epoch": 0.01817483681044413,
      "grad_norm": 1.515625,
      "learning_rate": 0.0012085106382978723,
      "loss": 8.7012,
      "step": 142
    },
    {
      "epoch": 0.018302828618968387,
      "grad_norm": 1.5625,
      "learning_rate": 0.0012170212765957448,
      "loss": 8.5763,
      "step": 143
    },
    {
      "epoch": 0.01843082042749264,
      "grad_norm": 1.7734375,
      "learning_rate": 0.001225531914893617,
      "loss": 9.1844,
      "step": 144
    },
    {
      "epoch": 0.018558812236016894,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0012340425531914894,
      "loss": 8.7128,
      "step": 145
    },
    {
      "epoch": 0.01868680404454115,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0012425531914893617,
      "loss": 8.6462,
      "step": 146
    },
    {
      "epoch": 0.018814795853065405,
      "grad_norm": 1.8984375,
      "learning_rate": 0.001251063829787234,
      "loss": 8.5614,
      "step": 147
    },
    {
      "epoch": 0.018942787661589657,
      "grad_norm": 2.96875,
      "learning_rate": 0.0012595744680851065,
      "loss": 8.9475,
      "step": 148
    },
    {
      "epoch": 0.019070779470113913,
      "grad_norm": 1.828125,
      "learning_rate": 0.0012680851063829788,
      "loss": 8.6816,
      "step": 149
    },
    {
      "epoch": 0.019198771278638168,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0012765957446808512,
      "loss": 8.9585,
      "step": 150
    },
    {
      "epoch": 0.01932676308716242,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0012851063829787234,
      "loss": 9.0933,
      "step": 151
    },
    {
      "epoch": 0.019454754895686675,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0012936170212765959,
      "loss": 8.6693,
      "step": 152
    },
    {
      "epoch": 0.01958274670421093,
      "grad_norm": 2.40625,
      "learning_rate": 0.0013021276595744683,
      "loss": 8.7413,
      "step": 153
    },
    {
      "epoch": 0.019710738512735186,
      "grad_norm": 1.625,
      "learning_rate": 0.0013106382978723403,
      "loss": 8.9221,
      "step": 154
    },
    {
      "epoch": 0.019838730321259438,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0013191489361702128,
      "loss": 8.8036,
      "step": 155
    },
    {
      "epoch": 0.019966722129783693,
      "grad_norm": 2.1875,
      "learning_rate": 0.001327659574468085,
      "loss": 8.2535,
      "step": 156
    },
    {
      "epoch": 0.02009471393830795,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0013361702127659574,
      "loss": 9.0529,
      "step": 157
    },
    {
      "epoch": 0.020222705746832204,
      "grad_norm": 1.578125,
      "learning_rate": 0.0013446808510638297,
      "loss": 9.0868,
      "step": 158
    },
    {
      "epoch": 0.020350697555356456,
      "grad_norm": 1.6328125,
      "learning_rate": 0.001353191489361702,
      "loss": 8.6749,
      "step": 159
    },
    {
      "epoch": 0.02047868936388071,
      "grad_norm": 2.5625,
      "learning_rate": 0.0013617021276595745,
      "loss": 8.9905,
      "step": 160
    },
    {
      "epoch": 0.020606681172404967,
      "grad_norm": 1.546875,
      "learning_rate": 0.0013702127659574468,
      "loss": 9.0812,
      "step": 161
    },
    {
      "epoch": 0.02073467298092922,
      "grad_norm": 2.109375,
      "learning_rate": 0.0013787234042553192,
      "loss": 8.9843,
      "step": 162
    },
    {
      "epoch": 0.020862664789453474,
      "grad_norm": 2.046875,
      "learning_rate": 0.0013872340425531914,
      "loss": 9.0285,
      "step": 163
    },
    {
      "epoch": 0.02099065659797773,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0013957446808510639,
      "loss": 9.0726,
      "step": 164
    },
    {
      "epoch": 0.021118648406501985,
      "grad_norm": 1.296875,
      "learning_rate": 0.0014042553191489363,
      "loss": 8.6532,
      "step": 165
    },
    {
      "epoch": 0.021246640215026237,
      "grad_norm": 3.578125,
      "learning_rate": 0.0014127659574468085,
      "loss": 9.1676,
      "step": 166
    },
    {
      "epoch": 0.021374632023550492,
      "grad_norm": 1.546875,
      "learning_rate": 0.001421276595744681,
      "loss": 8.7166,
      "step": 167
    },
    {
      "epoch": 0.021502623832074748,
      "grad_norm": 1.796875,
      "learning_rate": 0.0014297872340425532,
      "loss": 8.8643,
      "step": 168
    },
    {
      "epoch": 0.021630615640599003,
      "grad_norm": 1.859375,
      "learning_rate": 0.0014382978723404256,
      "loss": 8.4166,
      "step": 169
    },
    {
      "epoch": 0.021758607449123255,
      "grad_norm": 1.203125,
      "learning_rate": 0.001446808510638298,
      "loss": 8.7252,
      "step": 170
    },
    {
      "epoch": 0.02188659925764751,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0014553191489361703,
      "loss": 8.3994,
      "step": 171
    },
    {
      "epoch": 0.022014591066171766,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0014638297872340425,
      "loss": 8.5685,
      "step": 172
    },
    {
      "epoch": 0.022142582874696018,
      "grad_norm": 1.9609375,
      "learning_rate": 0.0014723404255319148,
      "loss": 8.4525,
      "step": 173
    },
    {
      "epoch": 0.022270574683220273,
      "grad_norm": 1.46875,
      "learning_rate": 0.0014808510638297872,
      "loss": 8.4888,
      "step": 174
    },
    {
      "epoch": 0.02239856649174453,
      "grad_norm": 2.03125,
      "learning_rate": 0.0014893617021276596,
      "loss": 8.6854,
      "step": 175
    },
    {
      "epoch": 0.022526558300268784,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0014978723404255319,
      "loss": 8.7856,
      "step": 176
    },
    {
      "epoch": 0.022654550108793036,
      "grad_norm": 1.515625,
      "learning_rate": 0.0015063829787234043,
      "loss": 9.0148,
      "step": 177
    },
    {
      "epoch": 0.02278254191731729,
      "grad_norm": 1.09375,
      "learning_rate": 0.0015148936170212765,
      "loss": 8.732,
      "step": 178
    },
    {
      "epoch": 0.022910533725841547,
      "grad_norm": 2.484375,
      "learning_rate": 0.001523404255319149,
      "loss": 8.8451,
      "step": 179
    },
    {
      "epoch": 0.023038525534365802,
      "grad_norm": 1.515625,
      "learning_rate": 0.0015319148936170212,
      "loss": 8.9048,
      "step": 180
    },
    {
      "epoch": 0.023166517342890054,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0015404255319148936,
      "loss": 8.2626,
      "step": 181
    },
    {
      "epoch": 0.02329450915141431,
      "grad_norm": 3.203125,
      "learning_rate": 0.001548936170212766,
      "loss": 9.0012,
      "step": 182
    },
    {
      "epoch": 0.023422500959938565,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0015574468085106383,
      "loss": 8.4391,
      "step": 183
    },
    {
      "epoch": 0.023550492768462817,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0015659574468085107,
      "loss": 8.6437,
      "step": 184
    },
    {
      "epoch": 0.023678484576987072,
      "grad_norm": 1.2734375,
      "learning_rate": 0.001574468085106383,
      "loss": 8.4913,
      "step": 185
    },
    {
      "epoch": 0.023806476385511328,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0015829787234042554,
      "loss": 8.2586,
      "step": 186
    },
    {
      "epoch": 0.023934468194035583,
      "grad_norm": 2.3125,
      "learning_rate": 0.0015914893617021279,
      "loss": 8.6049,
      "step": 187
    },
    {
      "epoch": 0.024062460002559835,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0016,
      "loss": 9.1021,
      "step": 188
    },
    {
      "epoch": 0.02419045181108409,
      "grad_norm": 1.515625,
      "learning_rate": 0.0016085106382978725,
      "loss": 8.7149,
      "step": 189
    },
    {
      "epoch": 0.024318443619608346,
      "grad_norm": 1.8359375,
      "learning_rate": 0.0016170212765957445,
      "loss": 8.8414,
      "step": 190
    },
    {
      "epoch": 0.024446435428132598,
      "grad_norm": 1.4375,
      "learning_rate": 0.001625531914893617,
      "loss": 8.7747,
      "step": 191
    },
    {
      "epoch": 0.024574427236656853,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0016340425531914894,
      "loss": 8.3365,
      "step": 192
    },
    {
      "epoch": 0.02470241904518111,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0016425531914893616,
      "loss": 8.3302,
      "step": 193
    },
    {
      "epoch": 0.024830410853705364,
      "grad_norm": 3.25,
      "learning_rate": 0.001651063829787234,
      "loss": 8.8121,
      "step": 194
    },
    {
      "epoch": 0.024958402662229616,
      "grad_norm": 1.28125,
      "learning_rate": 0.0016595744680851063,
      "loss": 8.7776,
      "step": 195
    },
    {
      "epoch": 0.02508639447075387,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0016680851063829787,
      "loss": 8.5077,
      "step": 196
    },
    {
      "epoch": 0.025214386279278127,
      "grad_norm": 1.515625,
      "learning_rate": 0.0016765957446808512,
      "loss": 8.94,
      "step": 197
    },
    {
      "epoch": 0.025342378087802382,
      "grad_norm": 1.171875,
      "learning_rate": 0.0016851063829787234,
      "loss": 8.5434,
      "step": 198
    },
    {
      "epoch": 0.025470369896326634,
      "grad_norm": 1.28125,
      "learning_rate": 0.0016936170212765958,
      "loss": 8.7839,
      "step": 199
    },
    {
      "epoch": 0.02559836170485089,
      "grad_norm": 1.1953125,
      "learning_rate": 0.001702127659574468,
      "loss": 8.6118,
      "step": 200
    },
    {
      "epoch": 0.025726353513375145,
      "grad_norm": 1.484375,
      "learning_rate": 0.0017106382978723405,
      "loss": 8.5147,
      "step": 201
    },
    {
      "epoch": 0.025854345321899397,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0017191489361702127,
      "loss": 8.9552,
      "step": 202
    },
    {
      "epoch": 0.025982337130423652,
      "grad_norm": 1.3125,
      "learning_rate": 0.0017276595744680852,
      "loss": 7.8895,
      "step": 203
    },
    {
      "epoch": 0.026110328938947908,
      "grad_norm": 1.8125,
      "learning_rate": 0.0017361702127659576,
      "loss": 8.7647,
      "step": 204
    },
    {
      "epoch": 0.026238320747472163,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0017446808510638298,
      "loss": 8.6143,
      "step": 205
    },
    {
      "epoch": 0.026366312555996415,
      "grad_norm": 1.28125,
      "learning_rate": 0.0017531914893617023,
      "loss": 8.629,
      "step": 206
    },
    {
      "epoch": 0.02649430436452067,
      "grad_norm": 1.0625,
      "learning_rate": 0.0017617021276595745,
      "loss": 8.4275,
      "step": 207
    },
    {
      "epoch": 0.026622296173044926,
      "grad_norm": 1.84375,
      "learning_rate": 0.0017702127659574467,
      "loss": 8.3776,
      "step": 208
    },
    {
      "epoch": 0.02675028798156918,
      "grad_norm": 1.375,
      "learning_rate": 0.0017787234042553192,
      "loss": 8.4555,
      "step": 209
    },
    {
      "epoch": 0.026878279790093433,
      "grad_norm": 1.546875,
      "learning_rate": 0.0017872340425531914,
      "loss": 8.9744,
      "step": 210
    },
    {
      "epoch": 0.02700627159861769,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0017957446808510638,
      "loss": 8.721,
      "step": 211
    },
    {
      "epoch": 0.027134263407141944,
      "grad_norm": 1.71875,
      "learning_rate": 0.001804255319148936,
      "loss": 8.7357,
      "step": 212
    },
    {
      "epoch": 0.027262255215666196,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0018127659574468085,
      "loss": 8.2169,
      "step": 213
    },
    {
      "epoch": 0.02739024702419045,
      "grad_norm": 1.2421875,
      "learning_rate": 0.001821276595744681,
      "loss": 8.5497,
      "step": 214
    },
    {
      "epoch": 0.027518238832714707,
      "grad_norm": 1.171875,
      "learning_rate": 0.0018297872340425532,
      "loss": 8.6894,
      "step": 215
    },
    {
      "epoch": 0.027646230641238962,
      "grad_norm": 1.21875,
      "learning_rate": 0.0018382978723404256,
      "loss": 8.6858,
      "step": 216
    },
    {
      "epoch": 0.027774222449763214,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0018468085106382978,
      "loss": 8.6044,
      "step": 217
    },
    {
      "epoch": 0.02790221425828747,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0018553191489361703,
      "loss": 8.4637,
      "step": 218
    },
    {
      "epoch": 0.028030206066811725,
      "grad_norm": 1.625,
      "learning_rate": 0.0018638297872340427,
      "loss": 8.7671,
      "step": 219
    },
    {
      "epoch": 0.02815819787533598,
      "grad_norm": 1.203125,
      "learning_rate": 0.001872340425531915,
      "loss": 8.6098,
      "step": 220
    },
    {
      "epoch": 0.028286189683860232,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0018808510638297874,
      "loss": 8.3476,
      "step": 221
    },
    {
      "epoch": 0.028414181492384488,
      "grad_norm": 2.203125,
      "learning_rate": 0.0018893617021276596,
      "loss": 8.5881,
      "step": 222
    },
    {
      "epoch": 0.028542173300908743,
      "grad_norm": 1.53125,
      "learning_rate": 0.001897872340425532,
      "loss": 9.3298,
      "step": 223
    },
    {
      "epoch": 0.028670165109432995,
      "grad_norm": 1.53125,
      "learning_rate": 0.0019063829787234045,
      "loss": 8.8214,
      "step": 224
    },
    {
      "epoch": 0.02879815691795725,
      "grad_norm": 1.234375,
      "learning_rate": 0.0019148936170212767,
      "loss": 8.4738,
      "step": 225
    },
    {
      "epoch": 0.028926148726481506,
      "grad_norm": 1.7109375,
      "learning_rate": 0.001923404255319149,
      "loss": 8.6569,
      "step": 226
    },
    {
      "epoch": 0.02905414053500576,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0019319148936170212,
      "loss": 8.6104,
      "step": 227
    },
    {
      "epoch": 0.029182132343530013,
      "grad_norm": 1.078125,
      "learning_rate": 0.0019404255319148936,
      "loss": 8.1561,
      "step": 228
    },
    {
      "epoch": 0.02931012415205427,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0019489361702127658,
      "loss": 8.3613,
      "step": 229
    },
    {
      "epoch": 0.029438115960578524,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0019574468085106385,
      "loss": 7.9761,
      "step": 230
    },
    {
      "epoch": 0.029566107769102776,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0019659574468085107,
      "loss": 7.9273,
      "step": 231
    },
    {
      "epoch": 0.02969409957762703,
      "grad_norm": 1.2734375,
      "learning_rate": 0.001974468085106383,
      "loss": 9.0582,
      "step": 232
    },
    {
      "epoch": 0.029822091386151287,
      "grad_norm": 1.0703125,
      "learning_rate": 0.001982978723404255,
      "loss": 8.6655,
      "step": 233
    },
    {
      "epoch": 0.029950083194675542,
      "grad_norm": 1.078125,
      "learning_rate": 0.001991489361702128,
      "loss": 8.4073,
      "step": 234
    },
    {
      "epoch": 0.030078075003199794,
      "grad_norm": 1.4609375,
      "learning_rate": 0.002,
      "loss": 8.0217,
      "step": 235
    },
    {
      "epoch": 0.03020606681172405,
      "grad_norm": 1.4296875,
      "learning_rate": 0.001999999914066888,
      "loss": 8.6746,
      "step": 236
    },
    {
      "epoch": 0.030334058620248305,
      "grad_norm": 1.0,
      "learning_rate": 0.001999999656267565,
      "loss": 8.306,
      "step": 237
    },
    {
      "epoch": 0.03046205042877256,
      "grad_norm": 0.87109375,
      "learning_rate": 0.001999999226602077,
      "loss": 8.6114,
      "step": 238
    },
    {
      "epoch": 0.030590042237296812,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0019999986250704973,
      "loss": 8.2038,
      "step": 239
    },
    {
      "epoch": 0.030718034045821067,
      "grad_norm": 2.109375,
      "learning_rate": 0.001999997851672929,
      "loss": 8.7213,
      "step": 240
    },
    {
      "epoch": 0.030846025854345323,
      "grad_norm": 1.0078125,
      "learning_rate": 0.001999996906409505,
      "loss": 8.6042,
      "step": 241
    },
    {
      "epoch": 0.030974017662869575,
      "grad_norm": 1.1015625,
      "learning_rate": 0.001999995789280388,
      "loss": 8.2493,
      "step": 242
    },
    {
      "epoch": 0.03110200947139383,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0019999945002857704,
      "loss": 8.5548,
      "step": 243
    },
    {
      "epoch": 0.031230001279918086,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0019999930394258725,
      "loss": 8.5153,
      "step": 244
    },
    {
      "epoch": 0.03135799308844234,
      "grad_norm": 1.5390625,
      "learning_rate": 0.001999991406700947,
      "loss": 8.0701,
      "step": 245
    },
    {
      "epoch": 0.03148598489696659,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0019999896021112725,
      "loss": 8.3185,
      "step": 246
    },
    {
      "epoch": 0.03161397670549085,
      "grad_norm": 1.0859375,
      "learning_rate": 0.001999987625657161,
      "loss": 8.3867,
      "step": 247
    },
    {
      "epoch": 0.031741968514015104,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0019999854773389515,
      "loss": 8.4251,
      "step": 248
    },
    {
      "epoch": 0.03186996032253936,
      "grad_norm": 0.953125,
      "learning_rate": 0.001999983157157013,
      "loss": 8.5273,
      "step": 249
    },
    {
      "epoch": 0.031997952131063614,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0019999806651117445,
      "loss": 9.0086,
      "step": 250
    },
    {
      "epoch": 0.03212594393958786,
      "grad_norm": 1.4375,
      "learning_rate": 0.001999978001203574,
      "loss": 8.4195,
      "step": 251
    },
    {
      "epoch": 0.03225393574811212,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00199997516543296,
      "loss": 8.1317,
      "step": 252
    },
    {
      "epoch": 0.032381927556636374,
      "grad_norm": 1.078125,
      "learning_rate": 0.001999972157800389,
      "loss": 8.2909,
      "step": 253
    },
    {
      "epoch": 0.03250991936516063,
      "grad_norm": 1.3515625,
      "learning_rate": 0.001999968978306379,
      "loss": 8.4478,
      "step": 254
    },
    {
      "epoch": 0.032637911173684885,
      "grad_norm": 0.96484375,
      "learning_rate": 0.001999965626951476,
      "loss": 8.4762,
      "step": 255
    },
    {
      "epoch": 0.03276590298220914,
      "grad_norm": 1.3203125,
      "learning_rate": 0.001999962103736255,
      "loss": 8.2412,
      "step": 256
    },
    {
      "epoch": 0.032893894790733395,
      "grad_norm": 1.09375,
      "learning_rate": 0.0019999584086613227,
      "loss": 8.489,
      "step": 257
    },
    {
      "epoch": 0.03302188659925765,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0019999545417273136,
      "loss": 8.2531,
      "step": 258
    },
    {
      "epoch": 0.0331498784077819,
      "grad_norm": 1.03125,
      "learning_rate": 0.0019999505029348924,
      "loss": 8.4081,
      "step": 259
    },
    {
      "epoch": 0.033277870216306155,
      "grad_norm": 1.8125,
      "learning_rate": 0.001999946292284754,
      "loss": 8.06,
      "step": 260
    },
    {
      "epoch": 0.03340586202483041,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0019999419097776207,
      "loss": 8.3496,
      "step": 261
    },
    {
      "epoch": 0.033533853833354665,
      "grad_norm": 1.171875,
      "learning_rate": 0.0019999373554142465,
      "loss": 7.6989,
      "step": 262
    },
    {
      "epoch": 0.03366184564187892,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0019999326291954144,
      "loss": 8.7002,
      "step": 263
    },
    {
      "epoch": 0.033789837450403176,
      "grad_norm": 1.2890625,
      "learning_rate": 0.001999927731121936,
      "loss": 8.364,
      "step": 264
    },
    {
      "epoch": 0.03391782925892743,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0019999226611946536,
      "loss": 8.5855,
      "step": 265
    },
    {
      "epoch": 0.03404582106745168,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0019999174194144384,
      "loss": 8.5421,
      "step": 266
    },
    {
      "epoch": 0.034173812875975935,
      "grad_norm": 1.15625,
      "learning_rate": 0.001999912005782191,
      "loss": 7.9455,
      "step": 267
    },
    {
      "epoch": 0.03430180468450019,
      "grad_norm": 1.265625,
      "learning_rate": 0.001999906420298842,
      "loss": 8.0294,
      "step": 268
    },
    {
      "epoch": 0.034429796493024446,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0019999006629653517,
      "loss": 7.9632,
      "step": 269
    },
    {
      "epoch": 0.0345577883015487,
      "grad_norm": 1.1015625,
      "learning_rate": 0.001999894733782709,
      "loss": 8.2043,
      "step": 270
    },
    {
      "epoch": 0.03468578011007296,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0019998886327519338,
      "loss": 8.4007,
      "step": 271
    },
    {
      "epoch": 0.03481377191859721,
      "grad_norm": 0.82421875,
      "learning_rate": 0.001999882359874074,
      "loss": 8.3737,
      "step": 272
    },
    {
      "epoch": 0.03494176372712146,
      "grad_norm": 1.0625,
      "learning_rate": 0.0019998759151502073,
      "loss": 8.2074,
      "step": 273
    },
    {
      "epoch": 0.035069755535645716,
      "grad_norm": 1.296875,
      "learning_rate": 0.001999869298581442,
      "loss": 7.8895,
      "step": 274
    },
    {
      "epoch": 0.03519774734416997,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0019998625101689156,
      "loss": 8.2668,
      "step": 275
    },
    {
      "epoch": 0.03532573915269423,
      "grad_norm": 1.3046875,
      "learning_rate": 0.001999855549913794,
      "loss": 8.3547,
      "step": 276
    },
    {
      "epoch": 0.03545373096121848,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0019998484178172732,
      "loss": 8.3062,
      "step": 277
    },
    {
      "epoch": 0.03558172276974274,
      "grad_norm": 0.859375,
      "learning_rate": 0.00199984111388058,
      "loss": 8.4267,
      "step": 278
    },
    {
      "epoch": 0.03570971457826699,
      "grad_norm": 0.94140625,
      "learning_rate": 0.001999833638104969,
      "loss": 8.2584,
      "step": 279
    },
    {
      "epoch": 0.03583770638679125,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0019998259904917257,
      "loss": 8.4601,
      "step": 280
    },
    {
      "epoch": 0.0359656981953155,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0019998181710421635,
      "loss": 7.9099,
      "step": 281
    },
    {
      "epoch": 0.03609369000383975,
      "grad_norm": 1.1875,
      "learning_rate": 0.001999810179757627,
      "loss": 8.3067,
      "step": 282
    },
    {
      "epoch": 0.03622168181236401,
      "grad_norm": 1.171875,
      "learning_rate": 0.00199980201663949,
      "loss": 7.9643,
      "step": 283
    },
    {
      "epoch": 0.03634967362088826,
      "grad_norm": 1.0546875,
      "learning_rate": 0.001999793681689154,
      "loss": 8.4406,
      "step": 284
    },
    {
      "epoch": 0.03647766542941252,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0019997851749080525,
      "loss": 8.4436,
      "step": 285
    },
    {
      "epoch": 0.036605657237936774,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0019997764962976475,
      "loss": 7.9318,
      "step": 286
    },
    {
      "epoch": 0.03673364904646103,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0019997676458594306,
      "loss": 8.3004,
      "step": 287
    },
    {
      "epoch": 0.03686164085498528,
      "grad_norm": 0.90625,
      "learning_rate": 0.001999758623594923,
      "loss": 8.4003,
      "step": 288
    },
    {
      "epoch": 0.036989632663509533,
      "grad_norm": 1.0078125,
      "learning_rate": 0.001999749429505675,
      "loss": 7.9758,
      "step": 289
    },
    {
      "epoch": 0.03711762447203379,
      "grad_norm": 0.8671875,
      "learning_rate": 0.001999740063593267,
      "loss": 8.1537,
      "step": 290
    },
    {
      "epoch": 0.037245616280558044,
      "grad_norm": 1.0078125,
      "learning_rate": 0.001999730525859308,
      "loss": 7.8036,
      "step": 291
    },
    {
      "epoch": 0.0373736080890823,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0019997208163054377,
      "loss": 8.3486,
      "step": 292
    },
    {
      "epoch": 0.037501599897606555,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0019997109349333252,
      "loss": 8.439,
      "step": 293
    },
    {
      "epoch": 0.03762959170613081,
      "grad_norm": 1.0,
      "learning_rate": 0.0019997008817446687,
      "loss": 7.7537,
      "step": 294
    },
    {
      "epoch": 0.03775758351465506,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0019996906567411954,
      "loss": 8.2072,
      "step": 295
    },
    {
      "epoch": 0.037885575323179314,
      "grad_norm": 1.03125,
      "learning_rate": 0.001999680259924663,
      "loss": 7.9602,
      "step": 296
    },
    {
      "epoch": 0.03801356713170357,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0019996696912968587,
      "loss": 8.1023,
      "step": 297
    },
    {
      "epoch": 0.038141558940227825,
      "grad_norm": 0.9375,
      "learning_rate": 0.001999658950859598,
      "loss": 8.3889,
      "step": 298
    },
    {
      "epoch": 0.03826955074875208,
      "grad_norm": 0.94921875,
      "learning_rate": 0.001999648038614728,
      "loss": 8.2012,
      "step": 299
    },
    {
      "epoch": 0.038397542557276336,
      "grad_norm": 1.1171875,
      "learning_rate": 0.001999636954564123,
      "loss": 8.5743,
      "step": 300
    },
    {
      "epoch": 0.03852553436580059,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0019996256987096887,
      "loss": 8.1439,
      "step": 301
    },
    {
      "epoch": 0.03865352617432484,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0019996142710533597,
      "loss": 8.3462,
      "step": 302
    },
    {
      "epoch": 0.038781517982849095,
      "grad_norm": 1.1640625,
      "learning_rate": 0.001999602671597099,
      "loss": 8.2519,
      "step": 303
    },
    {
      "epoch": 0.03890950979137335,
      "grad_norm": 1.0625,
      "learning_rate": 0.0019995909003429017,
      "loss": 7.9275,
      "step": 304
    },
    {
      "epoch": 0.039037501599897606,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0019995789572927894,
      "loss": 7.8728,
      "step": 305
    },
    {
      "epoch": 0.03916549340842186,
      "grad_norm": 0.75390625,
      "learning_rate": 0.001999566842448816,
      "loss": 7.7588,
      "step": 306
    },
    {
      "epoch": 0.03929348521694612,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0019995545558130624,
      "loss": 7.8929,
      "step": 307
    },
    {
      "epoch": 0.03942147702547037,
      "grad_norm": 1.0390625,
      "learning_rate": 0.001999542097387641,
      "loss": 7.9069,
      "step": 308
    },
    {
      "epoch": 0.03954946883399463,
      "grad_norm": 0.90234375,
      "learning_rate": 0.001999529467174693,
      "loss": 8.1361,
      "step": 309
    },
    {
      "epoch": 0.039677460642518876,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0019995166651763886,
      "loss": 8.17,
      "step": 310
    },
    {
      "epoch": 0.03980545245104313,
      "grad_norm": 1.0078125,
      "learning_rate": 0.001999503691394929,
      "loss": 8.1647,
      "step": 311
    },
    {
      "epoch": 0.03993344425956739,
      "grad_norm": 1.0625,
      "learning_rate": 0.0019994905458325424,
      "loss": 8.086,
      "step": 312
    },
    {
      "epoch": 0.04006143606809164,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0019994772284914896,
      "loss": 8.39,
      "step": 313
    },
    {
      "epoch": 0.0401894278766159,
      "grad_norm": 0.890625,
      "learning_rate": 0.001999463739374059,
      "loss": 8.5374,
      "step": 314
    },
    {
      "epoch": 0.04031741968514015,
      "grad_norm": 1.078125,
      "learning_rate": 0.0019994500784825684,
      "loss": 7.6325,
      "step": 315
    },
    {
      "epoch": 0.04044541149366441,
      "grad_norm": 0.84765625,
      "learning_rate": 0.001999436245819366,
      "loss": 8.1332,
      "step": 316
    },
    {
      "epoch": 0.04057340330218866,
      "grad_norm": 1.1484375,
      "learning_rate": 0.001999422241386829,
      "loss": 8.4218,
      "step": 317
    },
    {
      "epoch": 0.04070139511071291,
      "grad_norm": 0.92578125,
      "learning_rate": 0.001999408065187365,
      "loss": 7.845,
      "step": 318
    },
    {
      "epoch": 0.04082938691923717,
      "grad_norm": 1.015625,
      "learning_rate": 0.0019993937172234095,
      "loss": 8.0856,
      "step": 319
    },
    {
      "epoch": 0.04095737872776142,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0019993791974974285,
      "loss": 7.8986,
      "step": 320
    },
    {
      "epoch": 0.04108537053628568,
      "grad_norm": 0.9453125,
      "learning_rate": 0.001999364506011918,
      "loss": 7.7555,
      "step": 321
    },
    {
      "epoch": 0.041213362344809934,
      "grad_norm": 1.0625,
      "learning_rate": 0.0019993496427694025,
      "loss": 8.1576,
      "step": 322
    },
    {
      "epoch": 0.04134135415333419,
      "grad_norm": 0.9140625,
      "learning_rate": 0.001999334607772437,
      "loss": 8.3152,
      "step": 323
    },
    {
      "epoch": 0.04146934596185844,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0019993194010236052,
      "loss": 7.9105,
      "step": 324
    },
    {
      "epoch": 0.04159733777038269,
      "grad_norm": 0.875,
      "learning_rate": 0.0019993040225255204,
      "loss": 8.0405,
      "step": 325
    },
    {
      "epoch": 0.04172532957890695,
      "grad_norm": 0.97265625,
      "learning_rate": 0.001999288472280826,
      "loss": 7.8964,
      "step": 326
    },
    {
      "epoch": 0.041853321387431204,
      "grad_norm": 0.96875,
      "learning_rate": 0.001999272750292194,
      "loss": 7.6787,
      "step": 327
    },
    {
      "epoch": 0.04198131319595546,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0019992568565623276,
      "loss": 8.4053,
      "step": 328
    },
    {
      "epoch": 0.042109305004479715,
      "grad_norm": 1.078125,
      "learning_rate": 0.0019992407910939575,
      "loss": 8.4506,
      "step": 329
    },
    {
      "epoch": 0.04223729681300397,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0019992245538898445,
      "loss": 8.1245,
      "step": 330
    },
    {
      "epoch": 0.04236528862152822,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0019992081449527804,
      "loss": 7.8487,
      "step": 331
    },
    {
      "epoch": 0.042493280430052474,
      "grad_norm": 1.015625,
      "learning_rate": 0.0019991915642855844,
      "loss": 7.8075,
      "step": 332
    },
    {
      "epoch": 0.04262127223857673,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0019991748118911063,
      "loss": 7.8185,
      "step": 333
    },
    {
      "epoch": 0.042749264047100985,
      "grad_norm": 1.203125,
      "learning_rate": 0.001999157887772225,
      "loss": 7.8235,
      "step": 334
    },
    {
      "epoch": 0.04287725585562524,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0019991407919318503,
      "loss": 7.9787,
      "step": 335
    },
    {
      "epoch": 0.043005247664149496,
      "grad_norm": 1.171875,
      "learning_rate": 0.00199912352437292,
      "loss": 7.2365,
      "step": 336
    },
    {
      "epoch": 0.04313323947267375,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0019991060850984006,
      "loss": 7.9981,
      "step": 337
    },
    {
      "epoch": 0.04326123128119801,
      "grad_norm": 1.28125,
      "learning_rate": 0.0019990884741112906,
      "loss": 8.5765,
      "step": 338
    },
    {
      "epoch": 0.043389223089722255,
      "grad_norm": 0.875,
      "learning_rate": 0.0019990706914146165,
      "loss": 8.0139,
      "step": 339
    },
    {
      "epoch": 0.04351721489824651,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0019990527370114344,
      "loss": 8.0336,
      "step": 340
    },
    {
      "epoch": 0.043645206706770766,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00199903461090483,
      "loss": 8.3171,
      "step": 341
    },
    {
      "epoch": 0.04377319851529502,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0019990163130979184,
      "loss": 7.6641,
      "step": 342
    },
    {
      "epoch": 0.04390119032381928,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0019989978435938447,
      "loss": 7.9447,
      "step": 343
    },
    {
      "epoch": 0.04402918213234353,
      "grad_norm": 1.3671875,
      "learning_rate": 0.001998979202395783,
      "loss": 8.3972,
      "step": 344
    },
    {
      "epoch": 0.04415717394086779,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0019989603895069373,
      "loss": 8.1053,
      "step": 345
    },
    {
      "epoch": 0.044285165749392036,
      "grad_norm": 1.03125,
      "learning_rate": 0.001998941404930541,
      "loss": 7.4835,
      "step": 346
    },
    {
      "epoch": 0.04441315755791629,
      "grad_norm": 0.92578125,
      "learning_rate": 0.001998922248669856,
      "loss": 8.315,
      "step": 347
    },
    {
      "epoch": 0.04454114936644055,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0019989029207281763,
      "loss": 7.7607,
      "step": 348
    },
    {
      "epoch": 0.0446691411749648,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0019988834211088218,
      "loss": 8.0765,
      "step": 349
    },
    {
      "epoch": 0.04479713298348906,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0019988637498151453,
      "loss": 7.8337,
      "step": 350
    },
    {
      "epoch": 0.04492512479201331,
      "grad_norm": 1.046875,
      "learning_rate": 0.0019988439068505267,
      "loss": 7.4024,
      "step": 351
    },
    {
      "epoch": 0.04505311660053757,
      "grad_norm": 1.09375,
      "learning_rate": 0.001998823892218377,
      "loss": 7.6323,
      "step": 352
    },
    {
      "epoch": 0.04518110840906182,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0019988037059221354,
      "loss": 7.9424,
      "step": 353
    },
    {
      "epoch": 0.04530910021758607,
      "grad_norm": 0.90234375,
      "learning_rate": 0.001998783347965272,
      "loss": 7.439,
      "step": 354
    },
    {
      "epoch": 0.04543709202611033,
      "grad_norm": 0.9765625,
      "learning_rate": 0.001998762818351285,
      "loss": 7.5989,
      "step": 355
    },
    {
      "epoch": 0.04556508383463458,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0019987421170837034,
      "loss": 8.237,
      "step": 356
    },
    {
      "epoch": 0.04569307564315884,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0019987212441660842,
      "loss": 7.6642,
      "step": 357
    },
    {
      "epoch": 0.045821067451683094,
      "grad_norm": 1.09375,
      "learning_rate": 0.001998700199602015,
      "loss": 7.5337,
      "step": 358
    },
    {
      "epoch": 0.04594905926020735,
      "grad_norm": 1.03125,
      "learning_rate": 0.001998678983395113,
      "loss": 7.8855,
      "step": 359
    },
    {
      "epoch": 0.046077051068731605,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0019986575955490246,
      "loss": 7.677,
      "step": 360
    },
    {
      "epoch": 0.04620504287725585,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0019986360360674253,
      "loss": 7.7139,
      "step": 361
    },
    {
      "epoch": 0.04633303468578011,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0019986143049540205,
      "loss": 7.895,
      "step": 362
    },
    {
      "epoch": 0.046461026494304364,
      "grad_norm": 1.1484375,
      "learning_rate": 0.001998592402212545,
      "loss": 8.281,
      "step": 363
    },
    {
      "epoch": 0.04658901830282862,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0019985703278467635,
      "loss": 7.3744,
      "step": 364
    },
    {
      "epoch": 0.046717010111352875,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0019985480818604695,
      "loss": 7.9484,
      "step": 365
    },
    {
      "epoch": 0.04684500191987713,
      "grad_norm": 1.125,
      "learning_rate": 0.0019985256642574864,
      "loss": 7.2767,
      "step": 366
    },
    {
      "epoch": 0.046972993728401385,
      "grad_norm": 0.953125,
      "learning_rate": 0.001998503075041667,
      "loss": 7.9239,
      "step": 367
    },
    {
      "epoch": 0.047100985536925634,
      "grad_norm": 1.0234375,
      "learning_rate": 0.001998480314216894,
      "loss": 7.4436,
      "step": 368
    },
    {
      "epoch": 0.04722897734544989,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0019984573817870784,
      "loss": 7.6712,
      "step": 369
    },
    {
      "epoch": 0.047356969153974145,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0019984342777561626,
      "loss": 7.953,
      "step": 370
    },
    {
      "epoch": 0.0474849609624984,
      "grad_norm": 1.0625,
      "learning_rate": 0.0019984110021281165,
      "loss": 7.39,
      "step": 371
    },
    {
      "epoch": 0.047612952771022655,
      "grad_norm": 1.03125,
      "learning_rate": 0.0019983875549069406,
      "loss": 7.7632,
      "step": 372
    },
    {
      "epoch": 0.04774094457954691,
      "grad_norm": 0.98828125,
      "learning_rate": 0.001998363936096665,
      "loss": 7.7866,
      "step": 373
    },
    {
      "epoch": 0.047868936388071166,
      "grad_norm": 1.078125,
      "learning_rate": 0.0019983401457013485,
      "loss": 7.5003,
      "step": 374
    },
    {
      "epoch": 0.047996928196595415,
      "grad_norm": 1.03125,
      "learning_rate": 0.0019983161837250807,
      "loss": 7.3998,
      "step": 375
    },
    {
      "epoch": 0.04812492000511967,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0019982920501719786,
      "loss": 7.1467,
      "step": 376
    },
    {
      "epoch": 0.048252911813643926,
      "grad_norm": 1.0078125,
      "learning_rate": 0.001998267745046191,
      "loss": 7.8815,
      "step": 377
    },
    {
      "epoch": 0.04838090362216818,
      "grad_norm": 1.125,
      "learning_rate": 0.0019982432683518943,
      "loss": 7.6482,
      "step": 378
    },
    {
      "epoch": 0.048508895430692436,
      "grad_norm": 1.28125,
      "learning_rate": 0.0019982186200932965,
      "loss": 7.4682,
      "step": 379
    },
    {
      "epoch": 0.04863688723921669,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0019981938002746327,
      "loss": 7.2471,
      "step": 380
    },
    {
      "epoch": 0.04876487904774095,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0019981688089001685,
      "loss": 8.0438,
      "step": 381
    },
    {
      "epoch": 0.048892870856265196,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0019981436459742,
      "loss": 7.0828,
      "step": 382
    },
    {
      "epoch": 0.04902086266478945,
      "grad_norm": 1.0625,
      "learning_rate": 0.0019981183115010508,
      "loss": 8.1522,
      "step": 383
    },
    {
      "epoch": 0.049148854473313706,
      "grad_norm": 1.171875,
      "learning_rate": 0.0019980928054850757,
      "loss": 7.5713,
      "step": 384
    },
    {
      "epoch": 0.04927684628183796,
      "grad_norm": 1.046875,
      "learning_rate": 0.001998067127930658,
      "loss": 7.7339,
      "step": 385
    },
    {
      "epoch": 0.04940483809036222,
      "grad_norm": 1.03125,
      "learning_rate": 0.0019980412788422114,
      "loss": 7.0573,
      "step": 386
    },
    {
      "epoch": 0.04953282989888647,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0019980152582241774,
      "loss": 8.1425,
      "step": 387
    },
    {
      "epoch": 0.04966082170741073,
      "grad_norm": 0.98828125,
      "learning_rate": 0.001997989066081029,
      "loss": 7.5571,
      "step": 388
    },
    {
      "epoch": 0.04978881351593498,
      "grad_norm": 0.87890625,
      "learning_rate": 0.001997962702417268,
      "loss": 7.9552,
      "step": 389
    },
    {
      "epoch": 0.04991680532445923,
      "grad_norm": 1.0703125,
      "learning_rate": 0.001997936167237424,
      "loss": 7.8074,
      "step": 390
    },
    {
      "epoch": 0.05004479713298349,
      "grad_norm": 1.296875,
      "learning_rate": 0.001997909460546059,
      "loss": 7.6329,
      "step": 391
    },
    {
      "epoch": 0.05017278894150774,
      "grad_norm": 1.046875,
      "learning_rate": 0.0019978825823477623,
      "loss": 7.607,
      "step": 392
    },
    {
      "epoch": 0.050300780750032,
      "grad_norm": 1.125,
      "learning_rate": 0.0019978555326471536,
      "loss": 7.5571,
      "step": 393
    },
    {
      "epoch": 0.050428772558556254,
      "grad_norm": 1.0078125,
      "learning_rate": 0.001997828311448881,
      "loss": 7.5765,
      "step": 394
    },
    {
      "epoch": 0.05055676436708051,
      "grad_norm": 1.0,
      "learning_rate": 0.001997800918757624,
      "loss": 7.6145,
      "step": 395
    },
    {
      "epoch": 0.050684756175604764,
      "grad_norm": 0.890625,
      "learning_rate": 0.00199777335457809,
      "loss": 7.7618,
      "step": 396
    },
    {
      "epoch": 0.05081274798412901,
      "grad_norm": 1.09375,
      "learning_rate": 0.0019977456189150164,
      "loss": 7.5914,
      "step": 397
    },
    {
      "epoch": 0.05094073979265327,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00199771771177317,
      "loss": 7.3126,
      "step": 398
    },
    {
      "epoch": 0.051068731601177524,
      "grad_norm": 1.1484375,
      "learning_rate": 0.001997689633157347,
      "loss": 7.1157,
      "step": 399
    },
    {
      "epoch": 0.05119672340970178,
      "grad_norm": 1.03125,
      "learning_rate": 0.0019976613830723733,
      "loss": 7.4847,
      "step": 400
    },
    {
      "epoch": 0.051324715218226034,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0019976329615231042,
      "loss": 7.9189,
      "step": 401
    },
    {
      "epoch": 0.05145270702675029,
      "grad_norm": 1.421875,
      "learning_rate": 0.0019976043685144243,
      "loss": 7.7608,
      "step": 402
    },
    {
      "epoch": 0.051580698835274545,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0019975756040512475,
      "loss": 7.4357,
      "step": 403
    },
    {
      "epoch": 0.051708690643798794,
      "grad_norm": 1.265625,
      "learning_rate": 0.0019975466681385183,
      "loss": 7.6623,
      "step": 404
    },
    {
      "epoch": 0.05183668245232305,
      "grad_norm": 1.578125,
      "learning_rate": 0.0019975175607812087,
      "loss": 7.4823,
      "step": 405
    },
    {
      "epoch": 0.051964674260847304,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0019974882819843217,
      "loss": 7.1651,
      "step": 406
    },
    {
      "epoch": 0.05209266606937156,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00199745883175289,
      "loss": 7.367,
      "step": 407
    },
    {
      "epoch": 0.052220657877895815,
      "grad_norm": 1.203125,
      "learning_rate": 0.001997429210091974,
      "loss": 7.6021,
      "step": 408
    },
    {
      "epoch": 0.05234864968642007,
      "grad_norm": 1.390625,
      "learning_rate": 0.0019973994170066653,
      "loss": 7.5098,
      "step": 409
    },
    {
      "epoch": 0.052476641494944326,
      "grad_norm": 1.09375,
      "learning_rate": 0.0019973694525020846,
      "loss": 7.162,
      "step": 410
    },
    {
      "epoch": 0.05260463330346858,
      "grad_norm": 1.640625,
      "learning_rate": 0.001997339316583381,
      "loss": 7.7455,
      "step": 411
    },
    {
      "epoch": 0.05273262511199283,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0019973090092557343,
      "loss": 7.2754,
      "step": 412
    },
    {
      "epoch": 0.052860616920517085,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0019972785305243535,
      "loss": 7.4922,
      "step": 413
    },
    {
      "epoch": 0.05298860872904134,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0019972478803944767,
      "loss": 7.5516,
      "step": 414
    },
    {
      "epoch": 0.053116600537565596,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0019972170588713712,
      "loss": 7.6435,
      "step": 415
    },
    {
      "epoch": 0.05324459234608985,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0019971860659603346,
      "loss": 7.5155,
      "step": 416
    },
    {
      "epoch": 0.05337258415461411,
      "grad_norm": 1.1875,
      "learning_rate": 0.0019971549016666937,
      "loss": 6.827,
      "step": 417
    },
    {
      "epoch": 0.05350057596313836,
      "grad_norm": 1.1875,
      "learning_rate": 0.0019971235659958044,
      "loss": 7.3864,
      "step": 418
    },
    {
      "epoch": 0.05362856777166261,
      "grad_norm": 1.4921875,
      "learning_rate": 0.001997092058953052,
      "loss": 7.1051,
      "step": 419
    },
    {
      "epoch": 0.053756559580186866,
      "grad_norm": 0.953125,
      "learning_rate": 0.0019970603805438514,
      "loss": 7.9602,
      "step": 420
    },
    {
      "epoch": 0.05388455138871112,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0019970285307736478,
      "loss": 7.4691,
      "step": 421
    },
    {
      "epoch": 0.05401254319723538,
      "grad_norm": 1.0859375,
      "learning_rate": 0.001996996509647914,
      "loss": 7.1073,
      "step": 422
    },
    {
      "epoch": 0.05414053500575963,
      "grad_norm": 1.1796875,
      "learning_rate": 0.001996964317172155,
      "loss": 7.0894,
      "step": 423
    },
    {
      "epoch": 0.05426852681428389,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0019969319533519017,
      "loss": 7.395,
      "step": 424
    },
    {
      "epoch": 0.05439651862280814,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0019968994181927178,
      "loss": 7.2452,
      "step": 425
    },
    {
      "epoch": 0.05452451043133239,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0019968667117001944,
      "loss": 6.8562,
      "step": 426
    },
    {
      "epoch": 0.05465250223985665,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0019968338338799524,
      "loss": 7.2328,
      "step": 427
    },
    {
      "epoch": 0.0547804940483809,
      "grad_norm": 1.59375,
      "learning_rate": 0.0019968007847376425,
      "loss": 7.0666,
      "step": 428
    },
    {
      "epoch": 0.05490848585690516,
      "grad_norm": 1.21875,
      "learning_rate": 0.001996767564278945,
      "loss": 7.1076,
      "step": 429
    },
    {
      "epoch": 0.05503647766542941,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00199673417250957,
      "loss": 7.576,
      "step": 430
    },
    {
      "epoch": 0.05516446947395367,
      "grad_norm": 1.3359375,
      "learning_rate": 0.001996700609435255,
      "loss": 7.3135,
      "step": 431
    },
    {
      "epoch": 0.055292461282477924,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0019966668750617685,
      "loss": 7.4828,
      "step": 432
    },
    {
      "epoch": 0.05542045309100217,
      "grad_norm": 1.140625,
      "learning_rate": 0.0019966329693949098,
      "loss": 7.2372,
      "step": 433
    },
    {
      "epoch": 0.05554844489952643,
      "grad_norm": 1.21875,
      "learning_rate": 0.0019965988924405048,
      "loss": 7.0769,
      "step": 434
    },
    {
      "epoch": 0.05567643670805068,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0019965646442044105,
      "loss": 7.2683,
      "step": 435
    },
    {
      "epoch": 0.05580442851657494,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0019965302246925136,
      "loss": 7.0883,
      "step": 436
    },
    {
      "epoch": 0.055932420325099194,
      "grad_norm": 1.25,
      "learning_rate": 0.001996495633910728,
      "loss": 6.5323,
      "step": 437
    },
    {
      "epoch": 0.05606041213362345,
      "grad_norm": 0.953125,
      "learning_rate": 0.001996460871865001,
      "loss": 6.7232,
      "step": 438
    },
    {
      "epoch": 0.056188403942147705,
      "grad_norm": 1.1484375,
      "learning_rate": 0.001996425938561305,
      "loss": 7.2941,
      "step": 439
    },
    {
      "epoch": 0.05631639575067196,
      "grad_norm": 1.4921875,
      "learning_rate": 0.001996390834005645,
      "loss": 7.2678,
      "step": 440
    },
    {
      "epoch": 0.05644438755919621,
      "grad_norm": 1.21875,
      "learning_rate": 0.0019963555582040546,
      "loss": 6.7167,
      "step": 441
    },
    {
      "epoch": 0.056572379367720464,
      "grad_norm": 1.0078125,
      "learning_rate": 0.001996320111162595,
      "loss": 6.805,
      "step": 442
    },
    {
      "epoch": 0.05670037117624472,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0019962844928873595,
      "loss": 7.389,
      "step": 443
    },
    {
      "epoch": 0.056828362984768975,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0019962487033844695,
      "loss": 6.7295,
      "step": 444
    },
    {
      "epoch": 0.05695635479329323,
      "grad_norm": 1.125,
      "learning_rate": 0.001996212742660076,
      "loss": 6.6994,
      "step": 445
    },
    {
      "epoch": 0.057084346601817486,
      "grad_norm": 1.265625,
      "learning_rate": 0.0019961766107203594,
      "loss": 7.1095,
      "step": 446
    },
    {
      "epoch": 0.05721233841034174,
      "grad_norm": 1.0234375,
      "learning_rate": 0.001996140307571529,
      "loss": 6.4544,
      "step": 447
    },
    {
      "epoch": 0.05734033021886599,
      "grad_norm": 1.03125,
      "learning_rate": 0.0019961038332198255,
      "loss": 6.6338,
      "step": 448
    },
    {
      "epoch": 0.057468322027390245,
      "grad_norm": 1.28125,
      "learning_rate": 0.001996067187671516,
      "loss": 7.0341,
      "step": 449
    },
    {
      "epoch": 0.0575963138359145,
      "grad_norm": 1.34375,
      "learning_rate": 0.0019960303709328996,
      "loss": 6.896,
      "step": 450
    },
    {
      "epoch": 0.057724305644438756,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0019959933830103033,
      "loss": 6.3652,
      "step": 451
    },
    {
      "epoch": 0.05785229745296301,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0019959562239100846,
      "loss": 6.6138,
      "step": 452
    },
    {
      "epoch": 0.05798028926148727,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00199591889363863,
      "loss": 6.873,
      "step": 453
    },
    {
      "epoch": 0.05810828107001152,
      "grad_norm": 1.046875,
      "learning_rate": 0.0019958813922023545,
      "loss": 7.1165,
      "step": 454
    },
    {
      "epoch": 0.05823627287853577,
      "grad_norm": 1.2421875,
      "learning_rate": 0.001995843719607704,
      "loss": 7.2551,
      "step": 455
    },
    {
      "epoch": 0.058364264687060026,
      "grad_norm": 1.0078125,
      "learning_rate": 0.001995805875861153,
      "loss": 6.3529,
      "step": 456
    },
    {
      "epoch": 0.05849225649558428,
      "grad_norm": 1.09375,
      "learning_rate": 0.0019957678609692055,
      "loss": 6.6698,
      "step": 457
    },
    {
      "epoch": 0.05862024830410854,
      "grad_norm": 1.484375,
      "learning_rate": 0.001995729674938395,
      "loss": 6.826,
      "step": 458
    },
    {
      "epoch": 0.05874824011263279,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0019956913177752843,
      "loss": 6.489,
      "step": 459
    },
    {
      "epoch": 0.05887623192115705,
      "grad_norm": 1.3671875,
      "learning_rate": 0.001995652789486466,
      "loss": 6.3699,
      "step": 460
    },
    {
      "epoch": 0.0590042237296813,
      "grad_norm": 1.328125,
      "learning_rate": 0.0019956140900785615,
      "loss": 6.7426,
      "step": 461
    },
    {
      "epoch": 0.05913221553820555,
      "grad_norm": 1.3515625,
      "learning_rate": 0.001995575219558222,
      "loss": 7.3886,
      "step": 462
    },
    {
      "epoch": 0.05926020734672981,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0019955361779321282,
      "loss": 6.8035,
      "step": 463
    },
    {
      "epoch": 0.05938819915525406,
      "grad_norm": 1.078125,
      "learning_rate": 0.0019954969652069897,
      "loss": 6.5063,
      "step": 464
    },
    {
      "epoch": 0.05951619096377832,
      "grad_norm": 1.2734375,
      "learning_rate": 0.001995457581389546,
      "loss": 6.6054,
      "step": 465
    },
    {
      "epoch": 0.05964418277230257,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0019954180264865656,
      "loss": 6.6185,
      "step": 466
    },
    {
      "epoch": 0.05977217458082683,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0019953783005048475,
      "loss": 6.3587,
      "step": 467
    },
    {
      "epoch": 0.059900166389351084,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0019953384034512183,
      "loss": 6.8593,
      "step": 468
    },
    {
      "epoch": 0.06002815819787534,
      "grad_norm": 1.0234375,
      "learning_rate": 0.001995298335332536,
      "loss": 7.2911,
      "step": 469
    },
    {
      "epoch": 0.06015615000639959,
      "grad_norm": 1.109375,
      "learning_rate": 0.0019952580961556856,
      "loss": 6.387,
      "step": 470
    },
    {
      "epoch": 0.06028414181492384,
      "grad_norm": 1.015625,
      "learning_rate": 0.0019952176859275835,
      "loss": 6.2956,
      "step": 471
    },
    {
      "epoch": 0.0604121336234481,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0019951771046551755,
      "loss": 6.6801,
      "step": 472
    },
    {
      "epoch": 0.060540125431972354,
      "grad_norm": 0.99609375,
      "learning_rate": 0.001995136352345435,
      "loss": 6.5915,
      "step": 473
    },
    {
      "epoch": 0.06066811724049661,
      "grad_norm": 1.4140625,
      "learning_rate": 0.001995095429005367,
      "loss": 6.4799,
      "step": 474
    },
    {
      "epoch": 0.060796109049020865,
      "grad_norm": 1.21875,
      "learning_rate": 0.0019950543346420042,
      "loss": 6.1293,
      "step": 475
    },
    {
      "epoch": 0.06092410085754512,
      "grad_norm": 1.5,
      "learning_rate": 0.001995013069262409,
      "loss": 6.9277,
      "step": 476
    },
    {
      "epoch": 0.06105209266606937,
      "grad_norm": 1.359375,
      "learning_rate": 0.0019949716328736744,
      "loss": 6.5933,
      "step": 477
    },
    {
      "epoch": 0.061180084474593624,
      "grad_norm": 1.125,
      "learning_rate": 0.0019949300254829216,
      "loss": 7.1021,
      "step": 478
    },
    {
      "epoch": 0.06130807628311788,
      "grad_norm": 1.3515625,
      "learning_rate": 0.001994888247097301,
      "loss": 6.2418,
      "step": 479
    },
    {
      "epoch": 0.061436068091642135,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0019948462977239937,
      "loss": 6.8638,
      "step": 480
    },
    {
      "epoch": 0.06156405990016639,
      "grad_norm": 1.3125,
      "learning_rate": 0.001994804177370209,
      "loss": 6.6927,
      "step": 481
    },
    {
      "epoch": 0.061692051708690646,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0019947618860431857,
      "loss": 5.5488,
      "step": 482
    },
    {
      "epoch": 0.0618200435172149,
      "grad_norm": 2.453125,
      "learning_rate": 0.0019947194237501928,
      "loss": 6.6076,
      "step": 483
    },
    {
      "epoch": 0.06194803532573915,
      "grad_norm": 1.1875,
      "learning_rate": 0.0019946767904985277,
      "loss": 6.5381,
      "step": 484
    },
    {
      "epoch": 0.062076027134263405,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0019946339862955175,
      "loss": 5.8563,
      "step": 485
    },
    {
      "epoch": 0.06220401894278766,
      "grad_norm": 1.578125,
      "learning_rate": 0.0019945910111485196,
      "loss": 6.5543,
      "step": 486
    },
    {
      "epoch": 0.062332010751311916,
      "grad_norm": 1.265625,
      "learning_rate": 0.001994547865064919,
      "loss": 6.2423,
      "step": 487
    },
    {
      "epoch": 0.06246000255983617,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0019945045480521317,
      "loss": 6.5485,
      "step": 488
    },
    {
      "epoch": 0.06258799436836042,
      "grad_norm": 1.0546875,
      "learning_rate": 0.001994461060117602,
      "loss": 6.746,
      "step": 489
    },
    {
      "epoch": 0.06271598617688467,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0019944174012688047,
      "loss": 6.1284,
      "step": 490
    },
    {
      "epoch": 0.06284397798540893,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0019943735715132424,
      "loss": 6.4353,
      "step": 491
    },
    {
      "epoch": 0.06297196979393319,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0019943295708584486,
      "loss": 6.1249,
      "step": 492
    },
    {
      "epoch": 0.06309996160245744,
      "grad_norm": 1.78125,
      "learning_rate": 0.001994285399311985,
      "loss": 6.3034,
      "step": 493
    },
    {
      "epoch": 0.0632279534109817,
      "grad_norm": 1.171875,
      "learning_rate": 0.001994241056881444,
      "loss": 6.1632,
      "step": 494
    },
    {
      "epoch": 0.06335594521950595,
      "grad_norm": 1.1875,
      "learning_rate": 0.0019941965435744456,
      "loss": 6.0351,
      "step": 495
    },
    {
      "epoch": 0.06348393702803021,
      "grad_norm": 1.0625,
      "learning_rate": 0.001994151859398641,
      "loss": 6.5344,
      "step": 496
    },
    {
      "epoch": 0.06361192883655446,
      "grad_norm": 0.99609375,
      "learning_rate": 0.001994107004361709,
      "loss": 6.8119,
      "step": 497
    },
    {
      "epoch": 0.06373992064507872,
      "grad_norm": 1.234375,
      "learning_rate": 0.001994061978471359,
      "loss": 6.022,
      "step": 498
    },
    {
      "epoch": 0.06386791245360297,
      "grad_norm": 1.734375,
      "learning_rate": 0.00199401678173533,
      "loss": 6.316,
      "step": 499
    },
    {
      "epoch": 0.06399590426212723,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0019939714141613894,
      "loss": 7.0251,
      "step": 500
    },
    {
      "epoch": 0.06412389607065148,
      "grad_norm": 1.6953125,
      "learning_rate": 0.0019939258757573344,
      "loss": 6.5996,
      "step": 501
    },
    {
      "epoch": 0.06425188787917573,
      "grad_norm": 1.328125,
      "learning_rate": 0.001993880166530991,
      "loss": 6.4581,
      "step": 502
    },
    {
      "epoch": 0.06437987968769998,
      "grad_norm": 1.625,
      "learning_rate": 0.001993834286490216,
      "loss": 6.393,
      "step": 503
    },
    {
      "epoch": 0.06450787149622424,
      "grad_norm": 1.2265625,
      "learning_rate": 0.001993788235642894,
      "loss": 5.9731,
      "step": 504
    },
    {
      "epoch": 0.06463586330474849,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0019937420139969396,
      "loss": 5.8187,
      "step": 505
    },
    {
      "epoch": 0.06476385511327275,
      "grad_norm": 1.2578125,
      "learning_rate": 0.001993695621560297,
      "loss": 6.6461,
      "step": 506
    },
    {
      "epoch": 0.064891846921797,
      "grad_norm": 1.296875,
      "learning_rate": 0.0019936490583409394,
      "loss": 6.7567,
      "step": 507
    },
    {
      "epoch": 0.06501983873032126,
      "grad_norm": 1.125,
      "learning_rate": 0.001993602324346869,
      "loss": 7.0416,
      "step": 508
    },
    {
      "epoch": 0.06514783053884551,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0019935554195861186,
      "loss": 6.1935,
      "step": 509
    },
    {
      "epoch": 0.06527582234736977,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0019935083440667492,
      "loss": 5.8587,
      "step": 510
    },
    {
      "epoch": 0.06540381415589402,
      "grad_norm": 1.0703125,
      "learning_rate": 0.001993461097796851,
      "loss": 6.0048,
      "step": 511
    },
    {
      "epoch": 0.06553180596441828,
      "grad_norm": 1.4375,
      "learning_rate": 0.001993413680784545,
      "loss": 6.3664,
      "step": 512
    },
    {
      "epoch": 0.06565979777294254,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00199336609303798,
      "loss": 5.6121,
      "step": 513
    },
    {
      "epoch": 0.06578778958146679,
      "grad_norm": 1.140625,
      "learning_rate": 0.0019933183345653346,
      "loss": 5.7348,
      "step": 514
    },
    {
      "epoch": 0.06591578138999105,
      "grad_norm": 1.109375,
      "learning_rate": 0.001993270405374817,
      "loss": 5.0338,
      "step": 515
    },
    {
      "epoch": 0.0660437731985153,
      "grad_norm": 1.296875,
      "learning_rate": 0.001993222305474665,
      "loss": 6.3537,
      "step": 516
    },
    {
      "epoch": 0.06617176500703954,
      "grad_norm": 1.265625,
      "learning_rate": 0.0019931740348731446,
      "loss": 6.5067,
      "step": 517
    },
    {
      "epoch": 0.0662997568155638,
      "grad_norm": 1.234375,
      "learning_rate": 0.0019931255935785523,
      "loss": 6.4928,
      "step": 518
    },
    {
      "epoch": 0.06642774862408805,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0019930769815992137,
      "loss": 6.11,
      "step": 519
    },
    {
      "epoch": 0.06655574043261231,
      "grad_norm": 1.109375,
      "learning_rate": 0.001993028198943483,
      "loss": 6.3656,
      "step": 520
    },
    {
      "epoch": 0.06668373224113656,
      "grad_norm": 1.734375,
      "learning_rate": 0.0019929792456197457,
      "loss": 5.7358,
      "step": 521
    },
    {
      "epoch": 0.06681172404966082,
      "grad_norm": 1.2265625,
      "learning_rate": 0.001992930121636413,
      "loss": 6.2351,
      "step": 522
    },
    {
      "epoch": 0.06693971585818508,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0019928808270019294,
      "loss": 6.9005,
      "step": 523
    },
    {
      "epoch": 0.06706770766670933,
      "grad_norm": 1.25,
      "learning_rate": 0.0019928313617247665,
      "loss": 6.2306,
      "step": 524
    },
    {
      "epoch": 0.06719569947523359,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0019927817258134257,
      "loss": 5.0563,
      "step": 525
    },
    {
      "epoch": 0.06732369128375784,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0019927319192764373,
      "loss": 6.1055,
      "step": 526
    },
    {
      "epoch": 0.0674516830922821,
      "grad_norm": 1.2578125,
      "learning_rate": 0.001992681942122362,
      "loss": 5.4302,
      "step": 527
    },
    {
      "epoch": 0.06757967490080635,
      "grad_norm": 1.1171875,
      "learning_rate": 0.001992631794359789,
      "loss": 5.721,
      "step": 528
    },
    {
      "epoch": 0.06770766670933061,
      "grad_norm": 2.296875,
      "learning_rate": 0.001992581475997337,
      "loss": 5.5592,
      "step": 529
    },
    {
      "epoch": 0.06783565851785486,
      "grad_norm": 1.25,
      "learning_rate": 0.0019925309870436536,
      "loss": 4.6098,
      "step": 530
    },
    {
      "epoch": 0.06796365032637912,
      "grad_norm": 1.78125,
      "learning_rate": 0.0019924803275074166,
      "loss": 6.1123,
      "step": 531
    },
    {
      "epoch": 0.06809164213490336,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0019924294973973325,
      "loss": 5.4104,
      "step": 532
    },
    {
      "epoch": 0.06821963394342762,
      "grad_norm": 1.125,
      "learning_rate": 0.0019923784967221378,
      "loss": 5.9449,
      "step": 533
    },
    {
      "epoch": 0.06834762575195187,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0019923273254905966,
      "loss": 6.1232,
      "step": 534
    },
    {
      "epoch": 0.06847561756047613,
      "grad_norm": 0.90234375,
      "learning_rate": 0.001992275983711505,
      "loss": 6.0036,
      "step": 535
    },
    {
      "epoch": 0.06860360936900038,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0019922244713936857,
      "loss": 6.3455,
      "step": 536
    },
    {
      "epoch": 0.06873160117752464,
      "grad_norm": 1.203125,
      "learning_rate": 0.0019921727885459928,
      "loss": 5.4057,
      "step": 537
    },
    {
      "epoch": 0.06885959298604889,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0019921209351773076,
      "loss": 5.4196,
      "step": 538
    },
    {
      "epoch": 0.06898758479457315,
      "grad_norm": 1.484375,
      "learning_rate": 0.0019920689112965435,
      "loss": 5.6834,
      "step": 539
    },
    {
      "epoch": 0.0691155766030974,
      "grad_norm": 1.03125,
      "learning_rate": 0.0019920167169126405,
      "loss": 6.2026,
      "step": 540
    },
    {
      "epoch": 0.06924356841162166,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0019919643520345696,
      "loss": 5.4044,
      "step": 541
    },
    {
      "epoch": 0.06937156022014591,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0019919118166713307,
      "loss": 6.3794,
      "step": 542
    },
    {
      "epoch": 0.06949955202867017,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0019918591108319525,
      "loss": 5.7818,
      "step": 543
    },
    {
      "epoch": 0.06962754383719442,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0019918062345254933,
      "loss": 4.6935,
      "step": 544
    },
    {
      "epoch": 0.06975553564571868,
      "grad_norm": 0.88671875,
      "learning_rate": 0.001991753187761041,
      "loss": 5.5346,
      "step": 545
    },
    {
      "epoch": 0.06988352745424292,
      "grad_norm": 1.125,
      "learning_rate": 0.001991699970547712,
      "loss": 5.6045,
      "step": 546
    },
    {
      "epoch": 0.07001151926276718,
      "grad_norm": 1.015625,
      "learning_rate": 0.0019916465828946532,
      "loss": 5.8812,
      "step": 547
    },
    {
      "epoch": 0.07013951107129143,
      "grad_norm": 1.046875,
      "learning_rate": 0.00199159302481104,
      "loss": 5.8691,
      "step": 548
    },
    {
      "epoch": 0.07026750287981569,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0019915392963060774,
      "loss": 6.3146,
      "step": 549
    },
    {
      "epoch": 0.07039549468833994,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0019914853973889987,
      "loss": 5.5429,
      "step": 550
    },
    {
      "epoch": 0.0705234864968642,
      "grad_norm": 0.828125,
      "learning_rate": 0.001991431328069068,
      "loss": 5.3731,
      "step": 551
    },
    {
      "epoch": 0.07065147830538845,
      "grad_norm": 0.94921875,
      "learning_rate": 0.001991377088355578,
      "loss": 5.0177,
      "step": 552
    },
    {
      "epoch": 0.07077947011391271,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0019913226782578507,
      "loss": 5.3153,
      "step": 553
    },
    {
      "epoch": 0.07090746192243697,
      "grad_norm": 1.078125,
      "learning_rate": 0.0019912680977852365,
      "loss": 5.8298,
      "step": 554
    },
    {
      "epoch": 0.07103545373096122,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0019912133469471172,
      "loss": 5.7878,
      "step": 555
    },
    {
      "epoch": 0.07116344553948548,
      "grad_norm": 0.87890625,
      "learning_rate": 0.001991158425752902,
      "loss": 5.3954,
      "step": 556
    },
    {
      "epoch": 0.07129143734800973,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0019911033342120297,
      "loss": 4.953,
      "step": 557
    },
    {
      "epoch": 0.07141942915653399,
      "grad_norm": 0.94921875,
      "learning_rate": 0.001991048072333969,
      "loss": 5.0921,
      "step": 558
    },
    {
      "epoch": 0.07154742096505824,
      "grad_norm": 1.2109375,
      "learning_rate": 0.001990992640128218,
      "loss": 5.314,
      "step": 559
    },
    {
      "epoch": 0.0716754127735825,
      "grad_norm": 1.0625,
      "learning_rate": 0.001990937037604303,
      "loss": 6.0037,
      "step": 560
    },
    {
      "epoch": 0.07180340458210674,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0019908812647717804,
      "loss": 5.4362,
      "step": 561
    },
    {
      "epoch": 0.071931396390631,
      "grad_norm": 1.0625,
      "learning_rate": 0.0019908253216402356,
      "loss": 5.5438,
      "step": 562
    },
    {
      "epoch": 0.07205938819915525,
      "grad_norm": 1.2109375,
      "learning_rate": 0.001990769208219283,
      "loss": 5.5765,
      "step": 563
    },
    {
      "epoch": 0.0721873800076795,
      "grad_norm": 1.2265625,
      "learning_rate": 0.001990712924518567,
      "loss": 5.669,
      "step": 564
    },
    {
      "epoch": 0.07231537181620376,
      "grad_norm": 0.890625,
      "learning_rate": 0.0019906564705477613,
      "loss": 5.5131,
      "step": 565
    },
    {
      "epoch": 0.07244336362472802,
      "grad_norm": 1.0546875,
      "learning_rate": 0.001990599846316568,
      "loss": 4.9516,
      "step": 566
    },
    {
      "epoch": 0.07257135543325227,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0019905430518347184,
      "loss": 6.1631,
      "step": 567
    },
    {
      "epoch": 0.07269934724177653,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0019904860871119744,
      "loss": 6.1165,
      "step": 568
    },
    {
      "epoch": 0.07282733905030078,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0019904289521581256,
      "loss": 4.8725,
      "step": 569
    },
    {
      "epoch": 0.07295533085882504,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0019903716469829922,
      "loss": 4.482,
      "step": 570
    },
    {
      "epoch": 0.07308332266734929,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0019903141715964232,
      "loss": 5.8826,
      "step": 571
    },
    {
      "epoch": 0.07321131447587355,
      "grad_norm": 1.03125,
      "learning_rate": 0.0019902565260082952,
      "loss": 5.0312,
      "step": 572
    },
    {
      "epoch": 0.0733393062843978,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0019901987102285174,
      "loss": 5.3519,
      "step": 573
    },
    {
      "epoch": 0.07346729809292206,
      "grad_norm": 0.98828125,
      "learning_rate": 0.001990140724267025,
      "loss": 5.8923,
      "step": 574
    },
    {
      "epoch": 0.0735952899014463,
      "grad_norm": 1.125,
      "learning_rate": 0.001990082568133784,
      "loss": 5.8639,
      "step": 575
    },
    {
      "epoch": 0.07372328170997056,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0019900242418387903,
      "loss": 5.8164,
      "step": 576
    },
    {
      "epoch": 0.07385127351849481,
      "grad_norm": 0.9375,
      "learning_rate": 0.0019899657453920677,
      "loss": 5.5802,
      "step": 577
    },
    {
      "epoch": 0.07397926532701907,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00198990707880367,
      "loss": 5.3828,
      "step": 578
    },
    {
      "epoch": 0.07410725713554332,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0019898482420836797,
      "loss": 5.7168,
      "step": 579
    },
    {
      "epoch": 0.07423524894406758,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0019897892352422086,
      "loss": 5.8738,
      "step": 580
    },
    {
      "epoch": 0.07436324075259183,
      "grad_norm": 1.046875,
      "learning_rate": 0.001989730058289399,
      "loss": 6.1339,
      "step": 581
    },
    {
      "epoch": 0.07449123256111609,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00198967071123542,
      "loss": 5.2345,
      "step": 582
    },
    {
      "epoch": 0.07461922436964034,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0019896111940904725,
      "loss": 5.3292,
      "step": 583
    },
    {
      "epoch": 0.0747472161781646,
      "grad_norm": 0.8515625,
      "learning_rate": 0.001989551506864785,
      "loss": 5.065,
      "step": 584
    },
    {
      "epoch": 0.07487520798668885,
      "grad_norm": 0.890625,
      "learning_rate": 0.001989491649568616,
      "loss": 4.7571,
      "step": 585
    },
    {
      "epoch": 0.07500319979521311,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0019894316222122525,
      "loss": 5.265,
      "step": 586
    },
    {
      "epoch": 0.07513119160373737,
      "grad_norm": 0.765625,
      "learning_rate": 0.0019893714248060113,
      "loss": 4.7523,
      "step": 587
    },
    {
      "epoch": 0.07525918341226162,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0019893110573602386,
      "loss": 5.5748,
      "step": 588
    },
    {
      "epoch": 0.07538717522078588,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00198925051988531,
      "loss": 4.8974,
      "step": 589
    },
    {
      "epoch": 0.07551516702931012,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0019891898123916285,
      "loss": 5.3623,
      "step": 590
    },
    {
      "epoch": 0.07564315883783437,
      "grad_norm": 0.83203125,
      "learning_rate": 0.001989128934889628,
      "loss": 5.249,
      "step": 591
    },
    {
      "epoch": 0.07577115064635863,
      "grad_norm": 0.890625,
      "learning_rate": 0.001989067887389772,
      "loss": 5.1926,
      "step": 592
    },
    {
      "epoch": 0.07589914245488288,
      "grad_norm": 0.98046875,
      "learning_rate": 0.001989006669902553,
      "loss": 5.4034,
      "step": 593
    },
    {
      "epoch": 0.07602713426340714,
      "grad_norm": 0.84375,
      "learning_rate": 0.0019889452824384908,
      "loss": 6.3733,
      "step": 594
    },
    {
      "epoch": 0.0761551260719314,
      "grad_norm": 0.96484375,
      "learning_rate": 0.001988883725008136,
      "loss": 5.3162,
      "step": 595
    },
    {
      "epoch": 0.07628311788045565,
      "grad_norm": 0.9140625,
      "learning_rate": 0.001988821997622069,
      "loss": 4.7493,
      "step": 596
    },
    {
      "epoch": 0.0764111096889799,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0019887601002908984,
      "loss": 5.9357,
      "step": 597
    },
    {
      "epoch": 0.07653910149750416,
      "grad_norm": 1.078125,
      "learning_rate": 0.001988698033025262,
      "loss": 5.8667,
      "step": 598
    },
    {
      "epoch": 0.07666709330602842,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0019886357958358272,
      "loss": 4.9038,
      "step": 599
    },
    {
      "epoch": 0.07679508511455267,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0019885733887332908,
      "loss": 5.0805,
      "step": 600
    },
    {
      "epoch": 0.07692307692307693,
      "grad_norm": 0.77734375,
      "learning_rate": 0.001988510811728378,
      "loss": 4.9848,
      "step": 601
    },
    {
      "epoch": 0.07705106873160118,
      "grad_norm": 0.953125,
      "learning_rate": 0.0019884480648318438,
      "loss": 4.7929,
      "step": 602
    },
    {
      "epoch": 0.07717906054012544,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0019883851480544724,
      "loss": 5.4275,
      "step": 603
    },
    {
      "epoch": 0.07730705234864968,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0019883220614070767,
      "loss": 5.4174,
      "step": 604
    },
    {
      "epoch": 0.07743504415717394,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0019882588049005,
      "loss": 5.7157,
      "step": 605
    },
    {
      "epoch": 0.07756303596569819,
      "grad_norm": 1.1171875,
      "learning_rate": 0.001988195378545613,
      "loss": 5.2198,
      "step": 606
    },
    {
      "epoch": 0.07769102777422245,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0019881317823533173,
      "loss": 5.3102,
      "step": 607
    },
    {
      "epoch": 0.0778190195827467,
      "grad_norm": 0.94921875,
      "learning_rate": 0.001988068016334542,
      "loss": 4.9465,
      "step": 608
    },
    {
      "epoch": 0.07794701139127096,
      "grad_norm": 1.109375,
      "learning_rate": 0.0019880040805002478,
      "loss": 5.0658,
      "step": 609
    },
    {
      "epoch": 0.07807500319979521,
      "grad_norm": 0.9453125,
      "learning_rate": 0.001987939974861422,
      "loss": 5.0365,
      "step": 610
    },
    {
      "epoch": 0.07820299500831947,
      "grad_norm": 0.81640625,
      "learning_rate": 0.001987875699429082,
      "loss": 5.0959,
      "step": 611
    },
    {
      "epoch": 0.07833098681684372,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0019878112542142754,
      "loss": 4.6309,
      "step": 612
    },
    {
      "epoch": 0.07845897862536798,
      "grad_norm": 0.78515625,
      "learning_rate": 0.001987746639228077,
      "loss": 4.9451,
      "step": 613
    },
    {
      "epoch": 0.07858697043389223,
      "grad_norm": 1.0703125,
      "learning_rate": 0.001987681854481594,
      "loss": 5.8263,
      "step": 614
    },
    {
      "epoch": 0.07871496224241649,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0019876168999859584,
      "loss": 4.6283,
      "step": 615
    },
    {
      "epoch": 0.07884295405094074,
      "grad_norm": 0.796875,
      "learning_rate": 0.001987551775752335,
      "loss": 5.0157,
      "step": 616
    },
    {
      "epoch": 0.078970945859465,
      "grad_norm": 0.98828125,
      "learning_rate": 0.001987486481791916,
      "loss": 5.3779,
      "step": 617
    },
    {
      "epoch": 0.07909893766798926,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0019874210181159238,
      "loss": 4.5538,
      "step": 618
    },
    {
      "epoch": 0.0792269294765135,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0019873553847356085,
      "loss": 4.5813,
      "step": 619
    },
    {
      "epoch": 0.07935492128503775,
      "grad_norm": 0.875,
      "learning_rate": 0.001987289581662251,
      "loss": 5.5369,
      "step": 620
    },
    {
      "epoch": 0.07948291309356201,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00198722360890716,
      "loss": 5.185,
      "step": 621
    },
    {
      "epoch": 0.07961090490208626,
      "grad_norm": 0.87890625,
      "learning_rate": 0.001987157466481675,
      "loss": 4.3254,
      "step": 622
    },
    {
      "epoch": 0.07973889671061052,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0019870911543971624,
      "loss": 4.7239,
      "step": 623
    },
    {
      "epoch": 0.07986688851913477,
      "grad_norm": 0.796875,
      "learning_rate": 0.00198702467266502,
      "loss": 4.8223,
      "step": 624
    },
    {
      "epoch": 0.07999488032765903,
      "grad_norm": 0.89453125,
      "learning_rate": 0.001986958021296673,
      "loss": 5.1434,
      "step": 625
    },
    {
      "epoch": 0.08012287213618328,
      "grad_norm": 0.828125,
      "learning_rate": 0.001986891200303577,
      "loss": 4.5331,
      "step": 626
    },
    {
      "epoch": 0.08025086394470754,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0019868242096972165,
      "loss": 4.4901,
      "step": 627
    },
    {
      "epoch": 0.0803788557532318,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0019867570494891042,
      "loss": 4.7608,
      "step": 628
    },
    {
      "epoch": 0.08050684756175605,
      "grad_norm": 0.921875,
      "learning_rate": 0.0019866897196907833,
      "loss": 5.24,
      "step": 629
    },
    {
      "epoch": 0.0806348393702803,
      "grad_norm": 0.81640625,
      "learning_rate": 0.001986622220313825,
      "loss": 5.4133,
      "step": 630
    },
    {
      "epoch": 0.08076283117880456,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0019865545513698306,
      "loss": 5.0365,
      "step": 631
    },
    {
      "epoch": 0.08089082298732882,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0019864867128704298,
      "loss": 4.9229,
      "step": 632
    },
    {
      "epoch": 0.08101881479585306,
      "grad_norm": 0.828125,
      "learning_rate": 0.001986418704827282,
      "loss": 5.6099,
      "step": 633
    },
    {
      "epoch": 0.08114680660437731,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0019863505272520753,
      "loss": 4.0732,
      "step": 634
    },
    {
      "epoch": 0.08127479841290157,
      "grad_norm": 1.1796875,
      "learning_rate": 0.001986282180156527,
      "loss": 4.8182,
      "step": 635
    },
    {
      "epoch": 0.08140279022142582,
      "grad_norm": 0.98046875,
      "learning_rate": 0.001986213663552384,
      "loss": 4.6104,
      "step": 636
    },
    {
      "epoch": 0.08153078202995008,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0019861449774514223,
      "loss": 5.0221,
      "step": 637
    },
    {
      "epoch": 0.08165877383847434,
      "grad_norm": 0.57421875,
      "learning_rate": 0.001986076121865446,
      "loss": 4.1284,
      "step": 638
    },
    {
      "epoch": 0.08178676564699859,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0019860070968062893,
      "loss": 4.1344,
      "step": 639
    },
    {
      "epoch": 0.08191475745552285,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0019859379022858148,
      "loss": 4.4689,
      "step": 640
    },
    {
      "epoch": 0.0820427492640471,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0019858685383159157,
      "loss": 4.8545,
      "step": 641
    },
    {
      "epoch": 0.08217074107257136,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0019857990049085127,
      "loss": 5.2177,
      "step": 642
    },
    {
      "epoch": 0.08229873288109561,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0019857293020755563,
      "loss": 4.6558,
      "step": 643
    },
    {
      "epoch": 0.08242672468961987,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0019856594298290264,
      "loss": 4.6432,
      "step": 644
    },
    {
      "epoch": 0.08255471649814412,
      "grad_norm": 0.80859375,
      "learning_rate": 0.001985589388180931,
      "loss": 5.1168,
      "step": 645
    },
    {
      "epoch": 0.08268270830666838,
      "grad_norm": 0.78125,
      "learning_rate": 0.0019855191771433085,
      "loss": 4.9751,
      "step": 646
    },
    {
      "epoch": 0.08281070011519263,
      "grad_norm": 0.828125,
      "learning_rate": 0.0019854487967282254,
      "loss": 4.6476,
      "step": 647
    },
    {
      "epoch": 0.08293869192371688,
      "grad_norm": 0.9765625,
      "learning_rate": 0.001985378246947778,
      "loss": 4.5548,
      "step": 648
    },
    {
      "epoch": 0.08306668373224113,
      "grad_norm": 0.8125,
      "learning_rate": 0.0019853075278140914,
      "loss": 5.4269,
      "step": 649
    },
    {
      "epoch": 0.08319467554076539,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00198523663933932,
      "loss": 4.3664,
      "step": 650
    },
    {
      "epoch": 0.08332266734928964,
      "grad_norm": 0.828125,
      "learning_rate": 0.0019851655815356466,
      "loss": 4.2599,
      "step": 651
    },
    {
      "epoch": 0.0834506591578139,
      "grad_norm": 0.921875,
      "learning_rate": 0.0019850943544152838,
      "loss": 4.4381,
      "step": 652
    },
    {
      "epoch": 0.08357865096633815,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0019850229579904733,
      "loss": 4.0914,
      "step": 653
    },
    {
      "epoch": 0.08370664277486241,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0019849513922734863,
      "loss": 4.7392,
      "step": 654
    },
    {
      "epoch": 0.08383463458338666,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0019848796572766213,
      "loss": 4.6851,
      "step": 655
    },
    {
      "epoch": 0.08396262639191092,
      "grad_norm": 1.8984375,
      "learning_rate": 0.001984807753012208,
      "loss": 4.8299,
      "step": 656
    },
    {
      "epoch": 0.08409061820043517,
      "grad_norm": 0.8359375,
      "learning_rate": 0.001984735679492604,
      "loss": 4.8623,
      "step": 657
    },
    {
      "epoch": 0.08421861000895943,
      "grad_norm": 0.9765625,
      "learning_rate": 0.001984663436730197,
      "loss": 4.7052,
      "step": 658
    },
    {
      "epoch": 0.08434660181748369,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0019845910247374013,
      "loss": 4.7845,
      "step": 659
    },
    {
      "epoch": 0.08447459362600794,
      "grad_norm": 0.82421875,
      "learning_rate": 0.001984518443526664,
      "loss": 5.0864,
      "step": 660
    },
    {
      "epoch": 0.0846025854345322,
      "grad_norm": 0.92578125,
      "learning_rate": 0.001984445693110459,
      "loss": 4.5949,
      "step": 661
    },
    {
      "epoch": 0.08473057724305644,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0019843727735012885,
      "loss": 5.2515,
      "step": 662
    },
    {
      "epoch": 0.08485856905158069,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0019842996847116855,
      "loss": 5.3414,
      "step": 663
    },
    {
      "epoch": 0.08498656086010495,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0019842264267542122,
      "loss": 4.752,
      "step": 664
    },
    {
      "epoch": 0.0851145526686292,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0019841529996414585,
      "loss": 4.4128,
      "step": 665
    },
    {
      "epoch": 0.08524254447715346,
      "grad_norm": 0.94140625,
      "learning_rate": 0.001984079403386044,
      "loss": 4.9011,
      "step": 666
    },
    {
      "epoch": 0.08537053628567771,
      "grad_norm": 0.85546875,
      "learning_rate": 0.001984005638000618,
      "loss": 4.722,
      "step": 667
    },
    {
      "epoch": 0.08549852809420197,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0019839317034978576,
      "loss": 4.8795,
      "step": 668
    },
    {
      "epoch": 0.08562651990272623,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0019838575998904697,
      "loss": 4.3069,
      "step": 669
    },
    {
      "epoch": 0.08575451171125048,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0019837833271911906,
      "loss": 4.3752,
      "step": 670
    },
    {
      "epoch": 0.08588250351977474,
      "grad_norm": 0.84765625,
      "learning_rate": 0.001983708885412785,
      "loss": 5.349,
      "step": 671
    },
    {
      "epoch": 0.08601049532829899,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0019836342745680473,
      "loss": 4.7595,
      "step": 672
    },
    {
      "epoch": 0.08613848713682325,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0019835594946698,
      "loss": 5.0233,
      "step": 673
    },
    {
      "epoch": 0.0862664789453475,
      "grad_norm": 0.859375,
      "learning_rate": 0.0019834845457308956,
      "loss": 3.208,
      "step": 674
    },
    {
      "epoch": 0.08639447075387176,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0019834094277642153,
      "loss": 5.8435,
      "step": 675
    },
    {
      "epoch": 0.08652246256239601,
      "grad_norm": 1.140625,
      "learning_rate": 0.0019833341407826687,
      "loss": 5.1078,
      "step": 676
    },
    {
      "epoch": 0.08665045437092025,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0019832586847991964,
      "loss": 4.787,
      "step": 677
    },
    {
      "epoch": 0.08677844617944451,
      "grad_norm": 0.80078125,
      "learning_rate": 0.001983183059826766,
      "loss": 4.0605,
      "step": 678
    },
    {
      "epoch": 0.08690643798796877,
      "grad_norm": 1.265625,
      "learning_rate": 0.001983107265878374,
      "loss": 4.4668,
      "step": 679
    },
    {
      "epoch": 0.08703442979649302,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0019830313029670483,
      "loss": 4.1574,
      "step": 680
    },
    {
      "epoch": 0.08716242160501728,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0019829551711058437,
      "loss": 4.8034,
      "step": 681
    },
    {
      "epoch": 0.08729041341354153,
      "grad_norm": 0.87890625,
      "learning_rate": 0.001982878870307845,
      "loss": 5.7189,
      "step": 682
    },
    {
      "epoch": 0.08741840522206579,
      "grad_norm": 1.0859375,
      "learning_rate": 0.001982802400586165,
      "loss": 4.3374,
      "step": 683
    },
    {
      "epoch": 0.08754639703059004,
      "grad_norm": 1.421875,
      "learning_rate": 0.001982725761953947,
      "loss": 5.032,
      "step": 684
    },
    {
      "epoch": 0.0876743888391143,
      "grad_norm": 0.96875,
      "learning_rate": 0.001982648954424362,
      "loss": 4.5456,
      "step": 685
    },
    {
      "epoch": 0.08780238064763855,
      "grad_norm": 1.34375,
      "learning_rate": 0.001982571978010611,
      "loss": 4.662,
      "step": 686
    },
    {
      "epoch": 0.08793037245616281,
      "grad_norm": 0.78515625,
      "learning_rate": 0.001982494832725924,
      "loss": 4.433,
      "step": 687
    },
    {
      "epoch": 0.08805836426468706,
      "grad_norm": 0.70703125,
      "learning_rate": 0.001982417518583559,
      "loss": 3.8644,
      "step": 688
    },
    {
      "epoch": 0.08818635607321132,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0019823400355968037,
      "loss": 4.7663,
      "step": 689
    },
    {
      "epoch": 0.08831434788173557,
      "grad_norm": 0.703125,
      "learning_rate": 0.0019822623837789754,
      "loss": 3.9368,
      "step": 690
    },
    {
      "epoch": 0.08844233969025983,
      "grad_norm": 1.1796875,
      "learning_rate": 0.001982184563143419,
      "loss": 4.4762,
      "step": 691
    },
    {
      "epoch": 0.08857033149878407,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0019821065737035105,
      "loss": 4.1119,
      "step": 692
    },
    {
      "epoch": 0.08869832330730833,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0019820284154726526,
      "loss": 5.4806,
      "step": 693
    },
    {
      "epoch": 0.08882631511583258,
      "grad_norm": 0.69140625,
      "learning_rate": 0.001981950088464278,
      "loss": 4.2745,
      "step": 694
    },
    {
      "epoch": 0.08895430692435684,
      "grad_norm": 0.796875,
      "learning_rate": 0.001981871592691849,
      "loss": 4.0399,
      "step": 695
    },
    {
      "epoch": 0.0890822987328811,
      "grad_norm": 0.77734375,
      "learning_rate": 0.001981792928168856,
      "loss": 4.6626,
      "step": 696
    },
    {
      "epoch": 0.08921029054140535,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0019817140949088194,
      "loss": 4.7864,
      "step": 697
    },
    {
      "epoch": 0.0893382823499296,
      "grad_norm": 0.9375,
      "learning_rate": 0.001981635092925287,
      "loss": 4.4117,
      "step": 698
    },
    {
      "epoch": 0.08946627415845386,
      "grad_norm": 0.84765625,
      "learning_rate": 0.001981555922231837,
      "loss": 4.7975,
      "step": 699
    },
    {
      "epoch": 0.08959426596697811,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0019814765828420765,
      "loss": 4.8129,
      "step": 700
    },
    {
      "epoch": 0.08972225777550237,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0019813970747696416,
      "loss": 5.0728,
      "step": 701
    },
    {
      "epoch": 0.08985024958402663,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0019813173980281957,
      "loss": 4.0898,
      "step": 702
    },
    {
      "epoch": 0.08997824139255088,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0019812375526314335,
      "loss": 4.3238,
      "step": 703
    },
    {
      "epoch": 0.09010623320107514,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0019811575385930776,
      "loss": 4.3598,
      "step": 704
    },
    {
      "epoch": 0.09023422500959939,
      "grad_norm": 1.015625,
      "learning_rate": 0.0019810773559268795,
      "loss": 4.0851,
      "step": 705
    },
    {
      "epoch": 0.09036221681812363,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0019809970046466203,
      "loss": 4.5524,
      "step": 706
    },
    {
      "epoch": 0.09049020862664789,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0019809164847661093,
      "loss": 4.7088,
      "step": 707
    },
    {
      "epoch": 0.09061820043517214,
      "grad_norm": 0.703125,
      "learning_rate": 0.0019808357962991852,
      "loss": 5.3506,
      "step": 708
    },
    {
      "epoch": 0.0907461922436964,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0019807549392597155,
      "loss": 5.1162,
      "step": 709
    },
    {
      "epoch": 0.09087418405222066,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0019806739136615974,
      "loss": 4.65,
      "step": 710
    },
    {
      "epoch": 0.09100217586074491,
      "grad_norm": 1.0859375,
      "learning_rate": 0.001980592719518756,
      "loss": 4.3019,
      "step": 711
    },
    {
      "epoch": 0.09113016766926917,
      "grad_norm": 0.78515625,
      "learning_rate": 0.001980511356845146,
      "loss": 4.2662,
      "step": 712
    },
    {
      "epoch": 0.09125815947779342,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00198042982565475,
      "loss": 4.3935,
      "step": 713
    },
    {
      "epoch": 0.09138615128631768,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0019803481259615817,
      "loss": 4.2838,
      "step": 714
    },
    {
      "epoch": 0.09151414309484193,
      "grad_norm": 0.73046875,
      "learning_rate": 0.001980266257779682,
      "loss": 3.8178,
      "step": 715
    },
    {
      "epoch": 0.09164213490336619,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0019801842211231214,
      "loss": 3.8339,
      "step": 716
    },
    {
      "epoch": 0.09177012671189044,
      "grad_norm": 0.88671875,
      "learning_rate": 0.001980102016005999,
      "loss": 4.864,
      "step": 717
    },
    {
      "epoch": 0.0918981185204147,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0019800196424424434,
      "loss": 4.958,
      "step": 718
    },
    {
      "epoch": 0.09202611032893895,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0019799371004466117,
      "loss": 4.9387,
      "step": 719
    },
    {
      "epoch": 0.09215410213746321,
      "grad_norm": 0.75,
      "learning_rate": 0.00197985439003269,
      "loss": 4.7775,
      "step": 720
    },
    {
      "epoch": 0.09228209394598745,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0019797715112148937,
      "loss": 4.7766,
      "step": 721
    },
    {
      "epoch": 0.0924100857545117,
      "grad_norm": 0.71875,
      "learning_rate": 0.001979688464007466,
      "loss": 4.7705,
      "step": 722
    },
    {
      "epoch": 0.09253807756303596,
      "grad_norm": 0.8125,
      "learning_rate": 0.001979605248424681,
      "loss": 4.4002,
      "step": 723
    },
    {
      "epoch": 0.09266606937156022,
      "grad_norm": 0.859375,
      "learning_rate": 0.0019795218644808404,
      "loss": 4.2165,
      "step": 724
    },
    {
      "epoch": 0.09279406118008447,
      "grad_norm": 0.82421875,
      "learning_rate": 0.001979438312190274,
      "loss": 4.1008,
      "step": 725
    },
    {
      "epoch": 0.09292205298860873,
      "grad_norm": 1.078125,
      "learning_rate": 0.0019793545915673433,
      "loss": 4.3263,
      "step": 726
    },
    {
      "epoch": 0.09305004479713298,
      "grad_norm": 0.8359375,
      "learning_rate": 0.001979270702626436,
      "loss": 3.8691,
      "step": 727
    },
    {
      "epoch": 0.09317803660565724,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00197918664538197,
      "loss": 4.0446,
      "step": 728
    },
    {
      "epoch": 0.0933060284141815,
      "grad_norm": 0.703125,
      "learning_rate": 0.001979102419848392,
      "loss": 3.8539,
      "step": 729
    },
    {
      "epoch": 0.09343402022270575,
      "grad_norm": 0.84375,
      "learning_rate": 0.0019790180260401777,
      "loss": 4.5445,
      "step": 730
    },
    {
      "epoch": 0.09356201203123,
      "grad_norm": 0.78125,
      "learning_rate": 0.001978933463971831,
      "loss": 4.5544,
      "step": 731
    },
    {
      "epoch": 0.09369000383975426,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0019788487336578854,
      "loss": 4.1572,
      "step": 732
    },
    {
      "epoch": 0.09381799564827852,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0019787638351129034,
      "loss": 4.1803,
      "step": 733
    },
    {
      "epoch": 0.09394598745680277,
      "grad_norm": 0.94140625,
      "learning_rate": 0.001978678768351477,
      "loss": 4.7113,
      "step": 734
    },
    {
      "epoch": 0.09407397926532701,
      "grad_norm": 0.875,
      "learning_rate": 0.0019785935333882245,
      "loss": 4.1744,
      "step": 735
    },
    {
      "epoch": 0.09420197107385127,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0019785081302377957,
      "loss": 4.7301,
      "step": 736
    },
    {
      "epoch": 0.09432996288237552,
      "grad_norm": 0.8125,
      "learning_rate": 0.0019784225589148693,
      "loss": 3.7484,
      "step": 737
    },
    {
      "epoch": 0.09445795469089978,
      "grad_norm": 0.89453125,
      "learning_rate": 0.001978336819434151,
      "loss": 4.2035,
      "step": 738
    },
    {
      "epoch": 0.09458594649942403,
      "grad_norm": 0.859375,
      "learning_rate": 0.001978250911810377,
      "loss": 4.1146,
      "step": 739
    },
    {
      "epoch": 0.09471393830794829,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0019781648360583126,
      "loss": 5.0394,
      "step": 740
    },
    {
      "epoch": 0.09484193011647254,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0019780785921927496,
      "loss": 3.593,
      "step": 741
    },
    {
      "epoch": 0.0949699219249968,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0019779921802285123,
      "loss": 4.7858,
      "step": 742
    },
    {
      "epoch": 0.09509791373352106,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0019779056001804508,
      "loss": 4.7605,
      "step": 743
    },
    {
      "epoch": 0.09522590554204531,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0019778188520634455,
      "loss": 3.895,
      "step": 744
    },
    {
      "epoch": 0.09535389735056957,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0019777319358924056,
      "loss": 4.1992,
      "step": 745
    },
    {
      "epoch": 0.09548188915909382,
      "grad_norm": 1.03125,
      "learning_rate": 0.001977644851682269,
      "loss": 3.8095,
      "step": 746
    },
    {
      "epoch": 0.09560988096761808,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0019775575994480034,
      "loss": 3.6871,
      "step": 747
    },
    {
      "epoch": 0.09573787277614233,
      "grad_norm": 0.73828125,
      "learning_rate": 0.001977470179204603,
      "loss": 4.58,
      "step": 748
    },
    {
      "epoch": 0.09586586458466659,
      "grad_norm": 0.7265625,
      "learning_rate": 0.001977382590967093,
      "loss": 4.2213,
      "step": 749
    },
    {
      "epoch": 0.09599385639319083,
      "grad_norm": 0.69921875,
      "learning_rate": 0.001977294834750527,
      "loss": 3.2659,
      "step": 750
    },
    {
      "epoch": 0.09612184820171508,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0019772069105699875,
      "loss": 3.6689,
      "step": 751
    },
    {
      "epoch": 0.09624984001023934,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0019771188184405856,
      "loss": 4.9027,
      "step": 752
    },
    {
      "epoch": 0.0963778318187636,
      "grad_norm": 0.67578125,
      "learning_rate": 0.001977030558377461,
      "loss": 3.7807,
      "step": 753
    },
    {
      "epoch": 0.09650582362728785,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0019769421303957825,
      "loss": 4.5244,
      "step": 754
    },
    {
      "epoch": 0.0966338154358121,
      "grad_norm": 0.74609375,
      "learning_rate": 0.001976853534510749,
      "loss": 4.5131,
      "step": 755
    },
    {
      "epoch": 0.09676180724433636,
      "grad_norm": 0.6953125,
      "learning_rate": 0.001976764770737586,
      "loss": 3.9611,
      "step": 756
    },
    {
      "epoch": 0.09688979905286062,
      "grad_norm": 0.82421875,
      "learning_rate": 0.001976675839091549,
      "loss": 3.9749,
      "step": 757
    },
    {
      "epoch": 0.09701779086138487,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0019765867395879233,
      "loss": 4.9185,
      "step": 758
    },
    {
      "epoch": 0.09714578266990913,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0019764974722420213,
      "loss": 3.7599,
      "step": 759
    },
    {
      "epoch": 0.09727377447843338,
      "grad_norm": 0.73046875,
      "learning_rate": 0.001976408037069185,
      "loss": 4.2933,
      "step": 760
    },
    {
      "epoch": 0.09740176628695764,
      "grad_norm": 0.921875,
      "learning_rate": 0.001976318434084786,
      "loss": 4.2614,
      "step": 761
    },
    {
      "epoch": 0.0975297580954819,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0019762286633042236,
      "loss": 4.316,
      "step": 762
    },
    {
      "epoch": 0.09765774990400615,
      "grad_norm": 0.68359375,
      "learning_rate": 0.001976138724742926,
      "loss": 3.9049,
      "step": 763
    },
    {
      "epoch": 0.09778574171253039,
      "grad_norm": 1.3671875,
      "learning_rate": 0.001976048618416351,
      "loss": 3.9992,
      "step": 764
    },
    {
      "epoch": 0.09791373352105465,
      "grad_norm": 0.87890625,
      "learning_rate": 0.001975958344339985,
      "loss": 4.1841,
      "step": 765
    },
    {
      "epoch": 0.0980417253295789,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0019758679025293426,
      "loss": 4.4145,
      "step": 766
    },
    {
      "epoch": 0.09816971713810316,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0019757772929999678,
      "loss": 4.5701,
      "step": 767
    },
    {
      "epoch": 0.09829770894662741,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0019756865157674335,
      "loss": 4.8722,
      "step": 768
    },
    {
      "epoch": 0.09842570075515167,
      "grad_norm": 1.140625,
      "learning_rate": 0.0019755955708473413,
      "loss": 4.5436,
      "step": 769
    },
    {
      "epoch": 0.09855369256367592,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0019755044582553216,
      "loss": 3.654,
      "step": 770
    },
    {
      "epoch": 0.09868168437220018,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0019754131780070332,
      "loss": 4.826,
      "step": 771
    },
    {
      "epoch": 0.09880967618072443,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0019753217301181646,
      "loss": 4.34,
      "step": 772
    },
    {
      "epoch": 0.09893766798924869,
      "grad_norm": 0.703125,
      "learning_rate": 0.001975230114604432,
      "loss": 4.0559,
      "step": 773
    },
    {
      "epoch": 0.09906565979777295,
      "grad_norm": 0.6875,
      "learning_rate": 0.0019751383314815817,
      "loss": 3.83,
      "step": 774
    },
    {
      "epoch": 0.0991936516062972,
      "grad_norm": 0.796875,
      "learning_rate": 0.001975046380765387,
      "loss": 4.0673,
      "step": 775
    },
    {
      "epoch": 0.09932164341482146,
      "grad_norm": 0.765625,
      "learning_rate": 0.0019749542624716526,
      "loss": 4.2356,
      "step": 776
    },
    {
      "epoch": 0.09944963522334571,
      "grad_norm": 0.71875,
      "learning_rate": 0.0019748619766162094,
      "loss": 3.3531,
      "step": 777
    },
    {
      "epoch": 0.09957762703186997,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0019747695232149192,
      "loss": 4.5328,
      "step": 778
    },
    {
      "epoch": 0.09970561884039421,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0019746769022836704,
      "loss": 4.4564,
      "step": 779
    },
    {
      "epoch": 0.09983361064891846,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0019745841138383824,
      "loss": 4.7321,
      "step": 780
    },
    {
      "epoch": 0.09996160245744272,
      "grad_norm": 0.7578125,
      "learning_rate": 0.001974491157895002,
      "loss": 4.0785,
      "step": 781
    },
    {
      "epoch": 0.10008959426596697,
      "grad_norm": 0.8359375,
      "learning_rate": 0.001974398034469505,
      "loss": 3.8121,
      "step": 782
    },
    {
      "epoch": 0.10021758607449123,
      "grad_norm": 0.796875,
      "learning_rate": 0.001974304743577897,
      "loss": 3.75,
      "step": 783
    },
    {
      "epoch": 0.10034557788301549,
      "grad_norm": 0.734375,
      "learning_rate": 0.00197421128523621,
      "loss": 4.1297,
      "step": 784
    },
    {
      "epoch": 0.10047356969153974,
      "grad_norm": 0.85546875,
      "learning_rate": 0.001974117659460508,
      "loss": 3.7657,
      "step": 785
    },
    {
      "epoch": 0.100601561500064,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0019740238662668812,
      "loss": 4.3152,
      "step": 786
    },
    {
      "epoch": 0.10072955330858825,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00197392990567145,
      "loss": 4.8603,
      "step": 787
    },
    {
      "epoch": 0.10085754511711251,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0019738357776903624,
      "loss": 4.3303,
      "step": 788
    },
    {
      "epoch": 0.10098553692563676,
      "grad_norm": 0.8671875,
      "learning_rate": 0.001973741482339796,
      "loss": 4.1671,
      "step": 789
    },
    {
      "epoch": 0.10111352873416102,
      "grad_norm": 0.97265625,
      "learning_rate": 0.001973647019635957,
      "loss": 4.0281,
      "step": 790
    },
    {
      "epoch": 0.10124152054268527,
      "grad_norm": 0.71484375,
      "learning_rate": 0.001973552389595081,
      "loss": 3.7274,
      "step": 791
    },
    {
      "epoch": 0.10136951235120953,
      "grad_norm": 0.9453125,
      "learning_rate": 0.001973457592233431,
      "loss": 4.1331,
      "step": 792
    },
    {
      "epoch": 0.10149750415973377,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0019733626275673,
      "loss": 3.1635,
      "step": 793
    },
    {
      "epoch": 0.10162549596825803,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0019732674956130084,
      "loss": 4.3879,
      "step": 794
    },
    {
      "epoch": 0.10175348777678228,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0019731721963869068,
      "loss": 4.3729,
      "step": 795
    },
    {
      "epoch": 0.10188147958530654,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0019730767299053734,
      "loss": 3.4163,
      "step": 796
    },
    {
      "epoch": 0.10200947139383079,
      "grad_norm": 1.140625,
      "learning_rate": 0.0019729810961848166,
      "loss": 3.9185,
      "step": 797
    },
    {
      "epoch": 0.10213746320235505,
      "grad_norm": 0.83203125,
      "learning_rate": 0.001972885295241672,
      "loss": 3.5473,
      "step": 798
    },
    {
      "epoch": 0.1022654550108793,
      "grad_norm": 0.66015625,
      "learning_rate": 0.001972789327092404,
      "loss": 4.071,
      "step": 799
    },
    {
      "epoch": 0.10239344681940356,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0019726931917535074,
      "loss": 3.8979,
      "step": 800
    },
    {
      "epoch": 0.10252143862792781,
      "grad_norm": 0.8046875,
      "learning_rate": 0.001972596889241504,
      "loss": 4.2525,
      "step": 801
    },
    {
      "epoch": 0.10264943043645207,
      "grad_norm": 0.7265625,
      "learning_rate": 0.001972500419572945,
      "loss": 4.0423,
      "step": 802
    },
    {
      "epoch": 0.10277742224497632,
      "grad_norm": 1.9453125,
      "learning_rate": 0.0019724037827644103,
      "loss": 4.3696,
      "step": 803
    },
    {
      "epoch": 0.10290541405350058,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0019723069788325086,
      "loss": 3.9245,
      "step": 804
    },
    {
      "epoch": 0.10303340586202483,
      "grad_norm": 0.75,
      "learning_rate": 0.001972210007793877,
      "loss": 3.7018,
      "step": 805
    },
    {
      "epoch": 0.10316139767054909,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0019721128696651817,
      "loss": 4.2149,
      "step": 806
    },
    {
      "epoch": 0.10328938947907335,
      "grad_norm": 0.796875,
      "learning_rate": 0.0019720155644631174,
      "loss": 4.0404,
      "step": 807
    },
    {
      "epoch": 0.10341738128759759,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0019719180922044078,
      "loss": 3.7965,
      "step": 808
    },
    {
      "epoch": 0.10354537309612184,
      "grad_norm": 0.859375,
      "learning_rate": 0.001971820452905805,
      "loss": 4.0452,
      "step": 809
    },
    {
      "epoch": 0.1036733649046461,
      "grad_norm": 1.71875,
      "learning_rate": 0.0019717226465840896,
      "loss": 4.3392,
      "step": 810
    },
    {
      "epoch": 0.10380135671317035,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0019716246732560715,
      "loss": 4.3587,
      "step": 811
    },
    {
      "epoch": 0.10392934852169461,
      "grad_norm": 0.8125,
      "learning_rate": 0.0019715265329385886,
      "loss": 3.2758,
      "step": 812
    },
    {
      "epoch": 0.10405734033021886,
      "grad_norm": 0.8515625,
      "learning_rate": 0.001971428225648509,
      "loss": 3.3008,
      "step": 813
    },
    {
      "epoch": 0.10418533213874312,
      "grad_norm": 0.76171875,
      "learning_rate": 0.001971329751402727,
      "loss": 4.4381,
      "step": 814
    },
    {
      "epoch": 0.10431332394726738,
      "grad_norm": 0.8671875,
      "learning_rate": 0.001971231110218168,
      "loss": 4.4996,
      "step": 815
    },
    {
      "epoch": 0.10444131575579163,
      "grad_norm": 0.765625,
      "learning_rate": 0.0019711323021117844,
      "loss": 3.783,
      "step": 816
    },
    {
      "epoch": 0.10456930756431589,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0019710333271005585,
      "loss": 4.3179,
      "step": 817
    },
    {
      "epoch": 0.10469729937284014,
      "grad_norm": 1.046875,
      "learning_rate": 0.0019709341852015005,
      "loss": 4.5188,
      "step": 818
    },
    {
      "epoch": 0.1048252911813644,
      "grad_norm": 0.9375,
      "learning_rate": 0.0019708348764316493,
      "loss": 4.0402,
      "step": 819
    },
    {
      "epoch": 0.10495328298988865,
      "grad_norm": 0.890625,
      "learning_rate": 0.0019707354008080733,
      "loss": 4.2469,
      "step": 820
    },
    {
      "epoch": 0.10508127479841291,
      "grad_norm": 0.9921875,
      "learning_rate": 0.001970635758347869,
      "loss": 4.6693,
      "step": 821
    },
    {
      "epoch": 0.10520926660693716,
      "grad_norm": 0.92578125,
      "learning_rate": 0.001970535949068161,
      "loss": 3.86,
      "step": 822
    },
    {
      "epoch": 0.1053372584154614,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0019704359729861036,
      "loss": 3.6659,
      "step": 823
    },
    {
      "epoch": 0.10546525022398566,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0019703358301188787,
      "loss": 3.8871,
      "step": 824
    },
    {
      "epoch": 0.10559324203250992,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0019702355204836985,
      "loss": 3.6283,
      "step": 825
    },
    {
      "epoch": 0.10572123384103417,
      "grad_norm": 0.97265625,
      "learning_rate": 0.001970135044097802,
      "loss": 4.9221,
      "step": 826
    },
    {
      "epoch": 0.10584922564955843,
      "grad_norm": 1.109375,
      "learning_rate": 0.0019700344009784578,
      "loss": 4.3436,
      "step": 827
    },
    {
      "epoch": 0.10597721745808268,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0019699335911429632,
      "loss": 3.1992,
      "step": 828
    },
    {
      "epoch": 0.10610520926660694,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0019698326146086443,
      "loss": 4.189,
      "step": 829
    },
    {
      "epoch": 0.10623320107513119,
      "grad_norm": 1.15625,
      "learning_rate": 0.0019697314713928555,
      "loss": 4.6437,
      "step": 830
    },
    {
      "epoch": 0.10636119288365545,
      "grad_norm": 0.8359375,
      "learning_rate": 0.001969630161512979,
      "loss": 3.8279,
      "step": 831
    },
    {
      "epoch": 0.1064891846921797,
      "grad_norm": 0.84765625,
      "learning_rate": 0.001969528684986428,
      "loss": 4.116,
      "step": 832
    },
    {
      "epoch": 0.10661717650070396,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0019694270418306415,
      "loss": 3.7031,
      "step": 833
    },
    {
      "epoch": 0.10674516830922821,
      "grad_norm": 0.75,
      "learning_rate": 0.0019693252320630893,
      "loss": 4.4006,
      "step": 834
    },
    {
      "epoch": 0.10687316011775247,
      "grad_norm": 0.75,
      "learning_rate": 0.001969223255701269,
      "loss": 4.0007,
      "step": 835
    },
    {
      "epoch": 0.10700115192627672,
      "grad_norm": 0.65234375,
      "learning_rate": 0.001969121112762707,
      "loss": 4.3481,
      "step": 836
    },
    {
      "epoch": 0.10712914373480097,
      "grad_norm": 0.79296875,
      "learning_rate": 0.001969018803264958,
      "loss": 4.1664,
      "step": 837
    },
    {
      "epoch": 0.10725713554332522,
      "grad_norm": 0.77734375,
      "learning_rate": 0.001968916327225605,
      "loss": 4.546,
      "step": 838
    },
    {
      "epoch": 0.10738512735184948,
      "grad_norm": 0.75,
      "learning_rate": 0.001968813684662261,
      "loss": 4.1408,
      "step": 839
    },
    {
      "epoch": 0.10751311916037373,
      "grad_norm": 0.671875,
      "learning_rate": 0.0019687108755925666,
      "loss": 3.9003,
      "step": 840
    },
    {
      "epoch": 0.10764111096889799,
      "grad_norm": 0.96875,
      "learning_rate": 0.001968607900034191,
      "loss": 3.8362,
      "step": 841
    },
    {
      "epoch": 0.10776910277742224,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0019685047580048328,
      "loss": 3.9856,
      "step": 842
    },
    {
      "epoch": 0.1078970945859465,
      "grad_norm": 0.71484375,
      "learning_rate": 0.001968401449522218,
      "loss": 4.0759,
      "step": 843
    },
    {
      "epoch": 0.10802508639447075,
      "grad_norm": 0.890625,
      "learning_rate": 0.0019682979746041018,
      "loss": 3.3251,
      "step": 844
    },
    {
      "epoch": 0.10815307820299501,
      "grad_norm": 1.0,
      "learning_rate": 0.001968194333268268,
      "loss": 3.3372,
      "step": 845
    },
    {
      "epoch": 0.10828107001151926,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0019680905255325303,
      "loss": 3.4541,
      "step": 846
    },
    {
      "epoch": 0.10840906182004352,
      "grad_norm": 0.66015625,
      "learning_rate": 0.001967986551414728,
      "loss": 3.8992,
      "step": 847
    },
    {
      "epoch": 0.10853705362856778,
      "grad_norm": 0.7421875,
      "learning_rate": 0.001967882410932731,
      "loss": 3.5374,
      "step": 848
    },
    {
      "epoch": 0.10866504543709203,
      "grad_norm": 0.72265625,
      "learning_rate": 0.001967778104104439,
      "loss": 3.0129,
      "step": 849
    },
    {
      "epoch": 0.10879303724561629,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0019676736309477776,
      "loss": 4.4006,
      "step": 850
    },
    {
      "epoch": 0.10892102905414054,
      "grad_norm": 1.125,
      "learning_rate": 0.001967568991480702,
      "loss": 3.6127,
      "step": 851
    },
    {
      "epoch": 0.10904902086266478,
      "grad_norm": 0.89453125,
      "learning_rate": 0.001967464185721197,
      "loss": 4.4741,
      "step": 852
    },
    {
      "epoch": 0.10917701267118904,
      "grad_norm": 0.671875,
      "learning_rate": 0.001967359213687275,
      "loss": 3.7461,
      "step": 853
    },
    {
      "epoch": 0.1093050044797133,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0019672540753969767,
      "loss": 3.8386,
      "step": 854
    },
    {
      "epoch": 0.10943299628823755,
      "grad_norm": 0.953125,
      "learning_rate": 0.0019671487708683723,
      "loss": 4.2028,
      "step": 855
    },
    {
      "epoch": 0.1095609880967618,
      "grad_norm": 0.796875,
      "learning_rate": 0.0019670433001195593,
      "loss": 2.9854,
      "step": 856
    },
    {
      "epoch": 0.10968897990528606,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0019669376631686656,
      "loss": 3.6348,
      "step": 857
    },
    {
      "epoch": 0.10981697171381032,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0019668318600338463,
      "loss": 3.4518,
      "step": 858
    },
    {
      "epoch": 0.10994496352233457,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0019667258907332848,
      "loss": 3.0528,
      "step": 859
    },
    {
      "epoch": 0.11007295533085883,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0019666197552851944,
      "loss": 4.175,
      "step": 860
    },
    {
      "epoch": 0.11020094713938308,
      "grad_norm": 1.046875,
      "learning_rate": 0.0019665134537078157,
      "loss": 3.7107,
      "step": 861
    },
    {
      "epoch": 0.11032893894790734,
      "grad_norm": 0.99609375,
      "learning_rate": 0.001966406986019419,
      "loss": 3.5773,
      "step": 862
    },
    {
      "epoch": 0.11045693075643159,
      "grad_norm": 1.0625,
      "learning_rate": 0.0019663003522383013,
      "loss": 3.7372,
      "step": 863
    },
    {
      "epoch": 0.11058492256495585,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0019661935523827903,
      "loss": 4.4449,
      "step": 864
    },
    {
      "epoch": 0.1107129143734801,
      "grad_norm": 0.98046875,
      "learning_rate": 0.001966086586471241,
      "loss": 3.0235,
      "step": 865
    },
    {
      "epoch": 0.11084090618200435,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0019659794545220377,
      "loss": 3.2442,
      "step": 866
    },
    {
      "epoch": 0.1109688979905286,
      "grad_norm": 0.6484375,
      "learning_rate": 0.001965872156553592,
      "loss": 3.7961,
      "step": 867
    },
    {
      "epoch": 0.11109688979905286,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0019657646925843454,
      "loss": 4.2102,
      "step": 868
    },
    {
      "epoch": 0.11122488160757711,
      "grad_norm": 0.76953125,
      "learning_rate": 0.001965657062632767,
      "loss": 4.6731,
      "step": 869
    },
    {
      "epoch": 0.11135287341610137,
      "grad_norm": 0.703125,
      "learning_rate": 0.001965549266717355,
      "loss": 3.9743,
      "step": 870
    },
    {
      "epoch": 0.11148086522462562,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0019654413048566352,
      "loss": 3.733,
      "step": 871
    },
    {
      "epoch": 0.11160885703314988,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0019653331770691637,
      "loss": 4.0487,
      "step": 872
    },
    {
      "epoch": 0.11173684884167413,
      "grad_norm": 0.734375,
      "learning_rate": 0.001965224883373523,
      "loss": 4.494,
      "step": 873
    },
    {
      "epoch": 0.11186484065019839,
      "grad_norm": 0.7578125,
      "learning_rate": 0.001965116423788326,
      "loss": 3.341,
      "step": 874
    },
    {
      "epoch": 0.11199283245872264,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0019650077983322124,
      "loss": 4.061,
      "step": 875
    },
    {
      "epoch": 0.1121208242672469,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0019648990070238515,
      "loss": 3.4607,
      "step": 876
    },
    {
      "epoch": 0.11224881607577115,
      "grad_norm": 0.9921875,
      "learning_rate": 0.001964790049881941,
      "loss": 3.4764,
      "step": 877
    },
    {
      "epoch": 0.11237680788429541,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0019646809269252075,
      "loss": 3.6304,
      "step": 878
    },
    {
      "epoch": 0.11250479969281967,
      "grad_norm": 0.984375,
      "learning_rate": 0.0019645716381724047,
      "loss": 4.4826,
      "step": 879
    },
    {
      "epoch": 0.11263279150134392,
      "grad_norm": 1.0546875,
      "learning_rate": 0.001964462183642316,
      "loss": 4.3733,
      "step": 880
    },
    {
      "epoch": 0.11276078330986816,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0019643525633537525,
      "loss": 3.9349,
      "step": 881
    },
    {
      "epoch": 0.11288877511839242,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0019642427773255548,
      "loss": 3.2893,
      "step": 882
    },
    {
      "epoch": 0.11301676692691667,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0019641328255765914,
      "loss": 3.0339,
      "step": 883
    },
    {
      "epoch": 0.11314475873544093,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0019640227081257587,
      "loss": 3.6765,
      "step": 884
    },
    {
      "epoch": 0.11327275054396518,
      "grad_norm": 0.79296875,
      "learning_rate": 0.001963912424991983,
      "loss": 3.2677,
      "step": 885
    },
    {
      "epoch": 0.11340074235248944,
      "grad_norm": 1.46875,
      "learning_rate": 0.0019638019761942177,
      "loss": 2.7998,
      "step": 886
    },
    {
      "epoch": 0.1135287341610137,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0019636913617514453,
      "loss": 4.5996,
      "step": 887
    },
    {
      "epoch": 0.11365672596953795,
      "grad_norm": 0.6875,
      "learning_rate": 0.0019635805816826768,
      "loss": 4.1142,
      "step": 888
    },
    {
      "epoch": 0.1137847177780622,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0019634696360069513,
      "loss": 3.8396,
      "step": 889
    },
    {
      "epoch": 0.11391270958658646,
      "grad_norm": 0.82421875,
      "learning_rate": 0.001963358524743337,
      "loss": 3.8086,
      "step": 890
    },
    {
      "epoch": 0.11404070139511072,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00196324724791093,
      "loss": 3.4109,
      "step": 891
    },
    {
      "epoch": 0.11416869320363497,
      "grad_norm": 0.75,
      "learning_rate": 0.0019631358055288545,
      "loss": 3.7986,
      "step": 892
    },
    {
      "epoch": 0.11429668501215923,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0019630241976162648,
      "loss": 3.6985,
      "step": 893
    },
    {
      "epoch": 0.11442467682068348,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0019629124241923417,
      "loss": 3.6048,
      "step": 894
    },
    {
      "epoch": 0.11455266862920772,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0019628004852762955,
      "loss": 3.5543,
      "step": 895
    },
    {
      "epoch": 0.11468066043773198,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0019626883808873644,
      "loss": 4.0461,
      "step": 896
    },
    {
      "epoch": 0.11480865224625623,
      "grad_norm": 0.65625,
      "learning_rate": 0.001962576111044816,
      "loss": 2.938,
      "step": 897
    },
    {
      "epoch": 0.11493664405478049,
      "grad_norm": 0.84375,
      "learning_rate": 0.0019624636757679453,
      "loss": 2.5969,
      "step": 898
    },
    {
      "epoch": 0.11506463586330475,
      "grad_norm": 0.75390625,
      "learning_rate": 0.001962351075076076,
      "loss": 5.0781,
      "step": 899
    },
    {
      "epoch": 0.115192627671829,
      "grad_norm": 1.28125,
      "learning_rate": 0.0019622383089885607,
      "loss": 3.4843,
      "step": 900
    },
    {
      "epoch": 0.11532061948035326,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00196212537752478,
      "loss": 3.7312,
      "step": 901
    },
    {
      "epoch": 0.11544861128887751,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0019620122807041426,
      "loss": 3.6822,
      "step": 902
    },
    {
      "epoch": 0.11557660309740177,
      "grad_norm": 0.78515625,
      "learning_rate": 0.001961899018546086,
      "loss": 3.4886,
      "step": 903
    },
    {
      "epoch": 0.11570459490592602,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0019617855910700776,
      "loss": 3.3682,
      "step": 904
    },
    {
      "epoch": 0.11583258671445028,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0019616719982956097,
      "loss": 3.3302,
      "step": 905
    },
    {
      "epoch": 0.11596057852297453,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0019615582402422067,
      "loss": 3.7622,
      "step": 906
    },
    {
      "epoch": 0.11608857033149879,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0019614443169294186,
      "loss": 3.6165,
      "step": 907
    },
    {
      "epoch": 0.11621656214002304,
      "grad_norm": 0.73828125,
      "learning_rate": 0.001961330228376825,
      "loss": 3.3886,
      "step": 908
    },
    {
      "epoch": 0.1163445539485473,
      "grad_norm": 0.6484375,
      "learning_rate": 0.001961215974604035,
      "loss": 3.6523,
      "step": 909
    },
    {
      "epoch": 0.11647254575707154,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0019611015556306845,
      "loss": 3.3891,
      "step": 910
    },
    {
      "epoch": 0.1166005375655958,
      "grad_norm": 0.796875,
      "learning_rate": 0.0019609869714764376,
      "loss": 3.6532,
      "step": 911
    },
    {
      "epoch": 0.11672852937412005,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0019608722221609877,
      "loss": 3.5814,
      "step": 912
    },
    {
      "epoch": 0.11685652118264431,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0019607573077040567,
      "loss": 3.7993,
      "step": 913
    },
    {
      "epoch": 0.11698451299116856,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0019606422281253943,
      "loss": 3.587,
      "step": 914
    },
    {
      "epoch": 0.11711250479969282,
      "grad_norm": 0.77734375,
      "learning_rate": 0.001960526983444779,
      "loss": 4.2087,
      "step": 915
    },
    {
      "epoch": 0.11724049660821707,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0019604115736820173,
      "loss": 3.4123,
      "step": 916
    },
    {
      "epoch": 0.11736848841674133,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0019602959988569437,
      "loss": 3.3716,
      "step": 917
    },
    {
      "epoch": 0.11749648022526558,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0019601802589894225,
      "loss": 3.3243,
      "step": 918
    },
    {
      "epoch": 0.11762447203378984,
      "grad_norm": 1.046875,
      "learning_rate": 0.001960064354099345,
      "loss": 3.7429,
      "step": 919
    },
    {
      "epoch": 0.1177524638423141,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0019599482842066313,
      "loss": 3.3318,
      "step": 920
    },
    {
      "epoch": 0.11788045565083835,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0019598320493312304,
      "loss": 4.0289,
      "step": 921
    },
    {
      "epoch": 0.1180084474593626,
      "grad_norm": 0.84375,
      "learning_rate": 0.0019597156494931185,
      "loss": 3.6527,
      "step": 922
    },
    {
      "epoch": 0.11813643926788686,
      "grad_norm": 1.0078125,
      "learning_rate": 0.001959599084712301,
      "loss": 3.641,
      "step": 923
    },
    {
      "epoch": 0.1182644310764111,
      "grad_norm": 0.81640625,
      "learning_rate": 0.001959482355008812,
      "loss": 3.377,
      "step": 924
    },
    {
      "epoch": 0.11839242288493536,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0019593654604027127,
      "loss": 4.121,
      "step": 925
    },
    {
      "epoch": 0.11852041469345961,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0019592484009140934,
      "loss": 4.4571,
      "step": 926
    },
    {
      "epoch": 0.11864840650198387,
      "grad_norm": 0.734375,
      "learning_rate": 0.001959131176563073,
      "loss": 3.3217,
      "step": 927
    },
    {
      "epoch": 0.11877639831050812,
      "grad_norm": 0.66015625,
      "learning_rate": 0.001959013787369798,
      "loss": 4.0584,
      "step": 928
    },
    {
      "epoch": 0.11890439011903238,
      "grad_norm": 0.625,
      "learning_rate": 0.001958896233354444,
      "loss": 3.2632,
      "step": 929
    },
    {
      "epoch": 0.11903238192755664,
      "grad_norm": 0.734375,
      "learning_rate": 0.0019587785145372145,
      "loss": 3.5658,
      "step": 930
    },
    {
      "epoch": 0.11916037373608089,
      "grad_norm": 0.90625,
      "learning_rate": 0.001958660630938341,
      "loss": 3.6855,
      "step": 931
    },
    {
      "epoch": 0.11928836554460515,
      "grad_norm": 0.765625,
      "learning_rate": 0.001958542582578085,
      "loss": 3.7663,
      "step": 932
    },
    {
      "epoch": 0.1194163573531294,
      "grad_norm": 1.09375,
      "learning_rate": 0.0019584243694767333,
      "loss": 3.7094,
      "step": 933
    },
    {
      "epoch": 0.11954434916165366,
      "grad_norm": 1.234375,
      "learning_rate": 0.0019583059916546035,
      "loss": 3.7818,
      "step": 934
    },
    {
      "epoch": 0.11967234097017791,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0019581874491320407,
      "loss": 3.2882,
      "step": 935
    },
    {
      "epoch": 0.11980033277870217,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0019580687419294184,
      "loss": 3.2619,
      "step": 936
    },
    {
      "epoch": 0.11992832458722642,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0019579498700671386,
      "loss": 4.2336,
      "step": 937
    },
    {
      "epoch": 0.12005631639575068,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0019578308335656313,
      "loss": 3.2063,
      "step": 938
    },
    {
      "epoch": 0.12018430820427492,
      "grad_norm": 0.7890625,
      "learning_rate": 0.001957711632445354,
      "loss": 3.2424,
      "step": 939
    },
    {
      "epoch": 0.12031230001279918,
      "grad_norm": 0.671875,
      "learning_rate": 0.0019575922667267945,
      "loss": 3.0782,
      "step": 940
    },
    {
      "epoch": 0.12044029182132343,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0019574727364304674,
      "loss": 3.9098,
      "step": 941
    },
    {
      "epoch": 0.12056828362984769,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0019573530415769157,
      "loss": 4.1458,
      "step": 942
    },
    {
      "epoch": 0.12069627543837194,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0019572331821867107,
      "loss": 3.5935,
      "step": 943
    },
    {
      "epoch": 0.1208242672468962,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0019571131582804525,
      "loss": 3.132,
      "step": 944
    },
    {
      "epoch": 0.12095225905542045,
      "grad_norm": 0.75,
      "learning_rate": 0.0019569929698787696,
      "loss": 4.0456,
      "step": 945
    },
    {
      "epoch": 0.12108025086394471,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0019568726170023174,
      "loss": 4.2372,
      "step": 946
    },
    {
      "epoch": 0.12120824267246896,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0019567520996717808,
      "loss": 4.0716,
      "step": 947
    },
    {
      "epoch": 0.12133623448099322,
      "grad_norm": 0.87109375,
      "learning_rate": 0.001956631417907873,
      "loss": 3.842,
      "step": 948
    },
    {
      "epoch": 0.12146422628951747,
      "grad_norm": 0.85546875,
      "learning_rate": 0.001956510571731335,
      "loss": 3.4917,
      "step": 949
    },
    {
      "epoch": 0.12159221809804173,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0019563895611629358,
      "loss": 3.5495,
      "step": 950
    },
    {
      "epoch": 0.12172020990656598,
      "grad_norm": 0.75390625,
      "learning_rate": 0.001956268386223474,
      "loss": 3.7297,
      "step": 951
    },
    {
      "epoch": 0.12184820171509024,
      "grad_norm": 0.921875,
      "learning_rate": 0.0019561470469337742,
      "loss": 3.8693,
      "step": 952
    },
    {
      "epoch": 0.12197619352361448,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0019560255433146915,
      "loss": 3.9551,
      "step": 953
    },
    {
      "epoch": 0.12210418533213874,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0019559038753871076,
      "loss": 4.2624,
      "step": 954
    },
    {
      "epoch": 0.12223217714066299,
      "grad_norm": 0.77734375,
      "learning_rate": 0.001955782043171933,
      "loss": 3.4484,
      "step": 955
    },
    {
      "epoch": 0.12236016894918725,
      "grad_norm": 0.65625,
      "learning_rate": 0.0019556600466901076,
      "loss": 3.3516,
      "step": 956
    },
    {
      "epoch": 0.1224881607577115,
      "grad_norm": 0.703125,
      "learning_rate": 0.001955537885962598,
      "loss": 4.0045,
      "step": 957
    },
    {
      "epoch": 0.12261615256623576,
      "grad_norm": 0.72265625,
      "learning_rate": 0.001955415561010399,
      "loss": 3.3169,
      "step": 958
    },
    {
      "epoch": 0.12274414437476001,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0019552930718545344,
      "loss": 3.5144,
      "step": 959
    },
    {
      "epoch": 0.12287213618328427,
      "grad_norm": 0.88671875,
      "learning_rate": 0.001955170418516056,
      "loss": 4.4434,
      "step": 960
    },
    {
      "epoch": 0.12300012799180852,
      "grad_norm": 0.8046875,
      "learning_rate": 0.001955047601016044,
      "loss": 3.717,
      "step": 961
    },
    {
      "epoch": 0.12312811980033278,
      "grad_norm": 0.6875,
      "learning_rate": 0.001954924619375606,
      "loss": 3.8565,
      "step": 962
    },
    {
      "epoch": 0.12325611160885704,
      "grad_norm": 1.0,
      "learning_rate": 0.0019548014736158787,
      "loss": 3.5405,
      "step": 963
    },
    {
      "epoch": 0.12338410341738129,
      "grad_norm": 0.640625,
      "learning_rate": 0.001954678163758027,
      "loss": 3.2001,
      "step": 964
    },
    {
      "epoch": 0.12351209522590555,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0019545546898232435,
      "loss": 3.7179,
      "step": 965
    },
    {
      "epoch": 0.1236400870344298,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0019544310518327494,
      "loss": 2.9281,
      "step": 966
    },
    {
      "epoch": 0.12376807884295406,
      "grad_norm": 0.7890625,
      "learning_rate": 0.001954307249807793,
      "loss": 3.5028,
      "step": 967
    },
    {
      "epoch": 0.1238960706514783,
      "grad_norm": 0.75,
      "learning_rate": 0.0019541832837696527,
      "loss": 2.6589,
      "step": 968
    },
    {
      "epoch": 0.12402406246000255,
      "grad_norm": 0.8125,
      "learning_rate": 0.0019540591537396336,
      "loss": 3.7952,
      "step": 969
    },
    {
      "epoch": 0.12415205426852681,
      "grad_norm": 0.71484375,
      "learning_rate": 0.001953934859739069,
      "loss": 3.2427,
      "step": 970
    },
    {
      "epoch": 0.12428004607705107,
      "grad_norm": 0.86328125,
      "learning_rate": 0.001953810401789322,
      "loss": 3.7648,
      "step": 971
    },
    {
      "epoch": 0.12440803788557532,
      "grad_norm": 0.83203125,
      "learning_rate": 0.001953685779911782,
      "loss": 3.4908,
      "step": 972
    },
    {
      "epoch": 0.12453602969409958,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0019535609941278677,
      "loss": 3.3113,
      "step": 973
    },
    {
      "epoch": 0.12466402150262383,
      "grad_norm": 0.83203125,
      "learning_rate": 0.001953436044459025,
      "loss": 4.4434,
      "step": 974
    },
    {
      "epoch": 0.12479201331114809,
      "grad_norm": 0.8125,
      "learning_rate": 0.0019533109309267287,
      "loss": 3.4783,
      "step": 975
    },
    {
      "epoch": 0.12492000511967234,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0019531856535524817,
      "loss": 2.8561,
      "step": 976
    },
    {
      "epoch": 0.1250479969281966,
      "grad_norm": 1.46875,
      "learning_rate": 0.0019530602123578149,
      "loss": 3.0004,
      "step": 977
    },
    {
      "epoch": 0.12517598873672084,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0019529346073642877,
      "loss": 3.2929,
      "step": 978
    },
    {
      "epoch": 0.1253039805452451,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0019528088385934869,
      "loss": 3.2741,
      "step": 979
    },
    {
      "epoch": 0.12543197235376935,
      "grad_norm": 0.76171875,
      "learning_rate": 0.001952682906067028,
      "loss": 3.7331,
      "step": 980
    },
    {
      "epoch": 0.12555996416229362,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0019525568098065548,
      "loss": 3.3573,
      "step": 981
    },
    {
      "epoch": 0.12568795597081786,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0019524305498337387,
      "loss": 2.9779,
      "step": 982
    },
    {
      "epoch": 0.12581594777934213,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00195230412617028,
      "loss": 3.574,
      "step": 983
    },
    {
      "epoch": 0.12594393958786637,
      "grad_norm": 0.765625,
      "learning_rate": 0.0019521775388379058,
      "loss": 3.7596,
      "step": 984
    },
    {
      "epoch": 0.12607193139639064,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0019520507878583728,
      "loss": 3.3743,
      "step": 985
    },
    {
      "epoch": 0.12619992320491488,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0019519238732534654,
      "loss": 3.1035,
      "step": 986
    },
    {
      "epoch": 0.12632791501343915,
      "grad_norm": 0.75,
      "learning_rate": 0.0019517967950449955,
      "loss": 3.5453,
      "step": 987
    },
    {
      "epoch": 0.1264559068219634,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0019516695532548036,
      "loss": 3.414,
      "step": 988
    },
    {
      "epoch": 0.12658389863048766,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0019515421479047587,
      "loss": 3.3166,
      "step": 989
    },
    {
      "epoch": 0.1267118904390119,
      "grad_norm": 0.875,
      "learning_rate": 0.0019514145790167565,
      "loss": 3.2411,
      "step": 990
    },
    {
      "epoch": 0.12683988224753615,
      "grad_norm": 0.890625,
      "learning_rate": 0.001951286846612723,
      "loss": 3.7565,
      "step": 991
    },
    {
      "epoch": 0.12696787405606041,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0019511589507146103,
      "loss": 3.4705,
      "step": 992
    },
    {
      "epoch": 0.12709586586458466,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0019510308913443996,
      "loss": 4.0593,
      "step": 993
    },
    {
      "epoch": 0.12722385767310893,
      "grad_norm": 1.46875,
      "learning_rate": 0.0019509026685241,
      "loss": 3.4814,
      "step": 994
    },
    {
      "epoch": 0.12735184948163317,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0019507742822757485,
      "loss": 3.6799,
      "step": 995
    },
    {
      "epoch": 0.12747984129015744,
      "grad_norm": 0.6875,
      "learning_rate": 0.0019506457326214108,
      "loss": 2.9499,
      "step": 996
    },
    {
      "epoch": 0.12760783309868168,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00195051701958318,
      "loss": 3.3708,
      "step": 997
    },
    {
      "epoch": 0.12773582490720595,
      "grad_norm": 0.859375,
      "learning_rate": 0.001950388143183177,
      "loss": 3.4441,
      "step": 998
    },
    {
      "epoch": 0.1278638167157302,
      "grad_norm": 0.78125,
      "learning_rate": 0.0019502591034435523,
      "loss": 3.6524,
      "step": 999
    },
    {
      "epoch": 0.12799180852425446,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0019501299003864827,
      "loss": 2.4237,
      "step": 1000
    },
    {
      "epoch": 0.1281198003327787,
      "grad_norm": 0.75390625,
      "learning_rate": 0.001950000534034174,
      "loss": 3.9275,
      "step": 1001
    },
    {
      "epoch": 0.12824779214130297,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0019498710044088604,
      "loss": 3.0411,
      "step": 1002
    },
    {
      "epoch": 0.1283757839498272,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0019497413115328027,
      "loss": 3.4542,
      "step": 1003
    },
    {
      "epoch": 0.12850377575835145,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0019496114554282916,
      "loss": 3.3073,
      "step": 1004
    },
    {
      "epoch": 0.12863176756687572,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0019494814361176447,
      "loss": 3.9253,
      "step": 1005
    },
    {
      "epoch": 0.12875975937539996,
      "grad_norm": 0.890625,
      "learning_rate": 0.001949351253623208,
      "loss": 3.639,
      "step": 1006
    },
    {
      "epoch": 0.12888775118392423,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0019492209079673552,
      "loss": 2.9475,
      "step": 1007
    },
    {
      "epoch": 0.12901574299244847,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0019490903991724881,
      "loss": 3.7006,
      "step": 1008
    },
    {
      "epoch": 0.12914373480097274,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0019489597272610375,
      "loss": 3.2428,
      "step": 1009
    },
    {
      "epoch": 0.12927172660949698,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0019488288922554612,
      "loss": 3.2863,
      "step": 1010
    },
    {
      "epoch": 0.12939971841802125,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0019486978941782454,
      "loss": 3.351,
      "step": 1011
    },
    {
      "epoch": 0.1295277102265455,
      "grad_norm": 1.015625,
      "learning_rate": 0.0019485667330519038,
      "loss": 3.6498,
      "step": 1012
    },
    {
      "epoch": 0.12965570203506976,
      "grad_norm": 0.796875,
      "learning_rate": 0.001948435408898979,
      "loss": 3.0958,
      "step": 1013
    },
    {
      "epoch": 0.129783693843594,
      "grad_norm": 0.734375,
      "learning_rate": 0.001948303921742041,
      "loss": 3.1954,
      "step": 1014
    },
    {
      "epoch": 0.12991168565211828,
      "grad_norm": 0.875,
      "learning_rate": 0.001948172271603688,
      "loss": 3.3837,
      "step": 1015
    },
    {
      "epoch": 0.13003967746064252,
      "grad_norm": 0.734375,
      "learning_rate": 0.0019480404585065464,
      "loss": 3.9452,
      "step": 1016
    },
    {
      "epoch": 0.13016766926916679,
      "grad_norm": 0.71875,
      "learning_rate": 0.00194790848247327,
      "loss": 2.9277,
      "step": 1017
    },
    {
      "epoch": 0.13029566107769103,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0019477763435265417,
      "loss": 3.8614,
      "step": 1018
    },
    {
      "epoch": 0.13042365288621527,
      "grad_norm": 0.64453125,
      "learning_rate": 0.001947644041689071,
      "loss": 2.7134,
      "step": 1019
    },
    {
      "epoch": 0.13055164469473954,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0019475115769835964,
      "loss": 3.3132,
      "step": 1020
    },
    {
      "epoch": 0.13067963650326378,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0019473789494328844,
      "loss": 3.5545,
      "step": 1021
    },
    {
      "epoch": 0.13080762831178805,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0019472461590597288,
      "loss": 4.4281,
      "step": 1022
    },
    {
      "epoch": 0.1309356201203123,
      "grad_norm": 0.76171875,
      "learning_rate": 0.001947113205886952,
      "loss": 3.2657,
      "step": 1023
    },
    {
      "epoch": 0.13106361192883656,
      "grad_norm": 0.7421875,
      "learning_rate": 0.001946980089937404,
      "loss": 3.9357,
      "step": 1024
    },
    {
      "epoch": 0.1311916037373608,
      "grad_norm": 0.83203125,
      "learning_rate": 0.001946846811233963,
      "loss": 3.8255,
      "step": 1025
    },
    {
      "epoch": 0.13131959554588507,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0019467133697995353,
      "loss": 3.7634,
      "step": 1026
    },
    {
      "epoch": 0.1314475873544093,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0019465797656570546,
      "loss": 3.7425,
      "step": 1027
    },
    {
      "epoch": 0.13157557916293358,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0019464459988294833,
      "loss": 4.4747,
      "step": 1028
    },
    {
      "epoch": 0.13170357097145782,
      "grad_norm": 0.71875,
      "learning_rate": 0.0019463120693398112,
      "loss": 3.3728,
      "step": 1029
    },
    {
      "epoch": 0.1318315627799821,
      "grad_norm": 0.6640625,
      "learning_rate": 0.001946177977211056,
      "loss": 3.3651,
      "step": 1030
    },
    {
      "epoch": 0.13195955458850633,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0019460437224662643,
      "loss": 3.2182,
      "step": 1031
    },
    {
      "epoch": 0.1320875463970306,
      "grad_norm": 1.078125,
      "learning_rate": 0.0019459093051285095,
      "loss": 3.6344,
      "step": 1032
    },
    {
      "epoch": 0.13221553820555484,
      "grad_norm": 0.765625,
      "learning_rate": 0.0019457747252208935,
      "loss": 3.5366,
      "step": 1033
    },
    {
      "epoch": 0.13234353001407909,
      "grad_norm": 0.828125,
      "learning_rate": 0.0019456399827665458,
      "loss": 2.5467,
      "step": 1034
    },
    {
      "epoch": 0.13247152182260336,
      "grad_norm": 0.734375,
      "learning_rate": 0.0019455050777886244,
      "loss": 3.0982,
      "step": 1035
    },
    {
      "epoch": 0.1325995136311276,
      "grad_norm": 0.80859375,
      "learning_rate": 0.001945370010310315,
      "loss": 3.695,
      "step": 1036
    },
    {
      "epoch": 0.13272750543965187,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0019452347803548308,
      "loss": 3.2746,
      "step": 1037
    },
    {
      "epoch": 0.1328554972481761,
      "grad_norm": 0.91796875,
      "learning_rate": 0.001945099387945413,
      "loss": 3.0667,
      "step": 1038
    },
    {
      "epoch": 0.13298348905670038,
      "grad_norm": 0.90234375,
      "learning_rate": 0.001944963833105332,
      "loss": 4.8267,
      "step": 1039
    },
    {
      "epoch": 0.13311148086522462,
      "grad_norm": 0.875,
      "learning_rate": 0.001944828115857884,
      "loss": 3.1858,
      "step": 1040
    },
    {
      "epoch": 0.1332394726737489,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0019446922362263948,
      "loss": 3.1982,
      "step": 1041
    },
    {
      "epoch": 0.13336746448227313,
      "grad_norm": 0.671875,
      "learning_rate": 0.0019445561942342173,
      "loss": 3.6458,
      "step": 1042
    },
    {
      "epoch": 0.1334954562907974,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0019444199899047328,
      "loss": 2.5773,
      "step": 1043
    },
    {
      "epoch": 0.13362344809932164,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00194428362326135,
      "loss": 3.7655,
      "step": 1044
    },
    {
      "epoch": 0.1337514399078459,
      "grad_norm": 0.81640625,
      "learning_rate": 0.001944147094327506,
      "loss": 3.6526,
      "step": 1045
    },
    {
      "epoch": 0.13387943171637015,
      "grad_norm": 0.6796875,
      "learning_rate": 0.001944010403126665,
      "loss": 2.7566,
      "step": 1046
    },
    {
      "epoch": 0.13400742352489442,
      "grad_norm": 0.828125,
      "learning_rate": 0.0019438735496823199,
      "loss": 3.2002,
      "step": 1047
    },
    {
      "epoch": 0.13413541533341866,
      "grad_norm": 1.0078125,
      "learning_rate": 0.001943736534017991,
      "loss": 4.015,
      "step": 1048
    },
    {
      "epoch": 0.1342634071419429,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0019435993561572273,
      "loss": 2.5212,
      "step": 1049
    },
    {
      "epoch": 0.13439139895046717,
      "grad_norm": 0.75,
      "learning_rate": 0.0019434620161236044,
      "loss": 3.9998,
      "step": 1050
    },
    {
      "epoch": 0.13451939075899141,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0019433245139407265,
      "loss": 3.5611,
      "step": 1051
    },
    {
      "epoch": 0.13464738256751568,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0019431868496322254,
      "loss": 2.9752,
      "step": 1052
    },
    {
      "epoch": 0.13477537437603992,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0019430490232217616,
      "loss": 3.282,
      "step": 1053
    },
    {
      "epoch": 0.1349033661845642,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0019429110347330223,
      "loss": 2.6788,
      "step": 1054
    },
    {
      "epoch": 0.13503135799308844,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0019427728841897231,
      "loss": 4.5743,
      "step": 1055
    },
    {
      "epoch": 0.1351593498016127,
      "grad_norm": 0.75,
      "learning_rate": 0.0019426345716156077,
      "loss": 2.6773,
      "step": 1056
    },
    {
      "epoch": 0.13528734161013695,
      "grad_norm": 0.98046875,
      "learning_rate": 0.001942496097034447,
      "loss": 3.6924,
      "step": 1057
    },
    {
      "epoch": 0.13541533341866122,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0019423574604700402,
      "loss": 4.4911,
      "step": 1058
    },
    {
      "epoch": 0.13554332522718546,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0019422186619462143,
      "loss": 3.733,
      "step": 1059
    },
    {
      "epoch": 0.13567131703570973,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0019420797014868243,
      "loss": 3.3893,
      "step": 1060
    },
    {
      "epoch": 0.13579930884423397,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0019419405791157526,
      "loss": 3.5231,
      "step": 1061
    },
    {
      "epoch": 0.13592730065275824,
      "grad_norm": 0.75,
      "learning_rate": 0.001941801294856909,
      "loss": 2.7523,
      "step": 1062
    },
    {
      "epoch": 0.13605529246128248,
      "grad_norm": 0.8125,
      "learning_rate": 0.001941661848734233,
      "loss": 2.8682,
      "step": 1063
    },
    {
      "epoch": 0.13618328426980672,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00194152224077169,
      "loss": 3.4219,
      "step": 1064
    },
    {
      "epoch": 0.136311276078331,
      "grad_norm": 1.015625,
      "learning_rate": 0.0019413824709932736,
      "loss": 3.1043,
      "step": 1065
    },
    {
      "epoch": 0.13643926788685523,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0019412425394230062,
      "loss": 2.6258,
      "step": 1066
    },
    {
      "epoch": 0.1365672596953795,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0019411024460849369,
      "loss": 3.6408,
      "step": 1067
    },
    {
      "epoch": 0.13669525150390374,
      "grad_norm": 0.7265625,
      "learning_rate": 0.001940962191003143,
      "loss": 3.8137,
      "step": 1068
    },
    {
      "epoch": 0.136823243312428,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0019408217742017298,
      "loss": 2.2635,
      "step": 1069
    },
    {
      "epoch": 0.13695123512095225,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0019406811957048298,
      "loss": 3.6414,
      "step": 1070
    },
    {
      "epoch": 0.13707922692947652,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0019405404555366045,
      "loss": 3.1282,
      "step": 1071
    },
    {
      "epoch": 0.13720721873800076,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0019403995537212416,
      "loss": 3.629,
      "step": 1072
    },
    {
      "epoch": 0.13733521054652503,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0019402584902829575,
      "loss": 3.1235,
      "step": 1073
    },
    {
      "epoch": 0.13746320235504927,
      "grad_norm": 0.75,
      "learning_rate": 0.0019401172652459967,
      "loss": 3.0351,
      "step": 1074
    },
    {
      "epoch": 0.13759119416357354,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0019399758786346304,
      "loss": 3.168,
      "step": 1075
    },
    {
      "epoch": 0.13771918597209779,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0019398343304731585,
      "loss": 2.8223,
      "step": 1076
    },
    {
      "epoch": 0.13784717778062203,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0019396926207859084,
      "loss": 3.45,
      "step": 1077
    },
    {
      "epoch": 0.1379751695891463,
      "grad_norm": 0.859375,
      "learning_rate": 0.0019395507495972351,
      "loss": 2.9948,
      "step": 1078
    },
    {
      "epoch": 0.13810316139767054,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0019394087169315216,
      "loss": 3.4169,
      "step": 1079
    },
    {
      "epoch": 0.1382311532061948,
      "grad_norm": 0.953125,
      "learning_rate": 0.0019392665228131784,
      "loss": 2.8572,
      "step": 1080
    },
    {
      "epoch": 0.13835914501471905,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0019391241672666439,
      "loss": 3.1166,
      "step": 1081
    },
    {
      "epoch": 0.13848713682324332,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0019389816503163839,
      "loss": 3.8743,
      "step": 1082
    },
    {
      "epoch": 0.13861512863176756,
      "grad_norm": 0.734375,
      "learning_rate": 0.001938838971986893,
      "loss": 3.135,
      "step": 1083
    },
    {
      "epoch": 0.13874312044029183,
      "grad_norm": 0.7578125,
      "learning_rate": 0.001938696132302692,
      "loss": 3.3525,
      "step": 1084
    },
    {
      "epoch": 0.13887111224881607,
      "grad_norm": 0.875,
      "learning_rate": 0.0019385531312883307,
      "loss": 2.9379,
      "step": 1085
    },
    {
      "epoch": 0.13899910405734034,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0019384099689683859,
      "loss": 2.9513,
      "step": 1086
    },
    {
      "epoch": 0.13912709586586458,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0019382666453674625,
      "loss": 2.7024,
      "step": 1087
    },
    {
      "epoch": 0.13925508767438885,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0019381231605101927,
      "loss": 2.989,
      "step": 1088
    },
    {
      "epoch": 0.1393830794829131,
      "grad_norm": 1.078125,
      "learning_rate": 0.0019379795144212374,
      "loss": 3.9851,
      "step": 1089
    },
    {
      "epoch": 0.13951107129143736,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0019378357071252838,
      "loss": 3.0601,
      "step": 1090
    },
    {
      "epoch": 0.1396390630999616,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0019376917386470477,
      "loss": 3.3714,
      "step": 1091
    },
    {
      "epoch": 0.13976705490848584,
      "grad_norm": 0.859375,
      "learning_rate": 0.0019375476090112726,
      "loss": 3.8742,
      "step": 1092
    },
    {
      "epoch": 0.1398950467170101,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0019374033182427296,
      "loss": 4.2344,
      "step": 1093
    },
    {
      "epoch": 0.14002303852553435,
      "grad_norm": 0.78125,
      "learning_rate": 0.001937258866366217,
      "loss": 3.7389,
      "step": 1094
    },
    {
      "epoch": 0.14015103033405862,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0019371142534065613,
      "loss": 3.0411,
      "step": 1095
    },
    {
      "epoch": 0.14027902214258287,
      "grad_norm": 0.78125,
      "learning_rate": 0.0019369694793886169,
      "loss": 3.3294,
      "step": 1096
    },
    {
      "epoch": 0.14040701395110713,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0019368245443372651,
      "loss": 2.6087,
      "step": 1097
    },
    {
      "epoch": 0.14053500575963138,
      "grad_norm": 0.9453125,
      "learning_rate": 0.001936679448277416,
      "loss": 2.9046,
      "step": 1098
    },
    {
      "epoch": 0.14066299756815565,
      "grad_norm": 0.5390625,
      "learning_rate": 0.001936534191234006,
      "loss": 2.6771,
      "step": 1099
    },
    {
      "epoch": 0.1407909893766799,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0019363887732320004,
      "loss": 3.3206,
      "step": 1100
    },
    {
      "epoch": 0.14091898118520416,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0019362431942963913,
      "loss": 2.7643,
      "step": 1101
    },
    {
      "epoch": 0.1410469729937284,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0019360974544521993,
      "loss": 3.333,
      "step": 1102
    },
    {
      "epoch": 0.14117496480225267,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0019359515537244714,
      "loss": 3.5429,
      "step": 1103
    },
    {
      "epoch": 0.1413029566107769,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0019358054921382833,
      "loss": 3.2524,
      "step": 1104
    },
    {
      "epoch": 0.14143094841930118,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0019356592697187384,
      "loss": 2.9627,
      "step": 1105
    },
    {
      "epoch": 0.14155894022782542,
      "grad_norm": 0.80078125,
      "learning_rate": 0.001935512886490967,
      "loss": 2.9596,
      "step": 1106
    },
    {
      "epoch": 0.14168693203634966,
      "grad_norm": 0.76171875,
      "learning_rate": 0.001935366342480128,
      "loss": 3.3662,
      "step": 1107
    },
    {
      "epoch": 0.14181492384487393,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0019352196377114064,
      "loss": 2.8613,
      "step": 1108
    },
    {
      "epoch": 0.14194291565339817,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0019350727722100168,
      "loss": 3.0555,
      "step": 1109
    },
    {
      "epoch": 0.14207090746192244,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0019349257460011997,
      "loss": 2.5011,
      "step": 1110
    },
    {
      "epoch": 0.14219889927044668,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0019347785591102243,
      "loss": 3.5901,
      "step": 1111
    },
    {
      "epoch": 0.14232689107897095,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0019346312115623868,
      "loss": 3.9179,
      "step": 1112
    },
    {
      "epoch": 0.1424548828874952,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0019344837033830116,
      "loss": 2.5353,
      "step": 1113
    },
    {
      "epoch": 0.14258287469601946,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0019343360345974504,
      "loss": 3.1144,
      "step": 1114
    },
    {
      "epoch": 0.1427108665045437,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0019341882052310818,
      "loss": 3.0103,
      "step": 1115
    },
    {
      "epoch": 0.14283885831306797,
      "grad_norm": 0.84375,
      "learning_rate": 0.0019340402153093135,
      "loss": 2.8897,
      "step": 1116
    },
    {
      "epoch": 0.14296685012159221,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0019338920648575796,
      "loss": 3.7469,
      "step": 1117
    },
    {
      "epoch": 0.14309484193011648,
      "grad_norm": 0.90625,
      "learning_rate": 0.001933743753901342,
      "loss": 3.8551,
      "step": 1118
    },
    {
      "epoch": 0.14322283373864073,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0019335952824660904,
      "loss": 3.0105,
      "step": 1119
    },
    {
      "epoch": 0.143350825547165,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0019334466505773426,
      "loss": 3.1833,
      "step": 1120
    },
    {
      "epoch": 0.14347881735568924,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0019332978582606427,
      "loss": 3.9847,
      "step": 1121
    },
    {
      "epoch": 0.14360680916421348,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0019331489055415632,
      "loss": 2.4844,
      "step": 1122
    },
    {
      "epoch": 0.14373480097273775,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0019329997924457042,
      "loss": 3.1257,
      "step": 1123
    },
    {
      "epoch": 0.143862792781262,
      "grad_norm": 0.703125,
      "learning_rate": 0.0019328505189986935,
      "loss": 3.5153,
      "step": 1124
    },
    {
      "epoch": 0.14399078458978626,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0019327010852261855,
      "loss": 2.6724,
      "step": 1125
    },
    {
      "epoch": 0.1441187763983105,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0019325514911538633,
      "loss": 3.6588,
      "step": 1126
    },
    {
      "epoch": 0.14424676820683477,
      "grad_norm": 0.828125,
      "learning_rate": 0.001932401736807437,
      "loss": 3.0896,
      "step": 1127
    },
    {
      "epoch": 0.144374760015359,
      "grad_norm": 0.84375,
      "learning_rate": 0.0019322518222126442,
      "loss": 3.5395,
      "step": 1128
    },
    {
      "epoch": 0.14450275182388328,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00193210174739525,
      "loss": 2.5968,
      "step": 1129
    },
    {
      "epoch": 0.14463074363240752,
      "grad_norm": 0.6875,
      "learning_rate": 0.0019319515123810476,
      "loss": 2.7883,
      "step": 1130
    },
    {
      "epoch": 0.1447587354409318,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0019318011171958569,
      "loss": 2.9237,
      "step": 1131
    },
    {
      "epoch": 0.14488672724945603,
      "grad_norm": 0.73828125,
      "learning_rate": 0.001931650561865526,
      "loss": 3.5064,
      "step": 1132
    },
    {
      "epoch": 0.1450147190579803,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0019314998464159303,
      "loss": 4.2771,
      "step": 1133
    },
    {
      "epoch": 0.14514271086650454,
      "grad_norm": 0.859375,
      "learning_rate": 0.0019313489708729728,
      "loss": 3.556,
      "step": 1134
    },
    {
      "epoch": 0.14527070267502878,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0019311979352625835,
      "loss": 3.1248,
      "step": 1135
    },
    {
      "epoch": 0.14539869448355305,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0019310467396107208,
      "loss": 3.4066,
      "step": 1136
    },
    {
      "epoch": 0.1455266862920773,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0019308953839433697,
      "loss": 3.0167,
      "step": 1137
    },
    {
      "epoch": 0.14565467810060156,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0019307438682865433,
      "loss": 3.4392,
      "step": 1138
    },
    {
      "epoch": 0.1457826699091258,
      "grad_norm": 0.6796875,
      "learning_rate": 0.001930592192666282,
      "loss": 3.042,
      "step": 1139
    },
    {
      "epoch": 0.14591066171765008,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0019304403571086541,
      "loss": 2.7384,
      "step": 1140
    },
    {
      "epoch": 0.14603865352617432,
      "grad_norm": 0.78125,
      "learning_rate": 0.0019302883616397544,
      "loss": 3.0887,
      "step": 1141
    },
    {
      "epoch": 0.14616664533469859,
      "grad_norm": 0.6796875,
      "learning_rate": 0.001930136206285706,
      "loss": 3.0912,
      "step": 1142
    },
    {
      "epoch": 0.14629463714322283,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0019299838910726592,
      "loss": 2.6067,
      "step": 1143
    },
    {
      "epoch": 0.1464226289517471,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0019298314160267921,
      "loss": 3.3939,
      "step": 1144
    },
    {
      "epoch": 0.14655062076027134,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0019296787811743101,
      "loss": 3.216,
      "step": 1145
    },
    {
      "epoch": 0.1466786125687956,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0019295259865414456,
      "loss": 3.2284,
      "step": 1146
    },
    {
      "epoch": 0.14680660437731985,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0019293730321544589,
      "loss": 3.2624,
      "step": 1147
    },
    {
      "epoch": 0.14693459618584412,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0019292199180396378,
      "loss": 3.297,
      "step": 1148
    },
    {
      "epoch": 0.14706258799436836,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0019290666442232977,
      "loss": 3.3239,
      "step": 1149
    },
    {
      "epoch": 0.1471905798028926,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0019289132107317803,
      "loss": 2.9973,
      "step": 1150
    },
    {
      "epoch": 0.14731857161141687,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0019287596175914567,
      "loss": 3.487,
      "step": 1151
    },
    {
      "epoch": 0.1474465634199411,
      "grad_norm": 0.6015625,
      "learning_rate": 0.001928605864828724,
      "loss": 2.6146,
      "step": 1152
    },
    {
      "epoch": 0.14757455522846538,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0019284519524700066,
      "loss": 3.4662,
      "step": 1153
    },
    {
      "epoch": 0.14770254703698962,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0019282978805417574,
      "loss": 3.3205,
      "step": 1154
    },
    {
      "epoch": 0.1478305388455139,
      "grad_norm": 0.86328125,
      "learning_rate": 0.001928143649070456,
      "loss": 3.1488,
      "step": 1155
    },
    {
      "epoch": 0.14795853065403813,
      "grad_norm": 0.671875,
      "learning_rate": 0.0019279892580826097,
      "loss": 3.4151,
      "step": 1156
    },
    {
      "epoch": 0.1480865224625624,
      "grad_norm": 0.60546875,
      "learning_rate": 0.001927834707604753,
      "loss": 3.2225,
      "step": 1157
    },
    {
      "epoch": 0.14821451427108664,
      "grad_norm": 0.84375,
      "learning_rate": 0.0019276799976634478,
      "loss": 3.3367,
      "step": 1158
    },
    {
      "epoch": 0.14834250607961091,
      "grad_norm": 0.7421875,
      "learning_rate": 0.001927525128285284,
      "loss": 2.9605,
      "step": 1159
    },
    {
      "epoch": 0.14847049788813516,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0019273700994968776,
      "loss": 2.9721,
      "step": 1160
    },
    {
      "epoch": 0.14859848969665942,
      "grad_norm": 0.6875,
      "learning_rate": 0.0019272149113248736,
      "loss": 3.028,
      "step": 1161
    },
    {
      "epoch": 0.14872648150518367,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0019270595637959427,
      "loss": 3.9092,
      "step": 1162
    },
    {
      "epoch": 0.14885447331370794,
      "grad_norm": 0.77734375,
      "learning_rate": 0.001926904056936785,
      "loss": 2.806,
      "step": 1163
    },
    {
      "epoch": 0.14898246512223218,
      "grad_norm": 0.83203125,
      "learning_rate": 0.001926748390774126,
      "loss": 3.1402,
      "step": 1164
    },
    {
      "epoch": 0.14911045693075642,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0019265925653347198,
      "loss": 3.7091,
      "step": 1165
    },
    {
      "epoch": 0.1492384487392807,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0019264365806453477,
      "loss": 3.5886,
      "step": 1166
    },
    {
      "epoch": 0.14936644054780493,
      "grad_norm": 0.76171875,
      "learning_rate": 0.001926280436732818,
      "loss": 2.6779,
      "step": 1167
    },
    {
      "epoch": 0.1494944323563292,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0019261241336239663,
      "loss": 2.7906,
      "step": 1168
    },
    {
      "epoch": 0.14962242416485344,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0019259676713456564,
      "loss": 2.8208,
      "step": 1169
    },
    {
      "epoch": 0.1497504159733777,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0019258110499247784,
      "loss": 3.129,
      "step": 1170
    },
    {
      "epoch": 0.14987840778190195,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0019256542693882504,
      "loss": 2.9581,
      "step": 1171
    },
    {
      "epoch": 0.15000639959042622,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0019254973297630179,
      "loss": 2.7063,
      "step": 1172
    },
    {
      "epoch": 0.15013439139895046,
      "grad_norm": 0.78515625,
      "learning_rate": 0.001925340231076053,
      "loss": 2.9345,
      "step": 1173
    },
    {
      "epoch": 0.15026238320747473,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0019251829733543562,
      "loss": 2.9553,
      "step": 1174
    },
    {
      "epoch": 0.15039037501599897,
      "grad_norm": 0.75,
      "learning_rate": 0.0019250255566249544,
      "loss": 3.142,
      "step": 1175
    },
    {
      "epoch": 0.15051836682452324,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0019248679809149025,
      "loss": 3.2804,
      "step": 1176
    },
    {
      "epoch": 0.15064635863304748,
      "grad_norm": 0.703125,
      "learning_rate": 0.0019247102462512822,
      "loss": 2.6575,
      "step": 1177
    },
    {
      "epoch": 0.15077435044157175,
      "grad_norm": 0.7890625,
      "learning_rate": 0.001924552352661203,
      "loss": 2.6658,
      "step": 1178
    },
    {
      "epoch": 0.150902342250096,
      "grad_norm": 0.8046875,
      "learning_rate": 0.001924394300171801,
      "loss": 2.4312,
      "step": 1179
    },
    {
      "epoch": 0.15103033405862024,
      "grad_norm": 0.78515625,
      "learning_rate": 0.001924236088810241,
      "loss": 3.1609,
      "step": 1180
    },
    {
      "epoch": 0.1511583258671445,
      "grad_norm": 0.71875,
      "learning_rate": 0.0019240777186037136,
      "loss": 2.6922,
      "step": 1181
    },
    {
      "epoch": 0.15128631767566875,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0019239191895794368,
      "loss": 3.1654,
      "step": 1182
    },
    {
      "epoch": 0.15141430948419302,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0019237605017646574,
      "loss": 2.8786,
      "step": 1183
    },
    {
      "epoch": 0.15154230129271726,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0019236016551866478,
      "loss": 2.8341,
      "step": 1184
    },
    {
      "epoch": 0.15167029310124153,
      "grad_norm": 0.703125,
      "learning_rate": 0.0019234426498727084,
      "loss": 3.9237,
      "step": 1185
    },
    {
      "epoch": 0.15179828490976577,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0019232834858501672,
      "loss": 2.6052,
      "step": 1186
    },
    {
      "epoch": 0.15192627671829004,
      "grad_norm": 0.90625,
      "learning_rate": 0.0019231241631463788,
      "loss": 2.8804,
      "step": 1187
    },
    {
      "epoch": 0.15205426852681428,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0019229646817887254,
      "loss": 2.9122,
      "step": 1188
    },
    {
      "epoch": 0.15218226033533855,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0019228050418046167,
      "loss": 2.3908,
      "step": 1189
    },
    {
      "epoch": 0.1523102521438628,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0019226452432214893,
      "loss": 2.6428,
      "step": 1190
    },
    {
      "epoch": 0.15243824395238706,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0019224852860668072,
      "loss": 3.1206,
      "step": 1191
    },
    {
      "epoch": 0.1525662357609113,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0019223251703680614,
      "loss": 3.2855,
      "step": 1192
    },
    {
      "epoch": 0.15269422756943557,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0019221648961527708,
      "loss": 3.9638,
      "step": 1193
    },
    {
      "epoch": 0.1528222193779598,
      "grad_norm": 0.8125,
      "learning_rate": 0.0019220044634484807,
      "loss": 3.7907,
      "step": 1194
    },
    {
      "epoch": 0.15295021118648405,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0019218438722827642,
      "loss": 3.2759,
      "step": 1195
    },
    {
      "epoch": 0.15307820299500832,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0019216831226832216,
      "loss": 2.4429,
      "step": 1196
    },
    {
      "epoch": 0.15320619480353256,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0019215222146774804,
      "loss": 2.9268,
      "step": 1197
    },
    {
      "epoch": 0.15333418661205683,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0019213611482931951,
      "loss": 3.4134,
      "step": 1198
    },
    {
      "epoch": 0.15346217842058107,
      "grad_norm": 0.59375,
      "learning_rate": 0.0019211999235580475,
      "loss": 2.5718,
      "step": 1199
    },
    {
      "epoch": 0.15359017022910534,
      "grad_norm": 0.546875,
      "learning_rate": 0.001921038540499747,
      "loss": 2.3142,
      "step": 1200
    },
    {
      "epoch": 0.15371816203762959,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0019208769991460298,
      "loss": 2.812,
      "step": 1201
    },
    {
      "epoch": 0.15384615384615385,
      "grad_norm": 0.78515625,
      "learning_rate": 0.001920715299524659,
      "loss": 3.1725,
      "step": 1202
    },
    {
      "epoch": 0.1539741456546781,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0019205534416634254,
      "loss": 2.6004,
      "step": 1203
    },
    {
      "epoch": 0.15410213746320237,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0019203914255901476,
      "loss": 3.0801,
      "step": 1204
    },
    {
      "epoch": 0.1542301292717266,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0019202292513326698,
      "loss": 3.1377,
      "step": 1205
    },
    {
      "epoch": 0.15435812108025088,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0019200669189188648,
      "loss": 2.4937,
      "step": 1206
    },
    {
      "epoch": 0.15448611288877512,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0019199044283766317,
      "loss": 3.0643,
      "step": 1207
    },
    {
      "epoch": 0.15461410469729936,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0019197417797338977,
      "loss": 3.5516,
      "step": 1208
    },
    {
      "epoch": 0.15474209650582363,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0019195789730186161,
      "loss": 2.5554,
      "step": 1209
    },
    {
      "epoch": 0.15487008831434787,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0019194160082587682,
      "loss": 2.9125,
      "step": 1210
    },
    {
      "epoch": 0.15499808012287214,
      "grad_norm": 0.84765625,
      "learning_rate": 0.001919252885482362,
      "loss": 3.3681,
      "step": 1211
    },
    {
      "epoch": 0.15512607193139638,
      "grad_norm": 1.203125,
      "learning_rate": 0.0019190896047174325,
      "loss": 3.2294,
      "step": 1212
    },
    {
      "epoch": 0.15525406373992065,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0019189261659920424,
      "loss": 2.887,
      "step": 1213
    },
    {
      "epoch": 0.1553820555484449,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0019187625693342816,
      "loss": 2.424,
      "step": 1214
    },
    {
      "epoch": 0.15551004735696916,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0019185988147722665,
      "loss": 3.0079,
      "step": 1215
    },
    {
      "epoch": 0.1556380391654934,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0019184349023341407,
      "loss": 2.8385,
      "step": 1216
    },
    {
      "epoch": 0.15576603097401767,
      "grad_norm": 0.7578125,
      "learning_rate": 0.001918270832048076,
      "loss": 2.5947,
      "step": 1217
    },
    {
      "epoch": 0.1558940227825419,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0019181066039422698,
      "loss": 3.0168,
      "step": 1218
    },
    {
      "epoch": 0.15602201459106618,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0019179422180449477,
      "loss": 2.724,
      "step": 1219
    },
    {
      "epoch": 0.15615000639959042,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0019177776743843622,
      "loss": 2.7551,
      "step": 1220
    },
    {
      "epoch": 0.1562779982081147,
      "grad_norm": 0.78125,
      "learning_rate": 0.0019176129729887924,
      "loss": 2.9282,
      "step": 1221
    },
    {
      "epoch": 0.15640599001663893,
      "grad_norm": 0.796875,
      "learning_rate": 0.0019174481138865454,
      "loss": 2.907,
      "step": 1222
    },
    {
      "epoch": 0.15653398182516318,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0019172830971059544,
      "loss": 3.1977,
      "step": 1223
    },
    {
      "epoch": 0.15666197363368745,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0019171179226753808,
      "loss": 2.5214,
      "step": 1224
    },
    {
      "epoch": 0.1567899654422117,
      "grad_norm": 0.8984375,
      "learning_rate": 0.001916952590623212,
      "loss": 2.4556,
      "step": 1225
    },
    {
      "epoch": 0.15691795725073596,
      "grad_norm": 0.80859375,
      "learning_rate": 0.001916787100977863,
      "loss": 3.6011,
      "step": 1226
    },
    {
      "epoch": 0.1570459490592602,
      "grad_norm": 0.890625,
      "learning_rate": 0.001916621453767776,
      "loss": 2.954,
      "step": 1227
    },
    {
      "epoch": 0.15717394086778447,
      "grad_norm": 0.71875,
      "learning_rate": 0.0019164556490214207,
      "loss": 3.195,
      "step": 1228
    },
    {
      "epoch": 0.1573019326763087,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0019162896867672924,
      "loss": 3.4031,
      "step": 1229
    },
    {
      "epoch": 0.15742992448483298,
      "grad_norm": 0.86328125,
      "learning_rate": 0.001916123567033915,
      "loss": 3.2088,
      "step": 1230
    },
    {
      "epoch": 0.15755791629335722,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0019159572898498386,
      "loss": 2.9164,
      "step": 1231
    },
    {
      "epoch": 0.1576859081018815,
      "grad_norm": 0.79296875,
      "learning_rate": 0.001915790855243641,
      "loss": 3.108,
      "step": 1232
    },
    {
      "epoch": 0.15781389991040573,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0019156242632439263,
      "loss": 2.9857,
      "step": 1233
    },
    {
      "epoch": 0.15794189171893,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0019154575138793262,
      "loss": 3.0548,
      "step": 1234
    },
    {
      "epoch": 0.15806988352745424,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0019152906071784992,
      "loss": 3.0768,
      "step": 1235
    },
    {
      "epoch": 0.1581978753359785,
      "grad_norm": 0.82421875,
      "learning_rate": 0.001915123543170131,
      "loss": 3.4871,
      "step": 1236
    },
    {
      "epoch": 0.15832586714450275,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0019149563218829341,
      "loss": 2.9833,
      "step": 1237
    },
    {
      "epoch": 0.158453858953027,
      "grad_norm": 0.875,
      "learning_rate": 0.0019147889433456485,
      "loss": 2.8796,
      "step": 1238
    },
    {
      "epoch": 0.15858185076155126,
      "grad_norm": 0.83203125,
      "learning_rate": 0.001914621407587041,
      "loss": 3.6634,
      "step": 1239
    },
    {
      "epoch": 0.1587098425700755,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0019144537146359047,
      "loss": 2.8981,
      "step": 1240
    },
    {
      "epoch": 0.15883783437859977,
      "grad_norm": 0.78125,
      "learning_rate": 0.001914285864521061,
      "loss": 2.8205,
      "step": 1241
    },
    {
      "epoch": 0.15896582618712402,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0019141178572713573,
      "loss": 3.5973,
      "step": 1242
    },
    {
      "epoch": 0.15909381799564828,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0019139496929156685,
      "loss": 2.8491,
      "step": 1243
    },
    {
      "epoch": 0.15922180980417253,
      "grad_norm": 0.734375,
      "learning_rate": 0.0019137813714828963,
      "loss": 2.3974,
      "step": 1244
    },
    {
      "epoch": 0.1593498016126968,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0019136128930019698,
      "loss": 2.7961,
      "step": 1245
    },
    {
      "epoch": 0.15947779342122104,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0019134442575018443,
      "loss": 2.7662,
      "step": 1246
    },
    {
      "epoch": 0.1596057852297453,
      "grad_norm": 0.8125,
      "learning_rate": 0.0019132754650115027,
      "loss": 3.2365,
      "step": 1247
    },
    {
      "epoch": 0.15973377703826955,
      "grad_norm": 0.6875,
      "learning_rate": 0.0019131065155599547,
      "loss": 3.1491,
      "step": 1248
    },
    {
      "epoch": 0.15986176884679382,
      "grad_norm": 0.9375,
      "learning_rate": 0.0019129374091762374,
      "loss": 2.2943,
      "step": 1249
    },
    {
      "epoch": 0.15998976065531806,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0019127681458894136,
      "loss": 3.297,
      "step": 1250
    },
    {
      "epoch": 0.16011775246384233,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0019125987257285748,
      "loss": 3.1192,
      "step": 1251
    },
    {
      "epoch": 0.16024574427236657,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0019124291487228385,
      "loss": 3.3185,
      "step": 1252
    },
    {
      "epoch": 0.1603737360808908,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0019122594149013487,
      "loss": 2.8022,
      "step": 1253
    },
    {
      "epoch": 0.16050172788941508,
      "grad_norm": 0.9375,
      "learning_rate": 0.0019120895242932775,
      "loss": 2.6998,
      "step": 1254
    },
    {
      "epoch": 0.16062971969793932,
      "grad_norm": 0.69140625,
      "learning_rate": 0.001911919476927823,
      "loss": 3.0829,
      "step": 1255
    },
    {
      "epoch": 0.1607577115064636,
      "grad_norm": 0.78125,
      "learning_rate": 0.0019117492728342106,
      "loss": 2.939,
      "step": 1256
    },
    {
      "epoch": 0.16088570331498783,
      "grad_norm": 0.69140625,
      "learning_rate": 0.001911578912041693,
      "loss": 2.9716,
      "step": 1257
    },
    {
      "epoch": 0.1610136951235121,
      "grad_norm": 0.76953125,
      "learning_rate": 0.001911408394579549,
      "loss": 3.2818,
      "step": 1258
    },
    {
      "epoch": 0.16114168693203634,
      "grad_norm": 1.296875,
      "learning_rate": 0.001911237720477085,
      "loss": 2.8096,
      "step": 1259
    },
    {
      "epoch": 0.1612696787405606,
      "grad_norm": 0.828125,
      "learning_rate": 0.0019110668897636342,
      "loss": 2.8765,
      "step": 1260
    },
    {
      "epoch": 0.16139767054908485,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0019108959024685568,
      "loss": 2.7983,
      "step": 1261
    },
    {
      "epoch": 0.16152566235760912,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0019107247586212393,
      "loss": 3.197,
      "step": 1262
    },
    {
      "epoch": 0.16165365416613336,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0019105534582510954,
      "loss": 2.7668,
      "step": 1263
    },
    {
      "epoch": 0.16178164597465763,
      "grad_norm": 0.64453125,
      "learning_rate": 0.001910382001387567,
      "loss": 2.8204,
      "step": 1264
    },
    {
      "epoch": 0.16190963778318188,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0019102103880601204,
      "loss": 2.8118,
      "step": 1265
    },
    {
      "epoch": 0.16203762959170612,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0019100386182982505,
      "loss": 2.9399,
      "step": 1266
    },
    {
      "epoch": 0.1621656214002304,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0019098666921314792,
      "loss": 2.491,
      "step": 1267
    },
    {
      "epoch": 0.16229361320875463,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0019096946095893544,
      "loss": 3.0527,
      "step": 1268
    },
    {
      "epoch": 0.1624216050172789,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0019095223707014514,
      "loss": 2.592,
      "step": 1269
    },
    {
      "epoch": 0.16254959682580314,
      "grad_norm": 0.78515625,
      "learning_rate": 0.001909349975497372,
      "loss": 2.4592,
      "step": 1270
    },
    {
      "epoch": 0.1626775886343274,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0019091774240067455,
      "loss": 3.2201,
      "step": 1271
    },
    {
      "epoch": 0.16280558044285165,
      "grad_norm": 3.03125,
      "learning_rate": 0.0019090047162592274,
      "loss": 2.8591,
      "step": 1272
    },
    {
      "epoch": 0.16293357225137592,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0019088318522845007,
      "loss": 2.8622,
      "step": 1273
    },
    {
      "epoch": 0.16306156405990016,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0019086588321122744,
      "loss": 2.4009,
      "step": 1274
    },
    {
      "epoch": 0.16318955586842443,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0019084856557722851,
      "loss": 3.1515,
      "step": 1275
    },
    {
      "epoch": 0.16331754767694867,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0019083123232942957,
      "loss": 2.7374,
      "step": 1276
    },
    {
      "epoch": 0.16344553948547294,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0019081388347080969,
      "loss": 2.9997,
      "step": 1277
    },
    {
      "epoch": 0.16357353129399718,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0019079651900435043,
      "loss": 2.8366,
      "step": 1278
    },
    {
      "epoch": 0.16370152310252145,
      "grad_norm": 0.67578125,
      "learning_rate": 0.001907791389330363,
      "loss": 2.8378,
      "step": 1279
    },
    {
      "epoch": 0.1638295149110457,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0019076174325985422,
      "loss": 3.6618,
      "step": 1280
    },
    {
      "epoch": 0.16395750671956993,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0019074433198779399,
      "loss": 2.6502,
      "step": 1281
    },
    {
      "epoch": 0.1640854985280942,
      "grad_norm": 0.78125,
      "learning_rate": 0.00190726905119848,
      "loss": 3.0741,
      "step": 1282
    },
    {
      "epoch": 0.16421349033661845,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0019070946265901136,
      "loss": 3.0057,
      "step": 1283
    },
    {
      "epoch": 0.16434148214514271,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0019069200460828182,
      "loss": 2.1336,
      "step": 1284
    },
    {
      "epoch": 0.16446947395366696,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0019067453097065983,
      "loss": 2.732,
      "step": 1285
    },
    {
      "epoch": 0.16459746576219123,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0019065704174914854,
      "loss": 3.4839,
      "step": 1286
    },
    {
      "epoch": 0.16472545757071547,
      "grad_norm": 1.015625,
      "learning_rate": 0.0019063953694675368,
      "loss": 2.4278,
      "step": 1287
    },
    {
      "epoch": 0.16485344937923974,
      "grad_norm": 0.765625,
      "learning_rate": 0.001906220165664838,
      "loss": 3.2791,
      "step": 1288
    },
    {
      "epoch": 0.16498144118776398,
      "grad_norm": 0.859375,
      "learning_rate": 0.001906044806113501,
      "loss": 2.9178,
      "step": 1289
    },
    {
      "epoch": 0.16510943299628825,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0019058692908436635,
      "loss": 3.374,
      "step": 1290
    },
    {
      "epoch": 0.1652374248048125,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0019056936198854908,
      "loss": 2.7851,
      "step": 1291
    },
    {
      "epoch": 0.16536541661333676,
      "grad_norm": 0.8125,
      "learning_rate": 0.001905517793269175,
      "loss": 2.7953,
      "step": 1292
    },
    {
      "epoch": 0.165493408421861,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0019053418110249343,
      "loss": 2.8631,
      "step": 1293
    },
    {
      "epoch": 0.16562140023038527,
      "grad_norm": 0.671875,
      "learning_rate": 0.0019051656731830149,
      "loss": 2.7794,
      "step": 1294
    },
    {
      "epoch": 0.1657493920389095,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0019049893797736881,
      "loss": 2.8277,
      "step": 1295
    },
    {
      "epoch": 0.16587738384743375,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0019048129308272533,
      "loss": 2.6165,
      "step": 1296
    },
    {
      "epoch": 0.16600537565595802,
      "grad_norm": 1.15625,
      "learning_rate": 0.0019046363263740358,
      "loss": 3.4371,
      "step": 1297
    },
    {
      "epoch": 0.16613336746448226,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0019044595664443884,
      "loss": 2.7327,
      "step": 1298
    },
    {
      "epoch": 0.16626135927300653,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0019042826510686894,
      "loss": 3.5604,
      "step": 1299
    },
    {
      "epoch": 0.16638935108153077,
      "grad_norm": 1.09375,
      "learning_rate": 0.0019041055802773456,
      "loss": 3.2296,
      "step": 1300
    },
    {
      "epoch": 0.16651734289005504,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0019039283541007887,
      "loss": 2.8312,
      "step": 1301
    },
    {
      "epoch": 0.16664533469857928,
      "grad_norm": 0.6328125,
      "learning_rate": 0.001903750972569478,
      "loss": 2.2702,
      "step": 1302
    },
    {
      "epoch": 0.16677332650710355,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0019035734357138998,
      "loss": 3.1609,
      "step": 1303
    },
    {
      "epoch": 0.1669013183156278,
      "grad_norm": 0.7265625,
      "learning_rate": 0.001903395743564566,
      "loss": 2.5168,
      "step": 1304
    },
    {
      "epoch": 0.16702931012415206,
      "grad_norm": 0.765625,
      "learning_rate": 0.0019032178961520168,
      "loss": 4.4863,
      "step": 1305
    },
    {
      "epoch": 0.1671573019326763,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0019030398935068172,
      "loss": 3.0494,
      "step": 1306
    },
    {
      "epoch": 0.16728529374120057,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0019028617356595607,
      "loss": 2.7109,
      "step": 1307
    },
    {
      "epoch": 0.16741328554972482,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0019026834226408657,
      "loss": 2.5678,
      "step": 1308
    },
    {
      "epoch": 0.16754127735824909,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0019025049544813793,
      "loss": 3.4073,
      "step": 1309
    },
    {
      "epoch": 0.16766926916677333,
      "grad_norm": 0.75390625,
      "learning_rate": 0.001902326331211773,
      "loss": 2.8455,
      "step": 1310
    },
    {
      "epoch": 0.16779726097529757,
      "grad_norm": 0.875,
      "learning_rate": 0.001902147552862747,
      "loss": 2.8332,
      "step": 1311
    },
    {
      "epoch": 0.16792525278382184,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0019019686194650268,
      "loss": 2.2093,
      "step": 1312
    },
    {
      "epoch": 0.16805324459234608,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0019017895310493652,
      "loss": 2.5917,
      "step": 1313
    },
    {
      "epoch": 0.16818123640087035,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0019016102876465411,
      "loss": 3.3356,
      "step": 1314
    },
    {
      "epoch": 0.1683092282093946,
      "grad_norm": 0.6875,
      "learning_rate": 0.0019014308892873612,
      "loss": 2.9746,
      "step": 1315
    },
    {
      "epoch": 0.16843722001791886,
      "grad_norm": 0.59375,
      "learning_rate": 0.0019012513360026572,
      "loss": 2.2447,
      "step": 1316
    },
    {
      "epoch": 0.1685652118264431,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0019010716278232882,
      "loss": 2.5181,
      "step": 1317
    },
    {
      "epoch": 0.16869320363496737,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0019008917647801407,
      "loss": 2.2988,
      "step": 1318
    },
    {
      "epoch": 0.1688211954434916,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0019007117469041265,
      "loss": 3.3072,
      "step": 1319
    },
    {
      "epoch": 0.16894918725201588,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0019005315742261848,
      "loss": 2.3308,
      "step": 1320
    },
    {
      "epoch": 0.16907717906054012,
      "grad_norm": 0.7109375,
      "learning_rate": 0.001900351246777281,
      "loss": 3.0191,
      "step": 1321
    },
    {
      "epoch": 0.1692051708690644,
      "grad_norm": 0.71875,
      "learning_rate": 0.0019001707645884076,
      "loss": 3.1153,
      "step": 1322
    },
    {
      "epoch": 0.16933316267758863,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0018999901276905834,
      "loss": 2.2271,
      "step": 1323
    },
    {
      "epoch": 0.16946115448611287,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0018998093361148533,
      "loss": 3.1218,
      "step": 1324
    },
    {
      "epoch": 0.16958914629463714,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0018996283898922899,
      "loss": 3.2087,
      "step": 1325
    },
    {
      "epoch": 0.16971713810316139,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0018994472890539913,
      "loss": 3.0561,
      "step": 1326
    },
    {
      "epoch": 0.16984512991168565,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0018992660336310827,
      "loss": 2.7923,
      "step": 1327
    },
    {
      "epoch": 0.1699731217202099,
      "grad_norm": 0.765625,
      "learning_rate": 0.001899084623654716,
      "loss": 2.9133,
      "step": 1328
    },
    {
      "epoch": 0.17010111352873417,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0018989030591560693,
      "loss": 2.528,
      "step": 1329
    },
    {
      "epoch": 0.1702291053372584,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0018987213401663475,
      "loss": 2.7901,
      "step": 1330
    },
    {
      "epoch": 0.17035709714578268,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0018985394667167816,
      "loss": 3.2481,
      "step": 1331
    },
    {
      "epoch": 0.17048508895430692,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0018983574388386298,
      "loss": 2.4459,
      "step": 1332
    },
    {
      "epoch": 0.1706130807628312,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0018981752565631767,
      "loss": 2.4015,
      "step": 1333
    },
    {
      "epoch": 0.17074107257135543,
      "grad_norm": 0.671875,
      "learning_rate": 0.001897992919921733,
      "loss": 2.535,
      "step": 1334
    },
    {
      "epoch": 0.1708690643798797,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0018978104289456362,
      "loss": 2.6201,
      "step": 1335
    },
    {
      "epoch": 0.17099705618840394,
      "grad_norm": 0.703125,
      "learning_rate": 0.0018976277836662506,
      "loss": 2.0525,
      "step": 1336
    },
    {
      "epoch": 0.1711250479969282,
      "grad_norm": 0.78125,
      "learning_rate": 0.001897444984114966,
      "loss": 3.2052,
      "step": 1337
    },
    {
      "epoch": 0.17125303980545245,
      "grad_norm": 2.015625,
      "learning_rate": 0.0018972620303232008,
      "loss": 3.0579,
      "step": 1338
    },
    {
      "epoch": 0.1713810316139767,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0018970789223223977,
      "loss": 2.5945,
      "step": 1339
    },
    {
      "epoch": 0.17150902342250096,
      "grad_norm": 1.0625,
      "learning_rate": 0.0018968956601440267,
      "loss": 2.5495,
      "step": 1340
    },
    {
      "epoch": 0.1716370152310252,
      "grad_norm": 0.9375,
      "learning_rate": 0.0018967122438195846,
      "loss": 2.9128,
      "step": 1341
    },
    {
      "epoch": 0.17176500703954947,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0018965286733805945,
      "loss": 3.3376,
      "step": 1342
    },
    {
      "epoch": 0.1718929988480737,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0018963449488586057,
      "loss": 3.4248,
      "step": 1343
    },
    {
      "epoch": 0.17202099065659798,
      "grad_norm": 0.640625,
      "learning_rate": 0.0018961610702851947,
      "loss": 2.7956,
      "step": 1344
    },
    {
      "epoch": 0.17214898246512222,
      "grad_norm": 0.765625,
      "learning_rate": 0.0018959770376919638,
      "loss": 2.8848,
      "step": 1345
    },
    {
      "epoch": 0.1722769742736465,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0018957928511105417,
      "loss": 2.2574,
      "step": 1346
    },
    {
      "epoch": 0.17240496608217074,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0018956085105725844,
      "loss": 2.9681,
      "step": 1347
    },
    {
      "epoch": 0.172532957890695,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0018954240161097736,
      "loss": 2.422,
      "step": 1348
    },
    {
      "epoch": 0.17266094969921925,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0018952393677538174,
      "loss": 2.5005,
      "step": 1349
    },
    {
      "epoch": 0.17278894150774352,
      "grad_norm": 0.75,
      "learning_rate": 0.0018950545655364508,
      "loss": 3.1804,
      "step": 1350
    },
    {
      "epoch": 0.17291693331626776,
      "grad_norm": 0.703125,
      "learning_rate": 0.0018948696094894352,
      "loss": 3.092,
      "step": 1351
    },
    {
      "epoch": 0.17304492512479203,
      "grad_norm": 0.71875,
      "learning_rate": 0.001894684499644558,
      "loss": 2.6194,
      "step": 1352
    },
    {
      "epoch": 0.17317291693331627,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0018944992360336336,
      "loss": 2.5943,
      "step": 1353
    },
    {
      "epoch": 0.1733009087418405,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0018943138186885025,
      "loss": 2.4157,
      "step": 1354
    },
    {
      "epoch": 0.17342890055036478,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0018941282476410315,
      "loss": 2.3782,
      "step": 1355
    },
    {
      "epoch": 0.17355689235888902,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0018939425229231141,
      "loss": 2.057,
      "step": 1356
    },
    {
      "epoch": 0.1736848841674133,
      "grad_norm": 0.875,
      "learning_rate": 0.0018937566445666706,
      "loss": 2.6457,
      "step": 1357
    },
    {
      "epoch": 0.17381287597593753,
      "grad_norm": 0.7734375,
      "learning_rate": 0.001893570612603646,
      "loss": 3.1611,
      "step": 1358
    },
    {
      "epoch": 0.1739408677844618,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0018933844270660144,
      "loss": 2.1942,
      "step": 1359
    },
    {
      "epoch": 0.17406885959298604,
      "grad_norm": 0.984375,
      "learning_rate": 0.0018931980879857736,
      "loss": 2.8986,
      "step": 1360
    },
    {
      "epoch": 0.1741968514015103,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0018930115953949495,
      "loss": 2.6838,
      "step": 1361
    },
    {
      "epoch": 0.17432484321003455,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0018928249493255938,
      "loss": 2.0056,
      "step": 1362
    },
    {
      "epoch": 0.17445283501855882,
      "grad_norm": 0.75,
      "learning_rate": 0.001892638149809785,
      "loss": 2.6713,
      "step": 1363
    },
    {
      "epoch": 0.17458082682708306,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0018924511968796268,
      "loss": 3.1198,
      "step": 1364
    },
    {
      "epoch": 0.17470881863560733,
      "grad_norm": 0.69921875,
      "learning_rate": 0.001892264090567251,
      "loss": 2.7258,
      "step": 1365
    },
    {
      "epoch": 0.17483681044413157,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0018920768309048143,
      "loss": 2.7344,
      "step": 1366
    },
    {
      "epoch": 0.17496480225265584,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0018918894179245005,
      "loss": 1.8712,
      "step": 1367
    },
    {
      "epoch": 0.17509279406118008,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0018917018516585195,
      "loss": 2.4761,
      "step": 1368
    },
    {
      "epoch": 0.17522078586970433,
      "grad_norm": 0.90234375,
      "learning_rate": 0.001891514132139108,
      "loss": 2.9963,
      "step": 1369
    },
    {
      "epoch": 0.1753487776782286,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0018913262593985279,
      "loss": 3.5638,
      "step": 1370
    },
    {
      "epoch": 0.17547676948675284,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0018911382334690688,
      "loss": 2.7291,
      "step": 1371
    },
    {
      "epoch": 0.1756047612952771,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0018909500543830456,
      "loss": 2.4118,
      "step": 1372
    },
    {
      "epoch": 0.17573275310380135,
      "grad_norm": 0.734375,
      "learning_rate": 0.0018907617221728002,
      "loss": 2.9008,
      "step": 1373
    },
    {
      "epoch": 0.17586074491232562,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0018905732368707003,
      "loss": 3.037,
      "step": 1374
    },
    {
      "epoch": 0.17598873672084986,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0018903845985091406,
      "loss": 2.4574,
      "step": 1375
    },
    {
      "epoch": 0.17611672852937413,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0018901958071205409,
      "loss": 2.9773,
      "step": 1376
    },
    {
      "epoch": 0.17624472033789837,
      "grad_norm": 0.69140625,
      "learning_rate": 0.001890006862737349,
      "loss": 2.6701,
      "step": 1377
    },
    {
      "epoch": 0.17637271214642264,
      "grad_norm": 0.78125,
      "learning_rate": 0.0018898177653920375,
      "loss": 3.5373,
      "step": 1378
    },
    {
      "epoch": 0.17650070395494688,
      "grad_norm": 0.7109375,
      "learning_rate": 0.001889628515117106,
      "loss": 2.758,
      "step": 1379
    },
    {
      "epoch": 0.17662869576347115,
      "grad_norm": 0.734375,
      "learning_rate": 0.0018894391119450799,
      "loss": 3.6988,
      "step": 1380
    },
    {
      "epoch": 0.1767566875719954,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0018892495559085117,
      "loss": 3.6966,
      "step": 1381
    },
    {
      "epoch": 0.17688467938051966,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0018890598470399792,
      "loss": 2.5999,
      "step": 1382
    },
    {
      "epoch": 0.1770126711890439,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0018888699853720874,
      "loss": 3.1421,
      "step": 1383
    },
    {
      "epoch": 0.17714066299756814,
      "grad_norm": 0.75,
      "learning_rate": 0.0018886799709374668,
      "loss": 2.849,
      "step": 1384
    },
    {
      "epoch": 0.1772686548060924,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0018884898037687748,
      "loss": 3.2061,
      "step": 1385
    },
    {
      "epoch": 0.17739664661461665,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0018882994838986945,
      "loss": 3.291,
      "step": 1386
    },
    {
      "epoch": 0.17752463842314092,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0018881090113599354,
      "loss": 3.1074,
      "step": 1387
    },
    {
      "epoch": 0.17765263023166517,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0018879183861852332,
      "loss": 2.3129,
      "step": 1388
    },
    {
      "epoch": 0.17778062204018943,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0018877276084073497,
      "loss": 2.5993,
      "step": 1389
    },
    {
      "epoch": 0.17790861384871368,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0018875366780590742,
      "loss": 2.6015,
      "step": 1390
    },
    {
      "epoch": 0.17803660565723795,
      "grad_norm": 0.640625,
      "learning_rate": 0.0018873455951732203,
      "loss": 2.6752,
      "step": 1391
    },
    {
      "epoch": 0.1781645974657622,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0018871543597826289,
      "loss": 2.6697,
      "step": 1392
    },
    {
      "epoch": 0.17829258927428646,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0018869629719201666,
      "loss": 2.7342,
      "step": 1393
    },
    {
      "epoch": 0.1784205810828107,
      "grad_norm": 0.84375,
      "learning_rate": 0.0018867714316187273,
      "loss": 3.1976,
      "step": 1394
    },
    {
      "epoch": 0.17854857289133497,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0018865797389112295,
      "loss": 2.5477,
      "step": 1395
    },
    {
      "epoch": 0.1786765646998592,
      "grad_norm": 0.796875,
      "learning_rate": 0.0018863878938306191,
      "loss": 2.923,
      "step": 1396
    },
    {
      "epoch": 0.17880455650838345,
      "grad_norm": 0.59375,
      "learning_rate": 0.0018861958964098678,
      "loss": 2.8756,
      "step": 1397
    },
    {
      "epoch": 0.17893254831690772,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0018860037466819732,
      "loss": 2.5697,
      "step": 1398
    },
    {
      "epoch": 0.17906054012543196,
      "grad_norm": 0.71875,
      "learning_rate": 0.00188581144467996,
      "loss": 2.9197,
      "step": 1399
    },
    {
      "epoch": 0.17918853193395623,
      "grad_norm": 0.6875,
      "learning_rate": 0.0018856189904368774,
      "loss": 2.3186,
      "step": 1400
    },
    {
      "epoch": 0.17931652374248047,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0018854263839858025,
      "loss": 2.8476,
      "step": 1401
    },
    {
      "epoch": 0.17944451555100474,
      "grad_norm": 0.64453125,
      "learning_rate": 0.001885233625359838,
      "loss": 2.2761,
      "step": 1402
    },
    {
      "epoch": 0.17957250735952898,
      "grad_norm": 0.640625,
      "learning_rate": 0.0018850407145921118,
      "loss": 1.9299,
      "step": 1403
    },
    {
      "epoch": 0.17970049916805325,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0018848476517157795,
      "loss": 2.463,
      "step": 1404
    },
    {
      "epoch": 0.1798284909765775,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0018846544367640218,
      "loss": 2.7922,
      "step": 1405
    },
    {
      "epoch": 0.17995648278510176,
      "grad_norm": 0.78125,
      "learning_rate": 0.0018844610697700457,
      "loss": 3.1053,
      "step": 1406
    },
    {
      "epoch": 0.180084474593626,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0018842675507670847,
      "loss": 2.8666,
      "step": 1407
    },
    {
      "epoch": 0.18021246640215027,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0018840738797883982,
      "loss": 2.0912,
      "step": 1408
    },
    {
      "epoch": 0.18034045821067451,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0018838800568672714,
      "loss": 2.5875,
      "step": 1409
    },
    {
      "epoch": 0.18046845001919878,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0018836860820370164,
      "loss": 2.6742,
      "step": 1410
    },
    {
      "epoch": 0.18059644182772303,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00188349195533097,
      "loss": 2.808,
      "step": 1411
    },
    {
      "epoch": 0.18072443363624727,
      "grad_norm": 0.7578125,
      "learning_rate": 0.001883297676782497,
      "loss": 2.5162,
      "step": 1412
    },
    {
      "epoch": 0.18085242544477154,
      "grad_norm": 0.71875,
      "learning_rate": 0.001883103246424987,
      "loss": 2.8881,
      "step": 1413
    },
    {
      "epoch": 0.18098041725329578,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0018829086642918559,
      "loss": 3.3223,
      "step": 1414
    },
    {
      "epoch": 0.18110840906182005,
      "grad_norm": 0.62890625,
      "learning_rate": 0.001882713930416546,
      "loss": 2.2337,
      "step": 1415
    },
    {
      "epoch": 0.1812364008703443,
      "grad_norm": 0.75390625,
      "learning_rate": 0.001882519044832525,
      "loss": 2.6395,
      "step": 1416
    },
    {
      "epoch": 0.18136439267886856,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0018823240075732878,
      "loss": 2.8588,
      "step": 1417
    },
    {
      "epoch": 0.1814923844873928,
      "grad_norm": 0.78125,
      "learning_rate": 0.0018821288186723541,
      "loss": 2.7788,
      "step": 1418
    },
    {
      "epoch": 0.18162037629591707,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0018819334781632708,
      "loss": 2.5142,
      "step": 1419
    },
    {
      "epoch": 0.1817483681044413,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0018817379860796102,
      "loss": 2.4458,
      "step": 1420
    },
    {
      "epoch": 0.18187635991296558,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0018815423424549708,
      "loss": 3.6767,
      "step": 1421
    },
    {
      "epoch": 0.18200435172148982,
      "grad_norm": 0.734375,
      "learning_rate": 0.0018813465473229765,
      "loss": 2.2514,
      "step": 1422
    },
    {
      "epoch": 0.1821323435300141,
      "grad_norm": 0.88671875,
      "learning_rate": 0.001881150600717279,
      "loss": 3.2388,
      "step": 1423
    },
    {
      "epoch": 0.18226033533853833,
      "grad_norm": 0.65625,
      "learning_rate": 0.001880954502671554,
      "loss": 2.5086,
      "step": 1424
    },
    {
      "epoch": 0.1823883271470626,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0018807582532195043,
      "loss": 2.1166,
      "step": 1425
    },
    {
      "epoch": 0.18251631895558684,
      "grad_norm": 0.921875,
      "learning_rate": 0.001880561852394859,
      "loss": 2.837,
      "step": 1426
    },
    {
      "epoch": 0.18264431076411108,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0018803653002313723,
      "loss": 2.4763,
      "step": 1427
    },
    {
      "epoch": 0.18277230257263535,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0018801685967628253,
      "loss": 2.5659,
      "step": 1428
    },
    {
      "epoch": 0.1829002943811596,
      "grad_norm": 0.66015625,
      "learning_rate": 0.001879971742023024,
      "loss": 2.2847,
      "step": 1429
    },
    {
      "epoch": 0.18302828618968386,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0018797747360458016,
      "loss": 2.5156,
      "step": 1430
    },
    {
      "epoch": 0.1831562779982081,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0018795775788650172,
      "loss": 3.3477,
      "step": 1431
    },
    {
      "epoch": 0.18328426980673238,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0018793802705145543,
      "loss": 2.5286,
      "step": 1432
    },
    {
      "epoch": 0.18341226161525662,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0018791828110283244,
      "loss": 2.8812,
      "step": 1433
    },
    {
      "epoch": 0.18354025342378089,
      "grad_norm": 0.90625,
      "learning_rate": 0.0018789852004402641,
      "loss": 2.9333,
      "step": 1434
    },
    {
      "epoch": 0.18366824523230513,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0018787874387843355,
      "loss": 3.843,
      "step": 1435
    },
    {
      "epoch": 0.1837962370408294,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0018785895260945275,
      "loss": 2.3381,
      "step": 1436
    },
    {
      "epoch": 0.18392422884935364,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0018783914624048545,
      "loss": 2.3582,
      "step": 1437
    },
    {
      "epoch": 0.1840522206578779,
      "grad_norm": 0.65625,
      "learning_rate": 0.001878193247749357,
      "loss": 2.1914,
      "step": 1438
    },
    {
      "epoch": 0.18418021246640215,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0018779948821621016,
      "loss": 2.2368,
      "step": 1439
    },
    {
      "epoch": 0.18430820427492642,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00187779636567718,
      "loss": 2.1871,
      "step": 1440
    },
    {
      "epoch": 0.18443619608345066,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0018775976983287114,
      "loss": 3.151,
      "step": 1441
    },
    {
      "epoch": 0.1845641878919749,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0018773988801508395,
      "loss": 3.6242,
      "step": 1442
    },
    {
      "epoch": 0.18469217970049917,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0018771999111777344,
      "loss": 2.7166,
      "step": 1443
    },
    {
      "epoch": 0.1848201715090234,
      "grad_norm": 0.76953125,
      "learning_rate": 0.001877000791443592,
      "loss": 3.1726,
      "step": 1444
    },
    {
      "epoch": 0.18494816331754768,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0018768015209826346,
      "loss": 2.6089,
      "step": 1445
    },
    {
      "epoch": 0.18507615512607192,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00187660209982911,
      "loss": 1.9994,
      "step": 1446
    },
    {
      "epoch": 0.1852041469345962,
      "grad_norm": 0.703125,
      "learning_rate": 0.0018764025280172919,
      "loss": 2.4597,
      "step": 1447
    },
    {
      "epoch": 0.18533213874312043,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0018762028055814798,
      "loss": 3.0217,
      "step": 1448
    },
    {
      "epoch": 0.1854601305516447,
      "grad_norm": 0.765625,
      "learning_rate": 0.0018760029325559993,
      "loss": 2.5456,
      "step": 1449
    },
    {
      "epoch": 0.18558812236016894,
      "grad_norm": 0.69140625,
      "learning_rate": 0.001875802908975202,
      "loss": 2.3325,
      "step": 1450
    },
    {
      "epoch": 0.1857161141686932,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0018756027348734651,
      "loss": 2.3227,
      "step": 1451
    },
    {
      "epoch": 0.18584410597721746,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0018754024102851919,
      "loss": 2.4312,
      "step": 1452
    },
    {
      "epoch": 0.18597209778574172,
      "grad_norm": 0.875,
      "learning_rate": 0.001875201935244811,
      "loss": 2.9671,
      "step": 1453
    },
    {
      "epoch": 0.18610008959426597,
      "grad_norm": 0.9609375,
      "learning_rate": 0.001875001309786778,
      "loss": 2.3042,
      "step": 1454
    },
    {
      "epoch": 0.1862280814027902,
      "grad_norm": 0.65625,
      "learning_rate": 0.0018748005339455731,
      "loss": 2.316,
      "step": 1455
    },
    {
      "epoch": 0.18635607321131448,
      "grad_norm": 0.859375,
      "learning_rate": 0.001874599607755703,
      "loss": 3.209,
      "step": 1456
    },
    {
      "epoch": 0.18648406501983872,
      "grad_norm": 0.765625,
      "learning_rate": 0.0018743985312517,
      "loss": 2.3886,
      "step": 1457
    },
    {
      "epoch": 0.186612056828363,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0018741973044681223,
      "loss": 2.1587,
      "step": 1458
    },
    {
      "epoch": 0.18674004863688723,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0018739959274395549,
      "loss": 2.5026,
      "step": 1459
    },
    {
      "epoch": 0.1868680404454115,
      "grad_norm": 1.046875,
      "learning_rate": 0.0018737944002006062,
      "loss": 3.3803,
      "step": 1460
    },
    {
      "epoch": 0.18699603225393574,
      "grad_norm": 0.671875,
      "learning_rate": 0.0018735927227859132,
      "loss": 3.0664,
      "step": 1461
    },
    {
      "epoch": 0.18712402406246,
      "grad_norm": 0.78125,
      "learning_rate": 0.001873390895230137,
      "loss": 3.0797,
      "step": 1462
    },
    {
      "epoch": 0.18725201587098425,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0018731889175679646,
      "loss": 2.2693,
      "step": 1463
    },
    {
      "epoch": 0.18738000767950852,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0018729867898341097,
      "loss": 3.7972,
      "step": 1464
    },
    {
      "epoch": 0.18750799948803276,
      "grad_norm": 0.734375,
      "learning_rate": 0.0018727845120633108,
      "loss": 1.7963,
      "step": 1465
    },
    {
      "epoch": 0.18763599129655703,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0018725820842903328,
      "loss": 2.7796,
      "step": 1466
    },
    {
      "epoch": 0.18776398310508127,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0018723795065499662,
      "loss": 2.2446,
      "step": 1467
    },
    {
      "epoch": 0.18789197491360554,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0018721767788770275,
      "loss": 1.9024,
      "step": 1468
    },
    {
      "epoch": 0.18801996672212978,
      "grad_norm": 1.2890625,
      "learning_rate": 0.001871973901306358,
      "loss": 2.7457,
      "step": 1469
    },
    {
      "epoch": 0.18814795853065402,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0018717708738728263,
      "loss": 2.1508,
      "step": 1470
    },
    {
      "epoch": 0.1882759503391783,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0018715676966113255,
      "loss": 3.3466,
      "step": 1471
    },
    {
      "epoch": 0.18840394214770254,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0018713643695567753,
      "loss": 3.7621,
      "step": 1472
    },
    {
      "epoch": 0.1885319339562268,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0018711608927441201,
      "loss": 2.8481,
      "step": 1473
    },
    {
      "epoch": 0.18865992576475105,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0018709572662083316,
      "loss": 2.7465,
      "step": 1474
    },
    {
      "epoch": 0.18878791757327532,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0018707534899844057,
      "loss": 2.9764,
      "step": 1475
    },
    {
      "epoch": 0.18891590938179956,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0018705495641073645,
      "loss": 2.4355,
      "step": 1476
    },
    {
      "epoch": 0.18904390119032383,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0018703454886122566,
      "loss": 3.0336,
      "step": 1477
    },
    {
      "epoch": 0.18917189299884807,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0018701412635341553,
      "loss": 3.1863,
      "step": 1478
    },
    {
      "epoch": 0.18929988480737234,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0018699368889081597,
      "loss": 2.3703,
      "step": 1479
    },
    {
      "epoch": 0.18942787661589658,
      "grad_norm": 0.90625,
      "learning_rate": 0.0018697323647693955,
      "loss": 2.6363,
      "step": 1480
    },
    {
      "epoch": 0.18955586842442085,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0018695276911530132,
      "loss": 1.9722,
      "step": 1481
    },
    {
      "epoch": 0.1896838602329451,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0018693228680941893,
      "loss": 3.1991,
      "step": 1482
    },
    {
      "epoch": 0.18981185204146936,
      "grad_norm": 0.91015625,
      "learning_rate": 0.001869117895628126,
      "loss": 2.3034,
      "step": 1483
    },
    {
      "epoch": 0.1899398438499936,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0018689127737900508,
      "loss": 2.9374,
      "step": 1484
    },
    {
      "epoch": 0.19006783565851784,
      "grad_norm": 0.703125,
      "learning_rate": 0.001868707502615218,
      "loss": 2.7309,
      "step": 1485
    },
    {
      "epoch": 0.1901958274670421,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0018685020821389059,
      "loss": 2.6579,
      "step": 1486
    },
    {
      "epoch": 0.19032381927556635,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00186829651239642,
      "loss": 2.4343,
      "step": 1487
    },
    {
      "epoch": 0.19045181108409062,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0018680907934230905,
      "loss": 2.795,
      "step": 1488
    },
    {
      "epoch": 0.19057980289261486,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0018678849252542734,
      "loss": 2.5973,
      "step": 1489
    },
    {
      "epoch": 0.19070779470113913,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0018676789079253508,
      "loss": 2.3545,
      "step": 1490
    },
    {
      "epoch": 0.19083578650966337,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0018674727414717302,
      "loss": 2.1406,
      "step": 1491
    },
    {
      "epoch": 0.19096377831818764,
      "grad_norm": 1.0703125,
      "learning_rate": 0.001867266425928844,
      "loss": 2.5358,
      "step": 1492
    },
    {
      "epoch": 0.19109177012671189,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0018670599613321518,
      "loss": 2.5614,
      "step": 1493
    },
    {
      "epoch": 0.19121976193523615,
      "grad_norm": 0.8125,
      "learning_rate": 0.001866853347717137,
      "loss": 2.8529,
      "step": 1494
    },
    {
      "epoch": 0.1913477537437604,
      "grad_norm": 0.78125,
      "learning_rate": 0.0018666465851193102,
      "loss": 2.5765,
      "step": 1495
    },
    {
      "epoch": 0.19147574555228467,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0018664396735742065,
      "loss": 2.7882,
      "step": 1496
    },
    {
      "epoch": 0.1916037373608089,
      "grad_norm": 0.65234375,
      "learning_rate": 0.001866232613117387,
      "loss": 2.9408,
      "step": 1497
    },
    {
      "epoch": 0.19173172916933318,
      "grad_norm": 0.59375,
      "learning_rate": 0.001866025403784439,
      "loss": 1.8605,
      "step": 1498
    },
    {
      "epoch": 0.19185972097785742,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0018658180456109738,
      "loss": 2.6042,
      "step": 1499
    },
    {
      "epoch": 0.19198771278638166,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0018656105386326298,
      "loss": 2.9134,
      "step": 1500
    },
    {
      "epoch": 0.19211570459490593,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0018654028828850703,
      "loss": 2.7119,
      "step": 1501
    },
    {
      "epoch": 0.19224369640343017,
      "grad_norm": 0.671875,
      "learning_rate": 0.0018651950784039847,
      "loss": 2.3791,
      "step": 1502
    },
    {
      "epoch": 0.19237168821195444,
      "grad_norm": 0.765625,
      "learning_rate": 0.0018649871252250875,
      "loss": 2.3598,
      "step": 1503
    },
    {
      "epoch": 0.19249968002047868,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0018647790233841182,
      "loss": 2.6392,
      "step": 1504
    },
    {
      "epoch": 0.19262767182900295,
      "grad_norm": 0.84765625,
      "learning_rate": 0.001864570772916843,
      "loss": 2.8219,
      "step": 1505
    },
    {
      "epoch": 0.1927556636375272,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0018643623738590534,
      "loss": 2.5432,
      "step": 1506
    },
    {
      "epoch": 0.19288365544605146,
      "grad_norm": 0.6875,
      "learning_rate": 0.0018641538262465654,
      "loss": 2.167,
      "step": 1507
    },
    {
      "epoch": 0.1930116472545757,
      "grad_norm": 0.875,
      "learning_rate": 0.0018639451301152216,
      "loss": 2.97,
      "step": 1508
    },
    {
      "epoch": 0.19313963906309997,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0018637362855008902,
      "loss": 3.1321,
      "step": 1509
    },
    {
      "epoch": 0.1932676308716242,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0018635272924394642,
      "loss": 2.8977,
      "step": 1510
    },
    {
      "epoch": 0.19339562268014848,
      "grad_norm": 0.921875,
      "learning_rate": 0.0018633181509668622,
      "loss": 3.0134,
      "step": 1511
    },
    {
      "epoch": 0.19352361448867272,
      "grad_norm": 0.68359375,
      "learning_rate": 0.001863108861119029,
      "loss": 2.2776,
      "step": 1512
    },
    {
      "epoch": 0.193651606297197,
      "grad_norm": 0.70703125,
      "learning_rate": 0.001862899422931934,
      "loss": 2.6121,
      "step": 1513
    },
    {
      "epoch": 0.19377959810572123,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0018626898364415732,
      "loss": 2.5245,
      "step": 1514
    },
    {
      "epoch": 0.19390758991424548,
      "grad_norm": 0.8828125,
      "learning_rate": 0.001862480101683967,
      "loss": 2.4596,
      "step": 1515
    },
    {
      "epoch": 0.19403558172276975,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0018622702186951617,
      "loss": 2.3409,
      "step": 1516
    },
    {
      "epoch": 0.194163573531294,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0018620601875112292,
      "loss": 2.0608,
      "step": 1517
    },
    {
      "epoch": 0.19429156533981826,
      "grad_norm": 0.84375,
      "learning_rate": 0.001861850008168267,
      "loss": 3.1202,
      "step": 1518
    },
    {
      "epoch": 0.1944195571483425,
      "grad_norm": 0.984375,
      "learning_rate": 0.0018616396807023974,
      "loss": 3.7863,
      "step": 1519
    },
    {
      "epoch": 0.19454754895686677,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0018614292051497686,
      "loss": 3.2588,
      "step": 1520
    },
    {
      "epoch": 0.194675540765391,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0018612185815465546,
      "loss": 2.4066,
      "step": 1521
    },
    {
      "epoch": 0.19480353257391528,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0018610078099289542,
      "loss": 3.1907,
      "step": 1522
    },
    {
      "epoch": 0.19493152438243952,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0018607968903331922,
      "loss": 3.1121,
      "step": 1523
    },
    {
      "epoch": 0.1950595161909638,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0018605858227955185,
      "loss": 1.7815,
      "step": 1524
    },
    {
      "epoch": 0.19518750799948803,
      "grad_norm": 0.6640625,
      "learning_rate": 0.001860374607352208,
      "loss": 2.689,
      "step": 1525
    },
    {
      "epoch": 0.1953154998080123,
      "grad_norm": 0.63671875,
      "learning_rate": 0.001860163244039562,
      "loss": 2.8225,
      "step": 1526
    },
    {
      "epoch": 0.19544349161653654,
      "grad_norm": 0.90625,
      "learning_rate": 0.0018599517328939067,
      "loss": 3.5599,
      "step": 1527
    },
    {
      "epoch": 0.19557148342506078,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0018597400739515932,
      "loss": 2.2174,
      "step": 1528
    },
    {
      "epoch": 0.19569947523358505,
      "grad_norm": 0.625,
      "learning_rate": 0.001859528267248999,
      "loss": 2.305,
      "step": 1529
    },
    {
      "epoch": 0.1958274670421093,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0018593163128225267,
      "loss": 2.0925,
      "step": 1530
    },
    {
      "epoch": 0.19595545885063356,
      "grad_norm": 0.671875,
      "learning_rate": 0.0018591042107086039,
      "loss": 2.0219,
      "step": 1531
    },
    {
      "epoch": 0.1960834506591578,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0018588919609436837,
      "loss": 2.6427,
      "step": 1532
    },
    {
      "epoch": 0.19621144246768207,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0018586795635642443,
      "loss": 2.4987,
      "step": 1533
    },
    {
      "epoch": 0.19633943427620631,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0018584670186067905,
      "loss": 1.7306,
      "step": 1534
    },
    {
      "epoch": 0.19646742608473058,
      "grad_norm": 0.65625,
      "learning_rate": 0.0018582543261078507,
      "loss": 2.872,
      "step": 1535
    },
    {
      "epoch": 0.19659541789325483,
      "grad_norm": 0.6875,
      "learning_rate": 0.0018580414861039804,
      "loss": 3.0412,
      "step": 1536
    },
    {
      "epoch": 0.1967234097017791,
      "grad_norm": 0.83984375,
      "learning_rate": 0.001857828498631759,
      "loss": 2.3755,
      "step": 1537
    },
    {
      "epoch": 0.19685140151030334,
      "grad_norm": 0.86328125,
      "learning_rate": 0.001857615363727792,
      "loss": 2.9821,
      "step": 1538
    },
    {
      "epoch": 0.1969793933188276,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0018574020814287103,
      "loss": 2.4472,
      "step": 1539
    },
    {
      "epoch": 0.19710738512735185,
      "grad_norm": 0.78125,
      "learning_rate": 0.0018571886517711699,
      "loss": 2.8286,
      "step": 1540
    },
    {
      "epoch": 0.19723537693587612,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0018569750747918516,
      "loss": 2.4344,
      "step": 1541
    },
    {
      "epoch": 0.19736336874440036,
      "grad_norm": 0.625,
      "learning_rate": 0.0018567613505274625,
      "loss": 2.3426,
      "step": 1542
    },
    {
      "epoch": 0.1974913605529246,
      "grad_norm": 0.7421875,
      "learning_rate": 0.001856547479014735,
      "loss": 2.6438,
      "step": 1543
    },
    {
      "epoch": 0.19761935236144887,
      "grad_norm": 0.84375,
      "learning_rate": 0.0018563334602904255,
      "loss": 3.1213,
      "step": 1544
    },
    {
      "epoch": 0.1977473441699731,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0018561192943913171,
      "loss": 2.881,
      "step": 1545
    },
    {
      "epoch": 0.19787533597849738,
      "grad_norm": 0.7109375,
      "learning_rate": 0.001855904981354218,
      "loss": 2.9395,
      "step": 1546
    },
    {
      "epoch": 0.19800332778702162,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0018556905212159607,
      "loss": 2.3717,
      "step": 1547
    },
    {
      "epoch": 0.1981313195955459,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0018554759140134041,
      "loss": 2.3607,
      "step": 1548
    },
    {
      "epoch": 0.19825931140407013,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0018552611597834317,
      "loss": 2.8245,
      "step": 1549
    },
    {
      "epoch": 0.1983873032125944,
      "grad_norm": 0.765625,
      "learning_rate": 0.0018550462585629527,
      "loss": 2.5345,
      "step": 1550
    },
    {
      "epoch": 0.19851529502111864,
      "grad_norm": 0.76953125,
      "learning_rate": 0.001854831210388901,
      "loss": 2.9771,
      "step": 1551
    },
    {
      "epoch": 0.1986432868296429,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0018546160152982365,
      "loss": 2.7484,
      "step": 1552
    },
    {
      "epoch": 0.19877127863816715,
      "grad_norm": 0.7421875,
      "learning_rate": 0.001854400673327944,
      "loss": 2.4425,
      "step": 1553
    },
    {
      "epoch": 0.19889927044669142,
      "grad_norm": 0.859375,
      "learning_rate": 0.0018541851845150332,
      "loss": 3.3745,
      "step": 1554
    },
    {
      "epoch": 0.19902726225521566,
      "grad_norm": 0.71875,
      "learning_rate": 0.0018539695488965396,
      "loss": 2.3475,
      "step": 1555
    },
    {
      "epoch": 0.19915525406373993,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0018537537665095233,
      "loss": 2.9094,
      "step": 1556
    },
    {
      "epoch": 0.19928324587226418,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0018535378373910705,
      "loss": 2.8408,
      "step": 1557
    },
    {
      "epoch": 0.19941123768078842,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0018533217615782918,
      "loss": 2.2625,
      "step": 1558
    },
    {
      "epoch": 0.19953922948931269,
      "grad_norm": 0.953125,
      "learning_rate": 0.0018531055391083236,
      "loss": 3.0588,
      "step": 1559
    },
    {
      "epoch": 0.19966722129783693,
      "grad_norm": 0.703125,
      "learning_rate": 0.0018528891700183268,
      "loss": 2.421,
      "step": 1560
    },
    {
      "epoch": 0.1997952131063612,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0018526726543454883,
      "loss": 2.2038,
      "step": 1561
    },
    {
      "epoch": 0.19992320491488544,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00185245599212702,
      "loss": 3.2602,
      "step": 1562
    },
    {
      "epoch": 0.2000511967234097,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0018522391834001582,
      "loss": 3.0152,
      "step": 1563
    },
    {
      "epoch": 0.20017918853193395,
      "grad_norm": 0.671875,
      "learning_rate": 0.0018520222282021655,
      "loss": 2.5999,
      "step": 1564
    },
    {
      "epoch": 0.20030718034045822,
      "grad_norm": 0.796875,
      "learning_rate": 0.001851805126570329,
      "loss": 2.734,
      "step": 1565
    },
    {
      "epoch": 0.20043517214898246,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0018515878785419612,
      "loss": 2.8684,
      "step": 1566
    },
    {
      "epoch": 0.20056316395750673,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0018513704841543997,
      "loss": 2.5574,
      "step": 1567
    },
    {
      "epoch": 0.20069115576603097,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0018511529434450075,
      "loss": 2.9679,
      "step": 1568
    },
    {
      "epoch": 0.20081914757455524,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0018509352564511717,
      "loss": 2.5509,
      "step": 1569
    },
    {
      "epoch": 0.20094713938307948,
      "grad_norm": 0.65234375,
      "learning_rate": 0.001850717423210306,
      "loss": 2.1389,
      "step": 1570
    },
    {
      "epoch": 0.20107513119160375,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0018504994437598486,
      "loss": 3.0495,
      "step": 1571
    },
    {
      "epoch": 0.201203123000128,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0018502813181372627,
      "loss": 2.8328,
      "step": 1572
    },
    {
      "epoch": 0.20133111480865223,
      "grad_norm": 0.796875,
      "learning_rate": 0.0018500630463800364,
      "loss": 2.8888,
      "step": 1573
    },
    {
      "epoch": 0.2014591066171765,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0018498446285256836,
      "loss": 1.7912,
      "step": 1574
    },
    {
      "epoch": 0.20158709842570074,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0018496260646117429,
      "loss": 2.1146,
      "step": 1575
    },
    {
      "epoch": 0.20171509023422501,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0018494073546757779,
      "loss": 2.7042,
      "step": 1576
    },
    {
      "epoch": 0.20184308204274926,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0018491884987553778,
      "loss": 2.4871,
      "step": 1577
    },
    {
      "epoch": 0.20197107385127352,
      "grad_norm": 0.66015625,
      "learning_rate": 0.001848969496888156,
      "loss": 2.5486,
      "step": 1578
    },
    {
      "epoch": 0.20209906565979777,
      "grad_norm": 0.6875,
      "learning_rate": 0.0018487503491117522,
      "loss": 2.3281,
      "step": 1579
    },
    {
      "epoch": 0.20222705746832204,
      "grad_norm": 0.75,
      "learning_rate": 0.00184853105546383,
      "loss": 2.8293,
      "step": 1580
    },
    {
      "epoch": 0.20235504927684628,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0018483116159820785,
      "loss": 1.8803,
      "step": 1581
    },
    {
      "epoch": 0.20248304108537055,
      "grad_norm": 0.703125,
      "learning_rate": 0.0018480920307042124,
      "loss": 2.9437,
      "step": 1582
    },
    {
      "epoch": 0.2026110328938948,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0018478722996679707,
      "loss": 2.1813,
      "step": 1583
    },
    {
      "epoch": 0.20273902470241906,
      "grad_norm": 0.71875,
      "learning_rate": 0.0018476524229111178,
      "loss": 2.6455,
      "step": 1584
    },
    {
      "epoch": 0.2028670165109433,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0018474324004714429,
      "loss": 2.2491,
      "step": 1585
    },
    {
      "epoch": 0.20299500831946754,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0018472122323867609,
      "loss": 2.0946,
      "step": 1586
    },
    {
      "epoch": 0.2031230001279918,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0018469919186949104,
      "loss": 1.9669,
      "step": 1587
    },
    {
      "epoch": 0.20325099193651605,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0018467714594337568,
      "loss": 2.7646,
      "step": 1588
    },
    {
      "epoch": 0.20337898374504032,
      "grad_norm": 0.703125,
      "learning_rate": 0.0018465508546411894,
      "loss": 3.0168,
      "step": 1589
    },
    {
      "epoch": 0.20350697555356456,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0018463301043551224,
      "loss": 3.0956,
      "step": 1590
    },
    {
      "epoch": 0.20363496736208883,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0018461092086134953,
      "loss": 3.3363,
      "step": 1591
    },
    {
      "epoch": 0.20376295917061307,
      "grad_norm": 0.828125,
      "learning_rate": 0.0018458881674542728,
      "loss": 2.804,
      "step": 1592
    },
    {
      "epoch": 0.20389095097913734,
      "grad_norm": 0.6875,
      "learning_rate": 0.0018456669809154445,
      "loss": 2.6905,
      "step": 1593
    },
    {
      "epoch": 0.20401894278766158,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0018454456490350252,
      "loss": 3.2209,
      "step": 1594
    },
    {
      "epoch": 0.20414693459618585,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0018452241718510533,
      "loss": 2.2401,
      "step": 1595
    },
    {
      "epoch": 0.2042749264047101,
      "grad_norm": 0.921875,
      "learning_rate": 0.001845002549401594,
      "loss": 2.722,
      "step": 1596
    },
    {
      "epoch": 0.20440291821323436,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0018447807817247369,
      "loss": 2.3071,
      "step": 1597
    },
    {
      "epoch": 0.2045309100217586,
      "grad_norm": 0.60546875,
      "learning_rate": 0.001844558868858596,
      "loss": 2.1875,
      "step": 1598
    },
    {
      "epoch": 0.20465890183028287,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0018443368108413108,
      "loss": 2.3423,
      "step": 1599
    },
    {
      "epoch": 0.20478689363880712,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0018441146077110453,
      "loss": 2.0963,
      "step": 1600
    },
    {
      "epoch": 0.20491488544733136,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0018438922595059892,
      "loss": 2.5987,
      "step": 1601
    },
    {
      "epoch": 0.20504287725585563,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0018436697662643558,
      "loss": 2.6389,
      "step": 1602
    },
    {
      "epoch": 0.20517086906437987,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0018434471280243852,
      "loss": 2.5303,
      "step": 1603
    },
    {
      "epoch": 0.20529886087290414,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0018432243448243408,
      "loss": 2.1759,
      "step": 1604
    },
    {
      "epoch": 0.20542685268142838,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0018430014167025118,
      "loss": 2.9475,
      "step": 1605
    },
    {
      "epoch": 0.20555484448995265,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0018427783436972115,
      "loss": 2.3167,
      "step": 1606
    },
    {
      "epoch": 0.2056828362984769,
      "grad_norm": 0.8125,
      "learning_rate": 0.0018425551258467792,
      "loss": 1.8225,
      "step": 1607
    },
    {
      "epoch": 0.20581082810700116,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0018423317631895782,
      "loss": 2.9559,
      "step": 1608
    },
    {
      "epoch": 0.2059388199155254,
      "grad_norm": 0.609375,
      "learning_rate": 0.0018421082557639973,
      "loss": 2.3132,
      "step": 1609
    },
    {
      "epoch": 0.20606681172404967,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0018418846036084493,
      "loss": 2.0487,
      "step": 1610
    },
    {
      "epoch": 0.2061948035325739,
      "grad_norm": 0.98828125,
      "learning_rate": 0.001841660806761373,
      "loss": 2.7038,
      "step": 1611
    },
    {
      "epoch": 0.20632279534109818,
      "grad_norm": 0.75,
      "learning_rate": 0.001841436865261231,
      "loss": 3.0643,
      "step": 1612
    },
    {
      "epoch": 0.20645078714962242,
      "grad_norm": 0.69921875,
      "learning_rate": 0.001841212779146512,
      "loss": 2.4436,
      "step": 1613
    },
    {
      "epoch": 0.2065787789581467,
      "grad_norm": 1.21875,
      "learning_rate": 0.0018409885484557283,
      "loss": 2.6367,
      "step": 1614
    },
    {
      "epoch": 0.20670677076667093,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0018407641732274174,
      "loss": 2.1925,
      "step": 1615
    },
    {
      "epoch": 0.20683476257519517,
      "grad_norm": 0.6875,
      "learning_rate": 0.0018405396535001426,
      "loss": 2.7517,
      "step": 1616
    },
    {
      "epoch": 0.20696275438371944,
      "grad_norm": 0.75,
      "learning_rate": 0.0018403149893124904,
      "loss": 2.9116,
      "step": 1617
    },
    {
      "epoch": 0.20709074619224369,
      "grad_norm": 0.703125,
      "learning_rate": 0.001840090180703074,
      "loss": 2.5406,
      "step": 1618
    },
    {
      "epoch": 0.20721873800076795,
      "grad_norm": 0.640625,
      "learning_rate": 0.0018398652277105292,
      "loss": 2.3838,
      "step": 1619
    },
    {
      "epoch": 0.2073467298092922,
      "grad_norm": 0.6875,
      "learning_rate": 0.0018396401303735184,
      "loss": 2.0291,
      "step": 1620
    },
    {
      "epoch": 0.20747472161781647,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0018394148887307285,
      "loss": 2.233,
      "step": 1621
    },
    {
      "epoch": 0.2076027134263407,
      "grad_norm": 0.828125,
      "learning_rate": 0.0018391895028208704,
      "loss": 3.2632,
      "step": 1622
    },
    {
      "epoch": 0.20773070523486498,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0018389639726826808,
      "loss": 2.7884,
      "step": 1623
    },
    {
      "epoch": 0.20785869704338922,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0018387382983549205,
      "loss": 2.2448,
      "step": 1624
    },
    {
      "epoch": 0.2079866888519135,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0018385124798763752,
      "loss": 2.3446,
      "step": 1625
    },
    {
      "epoch": 0.20811468066043773,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0018382865172858558,
      "loss": 2.4702,
      "step": 1626
    },
    {
      "epoch": 0.208242672468962,
      "grad_norm": 0.65625,
      "learning_rate": 0.0018380604106221971,
      "loss": 2.4118,
      "step": 1627
    },
    {
      "epoch": 0.20837066427748624,
      "grad_norm": 0.734375,
      "learning_rate": 0.00183783415992426,
      "loss": 2.8217,
      "step": 1628
    },
    {
      "epoch": 0.2084986560860105,
      "grad_norm": 0.71875,
      "learning_rate": 0.0018376077652309285,
      "loss": 2.3049,
      "step": 1629
    },
    {
      "epoch": 0.20862664789453475,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0018373812265811125,
      "loss": 3.035,
      "step": 1630
    },
    {
      "epoch": 0.208754639703059,
      "grad_norm": 0.609375,
      "learning_rate": 0.0018371545440137465,
      "loss": 2.3663,
      "step": 1631
    },
    {
      "epoch": 0.20888263151158326,
      "grad_norm": 0.84375,
      "learning_rate": 0.0018369277175677894,
      "loss": 1.9331,
      "step": 1632
    },
    {
      "epoch": 0.2090106233201075,
      "grad_norm": 0.671875,
      "learning_rate": 0.0018367007472822252,
      "loss": 2.5558,
      "step": 1633
    },
    {
      "epoch": 0.20913861512863177,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0018364736331960623,
      "loss": 1.6706,
      "step": 1634
    },
    {
      "epoch": 0.209266606937156,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0018362463753483337,
      "loss": 2.6203,
      "step": 1635
    },
    {
      "epoch": 0.20939459874568028,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0018360189737780981,
      "loss": 2.2446,
      "step": 1636
    },
    {
      "epoch": 0.20952259055420452,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0018357914285244373,
      "loss": 2.0096,
      "step": 1637
    },
    {
      "epoch": 0.2096505823627288,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0018355637396264594,
      "loss": 2.3997,
      "step": 1638
    },
    {
      "epoch": 0.20977857417125303,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0018353359071232953,
      "loss": 2.318,
      "step": 1639
    },
    {
      "epoch": 0.2099065659797773,
      "grad_norm": 0.796875,
      "learning_rate": 0.001835107931054103,
      "loss": 2.1892,
      "step": 1640
    },
    {
      "epoch": 0.21003455778830155,
      "grad_norm": 0.7578125,
      "learning_rate": 0.001834879811458063,
      "loss": 2.743,
      "step": 1641
    },
    {
      "epoch": 0.21016254959682582,
      "grad_norm": 0.68359375,
      "learning_rate": 0.001834651548374382,
      "loss": 2.3935,
      "step": 1642
    },
    {
      "epoch": 0.21029054140535006,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00183442314184229,
      "loss": 2.422,
      "step": 1643
    },
    {
      "epoch": 0.21041853321387433,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0018341945919010433,
      "loss": 2.4298,
      "step": 1644
    },
    {
      "epoch": 0.21054652502239857,
      "grad_norm": 0.9765625,
      "learning_rate": 0.001833965898589921,
      "loss": 2.5201,
      "step": 1645
    },
    {
      "epoch": 0.2106745168309228,
      "grad_norm": 0.85546875,
      "learning_rate": 0.001833737061948228,
      "loss": 3.0443,
      "step": 1646
    },
    {
      "epoch": 0.21080250863944708,
      "grad_norm": 0.7578125,
      "learning_rate": 0.001833508082015294,
      "loss": 2.2609,
      "step": 1647
    },
    {
      "epoch": 0.21093050044797132,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0018332789588304725,
      "loss": 2.3134,
      "step": 1648
    },
    {
      "epoch": 0.2110584922564956,
      "grad_norm": 0.8125,
      "learning_rate": 0.0018330496924331423,
      "loss": 2.7518,
      "step": 1649
    },
    {
      "epoch": 0.21118648406501983,
      "grad_norm": 0.6875,
      "learning_rate": 0.0018328202828627065,
      "loss": 2.1195,
      "step": 1650
    },
    {
      "epoch": 0.2113144758735441,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0018325907301585925,
      "loss": 2.7252,
      "step": 1651
    },
    {
      "epoch": 0.21144246768206834,
      "grad_norm": 0.671875,
      "learning_rate": 0.001832361034360253,
      "loss": 2.8943,
      "step": 1652
    },
    {
      "epoch": 0.2115704594905926,
      "grad_norm": 0.7109375,
      "learning_rate": 0.001832131195507165,
      "loss": 2.7617,
      "step": 1653
    },
    {
      "epoch": 0.21169845129911685,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00183190121363883,
      "loss": 2.2553,
      "step": 1654
    },
    {
      "epoch": 0.21182644310764112,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0018316710887947738,
      "loss": 2.5289,
      "step": 1655
    },
    {
      "epoch": 0.21195443491616536,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0018314408210145477,
      "loss": 2.7355,
      "step": 1656
    },
    {
      "epoch": 0.21208242672468963,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0018312104103377263,
      "loss": 2.4167,
      "step": 1657
    },
    {
      "epoch": 0.21221041853321387,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0018309798568039099,
      "loss": 2.6566,
      "step": 1658
    },
    {
      "epoch": 0.21233841034173812,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0018307491604527223,
      "loss": 2.1224,
      "step": 1659
    },
    {
      "epoch": 0.21246640215026238,
      "grad_norm": 0.625,
      "learning_rate": 0.001830518321323813,
      "loss": 2.2352,
      "step": 1660
    },
    {
      "epoch": 0.21259439395878663,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0018302873394568553,
      "loss": 1.8442,
      "step": 1661
    },
    {
      "epoch": 0.2127223857673109,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0018300562148915467,
      "loss": 2.8491,
      "step": 1662
    },
    {
      "epoch": 0.21285037757583514,
      "grad_norm": 0.84375,
      "learning_rate": 0.0018298249476676105,
      "loss": 2.6131,
      "step": 1663
    },
    {
      "epoch": 0.2129783693843594,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0018295935378247932,
      "loss": 2.3656,
      "step": 1664
    },
    {
      "epoch": 0.21310636119288365,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0018293619854028666,
      "loss": 2.1728,
      "step": 1665
    },
    {
      "epoch": 0.21323435300140792,
      "grad_norm": 0.8125,
      "learning_rate": 0.0018291302904416265,
      "loss": 2.6414,
      "step": 1666
    },
    {
      "epoch": 0.21336234480993216,
      "grad_norm": 0.9375,
      "learning_rate": 0.0018288984529808937,
      "loss": 2.4337,
      "step": 1667
    },
    {
      "epoch": 0.21349033661845643,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0018286664730605127,
      "loss": 1.952,
      "step": 1668
    },
    {
      "epoch": 0.21361832842698067,
      "grad_norm": 0.8046875,
      "learning_rate": 0.001828434350720354,
      "loss": 2.8447,
      "step": 1669
    },
    {
      "epoch": 0.21374632023550494,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0018282020860003105,
      "loss": 2.0582,
      "step": 1670
    },
    {
      "epoch": 0.21387431204402918,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0018279696789403014,
      "loss": 2.3149,
      "step": 1671
    },
    {
      "epoch": 0.21400230385255345,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0018277371295802691,
      "loss": 2.3914,
      "step": 1672
    },
    {
      "epoch": 0.2141302956610777,
      "grad_norm": 0.671875,
      "learning_rate": 0.0018275044379601813,
      "loss": 2.0256,
      "step": 1673
    },
    {
      "epoch": 0.21425828746960193,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00182727160412003,
      "loss": 3.1432,
      "step": 1674
    },
    {
      "epoch": 0.2143862792781262,
      "grad_norm": 0.65625,
      "learning_rate": 0.001827038628099831,
      "loss": 2.5148,
      "step": 1675
    },
    {
      "epoch": 0.21451427108665044,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0018268055099396254,
      "loss": 2.1723,
      "step": 1676
    },
    {
      "epoch": 0.2146422628951747,
      "grad_norm": 0.7421875,
      "learning_rate": 0.001826572249679478,
      "loss": 2.1861,
      "step": 1677
    },
    {
      "epoch": 0.21477025470369895,
      "grad_norm": 0.703125,
      "learning_rate": 0.0018263388473594789,
      "loss": 2.1927,
      "step": 1678
    },
    {
      "epoch": 0.21489824651222322,
      "grad_norm": 0.734375,
      "learning_rate": 0.0018261053030197414,
      "loss": 2.2532,
      "step": 1679
    },
    {
      "epoch": 0.21502623832074746,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0018258716167004045,
      "loss": 1.9148,
      "step": 1680
    },
    {
      "epoch": 0.21515423012927173,
      "grad_norm": 0.78125,
      "learning_rate": 0.0018256377884416307,
      "loss": 2.2142,
      "step": 1681
    },
    {
      "epoch": 0.21528222193779598,
      "grad_norm": 1.078125,
      "learning_rate": 0.0018254038182836069,
      "loss": 3.1401,
      "step": 1682
    },
    {
      "epoch": 0.21541021374632024,
      "grad_norm": 0.9296875,
      "learning_rate": 0.001825169706266545,
      "loss": 2.7409,
      "step": 1683
    },
    {
      "epoch": 0.2155382055548445,
      "grad_norm": 0.79296875,
      "learning_rate": 0.001824935452430681,
      "loss": 1.7577,
      "step": 1684
    },
    {
      "epoch": 0.21566619736336876,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0018247010568162752,
      "loss": 1.7475,
      "step": 1685
    },
    {
      "epoch": 0.215794189171893,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0018244665194636122,
      "loss": 2.1506,
      "step": 1686
    },
    {
      "epoch": 0.21592218098041727,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0018242318404130009,
      "loss": 2.3686,
      "step": 1687
    },
    {
      "epoch": 0.2160501727889415,
      "grad_norm": 0.76171875,
      "learning_rate": 0.001823997019704775,
      "loss": 2.0091,
      "step": 1688
    },
    {
      "epoch": 0.21617816459746575,
      "grad_norm": 0.70703125,
      "learning_rate": 0.001823762057379292,
      "loss": 2.331,
      "step": 1689
    },
    {
      "epoch": 0.21630615640599002,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0018235269534769341,
      "loss": 2.555,
      "step": 1690
    },
    {
      "epoch": 0.21643414821451426,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0018232917080381078,
      "loss": 2.1175,
      "step": 1691
    },
    {
      "epoch": 0.21656214002303853,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0018230563211032438,
      "loss": 2.5716,
      "step": 1692
    },
    {
      "epoch": 0.21669013183156277,
      "grad_norm": 0.703125,
      "learning_rate": 0.001822820792712797,
      "loss": 2.6372,
      "step": 1693
    },
    {
      "epoch": 0.21681812364008704,
      "grad_norm": 0.890625,
      "learning_rate": 0.0018225851229072471,
      "loss": 2.1453,
      "step": 1694
    },
    {
      "epoch": 0.21694611544861128,
      "grad_norm": 0.828125,
      "learning_rate": 0.0018223493117270975,
      "loss": 1.935,
      "step": 1695
    },
    {
      "epoch": 0.21707410725713555,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0018221133592128762,
      "loss": 2.7957,
      "step": 1696
    },
    {
      "epoch": 0.2172020990656598,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0018218772654051353,
      "loss": 2.1966,
      "step": 1697
    },
    {
      "epoch": 0.21733009087418406,
      "grad_norm": 0.68359375,
      "learning_rate": 0.001821641030344452,
      "loss": 2.2411,
      "step": 1698
    },
    {
      "epoch": 0.2174580826827083,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0018214046540714266,
      "loss": 2.3865,
      "step": 1699
    },
    {
      "epoch": 0.21758607449123257,
      "grad_norm": 0.96875,
      "learning_rate": 0.0018211681366266843,
      "loss": 1.9136,
      "step": 1700
    },
    {
      "epoch": 0.21771406629975681,
      "grad_norm": 0.65625,
      "learning_rate": 0.0018209314780508743,
      "loss": 1.9783,
      "step": 1701
    },
    {
      "epoch": 0.21784205810828108,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0018206946783846708,
      "loss": 2.8464,
      "step": 1702
    },
    {
      "epoch": 0.21797004991680533,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0018204577376687708,
      "loss": 1.816,
      "step": 1703
    },
    {
      "epoch": 0.21809804172532957,
      "grad_norm": 0.6796875,
      "learning_rate": 0.001820220655943897,
      "loss": 1.8659,
      "step": 1704
    },
    {
      "epoch": 0.21822603353385384,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0018199834332507953,
      "loss": 2.701,
      "step": 1705
    },
    {
      "epoch": 0.21835402534237808,
      "grad_norm": 0.81640625,
      "learning_rate": 0.001819746069630237,
      "loss": 2.6888,
      "step": 1706
    },
    {
      "epoch": 0.21848201715090235,
      "grad_norm": 0.78125,
      "learning_rate": 0.0018195085651230164,
      "loss": 2.4091,
      "step": 1707
    },
    {
      "epoch": 0.2186100089594266,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0018192709197699527,
      "loss": 2.0764,
      "step": 1708
    },
    {
      "epoch": 0.21873800076795086,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0018190331336118883,
      "loss": 2.4103,
      "step": 1709
    },
    {
      "epoch": 0.2188659925764751,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0018187952066896918,
      "loss": 1.6044,
      "step": 1710
    },
    {
      "epoch": 0.21899398438499937,
      "grad_norm": 0.859375,
      "learning_rate": 0.0018185571390442541,
      "loss": 2.733,
      "step": 1711
    },
    {
      "epoch": 0.2191219761935236,
      "grad_norm": 0.6875,
      "learning_rate": 0.0018183189307164913,
      "loss": 2.6466,
      "step": 1712
    },
    {
      "epoch": 0.21924996800204788,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0018180805817473434,
      "loss": 2.4262,
      "step": 1713
    },
    {
      "epoch": 0.21937795981057212,
      "grad_norm": 0.734375,
      "learning_rate": 0.0018178420921777742,
      "loss": 2.2208,
      "step": 1714
    },
    {
      "epoch": 0.2195059516190964,
      "grad_norm": 0.84375,
      "learning_rate": 0.0018176034620487717,
      "loss": 1.8573,
      "step": 1715
    },
    {
      "epoch": 0.21963394342762063,
      "grad_norm": 0.796875,
      "learning_rate": 0.0018173646914013495,
      "loss": 2.4445,
      "step": 1716
    },
    {
      "epoch": 0.21976193523614487,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0018171257802765432,
      "loss": 2.8525,
      "step": 1717
    },
    {
      "epoch": 0.21988992704466914,
      "grad_norm": 0.84375,
      "learning_rate": 0.001816886728715414,
      "loss": 2.616,
      "step": 1718
    },
    {
      "epoch": 0.22001791885319338,
      "grad_norm": 0.765625,
      "learning_rate": 0.0018166475367590467,
      "loss": 2.9834,
      "step": 1719
    },
    {
      "epoch": 0.22014591066171765,
      "grad_norm": 1.015625,
      "learning_rate": 0.0018164082044485501,
      "loss": 2.4314,
      "step": 1720
    },
    {
      "epoch": 0.2202739024702419,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0018161687318250578,
      "loss": 2.7677,
      "step": 1721
    },
    {
      "epoch": 0.22040189427876616,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0018159291189297267,
      "loss": 2.5941,
      "step": 1722
    },
    {
      "epoch": 0.2205298860872904,
      "grad_norm": 0.640625,
      "learning_rate": 0.0018156893658037383,
      "loss": 2.5202,
      "step": 1723
    },
    {
      "epoch": 0.22065787789581467,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0018154494724882978,
      "loss": 2.3138,
      "step": 1724
    },
    {
      "epoch": 0.22078586970433892,
      "grad_norm": 0.703125,
      "learning_rate": 0.0018152094390246351,
      "loss": 1.8814,
      "step": 1725
    },
    {
      "epoch": 0.22091386151286319,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0018149692654540037,
      "loss": 2.2193,
      "step": 1726
    },
    {
      "epoch": 0.22104185332138743,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0018147289518176818,
      "loss": 2.1171,
      "step": 1727
    },
    {
      "epoch": 0.2211698451299117,
      "grad_norm": 0.65625,
      "learning_rate": 0.0018144884981569703,
      "loss": 2.4629,
      "step": 1728
    },
    {
      "epoch": 0.22129783693843594,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0018142479045131954,
      "loss": 1.9991,
      "step": 1729
    },
    {
      "epoch": 0.2214258287469602,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0018140071709277072,
      "loss": 1.9095,
      "step": 1730
    },
    {
      "epoch": 0.22155382055548445,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0018137662974418798,
      "loss": 3.4107,
      "step": 1731
    },
    {
      "epoch": 0.2216818123640087,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0018135252840971111,
      "loss": 1.5575,
      "step": 1732
    },
    {
      "epoch": 0.22180980417253296,
      "grad_norm": 0.88671875,
      "learning_rate": 0.001813284130934823,
      "loss": 2.7152,
      "step": 1733
    },
    {
      "epoch": 0.2219377959810572,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0018130428379964615,
      "loss": 2.6445,
      "step": 1734
    },
    {
      "epoch": 0.22206578778958147,
      "grad_norm": 0.69140625,
      "learning_rate": 0.001812801405323497,
      "loss": 1.793,
      "step": 1735
    },
    {
      "epoch": 0.2221937795981057,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0018125598329574233,
      "loss": 2.1555,
      "step": 1736
    },
    {
      "epoch": 0.22232177140662998,
      "grad_norm": 0.69140625,
      "learning_rate": 0.001812318120939759,
      "loss": 2.2048,
      "step": 1737
    },
    {
      "epoch": 0.22244976321515422,
      "grad_norm": 0.68359375,
      "learning_rate": 0.001812076269312046,
      "loss": 2.3838,
      "step": 1738
    },
    {
      "epoch": 0.2225777550236785,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00181183427811585,
      "loss": 2.3338,
      "step": 1739
    },
    {
      "epoch": 0.22270574683220273,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0018115921473927617,
      "loss": 2.952,
      "step": 1740
    },
    {
      "epoch": 0.222833738640727,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0018113498771843949,
      "loss": 2.3862,
      "step": 1741
    },
    {
      "epoch": 0.22296173044925124,
      "grad_norm": 0.609375,
      "learning_rate": 0.0018111074675323878,
      "loss": 2.0935,
      "step": 1742
    },
    {
      "epoch": 0.2230897222577755,
      "grad_norm": 0.625,
      "learning_rate": 0.0018108649184784022,
      "loss": 1.8587,
      "step": 1743
    },
    {
      "epoch": 0.22321771406629975,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0018106222300641247,
      "loss": 2.6438,
      "step": 1744
    },
    {
      "epoch": 0.22334570587482402,
      "grad_norm": 0.765625,
      "learning_rate": 0.0018103794023312644,
      "loss": 2.1204,
      "step": 1745
    },
    {
      "epoch": 0.22347369768334827,
      "grad_norm": 1.484375,
      "learning_rate": 0.0018101364353215559,
      "loss": 2.2933,
      "step": 1746
    },
    {
      "epoch": 0.2236016894918725,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0018098933290767566,
      "loss": 2.2464,
      "step": 1747
    },
    {
      "epoch": 0.22372968130039678,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0018096500836386484,
      "loss": 2.2585,
      "step": 1748
    },
    {
      "epoch": 0.22385767310892102,
      "grad_norm": 0.59765625,
      "learning_rate": 0.001809406699049037,
      "loss": 1.7101,
      "step": 1749
    },
    {
      "epoch": 0.2239856649174453,
      "grad_norm": 0.734375,
      "learning_rate": 0.001809163175349752,
      "loss": 1.9935,
      "step": 1750
    },
    {
      "epoch": 0.22411365672596953,
      "grad_norm": 0.703125,
      "learning_rate": 0.0018089195125826464,
      "loss": 1.9987,
      "step": 1751
    },
    {
      "epoch": 0.2242416485344938,
      "grad_norm": 1.765625,
      "learning_rate": 0.0018086757107895984,
      "loss": 2.2874,
      "step": 1752
    },
    {
      "epoch": 0.22436964034301804,
      "grad_norm": 0.67578125,
      "learning_rate": 0.001808431770012509,
      "loss": 2.0606,
      "step": 1753
    },
    {
      "epoch": 0.2244976321515423,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0018081876902933031,
      "loss": 2.2475,
      "step": 1754
    },
    {
      "epoch": 0.22462562396006655,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00180794347167393,
      "loss": 2.2051,
      "step": 1755
    },
    {
      "epoch": 0.22475361576859082,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0018076991141963627,
      "loss": 2.2479,
      "step": 1756
    },
    {
      "epoch": 0.22488160757711506,
      "grad_norm": 0.74609375,
      "learning_rate": 0.001807454617902598,
      "loss": 2.3265,
      "step": 1757
    },
    {
      "epoch": 0.22500959938563933,
      "grad_norm": 1.65625,
      "learning_rate": 0.0018072099828346563,
      "loss": 2.5512,
      "step": 1758
    },
    {
      "epoch": 0.22513759119416357,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0018069652090345824,
      "loss": 2.1756,
      "step": 1759
    },
    {
      "epoch": 0.22526558300268784,
      "grad_norm": 1.40625,
      "learning_rate": 0.0018067202965444445,
      "loss": 2.7058,
      "step": 1760
    },
    {
      "epoch": 0.22539357481121208,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0018064752454063347,
      "loss": 2.8025,
      "step": 1761
    },
    {
      "epoch": 0.22552156661973632,
      "grad_norm": 0.6875,
      "learning_rate": 0.001806230055662369,
      "loss": 2.4627,
      "step": 1762
    },
    {
      "epoch": 0.2256495584282606,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0018059847273546874,
      "loss": 2.1257,
      "step": 1763
    },
    {
      "epoch": 0.22577755023678484,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0018057392605254536,
      "loss": 2.1263,
      "step": 1764
    },
    {
      "epoch": 0.2259055420453091,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0018054936552168548,
      "loss": 2.4031,
      "step": 1765
    },
    {
      "epoch": 0.22603353385383335,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0018052479114711026,
      "loss": 3.4957,
      "step": 1766
    },
    {
      "epoch": 0.22616152566235762,
      "grad_norm": 1.046875,
      "learning_rate": 0.001805002029330432,
      "loss": 3.1277,
      "step": 1767
    },
    {
      "epoch": 0.22628951747088186,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0018047560088371012,
      "loss": 2.2773,
      "step": 1768
    },
    {
      "epoch": 0.22641750927940613,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0018045098500333939,
      "loss": 2.4087,
      "step": 1769
    },
    {
      "epoch": 0.22654550108793037,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0018042635529616155,
      "loss": 1.6272,
      "step": 1770
    },
    {
      "epoch": 0.22667349289645464,
      "grad_norm": 0.765625,
      "learning_rate": 0.0018040171176640966,
      "loss": 2.1152,
      "step": 1771
    },
    {
      "epoch": 0.22680148470497888,
      "grad_norm": 0.66015625,
      "learning_rate": 0.001803770544183191,
      "loss": 1.9902,
      "step": 1772
    },
    {
      "epoch": 0.22692947651350315,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0018035238325612764,
      "loss": 2.1461,
      "step": 1773
    },
    {
      "epoch": 0.2270574683220274,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0018032769828407542,
      "loss": 1.7703,
      "step": 1774
    },
    {
      "epoch": 0.22718546013055166,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0018030299950640497,
      "loss": 2.6201,
      "step": 1775
    },
    {
      "epoch": 0.2273134519390759,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0018027828692736114,
      "loss": 1.8118,
      "step": 1776
    },
    {
      "epoch": 0.22744144374760014,
      "grad_norm": 0.734375,
      "learning_rate": 0.0018025356055119122,
      "loss": 2.0372,
      "step": 1777
    },
    {
      "epoch": 0.2275694355561244,
      "grad_norm": 0.78125,
      "learning_rate": 0.0018022882038214482,
      "loss": 2.617,
      "step": 1778
    },
    {
      "epoch": 0.22769742736464865,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0018020406642447393,
      "loss": 2.8302,
      "step": 1779
    },
    {
      "epoch": 0.22782541917317292,
      "grad_norm": 0.828125,
      "learning_rate": 0.0018017929868243297,
      "loss": 2.7023,
      "step": 1780
    },
    {
      "epoch": 0.22795341098169716,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0018015451716027861,
      "loss": 1.9147,
      "step": 1781
    },
    {
      "epoch": 0.22808140279022143,
      "grad_norm": 0.75,
      "learning_rate": 0.0018012972186227,
      "loss": 2.1965,
      "step": 1782
    },
    {
      "epoch": 0.22820939459874567,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0018010491279266858,
      "loss": 2.2661,
      "step": 1783
    },
    {
      "epoch": 0.22833738640726994,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0018008008995573823,
      "loss": 2.7743,
      "step": 1784
    },
    {
      "epoch": 0.22846537821579418,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0018005525335574512,
      "loss": 2.0464,
      "step": 1785
    },
    {
      "epoch": 0.22859337002431845,
      "grad_norm": 0.76171875,
      "learning_rate": 0.001800304029969579,
      "loss": 1.859,
      "step": 1786
    },
    {
      "epoch": 0.2287213618328427,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0018000553888364741,
      "loss": 2.5598,
      "step": 1787
    },
    {
      "epoch": 0.22884935364136696,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00179980661020087,
      "loss": 1.5413,
      "step": 1788
    },
    {
      "epoch": 0.2289773454498912,
      "grad_norm": 0.671875,
      "learning_rate": 0.0017995576941055233,
      "loss": 2.4955,
      "step": 1789
    },
    {
      "epoch": 0.22910533725841545,
      "grad_norm": 0.69921875,
      "learning_rate": 0.001799308640593214,
      "loss": 2.5884,
      "step": 1790
    },
    {
      "epoch": 0.22923332906693972,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0017990594497067467,
      "loss": 2.3885,
      "step": 1791
    },
    {
      "epoch": 0.22936132087546396,
      "grad_norm": 0.7421875,
      "learning_rate": 0.001798810121488948,
      "loss": 2.7853,
      "step": 1792
    },
    {
      "epoch": 0.22948931268398823,
      "grad_norm": 0.765625,
      "learning_rate": 0.0017985606559826697,
      "loss": 2.3762,
      "step": 1793
    },
    {
      "epoch": 0.22961730449251247,
      "grad_norm": 0.6953125,
      "learning_rate": 0.001798311053230786,
      "loss": 2.3493,
      "step": 1794
    },
    {
      "epoch": 0.22974529630103674,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0017980613132761955,
      "loss": 2.6752,
      "step": 1795
    },
    {
      "epoch": 0.22987328810956098,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0017978114361618198,
      "loss": 2.8626,
      "step": 1796
    },
    {
      "epoch": 0.23000127991808525,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0017975614219306046,
      "loss": 2.5789,
      "step": 1797
    },
    {
      "epoch": 0.2301292717266095,
      "grad_norm": 0.76953125,
      "learning_rate": 0.001797311270625519,
      "loss": 1.89,
      "step": 1798
    },
    {
      "epoch": 0.23025726353513376,
      "grad_norm": 0.765625,
      "learning_rate": 0.001797060982289555,
      "loss": 1.8042,
      "step": 1799
    },
    {
      "epoch": 0.230385255343658,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0017968105569657294,
      "loss": 2.3328,
      "step": 1800
    },
    {
      "epoch": 0.23051324715218227,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0017965599946970815,
      "loss": 2.253,
      "step": 1801
    },
    {
      "epoch": 0.2306412389607065,
      "grad_norm": 0.640625,
      "learning_rate": 0.0017963092955266742,
      "loss": 2.0552,
      "step": 1802
    },
    {
      "epoch": 0.23076923076923078,
      "grad_norm": 0.65625,
      "learning_rate": 0.001796058459497595,
      "loss": 2.3548,
      "step": 1803
    },
    {
      "epoch": 0.23089722257775502,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0017958074866529534,
      "loss": 2.136,
      "step": 1804
    },
    {
      "epoch": 0.23102521438627927,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0017955563770358834,
      "loss": 1.7016,
      "step": 1805
    },
    {
      "epoch": 0.23115320619480353,
      "grad_norm": 0.78125,
      "learning_rate": 0.0017953051306895422,
      "loss": 2.4346,
      "step": 1806
    },
    {
      "epoch": 0.23128119800332778,
      "grad_norm": 0.765625,
      "learning_rate": 0.001795053747657111,
      "loss": 2.1799,
      "step": 1807
    },
    {
      "epoch": 0.23140918981185205,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0017948022279817935,
      "loss": 2.3423,
      "step": 1808
    },
    {
      "epoch": 0.2315371816203763,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0017945505717068175,
      "loss": 2.1734,
      "step": 1809
    },
    {
      "epoch": 0.23166517342890056,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0017942987788754348,
      "loss": 2.6265,
      "step": 1810
    },
    {
      "epoch": 0.2317931652374248,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0017940468495309192,
      "loss": 1.864,
      "step": 1811
    },
    {
      "epoch": 0.23192115704594907,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0017937947837165695,
      "loss": 2.2471,
      "step": 1812
    },
    {
      "epoch": 0.2320491488544733,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0017935425814757072,
      "loss": 1.803,
      "step": 1813
    },
    {
      "epoch": 0.23217714066299758,
      "grad_norm": 1.0546875,
      "learning_rate": 0.001793290242851677,
      "loss": 2.9665,
      "step": 1814
    },
    {
      "epoch": 0.23230513247152182,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0017930377678878479,
      "loss": 3.1015,
      "step": 1815
    },
    {
      "epoch": 0.2324331242800461,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0017927851566276111,
      "loss": 2.2253,
      "step": 1816
    },
    {
      "epoch": 0.23256111608857033,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0017925324091143829,
      "loss": 1.6026,
      "step": 1817
    },
    {
      "epoch": 0.2326891078970946,
      "grad_norm": 0.94921875,
      "learning_rate": 0.001792279525391601,
      "loss": 3.0021,
      "step": 1818
    },
    {
      "epoch": 0.23281709970561884,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0017920265055027286,
      "loss": 3.3357,
      "step": 1819
    },
    {
      "epoch": 0.23294509151414308,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0017917733494912508,
      "loss": 2.4859,
      "step": 1820
    },
    {
      "epoch": 0.23307308332266735,
      "grad_norm": 0.796875,
      "learning_rate": 0.001791520057400676,
      "loss": 2.3131,
      "step": 1821
    },
    {
      "epoch": 0.2332010751311916,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0017912666292745376,
      "loss": 2.0643,
      "step": 1822
    },
    {
      "epoch": 0.23332906693971586,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0017910130651563909,
      "loss": 2.5073,
      "step": 1823
    },
    {
      "epoch": 0.2334570587482401,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0017907593650898146,
      "loss": 3.3201,
      "step": 1824
    },
    {
      "epoch": 0.23358505055676437,
      "grad_norm": 0.703125,
      "learning_rate": 0.0017905055291184118,
      "loss": 1.8804,
      "step": 1825
    },
    {
      "epoch": 0.23371304236528861,
      "grad_norm": 0.65625,
      "learning_rate": 0.0017902515572858079,
      "loss": 2.662,
      "step": 1826
    },
    {
      "epoch": 0.23384103417381288,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0017899974496356523,
      "loss": 2.985,
      "step": 1827
    },
    {
      "epoch": 0.23396902598233713,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0017897432062116178,
      "loss": 1.7263,
      "step": 1828
    },
    {
      "epoch": 0.2340970177908614,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0017894888270573994,
      "loss": 2.3007,
      "step": 1829
    },
    {
      "epoch": 0.23422500959938564,
      "grad_norm": 0.75390625,
      "learning_rate": 0.001789234312216717,
      "loss": 2.5953,
      "step": 1830
    },
    {
      "epoch": 0.2343530014079099,
      "grad_norm": 0.625,
      "learning_rate": 0.001788979661733313,
      "loss": 1.9172,
      "step": 1831
    },
    {
      "epoch": 0.23448099321643415,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0017887248756509528,
      "loss": 1.7856,
      "step": 1832
    },
    {
      "epoch": 0.23460898502495842,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0017884699540134262,
      "loss": 1.6776,
      "step": 1833
    },
    {
      "epoch": 0.23473697683348266,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0017882148968645453,
      "loss": 2.0662,
      "step": 1834
    },
    {
      "epoch": 0.2348649686420069,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0017879597042481455,
      "loss": 2.0161,
      "step": 1835
    },
    {
      "epoch": 0.23499296045053117,
      "grad_norm": 0.84375,
      "learning_rate": 0.0017877043762080862,
      "loss": 2.0219,
      "step": 1836
    },
    {
      "epoch": 0.2351209522590554,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0017874489127882494,
      "loss": 1.6734,
      "step": 1837
    },
    {
      "epoch": 0.23524894406757968,
      "grad_norm": 0.703125,
      "learning_rate": 0.0017871933140325411,
      "loss": 2.414,
      "step": 1838
    },
    {
      "epoch": 0.23537693587610392,
      "grad_norm": 0.703125,
      "learning_rate": 0.0017869375799848895,
      "loss": 2.1322,
      "step": 1839
    },
    {
      "epoch": 0.2355049276846282,
      "grad_norm": 0.69921875,
      "learning_rate": 0.001786681710689247,
      "loss": 2.4683,
      "step": 1840
    },
    {
      "epoch": 0.23563291949315243,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0017864257061895887,
      "loss": 2.3969,
      "step": 1841
    },
    {
      "epoch": 0.2357609113016767,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0017861695665299133,
      "loss": 1.9364,
      "step": 1842
    },
    {
      "epoch": 0.23588890311020094,
      "grad_norm": 0.703125,
      "learning_rate": 0.0017859132917542424,
      "loss": 2.4063,
      "step": 1843
    },
    {
      "epoch": 0.2360168949187252,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0017856568819066211,
      "loss": 2.4707,
      "step": 1844
    },
    {
      "epoch": 0.23614488672724945,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0017854003370311178,
      "loss": 2.5788,
      "step": 1845
    },
    {
      "epoch": 0.23627287853577372,
      "grad_norm": 0.828125,
      "learning_rate": 0.0017851436571718234,
      "loss": 2.6901,
      "step": 1846
    },
    {
      "epoch": 0.23640087034429796,
      "grad_norm": 0.609375,
      "learning_rate": 0.0017848868423728525,
      "loss": 1.9484,
      "step": 1847
    },
    {
      "epoch": 0.2365288621528222,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0017846298926783438,
      "loss": 1.8713,
      "step": 1848
    },
    {
      "epoch": 0.23665685396134648,
      "grad_norm": 0.6875,
      "learning_rate": 0.0017843728081324572,
      "loss": 1.9743,
      "step": 1849
    },
    {
      "epoch": 0.23678484576987072,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0017841155887793772,
      "loss": 2.1513,
      "step": 1850
    },
    {
      "epoch": 0.23691283757839499,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0017838582346633115,
      "loss": 2.3248,
      "step": 1851
    },
    {
      "epoch": 0.23704082938691923,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0017836007458284899,
      "loss": 2.2742,
      "step": 1852
    },
    {
      "epoch": 0.2371688211954435,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0017833431223191665,
      "loss": 1.4592,
      "step": 1853
    },
    {
      "epoch": 0.23729681300396774,
      "grad_norm": 0.921875,
      "learning_rate": 0.0017830853641796181,
      "loss": 2.1433,
      "step": 1854
    },
    {
      "epoch": 0.237424804812492,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0017828274714541445,
      "loss": 2.3519,
      "step": 1855
    },
    {
      "epoch": 0.23755279662101625,
      "grad_norm": 0.71875,
      "learning_rate": 0.0017825694441870685,
      "loss": 2.7425,
      "step": 1856
    },
    {
      "epoch": 0.23768078842954052,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0017823112824227369,
      "loss": 2.6476,
      "step": 1857
    },
    {
      "epoch": 0.23780878023806476,
      "grad_norm": 0.84375,
      "learning_rate": 0.0017820529862055182,
      "loss": 2.7962,
      "step": 1858
    },
    {
      "epoch": 0.23793677204658903,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0017817945555798054,
      "loss": 1.4136,
      "step": 1859
    },
    {
      "epoch": 0.23806476385511327,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0017815359905900138,
      "loss": 1.4896,
      "step": 1860
    },
    {
      "epoch": 0.23819275566363754,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0017812772912805818,
      "loss": 2.8909,
      "step": 1861
    },
    {
      "epoch": 0.23832074747216178,
      "grad_norm": 1.25,
      "learning_rate": 0.0017810184576959715,
      "loss": 2.7408,
      "step": 1862
    },
    {
      "epoch": 0.23844873928068602,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0017807594898806673,
      "loss": 2.2106,
      "step": 1863
    },
    {
      "epoch": 0.2385767310892103,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0017805003878791773,
      "loss": 2.0883,
      "step": 1864
    },
    {
      "epoch": 0.23870472289773453,
      "grad_norm": 0.59375,
      "learning_rate": 0.001780241151736032,
      "loss": 1.4579,
      "step": 1865
    },
    {
      "epoch": 0.2388327147062588,
      "grad_norm": 0.84375,
      "learning_rate": 0.0017799817814957858,
      "loss": 1.875,
      "step": 1866
    },
    {
      "epoch": 0.23896070651478304,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0017797222772030152,
      "loss": 2.0201,
      "step": 1867
    },
    {
      "epoch": 0.2390886983233073,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0017794626389023205,
      "loss": 2.225,
      "step": 1868
    },
    {
      "epoch": 0.23921669013183156,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0017792028666383246,
      "loss": 1.9855,
      "step": 1869
    },
    {
      "epoch": 0.23934468194035582,
      "grad_norm": 0.79296875,
      "learning_rate": 0.001778942960455674,
      "loss": 1.8229,
      "step": 1870
    },
    {
      "epoch": 0.23947267374888007,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0017786829203990373,
      "loss": 2.3429,
      "step": 1871
    },
    {
      "epoch": 0.23960066555740434,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0017784227465131067,
      "loss": 2.1904,
      "step": 1872
    },
    {
      "epoch": 0.23972865736592858,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0017781624388425975,
      "loss": 2.8277,
      "step": 1873
    },
    {
      "epoch": 0.23985664917445285,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0017779019974322474,
      "loss": 2.1523,
      "step": 1874
    },
    {
      "epoch": 0.2399846409829771,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0017776414223268182,
      "loss": 2.7125,
      "step": 1875
    },
    {
      "epoch": 0.24011263279150136,
      "grad_norm": 0.65625,
      "learning_rate": 0.0017773807135710933,
      "loss": 1.8931,
      "step": 1876
    },
    {
      "epoch": 0.2402406246000256,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0017771198712098802,
      "loss": 1.5403,
      "step": 1877
    },
    {
      "epoch": 0.24036861640854984,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0017768588952880085,
      "loss": 1.9959,
      "step": 1878
    },
    {
      "epoch": 0.2404966082170741,
      "grad_norm": 0.640625,
      "learning_rate": 0.0017765977858503312,
      "loss": 1.6762,
      "step": 1879
    },
    {
      "epoch": 0.24062460002559835,
      "grad_norm": 0.734375,
      "learning_rate": 0.0017763365429417242,
      "loss": 2.1619,
      "step": 1880
    },
    {
      "epoch": 0.24075259183412262,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0017760751666070866,
      "loss": 2.1427,
      "step": 1881
    },
    {
      "epoch": 0.24088058364264686,
      "grad_norm": 0.890625,
      "learning_rate": 0.0017758136568913402,
      "loss": 2.4352,
      "step": 1882
    },
    {
      "epoch": 0.24100857545117113,
      "grad_norm": 0.68359375,
      "learning_rate": 0.001775552013839429,
      "loss": 1.7955,
      "step": 1883
    },
    {
      "epoch": 0.24113656725969537,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0017752902374963213,
      "loss": 1.8275,
      "step": 1884
    },
    {
      "epoch": 0.24126455906821964,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0017750283279070076,
      "loss": 2.0326,
      "step": 1885
    },
    {
      "epoch": 0.24139255087674388,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0017747662851165008,
      "loss": 2.526,
      "step": 1886
    },
    {
      "epoch": 0.24152054268526815,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0017745041091698378,
      "loss": 1.8035,
      "step": 1887
    },
    {
      "epoch": 0.2416485344937924,
      "grad_norm": 0.7109375,
      "learning_rate": 0.001774241800112077,
      "loss": 2.3406,
      "step": 1888
    },
    {
      "epoch": 0.24177652630231666,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0017739793579883015,
      "loss": 3.0092,
      "step": 1889
    },
    {
      "epoch": 0.2419045181108409,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0017737167828436155,
      "loss": 2.4114,
      "step": 1890
    },
    {
      "epoch": 0.24203250991936517,
      "grad_norm": 0.84375,
      "learning_rate": 0.001773454074723147,
      "loss": 2.4877,
      "step": 1891
    },
    {
      "epoch": 0.24216050172788942,
      "grad_norm": 0.703125,
      "learning_rate": 0.0017731912336720467,
      "loss": 2.2426,
      "step": 1892
    },
    {
      "epoch": 0.24228849353641366,
      "grad_norm": 0.81640625,
      "learning_rate": 0.001772928259735488,
      "loss": 2.891,
      "step": 1893
    },
    {
      "epoch": 0.24241648534493793,
      "grad_norm": 0.66015625,
      "learning_rate": 0.001772665152958667,
      "loss": 2.5282,
      "step": 1894
    },
    {
      "epoch": 0.24254447715346217,
      "grad_norm": 0.6875,
      "learning_rate": 0.0017724019133868035,
      "loss": 2.0409,
      "step": 1895
    },
    {
      "epoch": 0.24267246896198644,
      "grad_norm": 0.8125,
      "learning_rate": 0.0017721385410651392,
      "loss": 2.1872,
      "step": 1896
    },
    {
      "epoch": 0.24280046077051068,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0017718750360389386,
      "loss": 2.6889,
      "step": 1897
    },
    {
      "epoch": 0.24292845257903495,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00177161139835349,
      "loss": 2.0247,
      "step": 1898
    },
    {
      "epoch": 0.2430564443875592,
      "grad_norm": 0.59375,
      "learning_rate": 0.001771347628054103,
      "loss": 1.234,
      "step": 1899
    },
    {
      "epoch": 0.24318443619608346,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0017710837251861112,
      "loss": 2.0191,
      "step": 1900
    },
    {
      "epoch": 0.2433124280046077,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0017708196897948706,
      "loss": 2.3977,
      "step": 1901
    },
    {
      "epoch": 0.24344041981313197,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00177055552192576,
      "loss": 2.53,
      "step": 1902
    },
    {
      "epoch": 0.2435684116216562,
      "grad_norm": 0.65625,
      "learning_rate": 0.001770291221624181,
      "loss": 1.7784,
      "step": 1903
    },
    {
      "epoch": 0.24369640343018048,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0017700267889355578,
      "loss": 2.3733,
      "step": 1904
    },
    {
      "epoch": 0.24382439523870472,
      "grad_norm": 0.734375,
      "learning_rate": 0.001769762223905337,
      "loss": 2.2235,
      "step": 1905
    },
    {
      "epoch": 0.24395238704722896,
      "grad_norm": 0.75,
      "learning_rate": 0.0017694975265789893,
      "loss": 2.177,
      "step": 1906
    },
    {
      "epoch": 0.24408037885575323,
      "grad_norm": 0.734375,
      "learning_rate": 0.0017692326970020066,
      "loss": 2.0946,
      "step": 1907
    },
    {
      "epoch": 0.24420837066427747,
      "grad_norm": 0.65625,
      "learning_rate": 0.001768967735219904,
      "loss": 2.5018,
      "step": 1908
    },
    {
      "epoch": 0.24433636247280174,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00176870264127822,
      "loss": 2.585,
      "step": 1909
    },
    {
      "epoch": 0.24446435428132599,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0017684374152225153,
      "loss": 1.9592,
      "step": 1910
    },
    {
      "epoch": 0.24459234608985025,
      "grad_norm": 0.671875,
      "learning_rate": 0.001768172057098373,
      "loss": 1.7779,
      "step": 1911
    },
    {
      "epoch": 0.2447203378983745,
      "grad_norm": 0.69140625,
      "learning_rate": 0.001767906566951399,
      "loss": 1.9582,
      "step": 1912
    },
    {
      "epoch": 0.24484832970689877,
      "grad_norm": 0.640625,
      "learning_rate": 0.0017676409448272228,
      "loss": 1.791,
      "step": 1913
    },
    {
      "epoch": 0.244976321515423,
      "grad_norm": 0.74609375,
      "learning_rate": 0.001767375190771495,
      "loss": 2.9113,
      "step": 1914
    },
    {
      "epoch": 0.24510431332394728,
      "grad_norm": 0.828125,
      "learning_rate": 0.0017671093048298905,
      "loss": 2.4855,
      "step": 1915
    },
    {
      "epoch": 0.24523230513247152,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0017668432870481057,
      "loss": 2.3397,
      "step": 1916
    },
    {
      "epoch": 0.2453602969409958,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0017665771374718601,
      "loss": 1.7093,
      "step": 1917
    },
    {
      "epoch": 0.24548828874952003,
      "grad_norm": 0.6484375,
      "learning_rate": 0.001766310856146896,
      "loss": 2.5175,
      "step": 1918
    },
    {
      "epoch": 0.2456162805580443,
      "grad_norm": 0.81640625,
      "learning_rate": 0.001766044443118978,
      "loss": 2.2833,
      "step": 1919
    },
    {
      "epoch": 0.24574427236656854,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0017657778984338937,
      "loss": 2.0257,
      "step": 1920
    },
    {
      "epoch": 0.24587226417509278,
      "grad_norm": 0.6484375,
      "learning_rate": 0.001765511222137453,
      "loss": 1.968,
      "step": 1921
    },
    {
      "epoch": 0.24600025598361705,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0017652444142754883,
      "loss": 2.1304,
      "step": 1922
    },
    {
      "epoch": 0.2461282477921413,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0017649774748938552,
      "loss": 2.5489,
      "step": 1923
    },
    {
      "epoch": 0.24625623960066556,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0017647104040384317,
      "loss": 2.9323,
      "step": 1924
    },
    {
      "epoch": 0.2463842314091898,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0017644432017551175,
      "loss": 2.5324,
      "step": 1925
    },
    {
      "epoch": 0.24651222321771407,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0017641758680898368,
      "loss": 1.8928,
      "step": 1926
    },
    {
      "epoch": 0.2466402150262383,
      "grad_norm": 0.72265625,
      "learning_rate": 0.001763908403088534,
      "loss": 1.9783,
      "step": 1927
    },
    {
      "epoch": 0.24676820683476258,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0017636408067971782,
      "loss": 2.6888,
      "step": 1928
    },
    {
      "epoch": 0.24689619864328682,
      "grad_norm": 0.71875,
      "learning_rate": 0.0017633730792617599,
      "loss": 2.2435,
      "step": 1929
    },
    {
      "epoch": 0.2470241904518111,
      "grad_norm": 0.734375,
      "learning_rate": 0.0017631052205282925,
      "loss": 2.698,
      "step": 1930
    },
    {
      "epoch": 0.24715218226033533,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0017628372306428112,
      "loss": 2.325,
      "step": 1931
    },
    {
      "epoch": 0.2472801740688596,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0017625691096513757,
      "loss": 2.3106,
      "step": 1932
    },
    {
      "epoch": 0.24740816587738385,
      "grad_norm": 1.6875,
      "learning_rate": 0.001762300857600066,
      "loss": 2.1838,
      "step": 1933
    },
    {
      "epoch": 0.24753615768590811,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0017620324745349857,
      "loss": 2.99,
      "step": 1934
    },
    {
      "epoch": 0.24766414949443236,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0017617639605022606,
      "loss": 2.394,
      "step": 1935
    },
    {
      "epoch": 0.2477921413029566,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0017614953155480398,
      "loss": 2.4231,
      "step": 1936
    },
    {
      "epoch": 0.24792013311148087,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0017612265397184937,
      "loss": 2.4116,
      "step": 1937
    },
    {
      "epoch": 0.2480481249200051,
      "grad_norm": 0.49609375,
      "learning_rate": 0.001760957633059816,
      "loss": 1.0815,
      "step": 1938
    },
    {
      "epoch": 0.24817611672852938,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0017606885956182228,
      "loss": 2.1413,
      "step": 1939
    },
    {
      "epoch": 0.24830410853705362,
      "grad_norm": 0.890625,
      "learning_rate": 0.0017604194274399526,
      "loss": 2.4926,
      "step": 1940
    },
    {
      "epoch": 0.2484321003455779,
      "grad_norm": 0.8515625,
      "learning_rate": 0.001760150128571266,
      "loss": 2.6715,
      "step": 1941
    },
    {
      "epoch": 0.24856009215410213,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0017598806990584467,
      "loss": 2.1951,
      "step": 1942
    },
    {
      "epoch": 0.2486880839626264,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0017596111389478002,
      "loss": 2.1095,
      "step": 1943
    },
    {
      "epoch": 0.24881607577115064,
      "grad_norm": 0.671875,
      "learning_rate": 0.0017593414482856551,
      "loss": 2.2912,
      "step": 1944
    },
    {
      "epoch": 0.2489440675796749,
      "grad_norm": 0.703125,
      "learning_rate": 0.001759071627118362,
      "loss": 2.397,
      "step": 1945
    },
    {
      "epoch": 0.24907205938819915,
      "grad_norm": 0.77734375,
      "learning_rate": 0.001758801675492294,
      "loss": 2.8487,
      "step": 1946
    },
    {
      "epoch": 0.24920005119672342,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0017585315934538466,
      "loss": 2.7305,
      "step": 1947
    },
    {
      "epoch": 0.24932804300524766,
      "grad_norm": 0.6875,
      "learning_rate": 0.001758261381049438,
      "loss": 2.0185,
      "step": 1948
    },
    {
      "epoch": 0.24945603481377193,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0017579910383255081,
      "loss": 2.5714,
      "step": 1949
    },
    {
      "epoch": 0.24958402662229617,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0017577205653285205,
      "loss": 2.0828,
      "step": 1950
    },
    {
      "epoch": 0.24971201843082041,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00175744996210496,
      "loss": 1.9161,
      "step": 1951
    },
    {
      "epoch": 0.24984001023934468,
      "grad_norm": 0.703125,
      "learning_rate": 0.0017571792287013337,
      "loss": 2.0155,
      "step": 1952
    },
    {
      "epoch": 0.24996800204786893,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0017569083651641719,
      "loss": 2.9505,
      "step": 1953
    },
    {
      "epoch": 0.2500959938563932,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0017566373715400272,
      "loss": 2.5936,
      "step": 1954
    },
    {
      "epoch": 0.25022398566491744,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0017563662478754737,
      "loss": 2.5969,
      "step": 1955
    },
    {
      "epoch": 0.2503519774734417,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0017560949942171087,
      "loss": 2.5183,
      "step": 1956
    },
    {
      "epoch": 0.250479969281966,
      "grad_norm": 0.71875,
      "learning_rate": 0.0017558236106115514,
      "loss": 2.3839,
      "step": 1957
    },
    {
      "epoch": 0.2506079610904902,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0017555520971054435,
      "loss": 2.2929,
      "step": 1958
    },
    {
      "epoch": 0.25073595289901446,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0017552804537454493,
      "loss": 1.8651,
      "step": 1959
    },
    {
      "epoch": 0.2508639447075387,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0017550086805782547,
      "loss": 2.2155,
      "step": 1960
    },
    {
      "epoch": 0.250991936516063,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0017547367776505687,
      "loss": 2.2835,
      "step": 1961
    },
    {
      "epoch": 0.25111992832458724,
      "grad_norm": 0.68359375,
      "learning_rate": 0.001754464745009122,
      "loss": 2.3756,
      "step": 1962
    },
    {
      "epoch": 0.2512479201331115,
      "grad_norm": 0.90625,
      "learning_rate": 0.001754192582700668,
      "loss": 2.7671,
      "step": 1963
    },
    {
      "epoch": 0.2513759119416357,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0017539202907719815,
      "loss": 2.199,
      "step": 1964
    },
    {
      "epoch": 0.25150390375015996,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0017536478692698615,
      "loss": 2.158,
      "step": 1965
    },
    {
      "epoch": 0.25163189555868426,
      "grad_norm": 0.66015625,
      "learning_rate": 0.001753375318241127,
      "loss": 1.554,
      "step": 1966
    },
    {
      "epoch": 0.2517598873672085,
      "grad_norm": 0.7109375,
      "learning_rate": 0.001753102637732621,
      "loss": 1.9319,
      "step": 1967
    },
    {
      "epoch": 0.25188787917573274,
      "grad_norm": 0.6875,
      "learning_rate": 0.0017528298277912076,
      "loss": 2.7169,
      "step": 1968
    },
    {
      "epoch": 0.252015870984257,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0017525568884637739,
      "loss": 2.8589,
      "step": 1969
    },
    {
      "epoch": 0.2521438627927813,
      "grad_norm": 0.609375,
      "learning_rate": 0.0017522838197972287,
      "loss": 1.7315,
      "step": 1970
    },
    {
      "epoch": 0.2522718546013055,
      "grad_norm": 1.140625,
      "learning_rate": 0.0017520106218385037,
      "loss": 2.052,
      "step": 1971
    },
    {
      "epoch": 0.25239984640982976,
      "grad_norm": 0.7578125,
      "learning_rate": 0.001751737294634552,
      "loss": 2.3838,
      "step": 1972
    },
    {
      "epoch": 0.252527838218354,
      "grad_norm": 1.328125,
      "learning_rate": 0.0017514638382323496,
      "loss": 2.0976,
      "step": 1973
    },
    {
      "epoch": 0.2526558300268783,
      "grad_norm": 0.859375,
      "learning_rate": 0.001751190252678894,
      "loss": 2.3574,
      "step": 1974
    },
    {
      "epoch": 0.25278382183540254,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0017509165380212059,
      "loss": 2.5292,
      "step": 1975
    },
    {
      "epoch": 0.2529118136439268,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0017506426943063272,
      "loss": 2.6169,
      "step": 1976
    },
    {
      "epoch": 0.253039805452451,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0017503687215813222,
      "loss": 1.9097,
      "step": 1977
    },
    {
      "epoch": 0.2531677972609753,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0017500946198932784,
      "loss": 2.3459,
      "step": 1978
    },
    {
      "epoch": 0.25329578906949957,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0017498203892893037,
      "loss": 2.5091,
      "step": 1979
    },
    {
      "epoch": 0.2534237808780238,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0017495460298165293,
      "loss": 1.6508,
      "step": 1980
    },
    {
      "epoch": 0.25355177268654805,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0017492715415221088,
      "loss": 2.7352,
      "step": 1981
    },
    {
      "epoch": 0.2536797644950723,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0017489969244532172,
      "loss": 1.9812,
      "step": 1982
    },
    {
      "epoch": 0.2538077563035966,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0017487221786570515,
      "loss": 2.0637,
      "step": 1983
    },
    {
      "epoch": 0.25393574811212083,
      "grad_norm": 0.703125,
      "learning_rate": 0.0017484473041808318,
      "loss": 2.0245,
      "step": 1984
    },
    {
      "epoch": 0.25406373992064507,
      "grad_norm": 1.15625,
      "learning_rate": 0.0017481723010717993,
      "loss": 2.5201,
      "step": 1985
    },
    {
      "epoch": 0.2541917317291693,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0017478971693772182,
      "loss": 1.5426,
      "step": 1986
    },
    {
      "epoch": 0.2543197235376936,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0017476219091443737,
      "loss": 2.4598,
      "step": 1987
    },
    {
      "epoch": 0.25444771534621785,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0017473465204205746,
      "loss": 2.2481,
      "step": 1988
    },
    {
      "epoch": 0.2545757071547421,
      "grad_norm": 0.6875,
      "learning_rate": 0.0017470710032531504,
      "loss": 2.2061,
      "step": 1989
    },
    {
      "epoch": 0.25470369896326633,
      "grad_norm": 0.66796875,
      "learning_rate": 0.001746795357689453,
      "loss": 2.437,
      "step": 1990
    },
    {
      "epoch": 0.25483169077179063,
      "grad_norm": 0.7265625,
      "learning_rate": 0.001746519583776857,
      "loss": 2.3337,
      "step": 1991
    },
    {
      "epoch": 0.2549596825803149,
      "grad_norm": 0.69140625,
      "learning_rate": 0.001746243681562758,
      "loss": 2.7445,
      "step": 1992
    },
    {
      "epoch": 0.2550876743888391,
      "grad_norm": 0.65625,
      "learning_rate": 0.001745967651094575,
      "loss": 2.015,
      "step": 1993
    },
    {
      "epoch": 0.25521566619736336,
      "grad_norm": 0.6484375,
      "learning_rate": 0.001745691492419748,
      "loss": 1.7049,
      "step": 1994
    },
    {
      "epoch": 0.2553436580058876,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0017454152055857396,
      "loss": 2.3196,
      "step": 1995
    },
    {
      "epoch": 0.2554716498144119,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0017451387906400337,
      "loss": 2.328,
      "step": 1996
    },
    {
      "epoch": 0.25559964162293614,
      "grad_norm": 0.65625,
      "learning_rate": 0.0017448622476301367,
      "loss": 1.8435,
      "step": 1997
    },
    {
      "epoch": 0.2557276334314604,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0017445855766035777,
      "loss": 2.4604,
      "step": 1998
    },
    {
      "epoch": 0.2558556252399846,
      "grad_norm": 0.75,
      "learning_rate": 0.0017443087776079064,
      "loss": 2.5163,
      "step": 1999
    },
    {
      "epoch": 0.2559836170485089,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0017440318506906957,
      "loss": 2.378,
      "step": 2000
    },
    {
      "epoch": 0.25611160885703316,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0017437547958995396,
      "loss": 1.6142,
      "step": 2001
    },
    {
      "epoch": 0.2562396006655574,
      "grad_norm": 0.796875,
      "learning_rate": 0.0017434776132820545,
      "loss": 1.9128,
      "step": 2002
    },
    {
      "epoch": 0.25636759247408164,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0017432003028858786,
      "loss": 1.5662,
      "step": 2003
    },
    {
      "epoch": 0.25649558428260594,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0017429228647586725,
      "loss": 2.1046,
      "step": 2004
    },
    {
      "epoch": 0.2566235760911302,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0017426452989481186,
      "loss": 3.4469,
      "step": 2005
    },
    {
      "epoch": 0.2567515678996544,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0017423676055019206,
      "loss": 2.3081,
      "step": 2006
    },
    {
      "epoch": 0.25687955970817866,
      "grad_norm": 1.0234375,
      "learning_rate": 0.001742089784467805,
      "loss": 1.9514,
      "step": 2007
    },
    {
      "epoch": 0.2570075515167029,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0017418118358935196,
      "loss": 2.2717,
      "step": 2008
    },
    {
      "epoch": 0.2571355433252272,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0017415337598268347,
      "loss": 2.0243,
      "step": 2009
    },
    {
      "epoch": 0.25726353513375144,
      "grad_norm": 0.65625,
      "learning_rate": 0.0017412555563155417,
      "loss": 1.7304,
      "step": 2010
    },
    {
      "epoch": 0.2573915269422757,
      "grad_norm": 0.703125,
      "learning_rate": 0.0017409772254074547,
      "loss": 2.1934,
      "step": 2011
    },
    {
      "epoch": 0.2575195187507999,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0017406987671504095,
      "loss": 2.1479,
      "step": 2012
    },
    {
      "epoch": 0.2576475105593242,
      "grad_norm": 0.703125,
      "learning_rate": 0.0017404201815922633,
      "loss": 1.8402,
      "step": 2013
    },
    {
      "epoch": 0.25777550236784846,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0017401414687808958,
      "loss": 2.8554,
      "step": 2014
    },
    {
      "epoch": 0.2579034941763727,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0017398626287642082,
      "loss": 2.2097,
      "step": 2015
    },
    {
      "epoch": 0.25803148598489695,
      "grad_norm": 0.7421875,
      "learning_rate": 0.001739583661590124,
      "loss": 2.0928,
      "step": 2016
    },
    {
      "epoch": 0.25815947779342124,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0017393045673065879,
      "loss": 2.405,
      "step": 2017
    },
    {
      "epoch": 0.2582874696019455,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0017390253459615668,
      "loss": 1.977,
      "step": 2018
    },
    {
      "epoch": 0.2584154614104697,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0017387459976030497,
      "loss": 2.4766,
      "step": 2019
    },
    {
      "epoch": 0.25854345321899397,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0017384665222790468,
      "loss": 2.0874,
      "step": 2020
    },
    {
      "epoch": 0.25867144502751827,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0017381869200375907,
      "loss": 1.8557,
      "step": 2021
    },
    {
      "epoch": 0.2587994368360425,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0017379071909267353,
      "loss": 2.7191,
      "step": 2022
    },
    {
      "epoch": 0.25892742864456675,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0017376273349945569,
      "loss": 2.3939,
      "step": 2023
    },
    {
      "epoch": 0.259055420453091,
      "grad_norm": 0.65234375,
      "learning_rate": 0.001737347352289153,
      "loss": 1.7861,
      "step": 2024
    },
    {
      "epoch": 0.25918341226161523,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0017370672428586435,
      "loss": 2.1994,
      "step": 2025
    },
    {
      "epoch": 0.25931140407013953,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0017367870067511696,
      "loss": 1.7971,
      "step": 2026
    },
    {
      "epoch": 0.25943939587866377,
      "grad_norm": 0.71875,
      "learning_rate": 0.0017365066440148942,
      "loss": 2.3634,
      "step": 2027
    },
    {
      "epoch": 0.259567387687188,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0017362261546980022,
      "loss": 2.4356,
      "step": 2028
    },
    {
      "epoch": 0.25969537949571225,
      "grad_norm": 0.75390625,
      "learning_rate": 0.001735945538848701,
      "loss": 2.2904,
      "step": 2029
    },
    {
      "epoch": 0.25982337130423655,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0017356647965152182,
      "loss": 1.8924,
      "step": 2030
    },
    {
      "epoch": 0.2599513631127608,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0017353839277458039,
      "loss": 1.8996,
      "step": 2031
    },
    {
      "epoch": 0.26007935492128503,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0017351029325887303,
      "loss": 1.676,
      "step": 2032
    },
    {
      "epoch": 0.2602073467298093,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0017348218110922909,
      "loss": 2.3148,
      "step": 2033
    },
    {
      "epoch": 0.26033533853833357,
      "grad_norm": 0.8046875,
      "learning_rate": 0.001734540563304801,
      "loss": 2.3384,
      "step": 2034
    },
    {
      "epoch": 0.2604633303468578,
      "grad_norm": 0.6875,
      "learning_rate": 0.0017342591892745977,
      "loss": 2.6498,
      "step": 2035
    },
    {
      "epoch": 0.26059132215538205,
      "grad_norm": 0.59765625,
      "learning_rate": 0.001733977689050039,
      "loss": 1.6003,
      "step": 2036
    },
    {
      "epoch": 0.2607193139639063,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0017336960626795062,
      "loss": 1.7326,
      "step": 2037
    },
    {
      "epoch": 0.26084730577243054,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0017334143102114012,
      "loss": 1.8704,
      "step": 2038
    },
    {
      "epoch": 0.26097529758095483,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0017331324316941473,
      "loss": 2.1407,
      "step": 2039
    },
    {
      "epoch": 0.2611032893894791,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00173285042717619,
      "loss": 2.6705,
      "step": 2040
    },
    {
      "epoch": 0.2612312811980033,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0017325682967059969,
      "loss": 2.7275,
      "step": 2041
    },
    {
      "epoch": 0.26135927300652756,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0017322860403320558,
      "loss": 2.2856,
      "step": 2042
    },
    {
      "epoch": 0.26148726481505186,
      "grad_norm": 0.65625,
      "learning_rate": 0.0017320036581028777,
      "loss": 2.1854,
      "step": 2043
    },
    {
      "epoch": 0.2616152566235761,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0017317211500669943,
      "loss": 2.0709,
      "step": 2044
    },
    {
      "epoch": 0.26174324843210034,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0017314385162729594,
      "loss": 1.8443,
      "step": 2045
    },
    {
      "epoch": 0.2618712402406246,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0017311557567693478,
      "loss": 2.0368,
      "step": 2046
    },
    {
      "epoch": 0.2619992320491489,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0017308728716047565,
      "loss": 1.3372,
      "step": 2047
    },
    {
      "epoch": 0.2621272238576731,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0017305898608278043,
      "loss": 1.7686,
      "step": 2048
    },
    {
      "epoch": 0.26225521566619736,
      "grad_norm": 0.66796875,
      "learning_rate": 0.001730306724487131,
      "loss": 1.8186,
      "step": 2049
    },
    {
      "epoch": 0.2623832074747216,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0017300234626313978,
      "loss": 2.2157,
      "step": 2050
    },
    {
      "epoch": 0.2625111992832459,
      "grad_norm": 0.671875,
      "learning_rate": 0.0017297400753092881,
      "loss": 2.1314,
      "step": 2051
    },
    {
      "epoch": 0.26263919109177014,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0017294565625695065,
      "loss": 1.6682,
      "step": 2052
    },
    {
      "epoch": 0.2627671829002944,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0017291729244607796,
      "loss": 2.0211,
      "step": 2053
    },
    {
      "epoch": 0.2628951747088186,
      "grad_norm": 0.78125,
      "learning_rate": 0.0017288891610318547,
      "loss": 1.9789,
      "step": 2054
    },
    {
      "epoch": 0.26302316651734287,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0017286052723315018,
      "loss": 2.8453,
      "step": 2055
    },
    {
      "epoch": 0.26315115832586716,
      "grad_norm": 1.140625,
      "learning_rate": 0.001728321258408511,
      "loss": 2.4635,
      "step": 2056
    },
    {
      "epoch": 0.2632791501343914,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0017280371193116953,
      "loss": 1.9552,
      "step": 2057
    },
    {
      "epoch": 0.26340714194291565,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0017277528550898884,
      "loss": 2.2466,
      "step": 2058
    },
    {
      "epoch": 0.2635351337514399,
      "grad_norm": 0.6796875,
      "learning_rate": 0.001727468465791946,
      "loss": 1.8553,
      "step": 2059
    },
    {
      "epoch": 0.2636631255599642,
      "grad_norm": 0.90625,
      "learning_rate": 0.0017271839514667446,
      "loss": 1.8293,
      "step": 2060
    },
    {
      "epoch": 0.2637911173684884,
      "grad_norm": 0.828125,
      "learning_rate": 0.0017268993121631825,
      "loss": 2.5633,
      "step": 2061
    },
    {
      "epoch": 0.26391910917701267,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00172661454793018,
      "loss": 2.3554,
      "step": 2062
    },
    {
      "epoch": 0.2640471009855369,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0017263296588166783,
      "loss": 2.4191,
      "step": 2063
    },
    {
      "epoch": 0.2641750927940612,
      "grad_norm": 0.90625,
      "learning_rate": 0.0017260446448716402,
      "loss": 2.2564,
      "step": 2064
    },
    {
      "epoch": 0.26430308460258545,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0017257595061440502,
      "loss": 2.5156,
      "step": 2065
    },
    {
      "epoch": 0.2644310764111097,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0017254742426829135,
      "loss": 2.6111,
      "step": 2066
    },
    {
      "epoch": 0.26455906821963393,
      "grad_norm": 0.765625,
      "learning_rate": 0.0017251888545372577,
      "loss": 2.8257,
      "step": 2067
    },
    {
      "epoch": 0.26468706002815817,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0017249033417561308,
      "loss": 2.1171,
      "step": 2068
    },
    {
      "epoch": 0.26481505183668247,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0017246177043886038,
      "loss": 1.7448,
      "step": 2069
    },
    {
      "epoch": 0.2649430436452067,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0017243319424837674,
      "loss": 2.3069,
      "step": 2070
    },
    {
      "epoch": 0.26507103545373095,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0017240460560907344,
      "loss": 2.2455,
      "step": 2071
    },
    {
      "epoch": 0.2651990272622552,
      "grad_norm": 0.828125,
      "learning_rate": 0.0017237600452586395,
      "loss": 1.6921,
      "step": 2072
    },
    {
      "epoch": 0.2653270190707795,
      "grad_norm": 0.671875,
      "learning_rate": 0.001723473910036638,
      "loss": 2.2602,
      "step": 2073
    },
    {
      "epoch": 0.26545501087930373,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0017231876504739067,
      "loss": 2.0165,
      "step": 2074
    },
    {
      "epoch": 0.265583002687828,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0017229012666196439,
      "loss": 2.0499,
      "step": 2075
    },
    {
      "epoch": 0.2657109944963522,
      "grad_norm": 0.703125,
      "learning_rate": 0.0017226147585230698,
      "loss": 2.1833,
      "step": 2076
    },
    {
      "epoch": 0.2658389863048765,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0017223281262334252,
      "loss": 1.8366,
      "step": 2077
    },
    {
      "epoch": 0.26596697811340075,
      "grad_norm": 0.671875,
      "learning_rate": 0.0017220413697999724,
      "loss": 2.4381,
      "step": 2078
    },
    {
      "epoch": 0.266094969921925,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0017217544892719951,
      "loss": 1.7741,
      "step": 2079
    },
    {
      "epoch": 0.26622296173044924,
      "grad_norm": 0.59765625,
      "learning_rate": 0.001721467484698799,
      "loss": 1.9059,
      "step": 2080
    },
    {
      "epoch": 0.2663509535389735,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0017211803561297096,
      "loss": 2.3177,
      "step": 2081
    },
    {
      "epoch": 0.2664789453474978,
      "grad_norm": 0.6484375,
      "learning_rate": 0.001720893103614075,
      "loss": 1.8283,
      "step": 2082
    },
    {
      "epoch": 0.266606937156022,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0017206057272012645,
      "loss": 1.8037,
      "step": 2083
    },
    {
      "epoch": 0.26673492896454626,
      "grad_norm": 0.7421875,
      "learning_rate": 0.001720318226940668,
      "loss": 2.4558,
      "step": 2084
    },
    {
      "epoch": 0.2668629207730705,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0017200306028816975,
      "loss": 2.0105,
      "step": 2085
    },
    {
      "epoch": 0.2669909125815948,
      "grad_norm": 0.921875,
      "learning_rate": 0.0017197428550737852,
      "loss": 2.3929,
      "step": 2086
    },
    {
      "epoch": 0.26711890439011904,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0017194549835663855,
      "loss": 1.6709,
      "step": 2087
    },
    {
      "epoch": 0.2672468961986433,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0017191669884089743,
      "loss": 2.5074,
      "step": 2088
    },
    {
      "epoch": 0.2673748880071675,
      "grad_norm": 0.65625,
      "learning_rate": 0.0017188788696510476,
      "loss": 1.8727,
      "step": 2089
    },
    {
      "epoch": 0.2675028798156918,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0017185906273421234,
      "loss": 1.591,
      "step": 2090
    },
    {
      "epoch": 0.26763087162421606,
      "grad_norm": 0.6875,
      "learning_rate": 0.0017183022615317413,
      "loss": 2.2147,
      "step": 2091
    },
    {
      "epoch": 0.2677588634327403,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0017180137722694608,
      "loss": 1.9292,
      "step": 2092
    },
    {
      "epoch": 0.26788685524126454,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0017177251596048645,
      "loss": 2.3658,
      "step": 2093
    },
    {
      "epoch": 0.26801484704978884,
      "grad_norm": 0.8359375,
      "learning_rate": 0.001717436423587554,
      "loss": 2.3886,
      "step": 2094
    },
    {
      "epoch": 0.2681428388583131,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0017171475642671543,
      "loss": 2.051,
      "step": 2095
    },
    {
      "epoch": 0.2682708306668373,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0017168585816933102,
      "loss": 1.7686,
      "step": 2096
    },
    {
      "epoch": 0.26839882247536156,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0017165694759156875,
      "loss": 2.26,
      "step": 2097
    },
    {
      "epoch": 0.2685268142838858,
      "grad_norm": 0.875,
      "learning_rate": 0.0017162802469839744,
      "loss": 2.0321,
      "step": 2098
    },
    {
      "epoch": 0.2686548060924101,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0017159908949478796,
      "loss": 2.2058,
      "step": 2099
    },
    {
      "epoch": 0.26878279790093434,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0017157014198571328,
      "loss": 2.7688,
      "step": 2100
    },
    {
      "epoch": 0.2689107897094586,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0017154118217614844,
      "loss": 2.7215,
      "step": 2101
    },
    {
      "epoch": 0.26903878151798283,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0017151221007107075,
      "loss": 2.1338,
      "step": 2102
    },
    {
      "epoch": 0.2691667733265071,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0017148322567545947,
      "loss": 2.135,
      "step": 2103
    },
    {
      "epoch": 0.26929476513503137,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0017145422899429606,
      "loss": 1.3878,
      "step": 2104
    },
    {
      "epoch": 0.2694227569435556,
      "grad_norm": 0.84765625,
      "learning_rate": 0.001714252200325641,
      "loss": 2.9014,
      "step": 2105
    },
    {
      "epoch": 0.26955074875207985,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0017139619879524917,
      "loss": 2.1395,
      "step": 2106
    },
    {
      "epoch": 0.26967874056060415,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0017136716528733914,
      "loss": 1.8354,
      "step": 2107
    },
    {
      "epoch": 0.2698067323691284,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0017133811951382382,
      "loss": 1.9129,
      "step": 2108
    },
    {
      "epoch": 0.26993472417765263,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0017130906147969519,
      "loss": 1.9945,
      "step": 2109
    },
    {
      "epoch": 0.27006271598617687,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0017127999118994742,
      "loss": 1.7722,
      "step": 2110
    },
    {
      "epoch": 0.2701907077947011,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0017125090864957661,
      "loss": 2.0798,
      "step": 2111
    },
    {
      "epoch": 0.2703186996032254,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0017122181386358118,
      "loss": 1.4662,
      "step": 2112
    },
    {
      "epoch": 0.27044669141174965,
      "grad_norm": 0.640625,
      "learning_rate": 0.0017119270683696142,
      "loss": 2.2357,
      "step": 2113
    },
    {
      "epoch": 0.2705746832202739,
      "grad_norm": 0.703125,
      "learning_rate": 0.0017116358757471996,
      "loss": 1.9167,
      "step": 2114
    },
    {
      "epoch": 0.27070267502879813,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0017113445608186134,
      "loss": 2.4054,
      "step": 2115
    },
    {
      "epoch": 0.27083066683732243,
      "grad_norm": 0.73828125,
      "learning_rate": 0.001711053123633923,
      "loss": 1.6191,
      "step": 2116
    },
    {
      "epoch": 0.2709586586458467,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0017107615642432167,
      "loss": 2.2597,
      "step": 2117
    },
    {
      "epoch": 0.2710866504543709,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0017104698826966036,
      "loss": 2.0519,
      "step": 2118
    },
    {
      "epoch": 0.27121464226289516,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0017101780790442138,
      "loss": 2.2515,
      "step": 2119
    },
    {
      "epoch": 0.27134263407141945,
      "grad_norm": 0.765625,
      "learning_rate": 0.0017098861533361989,
      "loss": 2.0212,
      "step": 2120
    },
    {
      "epoch": 0.2714706258799437,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0017095941056227307,
      "loss": 2.0456,
      "step": 2121
    },
    {
      "epoch": 0.27159861768846794,
      "grad_norm": 0.65625,
      "learning_rate": 0.0017093019359540025,
      "loss": 2.2132,
      "step": 2122
    },
    {
      "epoch": 0.2717266094969922,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0017090096443802283,
      "loss": 2.5875,
      "step": 2123
    },
    {
      "epoch": 0.2718546013055165,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0017087172309516432,
      "loss": 1.9798,
      "step": 2124
    },
    {
      "epoch": 0.2719825931140407,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0017084246957185035,
      "loss": 2.8135,
      "step": 2125
    },
    {
      "epoch": 0.27211058492256496,
      "grad_norm": 0.59375,
      "learning_rate": 0.0017081320387310854,
      "loss": 1.9758,
      "step": 2126
    },
    {
      "epoch": 0.2722385767310892,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0017078392600396875,
      "loss": 1.7119,
      "step": 2127
    },
    {
      "epoch": 0.27236656853961344,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0017075463596946281,
      "loss": 1.6204,
      "step": 2128
    },
    {
      "epoch": 0.27249456034813774,
      "grad_norm": 0.875,
      "learning_rate": 0.001707253337746247,
      "loss": 2.6273,
      "step": 2129
    },
    {
      "epoch": 0.272622552156662,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0017069601942449047,
      "loss": 2.4039,
      "step": 2130
    },
    {
      "epoch": 0.2727505439651862,
      "grad_norm": 0.76171875,
      "learning_rate": 0.001706666929240983,
      "loss": 1.4214,
      "step": 2131
    },
    {
      "epoch": 0.27287853577371046,
      "grad_norm": 0.671875,
      "learning_rate": 0.0017063735427848836,
      "loss": 2.1941,
      "step": 2132
    },
    {
      "epoch": 0.27300652758223476,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0017060800349270307,
      "loss": 1.7908,
      "step": 2133
    },
    {
      "epoch": 0.273134519390759,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0017057864057178672,
      "loss": 1.914,
      "step": 2134
    },
    {
      "epoch": 0.27326251119928324,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0017054926552078592,
      "loss": 1.9833,
      "step": 2135
    },
    {
      "epoch": 0.2733905030078075,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0017051987834474916,
      "loss": 2.7316,
      "step": 2136
    },
    {
      "epoch": 0.2735184948163318,
      "grad_norm": 0.640625,
      "learning_rate": 0.0017049047904872715,
      "loss": 1.7264,
      "step": 2137
    },
    {
      "epoch": 0.273646486624856,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0017046106763777265,
      "loss": 1.7297,
      "step": 2138
    },
    {
      "epoch": 0.27377447843338026,
      "grad_norm": 0.8125,
      "learning_rate": 0.0017043164411694042,
      "loss": 1.8993,
      "step": 2139
    },
    {
      "epoch": 0.2739024702419045,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0017040220849128742,
      "loss": 2.0728,
      "step": 2140
    },
    {
      "epoch": 0.27403046205042875,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0017037276076587267,
      "loss": 1.5208,
      "step": 2141
    },
    {
      "epoch": 0.27415845385895304,
      "grad_norm": 0.640625,
      "learning_rate": 0.0017034330094575715,
      "loss": 2.0964,
      "step": 2142
    },
    {
      "epoch": 0.2742864456674773,
      "grad_norm": 1.078125,
      "learning_rate": 0.001703138290360041,
      "loss": 1.6441,
      "step": 2143
    },
    {
      "epoch": 0.2744144374760015,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0017028434504167869,
      "loss": 2.101,
      "step": 2144
    },
    {
      "epoch": 0.27454242928452577,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0017025484896784824,
      "loss": 1.808,
      "step": 2145
    },
    {
      "epoch": 0.27467042109305007,
      "grad_norm": 0.83203125,
      "learning_rate": 0.001702253408195821,
      "loss": 2.0066,
      "step": 2146
    },
    {
      "epoch": 0.2747984129015743,
      "grad_norm": 0.671875,
      "learning_rate": 0.0017019582060195178,
      "loss": 2.3131,
      "step": 2147
    },
    {
      "epoch": 0.27492640471009855,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0017016628832003079,
      "loss": 2.5719,
      "step": 2148
    },
    {
      "epoch": 0.2750543965186228,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0017013674397889469,
      "loss": 2.2379,
      "step": 2149
    },
    {
      "epoch": 0.2751823883271471,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0017010718758362123,
      "loss": 2.2928,
      "step": 2150
    },
    {
      "epoch": 0.27531038013567133,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0017007761913929004,
      "loss": 1.627,
      "step": 2151
    },
    {
      "epoch": 0.27543837194419557,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0017004803865098305,
      "loss": 1.2555,
      "step": 2152
    },
    {
      "epoch": 0.2755663637527198,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0017001844612378415,
      "loss": 2.4845,
      "step": 2153
    },
    {
      "epoch": 0.27569435556124405,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0016998884156277918,
      "loss": 2.2966,
      "step": 2154
    },
    {
      "epoch": 0.27582234736976835,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0016995922497305628,
      "loss": 2.2351,
      "step": 2155
    },
    {
      "epoch": 0.2759503391782926,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0016992959635970548,
      "loss": 2.2393,
      "step": 2156
    },
    {
      "epoch": 0.27607833098681683,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0016989995572781896,
      "loss": 2.0211,
      "step": 2157
    },
    {
      "epoch": 0.2762063227953411,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0016987030308249094,
      "loss": 2.1111,
      "step": 2158
    },
    {
      "epoch": 0.27633431460386537,
      "grad_norm": 0.71875,
      "learning_rate": 0.0016984063842881768,
      "loss": 1.897,
      "step": 2159
    },
    {
      "epoch": 0.2764623064123896,
      "grad_norm": 0.85546875,
      "learning_rate": 0.001698109617718976,
      "loss": 2.2141,
      "step": 2160
    },
    {
      "epoch": 0.27659029822091385,
      "grad_norm": 0.65625,
      "learning_rate": 0.0016978127311683103,
      "loss": 1.8505,
      "step": 2161
    },
    {
      "epoch": 0.2767182900294381,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0016975157246872051,
      "loss": 2.6878,
      "step": 2162
    },
    {
      "epoch": 0.2768462818379624,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0016972185983267056,
      "loss": 2.4146,
      "step": 2163
    },
    {
      "epoch": 0.27697427364648664,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0016969213521378778,
      "loss": 1.8581,
      "step": 2164
    },
    {
      "epoch": 0.2771022654550109,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0016966239861718082,
      "loss": 2.5136,
      "step": 2165
    },
    {
      "epoch": 0.2772302572635351,
      "grad_norm": 0.546875,
      "learning_rate": 0.0016963265004796043,
      "loss": 1.6245,
      "step": 2166
    },
    {
      "epoch": 0.2773582490720594,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0016960288951123933,
      "loss": 2.2333,
      "step": 2167
    },
    {
      "epoch": 0.27748624088058366,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0016957311701213235,
      "loss": 2.2131,
      "step": 2168
    },
    {
      "epoch": 0.2776142326891079,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0016954333255575645,
      "loss": 1.6567,
      "step": 2169
    },
    {
      "epoch": 0.27774222449763214,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0016951353614723052,
      "loss": 2.1451,
      "step": 2170
    },
    {
      "epoch": 0.2778702163061564,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0016948372779167556,
      "loss": 1.6597,
      "step": 2171
    },
    {
      "epoch": 0.2779982081146807,
      "grad_norm": 0.65625,
      "learning_rate": 0.0016945390749421461,
      "loss": 2.5175,
      "step": 2172
    },
    {
      "epoch": 0.2781261999232049,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0016942407525997276,
      "loss": 1.9936,
      "step": 2173
    },
    {
      "epoch": 0.27825419173172916,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0016939423109407724,
      "loss": 2.2665,
      "step": 2174
    },
    {
      "epoch": 0.2783821835402534,
      "grad_norm": 0.625,
      "learning_rate": 0.0016936437500165717,
      "loss": 1.833,
      "step": 2175
    },
    {
      "epoch": 0.2785101753487777,
      "grad_norm": 0.703125,
      "learning_rate": 0.0016933450698784385,
      "loss": 2.2214,
      "step": 2176
    },
    {
      "epoch": 0.27863816715730194,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0016930462705777053,
      "loss": 2.2661,
      "step": 2177
    },
    {
      "epoch": 0.2787661589658262,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0016927473521657266,
      "loss": 1.7815,
      "step": 2178
    },
    {
      "epoch": 0.2788941507743504,
      "grad_norm": 0.65625,
      "learning_rate": 0.0016924483146938756,
      "loss": 2.374,
      "step": 2179
    },
    {
      "epoch": 0.2790221425828747,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0016921491582135469,
      "loss": 2.2158,
      "step": 2180
    },
    {
      "epoch": 0.27915013439139896,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0016918498827761552,
      "loss": 2.3447,
      "step": 2181
    },
    {
      "epoch": 0.2792781261999232,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0016915504884331364,
      "loss": 2.7944,
      "step": 2182
    },
    {
      "epoch": 0.27940611800844745,
      "grad_norm": 0.71875,
      "learning_rate": 0.0016912509752359454,
      "loss": 2.5773,
      "step": 2183
    },
    {
      "epoch": 0.2795341098169717,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0016909513432360594,
      "loss": 1.7594,
      "step": 2184
    },
    {
      "epoch": 0.279662101625496,
      "grad_norm": 0.625,
      "learning_rate": 0.0016906515924849743,
      "loss": 1.699,
      "step": 2185
    },
    {
      "epoch": 0.2797900934340202,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0016903517230342077,
      "loss": 1.9426,
      "step": 2186
    },
    {
      "epoch": 0.27991808524254447,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0016900517349352961,
      "loss": 2.0168,
      "step": 2187
    },
    {
      "epoch": 0.2800460770510687,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0016897516282397983,
      "loss": 2.2125,
      "step": 2188
    },
    {
      "epoch": 0.280174068859593,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0016894514029992922,
      "loss": 1.9918,
      "step": 2189
    },
    {
      "epoch": 0.28030206066811725,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0016891510592653763,
      "loss": 1.6391,
      "step": 2190
    },
    {
      "epoch": 0.2804300524766415,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0016888505970896696,
      "loss": 2.5625,
      "step": 2191
    },
    {
      "epoch": 0.28055804428516573,
      "grad_norm": 0.7109375,
      "learning_rate": 0.001688550016523811,
      "loss": 2.9778,
      "step": 2192
    },
    {
      "epoch": 0.28068603609369003,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0016882493176194609,
      "loss": 2.677,
      "step": 2193
    },
    {
      "epoch": 0.28081402790221427,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0016879485004282987,
      "loss": 1.5124,
      "step": 2194
    },
    {
      "epoch": 0.2809420197107385,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0016876475650020248,
      "loss": 2.4775,
      "step": 2195
    },
    {
      "epoch": 0.28107001151926275,
      "grad_norm": 0.703125,
      "learning_rate": 0.00168734651139236,
      "loss": 1.8563,
      "step": 2196
    },
    {
      "epoch": 0.281198003327787,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0016870453396510454,
      "loss": 1.1608,
      "step": 2197
    },
    {
      "epoch": 0.2813259951363113,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0016867440498298419,
      "loss": 2.3063,
      "step": 2198
    },
    {
      "epoch": 0.28145398694483553,
      "grad_norm": 0.78515625,
      "learning_rate": 0.001686442641980531,
      "loss": 1.6122,
      "step": 2199
    },
    {
      "epoch": 0.2815819787533598,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0016861411161549153,
      "loss": 2.436,
      "step": 2200
    },
    {
      "epoch": 0.281709970561884,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0016858394724048155,
      "loss": 2.1263,
      "step": 2201
    },
    {
      "epoch": 0.2818379623704083,
      "grad_norm": 0.640625,
      "learning_rate": 0.0016855377107820755,
      "loss": 2.4547,
      "step": 2202
    },
    {
      "epoch": 0.28196595417893255,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0016852358313385568,
      "loss": 1.8633,
      "step": 2203
    },
    {
      "epoch": 0.2820939459874568,
      "grad_norm": 0.703125,
      "learning_rate": 0.001684933834126143,
      "loss": 2.0092,
      "step": 2204
    },
    {
      "epoch": 0.28222193779598104,
      "grad_norm": 0.671875,
      "learning_rate": 0.0016846317191967365,
      "loss": 2.0841,
      "step": 2205
    },
    {
      "epoch": 0.28234992960450533,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0016843294866022611,
      "loss": 2.2062,
      "step": 2206
    },
    {
      "epoch": 0.2824779214130296,
      "grad_norm": 0.6875,
      "learning_rate": 0.0016840271363946606,
      "loss": 1.8663,
      "step": 2207
    },
    {
      "epoch": 0.2826059132215538,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0016837246686258987,
      "loss": 1.8661,
      "step": 2208
    },
    {
      "epoch": 0.28273390503007806,
      "grad_norm": 0.625,
      "learning_rate": 0.001683422083347959,
      "loss": 1.7528,
      "step": 2209
    },
    {
      "epoch": 0.28286189683860236,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0016831193806128458,
      "loss": 1.7468,
      "step": 2210
    },
    {
      "epoch": 0.2829898886471266,
      "grad_norm": 0.671875,
      "learning_rate": 0.0016828165604725839,
      "loss": 1.9191,
      "step": 2211
    },
    {
      "epoch": 0.28311788045565084,
      "grad_norm": 1.046875,
      "learning_rate": 0.0016825136229792173,
      "loss": 2.3941,
      "step": 2212
    },
    {
      "epoch": 0.2832458722641751,
      "grad_norm": 0.72265625,
      "learning_rate": 0.001682210568184811,
      "loss": 2.3444,
      "step": 2213
    },
    {
      "epoch": 0.2833738640726993,
      "grad_norm": 0.8125,
      "learning_rate": 0.0016819073961414498,
      "loss": 1.9588,
      "step": 2214
    },
    {
      "epoch": 0.2835018558812236,
      "grad_norm": 0.69921875,
      "learning_rate": 0.001681604106901239,
      "loss": 1.7908,
      "step": 2215
    },
    {
      "epoch": 0.28362984768974786,
      "grad_norm": 0.640625,
      "learning_rate": 0.0016813007005163035,
      "loss": 1.5282,
      "step": 2216
    },
    {
      "epoch": 0.2837578394982721,
      "grad_norm": 0.53125,
      "learning_rate": 0.0016809971770387886,
      "loss": 1.4676,
      "step": 2217
    },
    {
      "epoch": 0.28388583130679634,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0016806935365208597,
      "loss": 2.1402,
      "step": 2218
    },
    {
      "epoch": 0.28401382311532064,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0016803897790147024,
      "loss": 2.2143,
      "step": 2219
    },
    {
      "epoch": 0.2841418149238449,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0016800859045725228,
      "loss": 1.8457,
      "step": 2220
    },
    {
      "epoch": 0.2842698067323691,
      "grad_norm": 0.60546875,
      "learning_rate": 0.001679781913246546,
      "loss": 1.6046,
      "step": 2221
    },
    {
      "epoch": 0.28439779854089337,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0016794778050890184,
      "loss": 1.6472,
      "step": 2222
    },
    {
      "epoch": 0.28452579034941766,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0016791735801522055,
      "loss": 2.4384,
      "step": 2223
    },
    {
      "epoch": 0.2846537821579419,
      "grad_norm": 0.625,
      "learning_rate": 0.0016788692384883932,
      "loss": 1.8868,
      "step": 2224
    },
    {
      "epoch": 0.28478177396646615,
      "grad_norm": 0.78125,
      "learning_rate": 0.001678564780149888,
      "loss": 2.2071,
      "step": 2225
    },
    {
      "epoch": 0.2849097657749904,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0016782602051890157,
      "loss": 2.0174,
      "step": 2226
    },
    {
      "epoch": 0.28503775758351463,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0016779555136581226,
      "loss": 1.6741,
      "step": 2227
    },
    {
      "epoch": 0.2851657493920389,
      "grad_norm": 0.703125,
      "learning_rate": 0.0016776507056095746,
      "loss": 1.7213,
      "step": 2228
    },
    {
      "epoch": 0.28529374120056317,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0016773457810957582,
      "loss": 1.9,
      "step": 2229
    },
    {
      "epoch": 0.2854217330090874,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0016770407401690798,
      "loss": 2.2314,
      "step": 2230
    },
    {
      "epoch": 0.28554972481761165,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0016767355828819653,
      "loss": 1.5668,
      "step": 2231
    },
    {
      "epoch": 0.28567771662613595,
      "grad_norm": 0.62890625,
      "learning_rate": 0.001676430309286861,
      "loss": 1.6114,
      "step": 2232
    },
    {
      "epoch": 0.2858057084346602,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0016761249194362329,
      "loss": 2.0352,
      "step": 2233
    },
    {
      "epoch": 0.28593370024318443,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0016758194133825676,
      "loss": 1.8338,
      "step": 2234
    },
    {
      "epoch": 0.28606169205170867,
      "grad_norm": 0.63671875,
      "learning_rate": 0.001675513791178371,
      "loss": 1.9193,
      "step": 2235
    },
    {
      "epoch": 0.28618968386023297,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0016752080528761695,
      "loss": 1.3698,
      "step": 2236
    },
    {
      "epoch": 0.2863176756687572,
      "grad_norm": 0.69140625,
      "learning_rate": 0.001674902198528509,
      "loss": 1.6389,
      "step": 2237
    },
    {
      "epoch": 0.28644566747728145,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0016745962281879555,
      "loss": 2.198,
      "step": 2238
    },
    {
      "epoch": 0.2865736592858057,
      "grad_norm": 0.96484375,
      "learning_rate": 0.001674290141907095,
      "loss": 2.5121,
      "step": 2239
    },
    {
      "epoch": 0.28670165109433,
      "grad_norm": 0.65625,
      "learning_rate": 0.0016739839397385335,
      "loss": 2.536,
      "step": 2240
    },
    {
      "epoch": 0.28682964290285423,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0016736776217348966,
      "loss": 2.4081,
      "step": 2241
    },
    {
      "epoch": 0.2869576347113785,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0016733711879488302,
      "loss": 1.7229,
      "step": 2242
    },
    {
      "epoch": 0.2870856265199027,
      "grad_norm": 0.63671875,
      "learning_rate": 0.001673064638433,
      "loss": 2.0983,
      "step": 2243
    },
    {
      "epoch": 0.28721361832842696,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0016727579732400911,
      "loss": 1.489,
      "step": 2244
    },
    {
      "epoch": 0.28734161013695125,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0016724511924228094,
      "loss": 1.8602,
      "step": 2245
    },
    {
      "epoch": 0.2874696019454755,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00167214429603388,
      "loss": 1.3103,
      "step": 2246
    },
    {
      "epoch": 0.28759759375399974,
      "grad_norm": 0.640625,
      "learning_rate": 0.0016718372841260476,
      "loss": 1.9089,
      "step": 2247
    },
    {
      "epoch": 0.287725585562524,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0016715301567520777,
      "loss": 1.7999,
      "step": 2248
    },
    {
      "epoch": 0.2878535773710483,
      "grad_norm": 0.94921875,
      "learning_rate": 0.001671222913964755,
      "loss": 2.4093,
      "step": 2249
    },
    {
      "epoch": 0.2879815691795725,
      "grad_norm": 0.78125,
      "learning_rate": 0.0016709155558168843,
      "loss": 2.1312,
      "step": 2250
    },
    {
      "epoch": 0.28810956098809676,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0016706080823612895,
      "loss": 1.8269,
      "step": 2251
    },
    {
      "epoch": 0.288237552796621,
      "grad_norm": 0.96875,
      "learning_rate": 0.0016703004936508155,
      "loss": 2.5678,
      "step": 2252
    },
    {
      "epoch": 0.2883655446051453,
      "grad_norm": 0.703125,
      "learning_rate": 0.001669992789738326,
      "loss": 2.1799,
      "step": 2253
    },
    {
      "epoch": 0.28849353641366954,
      "grad_norm": 0.875,
      "learning_rate": 0.0016696849706767055,
      "loss": 2.1193,
      "step": 2254
    },
    {
      "epoch": 0.2886215282221938,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0016693770365188573,
      "loss": 1.6492,
      "step": 2255
    },
    {
      "epoch": 0.288749520030718,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0016690689873177043,
      "loss": 1.766,
      "step": 2256
    },
    {
      "epoch": 0.28887751183924226,
      "grad_norm": 0.65625,
      "learning_rate": 0.0016687608231261907,
      "loss": 1.9411,
      "step": 2257
    },
    {
      "epoch": 0.28900550364776656,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0016684525439972793,
      "loss": 2.3292,
      "step": 2258
    },
    {
      "epoch": 0.2891334954562908,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0016681441499839527,
      "loss": 1.2673,
      "step": 2259
    },
    {
      "epoch": 0.28926148726481504,
      "grad_norm": 0.765625,
      "learning_rate": 0.0016678356411392132,
      "loss": 1.8628,
      "step": 2260
    },
    {
      "epoch": 0.2893894790733393,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0016675270175160834,
      "loss": 2.0259,
      "step": 2261
    },
    {
      "epoch": 0.2895174708818636,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0016672182791676055,
      "loss": 2.522,
      "step": 2262
    },
    {
      "epoch": 0.2896454626903878,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0016669094261468405,
      "loss": 2.7048,
      "step": 2263
    },
    {
      "epoch": 0.28977345449891206,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0016666004585068703,
      "loss": 1.595,
      "step": 2264
    },
    {
      "epoch": 0.2899014463074363,
      "grad_norm": 0.578125,
      "learning_rate": 0.0016662913763007962,
      "loss": 1.6943,
      "step": 2265
    },
    {
      "epoch": 0.2900294381159606,
      "grad_norm": 0.6484375,
      "learning_rate": 0.001665982179581738,
      "loss": 1.8181,
      "step": 2266
    },
    {
      "epoch": 0.29015742992448484,
      "grad_norm": 0.7421875,
      "learning_rate": 0.001665672868402837,
      "loss": 2.3454,
      "step": 2267
    },
    {
      "epoch": 0.2902854217330091,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0016653634428172538,
      "loss": 1.8072,
      "step": 2268
    },
    {
      "epoch": 0.2904134135415333,
      "grad_norm": 0.6875,
      "learning_rate": 0.0016650539028781669,
      "loss": 2.2904,
      "step": 2269
    },
    {
      "epoch": 0.29054140535005757,
      "grad_norm": 0.6328125,
      "learning_rate": 0.001664744248638777,
      "loss": 1.7903,
      "step": 2270
    },
    {
      "epoch": 0.29066939715858187,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0016644344801523018,
      "loss": 2.4759,
      "step": 2271
    },
    {
      "epoch": 0.2907973889671061,
      "grad_norm": 0.828125,
      "learning_rate": 0.0016641245974719817,
      "loss": 2.28,
      "step": 2272
    },
    {
      "epoch": 0.29092538077563035,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0016638146006510742,
      "loss": 2.1623,
      "step": 2273
    },
    {
      "epoch": 0.2910533725841546,
      "grad_norm": 0.8203125,
      "learning_rate": 0.001663504489742857,
      "loss": 1.7928,
      "step": 2274
    },
    {
      "epoch": 0.2911813643926789,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0016631942648006283,
      "loss": 2.806,
      "step": 2275
    },
    {
      "epoch": 0.29130935620120313,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0016628839258777046,
      "loss": 2.6083,
      "step": 2276
    },
    {
      "epoch": 0.29143734800972737,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0016625734730274235,
      "loss": 1.6767,
      "step": 2277
    },
    {
      "epoch": 0.2915653398182516,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0016622629063031411,
      "loss": 1.733,
      "step": 2278
    },
    {
      "epoch": 0.2916933316267759,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0016619522257582328,
      "loss": 2.1218,
      "step": 2279
    },
    {
      "epoch": 0.29182132343530015,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0016616414314460947,
      "loss": 1.5823,
      "step": 2280
    },
    {
      "epoch": 0.2919493152438244,
      "grad_norm": 0.7421875,
      "learning_rate": 0.001661330523420142,
      "loss": 2.4062,
      "step": 2281
    },
    {
      "epoch": 0.29207730705234863,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0016610195017338083,
      "loss": 1.9095,
      "step": 2282
    },
    {
      "epoch": 0.29220529886087293,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0016607083664405484,
      "loss": 1.8142,
      "step": 2283
    },
    {
      "epoch": 0.29233329066939717,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0016603971175938362,
      "loss": 2.1675,
      "step": 2284
    },
    {
      "epoch": 0.2924612824779214,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0016600857552471645,
      "loss": 2.121,
      "step": 2285
    },
    {
      "epoch": 0.29258927428644566,
      "grad_norm": 0.609375,
      "learning_rate": 0.001659774279454046,
      "loss": 1.4831,
      "step": 2286
    },
    {
      "epoch": 0.2927172660949699,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0016594626902680127,
      "loss": 2.33,
      "step": 2287
    },
    {
      "epoch": 0.2928452579034942,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0016591509877426167,
      "loss": 1.6101,
      "step": 2288
    },
    {
      "epoch": 0.29297324971201844,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0016588391719314286,
      "loss": 1.8637,
      "step": 2289
    },
    {
      "epoch": 0.2931012415205427,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00165852724288804,
      "loss": 1.3932,
      "step": 2290
    },
    {
      "epoch": 0.2932292333290669,
      "grad_norm": 0.703125,
      "learning_rate": 0.0016582152006660595,
      "loss": 2.0144,
      "step": 2291
    },
    {
      "epoch": 0.2933572251375912,
      "grad_norm": 0.84375,
      "learning_rate": 0.0016579030453191177,
      "loss": 2.4574,
      "step": 2292
    },
    {
      "epoch": 0.29348521694611546,
      "grad_norm": 0.7109375,
      "learning_rate": 0.001657590776900863,
      "loss": 2.0244,
      "step": 2293
    },
    {
      "epoch": 0.2936132087546397,
      "grad_norm": 0.75,
      "learning_rate": 0.0016572783954649643,
      "loss": 2.1161,
      "step": 2294
    },
    {
      "epoch": 0.29374120056316394,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0016569659010651093,
      "loss": 2.3055,
      "step": 2295
    },
    {
      "epoch": 0.29386919237168824,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0016566532937550048,
      "loss": 1.7893,
      "step": 2296
    },
    {
      "epoch": 0.2939971841802125,
      "grad_norm": 0.640625,
      "learning_rate": 0.0016563405735883778,
      "loss": 1.7875,
      "step": 2297
    },
    {
      "epoch": 0.2941251759887367,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0016560277406189742,
      "loss": 1.791,
      "step": 2298
    },
    {
      "epoch": 0.29425316779726096,
      "grad_norm": 0.78125,
      "learning_rate": 0.0016557147949005596,
      "loss": 1.9242,
      "step": 2299
    },
    {
      "epoch": 0.2943811596057852,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0016554017364869185,
      "loss": 1.9023,
      "step": 2300
    },
    {
      "epoch": 0.2945091514143095,
      "grad_norm": 0.703125,
      "learning_rate": 0.0016550885654318555,
      "loss": 2.4082,
      "step": 2301
    },
    {
      "epoch": 0.29463714322283374,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0016547752817891936,
      "loss": 1.8531,
      "step": 2302
    },
    {
      "epoch": 0.294765135031358,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0016544618856127761,
      "loss": 1.8782,
      "step": 2303
    },
    {
      "epoch": 0.2948931268398822,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0016541483769564654,
      "loss": 1.7767,
      "step": 2304
    },
    {
      "epoch": 0.2950211186484065,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0016538347558741422,
      "loss": 2.0353,
      "step": 2305
    },
    {
      "epoch": 0.29514911045693076,
      "grad_norm": 0.7109375,
      "learning_rate": 0.001653521022419708,
      "loss": 2.4879,
      "step": 2306
    },
    {
      "epoch": 0.295277102265455,
      "grad_norm": 0.62109375,
      "learning_rate": 0.001653207176647083,
      "loss": 2.1444,
      "step": 2307
    },
    {
      "epoch": 0.29540509407397925,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0016528932186102067,
      "loss": 2.059,
      "step": 2308
    },
    {
      "epoch": 0.29553308588250354,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0016525791483630375,
      "loss": 2.133,
      "step": 2309
    },
    {
      "epoch": 0.2956610776910278,
      "grad_norm": 0.62890625,
      "learning_rate": 0.001652264965959554,
      "loss": 1.8686,
      "step": 2310
    },
    {
      "epoch": 0.295789069499552,
      "grad_norm": 0.8671875,
      "learning_rate": 0.001651950671453753,
      "loss": 2.5475,
      "step": 2311
    },
    {
      "epoch": 0.29591706130807627,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0016516362648996514,
      "loss": 2.5135,
      "step": 2312
    },
    {
      "epoch": 0.29604505311660056,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0016513217463512851,
      "loss": 2.1897,
      "step": 2313
    },
    {
      "epoch": 0.2961730449251248,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0016510071158627089,
      "loss": 1.9733,
      "step": 2314
    },
    {
      "epoch": 0.29630103673364905,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0016506923734879977,
      "loss": 1.6965,
      "step": 2315
    },
    {
      "epoch": 0.2964290285421733,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0016503775192812447,
      "loss": 2.1632,
      "step": 2316
    },
    {
      "epoch": 0.29655702035069753,
      "grad_norm": 0.8125,
      "learning_rate": 0.001650062553296563,
      "loss": 2.12,
      "step": 2317
    },
    {
      "epoch": 0.29668501215922183,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0016497474755880841,
      "loss": 1.7067,
      "step": 2318
    },
    {
      "epoch": 0.29681300396774607,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0016494322862099598,
      "loss": 2.3262,
      "step": 2319
    },
    {
      "epoch": 0.2969409957762703,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0016491169852163598,
      "loss": 1.3074,
      "step": 2320
    },
    {
      "epoch": 0.29706898758479455,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0016488015726614745,
      "loss": 2.0108,
      "step": 2321
    },
    {
      "epoch": 0.29719697939331885,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0016484860485995122,
      "loss": 1.7387,
      "step": 2322
    },
    {
      "epoch": 0.2973249712018431,
      "grad_norm": 0.75390625,
      "learning_rate": 0.001648170413084701,
      "loss": 2.0087,
      "step": 2323
    },
    {
      "epoch": 0.29745296301036733,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0016478546661712876,
      "loss": 1.9789,
      "step": 2324
    },
    {
      "epoch": 0.2975809548188916,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0016475388079135386,
      "loss": 2.2509,
      "step": 2325
    },
    {
      "epoch": 0.29770894662741587,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0016472228383657397,
      "loss": 2.175,
      "step": 2326
    },
    {
      "epoch": 0.2978369384359401,
      "grad_norm": 0.828125,
      "learning_rate": 0.0016469067575821945,
      "loss": 2.0688,
      "step": 2327
    },
    {
      "epoch": 0.29796493024446435,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0016465905656172276,
      "loss": 1.6159,
      "step": 2328
    },
    {
      "epoch": 0.2980929220529886,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0016462742625251807,
      "loss": 2.6066,
      "step": 2329
    },
    {
      "epoch": 0.29822091386151284,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0016459578483604168,
      "loss": 1.7553,
      "step": 2330
    },
    {
      "epoch": 0.29834890567003713,
      "grad_norm": 0.6875,
      "learning_rate": 0.0016456413231773156,
      "loss": 1.9752,
      "step": 2331
    },
    {
      "epoch": 0.2984768974785614,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0016453246870302776,
      "loss": 2.1476,
      "step": 2332
    },
    {
      "epoch": 0.2986048892870856,
      "grad_norm": 0.67578125,
      "learning_rate": 0.001645007939973722,
      "loss": 1.6078,
      "step": 2333
    },
    {
      "epoch": 0.29873288109560986,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0016446910820620874,
      "loss": 2.376,
      "step": 2334
    },
    {
      "epoch": 0.29886087290413416,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0016443741133498296,
      "loss": 1.7918,
      "step": 2335
    },
    {
      "epoch": 0.2989888647126584,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0016440570338914259,
      "loss": 1.7562,
      "step": 2336
    },
    {
      "epoch": 0.29911685652118264,
      "grad_norm": 0.6875,
      "learning_rate": 0.001643739843741371,
      "loss": 2.2888,
      "step": 2337
    },
    {
      "epoch": 0.2992448483297069,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00164342254295418,
      "loss": 1.587,
      "step": 2338
    },
    {
      "epoch": 0.2993728401382312,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0016431051315843853,
      "loss": 1.816,
      "step": 2339
    },
    {
      "epoch": 0.2995008319467554,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0016427876096865393,
      "loss": 2.1615,
      "step": 2340
    },
    {
      "epoch": 0.29962882375527966,
      "grad_norm": 0.6796875,
      "learning_rate": 0.001642469977315214,
      "loss": 2.2574,
      "step": 2341
    },
    {
      "epoch": 0.2997568155638039,
      "grad_norm": 0.71875,
      "learning_rate": 0.0016421522345249986,
      "loss": 2.2724,
      "step": 2342
    },
    {
      "epoch": 0.29988480737232814,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0016418343813705035,
      "loss": 2.1061,
      "step": 2343
    },
    {
      "epoch": 0.30001279918085244,
      "grad_norm": 0.69140625,
      "learning_rate": 0.001641516417906356,
      "loss": 1.4284,
      "step": 2344
    },
    {
      "epoch": 0.3001407909893767,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0016411983441872034,
      "loss": 2.0532,
      "step": 2345
    },
    {
      "epoch": 0.3002687827979009,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0016408801602677127,
      "loss": 2.3173,
      "step": 2346
    },
    {
      "epoch": 0.30039677460642517,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0016405618662025678,
      "loss": 1.599,
      "step": 2347
    },
    {
      "epoch": 0.30052476641494946,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0016402434620464736,
      "loss": 1.926,
      "step": 2348
    },
    {
      "epoch": 0.3006527582234737,
      "grad_norm": 0.75,
      "learning_rate": 0.0016399249478541524,
      "loss": 2.1439,
      "step": 2349
    },
    {
      "epoch": 0.30078075003199795,
      "grad_norm": 0.703125,
      "learning_rate": 0.0016396063236803465,
      "loss": 2.1182,
      "step": 2350
    },
    {
      "epoch": 0.3009087418405222,
      "grad_norm": 0.640625,
      "learning_rate": 0.001639287589579816,
      "loss": 1.9703,
      "step": 2351
    },
    {
      "epoch": 0.3010367336490465,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0016389687456073415,
      "loss": 2.249,
      "step": 2352
    },
    {
      "epoch": 0.3011647254575707,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0016386497918177206,
      "loss": 1.7458,
      "step": 2353
    },
    {
      "epoch": 0.30129271726609497,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0016383307282657712,
      "loss": 1.6477,
      "step": 2354
    },
    {
      "epoch": 0.3014207090746192,
      "grad_norm": 0.78125,
      "learning_rate": 0.0016380115550063293,
      "loss": 2.0693,
      "step": 2355
    },
    {
      "epoch": 0.3015487008831435,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00163769227209425,
      "loss": 1.9739,
      "step": 2356
    },
    {
      "epoch": 0.30167669269166775,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0016373728795844077,
      "loss": 2.0428,
      "step": 2357
    },
    {
      "epoch": 0.301804684500192,
      "grad_norm": 0.640625,
      "learning_rate": 0.0016370533775316944,
      "loss": 1.4827,
      "step": 2358
    },
    {
      "epoch": 0.30193267630871623,
      "grad_norm": 1.09375,
      "learning_rate": 0.0016367337659910222,
      "loss": 2.7486,
      "step": 2359
    },
    {
      "epoch": 0.30206066811724047,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0016364140450173218,
      "loss": 1.3969,
      "step": 2360
    },
    {
      "epoch": 0.30218865992576477,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0016360942146655417,
      "loss": 1.9208,
      "step": 2361
    },
    {
      "epoch": 0.302316651734289,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0016357742749906503,
      "loss": 2.0457,
      "step": 2362
    },
    {
      "epoch": 0.30244464354281325,
      "grad_norm": 0.703125,
      "learning_rate": 0.0016354542260476348,
      "loss": 2.0139,
      "step": 2363
    },
    {
      "epoch": 0.3025726353513375,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0016351340678915003,
      "loss": 1.7628,
      "step": 2364
    },
    {
      "epoch": 0.3027006271598618,
      "grad_norm": 0.59375,
      "learning_rate": 0.0016348138005772715,
      "loss": 1.1659,
      "step": 2365
    },
    {
      "epoch": 0.30282861896838603,
      "grad_norm": 0.71484375,
      "learning_rate": 0.001634493424159991,
      "loss": 1.6071,
      "step": 2366
    },
    {
      "epoch": 0.3029566107769103,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0016341729386947213,
      "loss": 1.888,
      "step": 2367
    },
    {
      "epoch": 0.3030846025854345,
      "grad_norm": 0.703125,
      "learning_rate": 0.001633852344236543,
      "loss": 2.1852,
      "step": 2368
    },
    {
      "epoch": 0.3032125943939588,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0016335316408405548,
      "loss": 1.9292,
      "step": 2369
    },
    {
      "epoch": 0.30334058620248305,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0016332108285618755,
      "loss": 2.2784,
      "step": 2370
    },
    {
      "epoch": 0.3034685780110073,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0016328899074556414,
      "loss": 1.9515,
      "step": 2371
    },
    {
      "epoch": 0.30359656981953154,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0016325688775770085,
      "loss": 1.3651,
      "step": 2372
    },
    {
      "epoch": 0.3037245616280558,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0016322477389811504,
      "loss": 2.4638,
      "step": 2373
    },
    {
      "epoch": 0.3038525534365801,
      "grad_norm": 0.734375,
      "learning_rate": 0.0016319264917232608,
      "loss": 2.0643,
      "step": 2374
    },
    {
      "epoch": 0.3039805452451043,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0016316051358585503,
      "loss": 1.3587,
      "step": 2375
    },
    {
      "epoch": 0.30410853705362856,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0016312836714422498,
      "loss": 2.0454,
      "step": 2376
    },
    {
      "epoch": 0.3042365288621528,
      "grad_norm": 0.78125,
      "learning_rate": 0.0016309620985296073,
      "loss": 2.2278,
      "step": 2377
    },
    {
      "epoch": 0.3043645206706771,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0016306404171758917,
      "loss": 1.349,
      "step": 2378
    },
    {
      "epoch": 0.30449251247920134,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0016303186274363882,
      "loss": 2.7405,
      "step": 2379
    },
    {
      "epoch": 0.3046205042877256,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0016299967293664018,
      "loss": 2.0896,
      "step": 2380
    },
    {
      "epoch": 0.3047484960962498,
      "grad_norm": 0.609375,
      "learning_rate": 0.001629674723021256,
      "loss": 1.4432,
      "step": 2381
    },
    {
      "epoch": 0.3048764879047741,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0016293526084562927,
      "loss": 2.7991,
      "step": 2382
    },
    {
      "epoch": 0.30500447971329836,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0016290303857268724,
      "loss": 2.5463,
      "step": 2383
    },
    {
      "epoch": 0.3051324715218226,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0016287080548883744,
      "loss": 1.2692,
      "step": 2384
    },
    {
      "epoch": 0.30526046333034684,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0016283856159961967,
      "loss": 2.2823,
      "step": 2385
    },
    {
      "epoch": 0.30538845513887114,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0016280630691057553,
      "loss": 1.6169,
      "step": 2386
    },
    {
      "epoch": 0.3055164469473954,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0016277404142724854,
      "loss": 1.395,
      "step": 2387
    },
    {
      "epoch": 0.3056444387559196,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0016274176515518406,
      "loss": 2.2512,
      "step": 2388
    },
    {
      "epoch": 0.30577243056444386,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0016270947809992925,
      "loss": 1.8281,
      "step": 2389
    },
    {
      "epoch": 0.3059004223729681,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0016267718026703318,
      "loss": 2.8004,
      "step": 2390
    },
    {
      "epoch": 0.3060284141814924,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0016264487166204678,
      "loss": 1.7855,
      "step": 2391
    },
    {
      "epoch": 0.30615640599001664,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0016261255229052275,
      "loss": 2.2853,
      "step": 2392
    },
    {
      "epoch": 0.3062843977985409,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0016258022215801576,
      "loss": 2.1509,
      "step": 2393
    },
    {
      "epoch": 0.3064123896070651,
      "grad_norm": 0.65625,
      "learning_rate": 0.0016254788127008225,
      "loss": 2.0921,
      "step": 2394
    },
    {
      "epoch": 0.3065403814155894,
      "grad_norm": 0.80859375,
      "learning_rate": 0.001625155296322805,
      "loss": 1.9499,
      "step": 2395
    },
    {
      "epoch": 0.30666837322411367,
      "grad_norm": 0.734375,
      "learning_rate": 0.0016248316725017066,
      "loss": 2.1459,
      "step": 2396
    },
    {
      "epoch": 0.3067963650326379,
      "grad_norm": 0.67578125,
      "learning_rate": 0.001624507941293148,
      "loss": 1.9803,
      "step": 2397
    },
    {
      "epoch": 0.30692435684116215,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0016241841027527669,
      "loss": 2.2875,
      "step": 2398
    },
    {
      "epoch": 0.30705234864968645,
      "grad_norm": 0.79296875,
      "learning_rate": 0.001623860156936221,
      "loss": 1.7612,
      "step": 2399
    },
    {
      "epoch": 0.3071803404582107,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0016235361038991845,
      "loss": 2.5925,
      "step": 2400
    },
    {
      "epoch": 0.30730833226673493,
      "grad_norm": 1.171875,
      "learning_rate": 0.0016232119436973523,
      "loss": 2.2328,
      "step": 2401
    },
    {
      "epoch": 0.30743632407525917,
      "grad_norm": 0.734375,
      "learning_rate": 0.0016228876763864357,
      "loss": 2.2595,
      "step": 2402
    },
    {
      "epoch": 0.3075643158837834,
      "grad_norm": 0.625,
      "learning_rate": 0.001622563302022166,
      "loss": 1.9799,
      "step": 2403
    },
    {
      "epoch": 0.3076923076923077,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0016222388206602921,
      "loss": 1.8567,
      "step": 2404
    },
    {
      "epoch": 0.30782029950083195,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0016219142323565807,
      "loss": 2.0031,
      "step": 2405
    },
    {
      "epoch": 0.3079482913093562,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0016215895371668183,
      "loss": 2.0958,
      "step": 2406
    },
    {
      "epoch": 0.30807628311788043,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0016212647351468087,
      "loss": 1.6928,
      "step": 2407
    },
    {
      "epoch": 0.30820427492640473,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0016209398263523746,
      "loss": 1.8195,
      "step": 2408
    },
    {
      "epoch": 0.308332266734929,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0016206148108393569,
      "loss": 2.2985,
      "step": 2409
    },
    {
      "epoch": 0.3084602585434532,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0016202896886636144,
      "loss": 1.6776,
      "step": 2410
    },
    {
      "epoch": 0.30858825035197746,
      "grad_norm": 0.640625,
      "learning_rate": 0.0016199644598810248,
      "loss": 1.9467,
      "step": 2411
    },
    {
      "epoch": 0.30871624216050175,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0016196391245474843,
      "loss": 1.9306,
      "step": 2412
    },
    {
      "epoch": 0.308844233969026,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0016193136827189066,
      "loss": 2.1253,
      "step": 2413
    },
    {
      "epoch": 0.30897222577755024,
      "grad_norm": 0.640625,
      "learning_rate": 0.0016189881344512243,
      "loss": 2.7192,
      "step": 2414
    },
    {
      "epoch": 0.3091002175860745,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0016186624798003879,
      "loss": 1.5971,
      "step": 2415
    },
    {
      "epoch": 0.3092282093945987,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0016183367188223672,
      "loss": 1.7307,
      "step": 2416
    },
    {
      "epoch": 0.309356201203123,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0016180108515731484,
      "loss": 2.052,
      "step": 2417
    },
    {
      "epoch": 0.30948419301164726,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0016176848781087384,
      "loss": 2.0716,
      "step": 2418
    },
    {
      "epoch": 0.3096121848201715,
      "grad_norm": 0.703125,
      "learning_rate": 0.0016173587984851597,
      "loss": 2.4103,
      "step": 2419
    },
    {
      "epoch": 0.30974017662869574,
      "grad_norm": 0.62109375,
      "learning_rate": 0.001617032612758455,
      "loss": 1.9785,
      "step": 2420
    },
    {
      "epoch": 0.30986816843722004,
      "grad_norm": 0.7890625,
      "learning_rate": 0.001616706320984685,
      "loss": 2.4731,
      "step": 2421
    },
    {
      "epoch": 0.3099961602457443,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0016163799232199274,
      "loss": 2.2947,
      "step": 2422
    },
    {
      "epoch": 0.3101241520542685,
      "grad_norm": 0.671875,
      "learning_rate": 0.00161605341952028,
      "loss": 2.1623,
      "step": 2423
    },
    {
      "epoch": 0.31025214386279276,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0016157268099418567,
      "loss": 1.8731,
      "step": 2424
    },
    {
      "epoch": 0.31038013567131706,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0016154000945407911,
      "loss": 1.9689,
      "step": 2425
    },
    {
      "epoch": 0.3105081274798413,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0016150732733732345,
      "loss": 1.6827,
      "step": 2426
    },
    {
      "epoch": 0.31063611928836554,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0016147463464953567,
      "loss": 2.0468,
      "step": 2427
    },
    {
      "epoch": 0.3107641110968898,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0016144193139633449,
      "loss": 2.2374,
      "step": 2428
    },
    {
      "epoch": 0.3108921029054141,
      "grad_norm": 0.62109375,
      "learning_rate": 0.001614092175833405,
      "loss": 1.5525,
      "step": 2429
    },
    {
      "epoch": 0.3110200947139383,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0016137649321617618,
      "loss": 1.8048,
      "step": 2430
    },
    {
      "epoch": 0.31114808652246256,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0016134375830046564,
      "loss": 1.862,
      "step": 2431
    },
    {
      "epoch": 0.3112760783309868,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0016131101284183496,
      "loss": 2.4082,
      "step": 2432
    },
    {
      "epoch": 0.31140407013951105,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0016127825684591195,
      "loss": 1.9432,
      "step": 2433
    },
    {
      "epoch": 0.31153206194803534,
      "grad_norm": 0.67578125,
      "learning_rate": 0.001612454903183263,
      "loss": 2.1635,
      "step": 2434
    },
    {
      "epoch": 0.3116600537565596,
      "grad_norm": 0.59375,
      "learning_rate": 0.0016121271326470944,
      "loss": 1.5595,
      "step": 2435
    },
    {
      "epoch": 0.3117880455650838,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0016117992569069464,
      "loss": 1.8602,
      "step": 2436
    },
    {
      "epoch": 0.31191603737360807,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0016114712760191697,
      "loss": 1.8645,
      "step": 2437
    },
    {
      "epoch": 0.31204402918213237,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0016111431900401331,
      "loss": 1.7426,
      "step": 2438
    },
    {
      "epoch": 0.3121720209906566,
      "grad_norm": 0.8046875,
      "learning_rate": 0.001610814999026224,
      "loss": 2.184,
      "step": 2439
    },
    {
      "epoch": 0.31230001279918085,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0016104867030338468,
      "loss": 1.8106,
      "step": 2440
    },
    {
      "epoch": 0.3124280046077051,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0016101583021194247,
      "loss": 2.0057,
      "step": 2441
    },
    {
      "epoch": 0.3125559964162294,
      "grad_norm": 0.640625,
      "learning_rate": 0.0016098297963393988,
      "loss": 2.1331,
      "step": 2442
    },
    {
      "epoch": 0.31268398822475363,
      "grad_norm": 0.71875,
      "learning_rate": 0.0016095011857502283,
      "loss": 2.1841,
      "step": 2443
    },
    {
      "epoch": 0.31281198003327787,
      "grad_norm": 0.703125,
      "learning_rate": 0.0016091724704083896,
      "loss": 1.8123,
      "step": 2444
    },
    {
      "epoch": 0.3129399718418021,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0016088436503703782,
      "loss": 1.5645,
      "step": 2445
    },
    {
      "epoch": 0.31306796365032635,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0016085147256927072,
      "loss": 2.1576,
      "step": 2446
    },
    {
      "epoch": 0.31319595545885065,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0016081856964319078,
      "loss": 2.5769,
      "step": 2447
    },
    {
      "epoch": 0.3133239472673749,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0016078565626445288,
      "loss": 2.5833,
      "step": 2448
    },
    {
      "epoch": 0.31345193907589913,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0016075273243871368,
      "loss": 2.0971,
      "step": 2449
    },
    {
      "epoch": 0.3135799308844234,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0016071979817163173,
      "loss": 2.2803,
      "step": 2450
    },
    {
      "epoch": 0.31370792269294767,
      "grad_norm": 0.65625,
      "learning_rate": 0.0016068685346886731,
      "loss": 1.572,
      "step": 2451
    },
    {
      "epoch": 0.3138359145014719,
      "grad_norm": 0.734375,
      "learning_rate": 0.001606538983360825,
      "loss": 1.9218,
      "step": 2452
    },
    {
      "epoch": 0.31396390630999615,
      "grad_norm": 0.734375,
      "learning_rate": 0.0016062093277894114,
      "loss": 2.2559,
      "step": 2453
    },
    {
      "epoch": 0.3140918981185204,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0016058795680310896,
      "loss": 1.5226,
      "step": 2454
    },
    {
      "epoch": 0.3142198899270447,
      "grad_norm": 0.65625,
      "learning_rate": 0.0016055497041425333,
      "loss": 2.114,
      "step": 2455
    },
    {
      "epoch": 0.31434788173556893,
      "grad_norm": 0.65625,
      "learning_rate": 0.0016052197361804358,
      "loss": 2.1792,
      "step": 2456
    },
    {
      "epoch": 0.3144758735440932,
      "grad_norm": 0.75390625,
      "learning_rate": 0.001604889664201507,
      "loss": 2.1734,
      "step": 2457
    },
    {
      "epoch": 0.3146038653526174,
      "grad_norm": 0.8125,
      "learning_rate": 0.001604559488262475,
      "loss": 1.8134,
      "step": 2458
    },
    {
      "epoch": 0.31473185716114166,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0016042292084200863,
      "loss": 2.4217,
      "step": 2459
    },
    {
      "epoch": 0.31485984896966596,
      "grad_norm": 0.71484375,
      "learning_rate": 0.001603898824731105,
      "loss": 1.483,
      "step": 2460
    },
    {
      "epoch": 0.3149878407781902,
      "grad_norm": 0.8984375,
      "learning_rate": 0.001603568337252312,
      "loss": 1.6852,
      "step": 2461
    },
    {
      "epoch": 0.31511583258671444,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0016032377460405076,
      "loss": 2.1321,
      "step": 2462
    },
    {
      "epoch": 0.3152438243952387,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0016029070511525095,
      "loss": 2.7639,
      "step": 2463
    },
    {
      "epoch": 0.315371816203763,
      "grad_norm": 0.6640625,
      "learning_rate": 0.001602576252645152,
      "loss": 1.8637,
      "step": 2464
    },
    {
      "epoch": 0.3154998080122872,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0016022453505752897,
      "loss": 2.017,
      "step": 2465
    },
    {
      "epoch": 0.31562779982081146,
      "grad_norm": 0.5703125,
      "learning_rate": 0.001601914344999792,
      "loss": 1.7595,
      "step": 2466
    },
    {
      "epoch": 0.3157557916293357,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0016015832359755483,
      "loss": 1.5627,
      "step": 2467
    },
    {
      "epoch": 0.31588378343786,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0016012520235594652,
      "loss": 1.0548,
      "step": 2468
    },
    {
      "epoch": 0.31601177524638424,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0016009207078084664,
      "loss": 1.8276,
      "step": 2469
    },
    {
      "epoch": 0.3161397670549085,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0016005892887794942,
      "loss": 1.805,
      "step": 2470
    },
    {
      "epoch": 0.3162677588634327,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0016002577665295084,
      "loss": 1.7979,
      "step": 2471
    },
    {
      "epoch": 0.316395750671957,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0015999261411154865,
      "loss": 1.8488,
      "step": 2472
    },
    {
      "epoch": 0.31652374248048126,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0015995944125944232,
      "loss": 2.2302,
      "step": 2473
    },
    {
      "epoch": 0.3166517342890055,
      "grad_norm": 0.62890625,
      "learning_rate": 0.001599262581023332,
      "loss": 1.8294,
      "step": 2474
    },
    {
      "epoch": 0.31677972609752975,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0015989306464592435,
      "loss": 2.2851,
      "step": 2475
    },
    {
      "epoch": 0.316907717906054,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0015985986089592059,
      "loss": 2.043,
      "step": 2476
    },
    {
      "epoch": 0.3170357097145783,
      "grad_norm": 0.70703125,
      "learning_rate": 0.001598266468580285,
      "loss": 1.8559,
      "step": 2477
    },
    {
      "epoch": 0.3171637015231025,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0015979342253795651,
      "loss": 1.7928,
      "step": 2478
    },
    {
      "epoch": 0.31729169333162677,
      "grad_norm": 0.69140625,
      "learning_rate": 0.001597601879414147,
      "loss": 2.0841,
      "step": 2479
    },
    {
      "epoch": 0.317419685140151,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0015972694307411501,
      "loss": 2.2738,
      "step": 2480
    },
    {
      "epoch": 0.3175476769486753,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0015969368794177107,
      "loss": 2.1731,
      "step": 2481
    },
    {
      "epoch": 0.31767566875719955,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0015966042255009835,
      "loss": 2.1865,
      "step": 2482
    },
    {
      "epoch": 0.3178036605657238,
      "grad_norm": 0.828125,
      "learning_rate": 0.0015962714690481406,
      "loss": 2.7187,
      "step": 2483
    },
    {
      "epoch": 0.31793165237424803,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0015959386101163714,
      "loss": 2.2262,
      "step": 2484
    },
    {
      "epoch": 0.3180596441827723,
      "grad_norm": 0.76953125,
      "learning_rate": 0.001595605648762883,
      "loss": 1.9918,
      "step": 2485
    },
    {
      "epoch": 0.31818763599129657,
      "grad_norm": 0.59375,
      "learning_rate": 0.0015952725850449001,
      "loss": 1.59,
      "step": 2486
    },
    {
      "epoch": 0.3183156277998208,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0015949394190196658,
      "loss": 1.8648,
      "step": 2487
    },
    {
      "epoch": 0.31844361960834505,
      "grad_norm": 0.76171875,
      "learning_rate": 0.001594606150744439,
      "loss": 1.8528,
      "step": 2488
    },
    {
      "epoch": 0.3185716114168693,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0015942727802764982,
      "loss": 1.7796,
      "step": 2489
    },
    {
      "epoch": 0.3186996032253936,
      "grad_norm": 0.578125,
      "learning_rate": 0.001593939307673138,
      "loss": 1.3752,
      "step": 2490
    },
    {
      "epoch": 0.31882759503391783,
      "grad_norm": 0.765625,
      "learning_rate": 0.0015936057329916716,
      "loss": 1.636,
      "step": 2491
    },
    {
      "epoch": 0.3189555868424421,
      "grad_norm": 0.6875,
      "learning_rate": 0.0015932720562894286,
      "loss": 1.9692,
      "step": 2492
    },
    {
      "epoch": 0.3190835786509663,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0015929382776237568,
      "loss": 1.6485,
      "step": 2493
    },
    {
      "epoch": 0.3192115704594906,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0015926043970520222,
      "loss": 2.2342,
      "step": 2494
    },
    {
      "epoch": 0.31933956226801485,
      "grad_norm": 0.71875,
      "learning_rate": 0.0015922704146316066,
      "loss": 1.9326,
      "step": 2495
    },
    {
      "epoch": 0.3194675540765391,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0015919363304199112,
      "loss": 1.4872,
      "step": 2496
    },
    {
      "epoch": 0.31959554588506334,
      "grad_norm": 0.671875,
      "learning_rate": 0.0015916021444743534,
      "loss": 1.4434,
      "step": 2497
    },
    {
      "epoch": 0.31972353769358763,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0015912678568523685,
      "loss": 2.0994,
      "step": 2498
    },
    {
      "epoch": 0.3198515295021119,
      "grad_norm": 0.67578125,
      "learning_rate": 0.001590933467611409,
      "loss": 1.659,
      "step": 2499
    },
    {
      "epoch": 0.3199795213106361,
      "grad_norm": 0.625,
      "learning_rate": 0.0015905989768089453,
      "loss": 1.5854,
      "step": 2500
    },
    {
      "epoch": 0.32010751311916036,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0015902643845024651,
      "loss": 2.3666,
      "step": 2501
    },
    {
      "epoch": 0.32023550492768466,
      "grad_norm": 0.796875,
      "learning_rate": 0.0015899296907494738,
      "loss": 1.8525,
      "step": 2502
    },
    {
      "epoch": 0.3203634967362089,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0015895948956074935,
      "loss": 1.7156,
      "step": 2503
    },
    {
      "epoch": 0.32049148854473314,
      "grad_norm": 0.6328125,
      "learning_rate": 0.001589259999134064,
      "loss": 1.3789,
      "step": 2504
    },
    {
      "epoch": 0.3206194803532574,
      "grad_norm": 0.578125,
      "learning_rate": 0.0015889250013867437,
      "loss": 1.61,
      "step": 2505
    },
    {
      "epoch": 0.3207474721617816,
      "grad_norm": 0.625,
      "learning_rate": 0.0015885899024231064,
      "loss": 1.8361,
      "step": 2506
    },
    {
      "epoch": 0.3208754639703059,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0015882547023007448,
      "loss": 2.7552,
      "step": 2507
    },
    {
      "epoch": 0.32100345577883016,
      "grad_norm": 0.93359375,
      "learning_rate": 0.001587919401077268,
      "loss": 2.1026,
      "step": 2508
    },
    {
      "epoch": 0.3211314475873544,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0015875839988103038,
      "loss": 2.1325,
      "step": 2509
    },
    {
      "epoch": 0.32125943939587864,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0015872484955574955,
      "loss": 2.1547,
      "step": 2510
    },
    {
      "epoch": 0.32138743120440294,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0015869128913765054,
      "loss": 1.0042,
      "step": 2511
    },
    {
      "epoch": 0.3215154230129272,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0015865771863250125,
      "loss": 1.9183,
      "step": 2512
    },
    {
      "epoch": 0.3216434148214514,
      "grad_norm": 0.6953125,
      "learning_rate": 0.001586241380460713,
      "loss": 1.7418,
      "step": 2513
    },
    {
      "epoch": 0.32177140662997566,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0015859054738413209,
      "loss": 1.9188,
      "step": 2514
    },
    {
      "epoch": 0.32189939843849996,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0015855694665245666,
      "loss": 2.4601,
      "step": 2515
    },
    {
      "epoch": 0.3220273902470242,
      "grad_norm": 0.6328125,
      "learning_rate": 0.001585233358568199,
      "loss": 1.9397,
      "step": 2516
    },
    {
      "epoch": 0.32215538205554844,
      "grad_norm": 0.59375,
      "learning_rate": 0.0015848971500299828,
      "loss": 1.9245,
      "step": 2517
    },
    {
      "epoch": 0.3222833738640727,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0015845608409677024,
      "loss": 1.6154,
      "step": 2518
    },
    {
      "epoch": 0.32241136567259693,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0015842244314391567,
      "loss": 2.3726,
      "step": 2519
    },
    {
      "epoch": 0.3225393574811212,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0015838879215021638,
      "loss": 2.3714,
      "step": 2520
    },
    {
      "epoch": 0.32266734928964547,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0015835513112145581,
      "loss": 1.7793,
      "step": 2521
    },
    {
      "epoch": 0.3227953410981697,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0015832146006341914,
      "loss": 1.4316,
      "step": 2522
    },
    {
      "epoch": 0.32292333290669395,
      "grad_norm": 0.81640625,
      "learning_rate": 0.001582877789818933,
      "loss": 1.6525,
      "step": 2523
    },
    {
      "epoch": 0.32305132471521825,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0015825408788266695,
      "loss": 1.7757,
      "step": 2524
    },
    {
      "epoch": 0.3231793165237425,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0015822038677153045,
      "loss": 2.8677,
      "step": 2525
    },
    {
      "epoch": 0.32330730833226673,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0015818667565427585,
      "loss": 1.7011,
      "step": 2526
    },
    {
      "epoch": 0.32343530014079097,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00158152954536697,
      "loss": 2.1351,
      "step": 2527
    },
    {
      "epoch": 0.32356329194931527,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0015811922342458937,
      "loss": 2.0418,
      "step": 2528
    },
    {
      "epoch": 0.3236912837578395,
      "grad_norm": 0.625,
      "learning_rate": 0.0015808548232375027,
      "loss": 1.4328,
      "step": 2529
    },
    {
      "epoch": 0.32381927556636375,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0015805173123997855,
      "loss": 1.9849,
      "step": 2530
    },
    {
      "epoch": 0.323947267374888,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0015801797017907498,
      "loss": 1.4235,
      "step": 2531
    },
    {
      "epoch": 0.32407525918341223,
      "grad_norm": 0.6171875,
      "learning_rate": 0.001579841991468419,
      "loss": 1.4632,
      "step": 2532
    },
    {
      "epoch": 0.32420325099193653,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0015795041814908339,
      "loss": 1.6792,
      "step": 2533
    },
    {
      "epoch": 0.3243312428004608,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0015791662719160532,
      "loss": 1.4653,
      "step": 2534
    },
    {
      "epoch": 0.324459234608985,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0015788282628021518,
      "loss": 1.6253,
      "step": 2535
    },
    {
      "epoch": 0.32458722641750926,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0015784901542072222,
      "loss": 2.49,
      "step": 2536
    },
    {
      "epoch": 0.32471521822603355,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0015781519461893732,
      "loss": 1.3281,
      "step": 2537
    },
    {
      "epoch": 0.3248432100345578,
      "grad_norm": 0.765625,
      "learning_rate": 0.0015778136388067323,
      "loss": 1.833,
      "step": 2538
    },
    {
      "epoch": 0.32497120184308204,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0015774752321174427,
      "loss": 1.7214,
      "step": 2539
    },
    {
      "epoch": 0.3250991936516063,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0015771367261796653,
      "loss": 1.3661,
      "step": 2540
    },
    {
      "epoch": 0.3252271854601306,
      "grad_norm": 0.640625,
      "learning_rate": 0.0015767981210515769,
      "loss": 1.6082,
      "step": 2541
    },
    {
      "epoch": 0.3253551772686548,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0015764594167913737,
      "loss": 1.6422,
      "step": 2542
    },
    {
      "epoch": 0.32548316907717906,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0015761206134572664,
      "loss": 1.3524,
      "step": 2543
    },
    {
      "epoch": 0.3256111608857033,
      "grad_norm": 0.78125,
      "learning_rate": 0.0015757817111074844,
      "loss": 2.2247,
      "step": 2544
    },
    {
      "epoch": 0.3257391526942276,
      "grad_norm": 0.75,
      "learning_rate": 0.0015754427098002733,
      "loss": 1.2803,
      "step": 2545
    },
    {
      "epoch": 0.32586714450275184,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0015751036095938962,
      "loss": 2.1646,
      "step": 2546
    },
    {
      "epoch": 0.3259951363112761,
      "grad_norm": 0.703125,
      "learning_rate": 0.0015747644105466327,
      "loss": 2.3063,
      "step": 2547
    },
    {
      "epoch": 0.3261231281198003,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00157442511271678,
      "loss": 1.6586,
      "step": 2548
    },
    {
      "epoch": 0.32625111992832456,
      "grad_norm": 0.62890625,
      "learning_rate": 0.001574085716162652,
      "loss": 1.6548,
      "step": 2549
    },
    {
      "epoch": 0.32637911173684886,
      "grad_norm": 0.6875,
      "learning_rate": 0.001573746220942579,
      "loss": 1.8254,
      "step": 2550
    },
    {
      "epoch": 0.3265071035453731,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0015734066271149091,
      "loss": 1.6683,
      "step": 2551
    },
    {
      "epoch": 0.32663509535389734,
      "grad_norm": 0.6953125,
      "learning_rate": 0.001573066934738007,
      "loss": 1.4089,
      "step": 2552
    },
    {
      "epoch": 0.3267630871624216,
      "grad_norm": 0.9140625,
      "learning_rate": 0.001572727143870254,
      "loss": 1.8392,
      "step": 2553
    },
    {
      "epoch": 0.3268910789709459,
      "grad_norm": 0.6640625,
      "learning_rate": 0.001572387254570049,
      "loss": 1.9895,
      "step": 2554
    },
    {
      "epoch": 0.3270190707794701,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0015720472668958078,
      "loss": 1.3728,
      "step": 2555
    },
    {
      "epoch": 0.32714706258799436,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0015717071809059622,
      "loss": 1.7139,
      "step": 2556
    },
    {
      "epoch": 0.3272750543965186,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0015713669966589618,
      "loss": 2.6925,
      "step": 2557
    },
    {
      "epoch": 0.3274030462050429,
      "grad_norm": 0.8671875,
      "learning_rate": 0.001571026714213273,
      "loss": 2.1207,
      "step": 2558
    },
    {
      "epoch": 0.32753103801356714,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0015706863336273783,
      "loss": 1.4671,
      "step": 2559
    },
    {
      "epoch": 0.3276590298220914,
      "grad_norm": 0.58984375,
      "learning_rate": 0.001570345854959778,
      "loss": 1.6438,
      "step": 2560
    },
    {
      "epoch": 0.3277870216306156,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0015700052782689888,
      "loss": 1.302,
      "step": 2561
    },
    {
      "epoch": 0.32791501343913987,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0015696646036135445,
      "loss": 1.9853,
      "step": 2562
    },
    {
      "epoch": 0.32804300524766417,
      "grad_norm": 0.69921875,
      "learning_rate": 0.001569323831051995,
      "loss": 1.5743,
      "step": 2563
    },
    {
      "epoch": 0.3281709970561884,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0015689829606429083,
      "loss": 1.9172,
      "step": 2564
    },
    {
      "epoch": 0.32829898886471265,
      "grad_norm": 0.63671875,
      "learning_rate": 0.001568641992444868,
      "loss": 1.8761,
      "step": 2565
    },
    {
      "epoch": 0.3284269806732369,
      "grad_norm": 0.703125,
      "learning_rate": 0.0015683009265164754,
      "loss": 1.736,
      "step": 2566
    },
    {
      "epoch": 0.3285549724817612,
      "grad_norm": 0.7890625,
      "learning_rate": 0.001567959762916348,
      "loss": 2.5999,
      "step": 2567
    },
    {
      "epoch": 0.32868296429028543,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0015676185017031202,
      "loss": 2.2335,
      "step": 2568
    },
    {
      "epoch": 0.32881095609880967,
      "grad_norm": 0.640625,
      "learning_rate": 0.0015672771429354436,
      "loss": 1.6185,
      "step": 2569
    },
    {
      "epoch": 0.3289389479073339,
      "grad_norm": 0.8984375,
      "learning_rate": 0.001566935686671986,
      "loss": 2.5729,
      "step": 2570
    },
    {
      "epoch": 0.3290669397158582,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0015665941329714322,
      "loss": 1.8442,
      "step": 2571
    },
    {
      "epoch": 0.32919493152438245,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0015662524818924838,
      "loss": 2.1598,
      "step": 2572
    },
    {
      "epoch": 0.3293229233329067,
      "grad_norm": 0.65625,
      "learning_rate": 0.0015659107334938592,
      "loss": 1.945,
      "step": 2573
    },
    {
      "epoch": 0.32945091514143093,
      "grad_norm": 0.65234375,
      "learning_rate": 0.001565568887834293,
      "loss": 1.7213,
      "step": 2574
    },
    {
      "epoch": 0.32957890694995523,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0015652269449725374,
      "loss": 1.8133,
      "step": 2575
    },
    {
      "epoch": 0.32970689875847947,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0015648849049673603,
      "loss": 1.5971,
      "step": 2576
    },
    {
      "epoch": 0.3298348905670037,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0015645427678775477,
      "loss": 1.5584,
      "step": 2577
    },
    {
      "epoch": 0.32996288237552795,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0015642005337619005,
      "loss": 2.3067,
      "step": 2578
    },
    {
      "epoch": 0.3300908741840522,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0015638582026792375,
      "loss": 2.2607,
      "step": 2579
    },
    {
      "epoch": 0.3302188659925765,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0015635157746883945,
      "loss": 1.41,
      "step": 2580
    },
    {
      "epoch": 0.33034685780110074,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0015631732498482224,
      "loss": 1.9094,
      "step": 2581
    },
    {
      "epoch": 0.330474849609625,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00156283062821759,
      "loss": 1.475,
      "step": 2582
    },
    {
      "epoch": 0.3306028414181492,
      "grad_norm": 0.75390625,
      "learning_rate": 0.001562487909855382,
      "loss": 1.8897,
      "step": 2583
    },
    {
      "epoch": 0.3307308332266735,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0015621450948205007,
      "loss": 1.7415,
      "step": 2584
    },
    {
      "epoch": 0.33085882503519776,
      "grad_norm": 0.71875,
      "learning_rate": 0.0015618021831718641,
      "loss": 1.5678,
      "step": 2585
    },
    {
      "epoch": 0.330986816843722,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0015614591749684075,
      "loss": 1.7243,
      "step": 2586
    },
    {
      "epoch": 0.33111480865224624,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0015611160702690818,
      "loss": 2.295,
      "step": 2587
    },
    {
      "epoch": 0.33124280046077054,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0015607728691328555,
      "loss": 1.8111,
      "step": 2588
    },
    {
      "epoch": 0.3313707922692948,
      "grad_norm": 0.59375,
      "learning_rate": 0.0015604295716187134,
      "loss": 1.6041,
      "step": 2589
    },
    {
      "epoch": 0.331498784077819,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0015600861777856564,
      "loss": 1.9043,
      "step": 2590
    },
    {
      "epoch": 0.33162677588634326,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0015597426876927027,
      "loss": 1.5951,
      "step": 2591
    },
    {
      "epoch": 0.3317547676948675,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0015593991013988862,
      "loss": 2.0831,
      "step": 2592
    },
    {
      "epoch": 0.3318827595033918,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0015590554189632581,
      "loss": 1.7642,
      "step": 2593
    },
    {
      "epoch": 0.33201075131191604,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0015587116404448859,
      "loss": 1.8615,
      "step": 2594
    },
    {
      "epoch": 0.3321387431204403,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0015583677659028532,
      "loss": 1.6543,
      "step": 2595
    },
    {
      "epoch": 0.3322667349289645,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0015580237953962603,
      "loss": 2.3826,
      "step": 2596
    },
    {
      "epoch": 0.3323947267374888,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0015576797289842245,
      "loss": 1.7929,
      "step": 2597
    },
    {
      "epoch": 0.33252271854601306,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0015573355667258793,
      "loss": 1.2576,
      "step": 2598
    },
    {
      "epoch": 0.3326507103545373,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0015569913086803742,
      "loss": 1.846,
      "step": 2599
    },
    {
      "epoch": 0.33277870216306155,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0015566469549068756,
      "loss": 1.8493,
      "step": 2600
    },
    {
      "epoch": 0.33290669397158584,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0015563025054645663,
      "loss": 2.0497,
      "step": 2601
    },
    {
      "epoch": 0.3330346857801101,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0015559579604126456,
      "loss": 1.8803,
      "step": 2602
    },
    {
      "epoch": 0.3331626775886343,
      "grad_norm": 0.8203125,
      "learning_rate": 0.001555613319810329,
      "loss": 2.11,
      "step": 2603
    },
    {
      "epoch": 0.33329066939715857,
      "grad_norm": 0.77734375,
      "learning_rate": 0.001555268583716849,
      "loss": 2.0013,
      "step": 2604
    },
    {
      "epoch": 0.3334186612056828,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0015549237521914536,
      "loss": 2.1363,
      "step": 2605
    },
    {
      "epoch": 0.3335466530142071,
      "grad_norm": 0.5859375,
      "learning_rate": 0.001554578825293408,
      "loss": 2.0327,
      "step": 2606
    },
    {
      "epoch": 0.33367464482273135,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0015542338030819932,
      "loss": 2.2868,
      "step": 2607
    },
    {
      "epoch": 0.3338026366312556,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0015538886856165072,
      "loss": 1.5871,
      "step": 2608
    },
    {
      "epoch": 0.33393062843977983,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0015535434729562637,
      "loss": 1.6191,
      "step": 2609
    },
    {
      "epoch": 0.33405862024830413,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0015531981651605933,
      "loss": 1.4873,
      "step": 2610
    },
    {
      "epoch": 0.33418661205682837,
      "grad_norm": 0.78125,
      "learning_rate": 0.001552852762288843,
      "loss": 2.2037,
      "step": 2611
    },
    {
      "epoch": 0.3343146038653526,
      "grad_norm": 0.65234375,
      "learning_rate": 0.001552507264400375,
      "loss": 1.4406,
      "step": 2612
    },
    {
      "epoch": 0.33444259567387685,
      "grad_norm": 0.65625,
      "learning_rate": 0.0015521616715545696,
      "loss": 1.5472,
      "step": 2613
    },
    {
      "epoch": 0.33457058748240115,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0015518159838108225,
      "loss": 2.2052,
      "step": 2614
    },
    {
      "epoch": 0.3346985792909254,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0015514702012285452,
      "loss": 1.8324,
      "step": 2615
    },
    {
      "epoch": 0.33482657109944963,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0015511243238671667,
      "loss": 2.1583,
      "step": 2616
    },
    {
      "epoch": 0.3349545629079739,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0015507783517861307,
      "loss": 1.9711,
      "step": 2617
    },
    {
      "epoch": 0.33508255471649817,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0015504322850448992,
      "loss": 2.0613,
      "step": 2618
    },
    {
      "epoch": 0.3352105465250224,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0015500861237029487,
      "loss": 1.4455,
      "step": 2619
    },
    {
      "epoch": 0.33533853833354665,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0015497398678197727,
      "loss": 1.5199,
      "step": 2620
    },
    {
      "epoch": 0.3354665301420709,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0015493935174548813,
      "loss": 1.422,
      "step": 2621
    },
    {
      "epoch": 0.33559452195059514,
      "grad_norm": 0.90625,
      "learning_rate": 0.0015490470726678002,
      "loss": 2.3619,
      "step": 2622
    },
    {
      "epoch": 0.33572251375911943,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0015487005335180712,
      "loss": 1.3433,
      "step": 2623
    },
    {
      "epoch": 0.3358505055676437,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0015483539000652532,
      "loss": 2.2245,
      "step": 2624
    },
    {
      "epoch": 0.3359784973761679,
      "grad_norm": 0.640625,
      "learning_rate": 0.0015480071723689203,
      "loss": 2.0421,
      "step": 2625
    },
    {
      "epoch": 0.33610648918469216,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0015476603504886637,
      "loss": 1.4549,
      "step": 2626
    },
    {
      "epoch": 0.33623448099321646,
      "grad_norm": 0.8125,
      "learning_rate": 0.0015473134344840902,
      "loss": 1.927,
      "step": 2627
    },
    {
      "epoch": 0.3363624728017407,
      "grad_norm": 0.625,
      "learning_rate": 0.0015469664244148233,
      "loss": 1.4221,
      "step": 2628
    },
    {
      "epoch": 0.33649046461026494,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0015466193203405015,
      "loss": 2.0512,
      "step": 2629
    },
    {
      "epoch": 0.3366184564187892,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0015462721223207807,
      "loss": 1.7954,
      "step": 2630
    },
    {
      "epoch": 0.3367464482273135,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0015459248304153328,
      "loss": 1.753,
      "step": 2631
    },
    {
      "epoch": 0.3368744400358377,
      "grad_norm": 0.59765625,
      "learning_rate": 0.001545577444683845,
      "loss": 1.4968,
      "step": 2632
    },
    {
      "epoch": 0.33700243184436196,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0015452299651860218,
      "loss": 1.9184,
      "step": 2633
    },
    {
      "epoch": 0.3371304236528862,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0015448823919815829,
      "loss": 2.206,
      "step": 2634
    },
    {
      "epoch": 0.33725841546141044,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0015445347251302642,
      "loss": 1.5815,
      "step": 2635
    },
    {
      "epoch": 0.33738640726993474,
      "grad_norm": 0.734375,
      "learning_rate": 0.0015441869646918179,
      "loss": 1.9161,
      "step": 2636
    },
    {
      "epoch": 0.337514399078459,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0015438391107260125,
      "loss": 2.0598,
      "step": 2637
    },
    {
      "epoch": 0.3376423908869832,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0015434911632926324,
      "loss": 1.7627,
      "step": 2638
    },
    {
      "epoch": 0.33777038269550747,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0015431431224514779,
      "loss": 2.4773,
      "step": 2639
    },
    {
      "epoch": 0.33789837450403176,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0015427949882623653,
      "loss": 1.8044,
      "step": 2640
    },
    {
      "epoch": 0.338026366312556,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0015424467607851274,
      "loss": 1.7626,
      "step": 2641
    },
    {
      "epoch": 0.33815435812108025,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0015420984400796125,
      "loss": 2.3309,
      "step": 2642
    },
    {
      "epoch": 0.3382823499296045,
      "grad_norm": 0.625,
      "learning_rate": 0.0015417500262056851,
      "loss": 1.7784,
      "step": 2643
    },
    {
      "epoch": 0.3384103417381288,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0015414015192232262,
      "loss": 2.0481,
      "step": 2644
    },
    {
      "epoch": 0.338538333546653,
      "grad_norm": 0.88671875,
      "learning_rate": 0.001541052919192132,
      "loss": 2.2609,
      "step": 2645
    },
    {
      "epoch": 0.33866632535517727,
      "grad_norm": 0.69140625,
      "learning_rate": 0.001540704226172315,
      "loss": 2.1834,
      "step": 2646
    },
    {
      "epoch": 0.3387943171637015,
      "grad_norm": 0.5703125,
      "learning_rate": 0.001540355440223704,
      "loss": 1.4493,
      "step": 2647
    },
    {
      "epoch": 0.33892230897222575,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0015400065614062438,
      "loss": 2.0324,
      "step": 2648
    },
    {
      "epoch": 0.33905030078075005,
      "grad_norm": 0.68359375,
      "learning_rate": 0.001539657589779894,
      "loss": 1.8173,
      "step": 2649
    },
    {
      "epoch": 0.3391782925892743,
      "grad_norm": 0.5703125,
      "learning_rate": 0.001539308525404632,
      "loss": 1.351,
      "step": 2650
    },
    {
      "epoch": 0.33930628439779853,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0015389593683404498,
      "loss": 1.9675,
      "step": 2651
    },
    {
      "epoch": 0.33943427620632277,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0015386101186473553,
      "loss": 1.6773,
      "step": 2652
    },
    {
      "epoch": 0.33956226801484707,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0015382607763853736,
      "loss": 1.9425,
      "step": 2653
    },
    {
      "epoch": 0.3396902598233713,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0015379113416145438,
      "loss": 1.9117,
      "step": 2654
    },
    {
      "epoch": 0.33981825163189555,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0015375618143949228,
      "loss": 1.9229,
      "step": 2655
    },
    {
      "epoch": 0.3399462434404198,
      "grad_norm": 0.66015625,
      "learning_rate": 0.001537212194786582,
      "loss": 2.0164,
      "step": 2656
    },
    {
      "epoch": 0.3400742352489441,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0015368624828496096,
      "loss": 1.8742,
      "step": 2657
    },
    {
      "epoch": 0.34020222705746833,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0015365126786441087,
      "loss": 1.9559,
      "step": 2658
    },
    {
      "epoch": 0.3403302188659926,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0015361627822301997,
      "loss": 1.3905,
      "step": 2659
    },
    {
      "epoch": 0.3404582106745168,
      "grad_norm": 0.71484375,
      "learning_rate": 0.001535812793668017,
      "loss": 2.0236,
      "step": 2660
    },
    {
      "epoch": 0.3405862024830411,
      "grad_norm": 0.8125,
      "learning_rate": 0.0015354627130177124,
      "loss": 1.9235,
      "step": 2661
    },
    {
      "epoch": 0.34071419429156535,
      "grad_norm": 0.578125,
      "learning_rate": 0.001535112540339453,
      "loss": 1.4778,
      "step": 2662
    },
    {
      "epoch": 0.3408421861000896,
      "grad_norm": 0.703125,
      "learning_rate": 0.0015347622756934213,
      "loss": 1.9269,
      "step": 2663
    },
    {
      "epoch": 0.34097017790861384,
      "grad_norm": 0.734375,
      "learning_rate": 0.0015344119191398162,
      "loss": 1.7719,
      "step": 2664
    },
    {
      "epoch": 0.3410981697171381,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0015340614707388517,
      "loss": 1.7263,
      "step": 2665
    },
    {
      "epoch": 0.3412261615256624,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0015337109305507593,
      "loss": 1.9125,
      "step": 2666
    },
    {
      "epoch": 0.3413541533341866,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0015333602986357833,
      "loss": 1.5131,
      "step": 2667
    },
    {
      "epoch": 0.34148214514271086,
      "grad_norm": 0.8125,
      "learning_rate": 0.001533009575054187,
      "loss": 2.1511,
      "step": 2668
    },
    {
      "epoch": 0.3416101369512351,
      "grad_norm": 0.65625,
      "learning_rate": 0.0015326587598662466,
      "loss": 1.4644,
      "step": 2669
    },
    {
      "epoch": 0.3417381287597594,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0015323078531322568,
      "loss": 1.684,
      "step": 2670
    },
    {
      "epoch": 0.34186612056828364,
      "grad_norm": 0.671875,
      "learning_rate": 0.0015319568549125254,
      "loss": 2.244,
      "step": 2671
    },
    {
      "epoch": 0.3419941123768079,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0015316057652673776,
      "loss": 2.4634,
      "step": 2672
    },
    {
      "epoch": 0.3421221041853321,
      "grad_norm": 0.78125,
      "learning_rate": 0.0015312545842571538,
      "loss": 1.6878,
      "step": 2673
    },
    {
      "epoch": 0.3422500959938564,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0015309033119422107,
      "loss": 1.876,
      "step": 2674
    },
    {
      "epoch": 0.34237808780238066,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0015305519483829193,
      "loss": 1.8942,
      "step": 2675
    },
    {
      "epoch": 0.3425060796109049,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0015302004936396673,
      "loss": 1.7484,
      "step": 2676
    },
    {
      "epoch": 0.34263407141942914,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0015298489477728586,
      "loss": 1.8411,
      "step": 2677
    },
    {
      "epoch": 0.3427620632279534,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0015294973108429112,
      "loss": 1.6062,
      "step": 2678
    },
    {
      "epoch": 0.3428900550364777,
      "grad_norm": 1.046875,
      "learning_rate": 0.0015291455829102602,
      "loss": 2.3653,
      "step": 2679
    },
    {
      "epoch": 0.3430180468450019,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0015287937640353552,
      "loss": 1.5391,
      "step": 2680
    },
    {
      "epoch": 0.34314603865352616,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0015284418542786625,
      "loss": 1.8314,
      "step": 2681
    },
    {
      "epoch": 0.3432740304620504,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0015280898537006634,
      "loss": 1.9002,
      "step": 2682
    },
    {
      "epoch": 0.3434020222705747,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0015277377623618548,
      "loss": 1.7858,
      "step": 2683
    },
    {
      "epoch": 0.34353001407909894,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0015273855803227493,
      "loss": 1.5348,
      "step": 2684
    },
    {
      "epoch": 0.3436580058876232,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0015270333076438747,
      "loss": 1.8471,
      "step": 2685
    },
    {
      "epoch": 0.3437859976961474,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0015266809443857756,
      "loss": 1.9142,
      "step": 2686
    },
    {
      "epoch": 0.3439139895046717,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0015263284906090106,
      "loss": 2.1114,
      "step": 2687
    },
    {
      "epoch": 0.34404198131319597,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0015259759463741554,
      "loss": 2.1347,
      "step": 2688
    },
    {
      "epoch": 0.3441699731217202,
      "grad_norm": 0.8125,
      "learning_rate": 0.0015256233117418,
      "loss": 1.7064,
      "step": 2689
    },
    {
      "epoch": 0.34429796493024445,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00152527058677255,
      "loss": 1.3559,
      "step": 2690
    },
    {
      "epoch": 0.34442595673876875,
      "grad_norm": 0.703125,
      "learning_rate": 0.0015249177715270276,
      "loss": 1.7366,
      "step": 2691
    },
    {
      "epoch": 0.344553948547293,
      "grad_norm": 0.7421875,
      "learning_rate": 0.001524564866065869,
      "loss": 1.9508,
      "step": 2692
    },
    {
      "epoch": 0.34468194035581723,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0015242118704497276,
      "loss": 1.2293,
      "step": 2693
    },
    {
      "epoch": 0.34480993216434147,
      "grad_norm": 0.87109375,
      "learning_rate": 0.001523858784739271,
      "loss": 1.5188,
      "step": 2694
    },
    {
      "epoch": 0.3449379239728657,
      "grad_norm": 0.72265625,
      "learning_rate": 0.001523505608995183,
      "loss": 2.4958,
      "step": 2695
    },
    {
      "epoch": 0.34506591578139,
      "grad_norm": 0.90625,
      "learning_rate": 0.001523152343278162,
      "loss": 2.0149,
      "step": 2696
    },
    {
      "epoch": 0.34519390758991425,
      "grad_norm": 0.734375,
      "learning_rate": 0.0015227989876489226,
      "loss": 1.9293,
      "step": 2697
    },
    {
      "epoch": 0.3453218993984385,
      "grad_norm": 0.5859375,
      "learning_rate": 0.001522445542168195,
      "loss": 1.8729,
      "step": 2698
    },
    {
      "epoch": 0.34544989120696273,
      "grad_norm": 0.92578125,
      "learning_rate": 0.001522092006896725,
      "loss": 2.6173,
      "step": 2699
    },
    {
      "epoch": 0.34557788301548703,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0015217383818952723,
      "loss": 1.7474,
      "step": 2700
    },
    {
      "epoch": 0.34570587482401127,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0015213846672246139,
      "loss": 1.6723,
      "step": 2701
    },
    {
      "epoch": 0.3458338666325355,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0015210308629455408,
      "loss": 1.8196,
      "step": 2702
    },
    {
      "epoch": 0.34596185844105976,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0015206769691188601,
      "loss": 1.8093,
      "step": 2703
    },
    {
      "epoch": 0.34608985024958405,
      "grad_norm": 0.67578125,
      "learning_rate": 0.001520322985805395,
      "loss": 1.6877,
      "step": 2704
    },
    {
      "epoch": 0.3462178420581083,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0015199689130659822,
      "loss": 1.9527,
      "step": 2705
    },
    {
      "epoch": 0.34634583386663254,
      "grad_norm": 0.734375,
      "learning_rate": 0.0015196147509614756,
      "loss": 1.9702,
      "step": 2706
    },
    {
      "epoch": 0.3464738256751568,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0015192604995527433,
      "loss": 1.997,
      "step": 2707
    },
    {
      "epoch": 0.346601817483681,
      "grad_norm": 0.6640625,
      "learning_rate": 0.001518906158900669,
      "loss": 1.9782,
      "step": 2708
    },
    {
      "epoch": 0.3467298092922053,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0015185517290661526,
      "loss": 1.4331,
      "step": 2709
    },
    {
      "epoch": 0.34685780110072956,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0015181972101101083,
      "loss": 1.6338,
      "step": 2710
    },
    {
      "epoch": 0.3469857929092538,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0015178426020934653,
      "loss": 1.7964,
      "step": 2711
    },
    {
      "epoch": 0.34711378471777804,
      "grad_norm": 0.890625,
      "learning_rate": 0.0015174879050771698,
      "loss": 1.9825,
      "step": 2712
    },
    {
      "epoch": 0.34724177652630234,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0015171331191221812,
      "loss": 1.7324,
      "step": 2713
    },
    {
      "epoch": 0.3473697683348266,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0015167782442894757,
      "loss": 1.8582,
      "step": 2714
    },
    {
      "epoch": 0.3474977601433508,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0015164232806400447,
      "loss": 2.2377,
      "step": 2715
    },
    {
      "epoch": 0.34762575195187506,
      "grad_norm": 0.703125,
      "learning_rate": 0.0015160682282348938,
      "loss": 1.9394,
      "step": 2716
    },
    {
      "epoch": 0.34775374376039936,
      "grad_norm": 0.703125,
      "learning_rate": 0.001515713087135045,
      "loss": 2.2358,
      "step": 2717
    },
    {
      "epoch": 0.3478817355689236,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0015153578574015347,
      "loss": 2.3023,
      "step": 2718
    },
    {
      "epoch": 0.34800972737744784,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0015150025390954151,
      "loss": 1.5012,
      "step": 2719
    },
    {
      "epoch": 0.3481377191859721,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0015146471322777532,
      "loss": 1.8386,
      "step": 2720
    },
    {
      "epoch": 0.3482657109944963,
      "grad_norm": 0.625,
      "learning_rate": 0.0015142916370096322,
      "loss": 1.533,
      "step": 2721
    },
    {
      "epoch": 0.3483937028030206,
      "grad_norm": 0.640625,
      "learning_rate": 0.0015139360533521487,
      "loss": 1.5585,
      "step": 2722
    },
    {
      "epoch": 0.34852169461154486,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0015135803813664157,
      "loss": 1.312,
      "step": 2723
    },
    {
      "epoch": 0.3486496864200691,
      "grad_norm": 0.6484375,
      "learning_rate": 0.001513224621113562,
      "loss": 2.0215,
      "step": 2724
    },
    {
      "epoch": 0.34877767822859335,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0015128687726547296,
      "loss": 1.3335,
      "step": 2725
    },
    {
      "epoch": 0.34890567003711764,
      "grad_norm": 0.61328125,
      "learning_rate": 0.001512512836051078,
      "loss": 1.4538,
      "step": 2726
    },
    {
      "epoch": 0.3490336618456419,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0015121568113637795,
      "loss": 1.5904,
      "step": 2727
    },
    {
      "epoch": 0.3491616536541661,
      "grad_norm": 0.67578125,
      "learning_rate": 0.001511800698654024,
      "loss": 1.5402,
      "step": 2728
    },
    {
      "epoch": 0.34928964546269037,
      "grad_norm": 0.625,
      "learning_rate": 0.0015114444979830144,
      "loss": 1.7386,
      "step": 2729
    },
    {
      "epoch": 0.34941763727121466,
      "grad_norm": 0.671875,
      "learning_rate": 0.0015110882094119696,
      "loss": 1.9228,
      "step": 2730
    },
    {
      "epoch": 0.3495456290797389,
      "grad_norm": 0.859375,
      "learning_rate": 0.0015107318330021236,
      "loss": 1.962,
      "step": 2731
    },
    {
      "epoch": 0.34967362088826315,
      "grad_norm": 0.58984375,
      "learning_rate": 0.001510375368814726,
      "loss": 1.2769,
      "step": 2732
    },
    {
      "epoch": 0.3498016126967874,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0015100188169110402,
      "loss": 2.0172,
      "step": 2733
    },
    {
      "epoch": 0.3499296045053117,
      "grad_norm": 0.8671875,
      "learning_rate": 0.001509662177352346,
      "loss": 1.4616,
      "step": 2734
    },
    {
      "epoch": 0.35005759631383593,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0015093054501999375,
      "loss": 1.592,
      "step": 2735
    },
    {
      "epoch": 0.35018558812236017,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0015089486355151238,
      "loss": 1.8327,
      "step": 2736
    },
    {
      "epoch": 0.3503135799308844,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0015085917333592298,
      "loss": 1.7719,
      "step": 2737
    },
    {
      "epoch": 0.35044157173940865,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0015082347437935944,
      "loss": 1.824,
      "step": 2738
    },
    {
      "epoch": 0.35056956354793295,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0015078776668795725,
      "loss": 2.6873,
      "step": 2739
    },
    {
      "epoch": 0.3506975553564572,
      "grad_norm": 0.8203125,
      "learning_rate": 0.001507520502678533,
      "loss": 2.5574,
      "step": 2740
    },
    {
      "epoch": 0.35082554716498143,
      "grad_norm": 0.828125,
      "learning_rate": 0.001507163251251861,
      "loss": 1.706,
      "step": 2741
    },
    {
      "epoch": 0.3509535389735057,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0015068059126609557,
      "loss": 1.4211,
      "step": 2742
    },
    {
      "epoch": 0.35108153078202997,
      "grad_norm": 0.671875,
      "learning_rate": 0.0015064484869672316,
      "loss": 1.6856,
      "step": 2743
    },
    {
      "epoch": 0.3512095225905542,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0015060909742321176,
      "loss": 1.3952,
      "step": 2744
    },
    {
      "epoch": 0.35133751439907845,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0015057333745170588,
      "loss": 1.8818,
      "step": 2745
    },
    {
      "epoch": 0.3514655062076027,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0015053756878835142,
      "loss": 1.6038,
      "step": 2746
    },
    {
      "epoch": 0.351593498016127,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0015050179143929578,
      "loss": 1.694,
      "step": 2747
    },
    {
      "epoch": 0.35172148982465123,
      "grad_norm": 0.625,
      "learning_rate": 0.0015046600541068792,
      "loss": 1.8077,
      "step": 2748
    },
    {
      "epoch": 0.3518494816331755,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0015043021070867825,
      "loss": 1.8925,
      "step": 2749
    },
    {
      "epoch": 0.3519774734416997,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0015039440733941862,
      "loss": 1.3568,
      "step": 2750
    },
    {
      "epoch": 0.35210546525022396,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0015035859530906246,
      "loss": 1.8097,
      "step": 2751
    },
    {
      "epoch": 0.35223345705874826,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0015032277462376468,
      "loss": 1.8094,
      "step": 2752
    },
    {
      "epoch": 0.3523614488672725,
      "grad_norm": 1.640625,
      "learning_rate": 0.001502869452896815,
      "loss": 1.7377,
      "step": 2753
    },
    {
      "epoch": 0.35248944067579674,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00150251107312971,
      "loss": 1.7214,
      "step": 2754
    },
    {
      "epoch": 0.352617432484321,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0015021526069979231,
      "loss": 1.1805,
      "step": 2755
    },
    {
      "epoch": 0.3527454242928453,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0015017940545630638,
      "loss": 1.9942,
      "step": 2756
    },
    {
      "epoch": 0.3528734161013695,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0015014354158867547,
      "loss": 2.0538,
      "step": 2757
    },
    {
      "epoch": 0.35300140790989376,
      "grad_norm": 0.84375,
      "learning_rate": 0.0015010766910306338,
      "loss": 1.4594,
      "step": 2758
    },
    {
      "epoch": 0.353129399718418,
      "grad_norm": 0.640625,
      "learning_rate": 0.0015007178800563535,
      "loss": 1.7864,
      "step": 2759
    },
    {
      "epoch": 0.3532573915269423,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0015003589830255813,
      "loss": 1.6187,
      "step": 2760
    },
    {
      "epoch": 0.35338538333546654,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0015,
      "loss": 1.5381,
      "step": 2761
    },
    {
      "epoch": 0.3535133751439908,
      "grad_norm": 0.5625,
      "learning_rate": 0.0014996409310413061,
      "loss": 1.3154,
      "step": 2762
    },
    {
      "epoch": 0.353641366952515,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0014992817762112122,
      "loss": 1.6695,
      "step": 2763
    },
    {
      "epoch": 0.3537693587610393,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0014989225355714435,
      "loss": 2.2739,
      "step": 2764
    },
    {
      "epoch": 0.35389735056956356,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0014985632091837427,
      "loss": 1.446,
      "step": 2765
    },
    {
      "epoch": 0.3540253423780878,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0014982037971098651,
      "loss": 2.0922,
      "step": 2766
    },
    {
      "epoch": 0.35415333418661205,
      "grad_norm": 0.7578125,
      "learning_rate": 0.001497844299411582,
      "loss": 2.176,
      "step": 2767
    },
    {
      "epoch": 0.3542813259951363,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0014974847161506786,
      "loss": 1.7577,
      "step": 2768
    },
    {
      "epoch": 0.3544093178036606,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0014971250473889551,
      "loss": 1.7986,
      "step": 2769
    },
    {
      "epoch": 0.3545373096121848,
      "grad_norm": 0.578125,
      "learning_rate": 0.001496765293188226,
      "loss": 1.4225,
      "step": 2770
    },
    {
      "epoch": 0.35466530142070907,
      "grad_norm": 0.59375,
      "learning_rate": 0.001496405453610322,
      "loss": 1.4218,
      "step": 2771
    },
    {
      "epoch": 0.3547932932292333,
      "grad_norm": 0.71875,
      "learning_rate": 0.0014960455287170867,
      "loss": 2.0929,
      "step": 2772
    },
    {
      "epoch": 0.3549212850377576,
      "grad_norm": 0.5625,
      "learning_rate": 0.0014956855185703785,
      "loss": 1.6467,
      "step": 2773
    },
    {
      "epoch": 0.35504927684628185,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0014953254232320723,
      "loss": 1.4599,
      "step": 2774
    },
    {
      "epoch": 0.3551772686548061,
      "grad_norm": 0.734375,
      "learning_rate": 0.001494965242764055,
      "loss": 2.088,
      "step": 2775
    },
    {
      "epoch": 0.35530526046333033,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0014946049772282303,
      "loss": 1.8337,
      "step": 2776
    },
    {
      "epoch": 0.3554332522718546,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0014942446266865152,
      "loss": 1.5227,
      "step": 2777
    },
    {
      "epoch": 0.35556124408037887,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0014938841912008423,
      "loss": 1.5772,
      "step": 2778
    },
    {
      "epoch": 0.3556892358889031,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0014935236708331582,
      "loss": 1.6489,
      "step": 2779
    },
    {
      "epoch": 0.35581722769742735,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0014931630656454234,
      "loss": 1.93,
      "step": 2780
    },
    {
      "epoch": 0.3559452195059516,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0014928023756996145,
      "loss": 2.2433,
      "step": 2781
    },
    {
      "epoch": 0.3560732113144759,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0014924416010577218,
      "loss": 1.7999,
      "step": 2782
    },
    {
      "epoch": 0.35620120312300013,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0014920807417817504,
      "loss": 1.908,
      "step": 2783
    },
    {
      "epoch": 0.3563291949315244,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0014917197979337194,
      "loss": 2.1507,
      "step": 2784
    },
    {
      "epoch": 0.3564571867400486,
      "grad_norm": 0.6953125,
      "learning_rate": 0.001491358769575663,
      "loss": 1.4236,
      "step": 2785
    },
    {
      "epoch": 0.3565851785485729,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0014909976567696302,
      "loss": 1.8716,
      "step": 2786
    },
    {
      "epoch": 0.35671317035709715,
      "grad_norm": 0.703125,
      "learning_rate": 0.0014906364595776832,
      "loss": 1.7736,
      "step": 2787
    },
    {
      "epoch": 0.3568411621656214,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0014902751780619008,
      "loss": 1.8803,
      "step": 2788
    },
    {
      "epoch": 0.35696915397414564,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0014899138122843741,
      "loss": 1.388,
      "step": 2789
    },
    {
      "epoch": 0.35709714578266993,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0014895523623072102,
      "loss": 2.3306,
      "step": 2790
    },
    {
      "epoch": 0.3572251375911942,
      "grad_norm": 0.65625,
      "learning_rate": 0.00148919082819253,
      "loss": 1.6838,
      "step": 2791
    },
    {
      "epoch": 0.3573531293997184,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0014888292100024686,
      "loss": 2.4202,
      "step": 2792
    },
    {
      "epoch": 0.35748112120824266,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0014884675077991765,
      "loss": 1.8399,
      "step": 2793
    },
    {
      "epoch": 0.3576091130167669,
      "grad_norm": 0.62890625,
      "learning_rate": 0.001488105721644818,
      "loss": 2.1898,
      "step": 2794
    },
    {
      "epoch": 0.3577371048252912,
      "grad_norm": 0.482421875,
      "learning_rate": 0.001487743851601572,
      "loss": 1.0444,
      "step": 2795
    },
    {
      "epoch": 0.35786509663381544,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0014873818977316313,
      "loss": 1.7961,
      "step": 2796
    },
    {
      "epoch": 0.3579930884423397,
      "grad_norm": 0.625,
      "learning_rate": 0.0014870198600972042,
      "loss": 1.4775,
      "step": 2797
    },
    {
      "epoch": 0.3581210802508639,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0014866577387605117,
      "loss": 1.676,
      "step": 2798
    },
    {
      "epoch": 0.3582490720593882,
      "grad_norm": 0.65625,
      "learning_rate": 0.0014862955337837916,
      "loss": 1.8587,
      "step": 2799
    },
    {
      "epoch": 0.35837706386791246,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0014859332452292935,
      "loss": 1.9806,
      "step": 2800
    },
    {
      "epoch": 0.3585050556764367,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0014855708731592835,
      "loss": 1.9042,
      "step": 2801
    },
    {
      "epoch": 0.35863304748496094,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0014852084176360403,
      "loss": 1.1844,
      "step": 2802
    },
    {
      "epoch": 0.35876103929348524,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0014848458787218581,
      "loss": 1.8746,
      "step": 2803
    },
    {
      "epoch": 0.3588890311020095,
      "grad_norm": 0.765625,
      "learning_rate": 0.0014844832564790455,
      "loss": 1.6459,
      "step": 2804
    },
    {
      "epoch": 0.3590170229105337,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0014841205509699242,
      "loss": 1.6936,
      "step": 2805
    },
    {
      "epoch": 0.35914501471905796,
      "grad_norm": 0.64453125,
      "learning_rate": 0.001483757762256832,
      "loss": 1.7737,
      "step": 2806
    },
    {
      "epoch": 0.35927300652758226,
      "grad_norm": 0.80078125,
      "learning_rate": 0.001483394890402119,
      "loss": 1.8089,
      "step": 2807
    },
    {
      "epoch": 0.3594009983361065,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0014830319354681514,
      "loss": 1.3573,
      "step": 2808
    },
    {
      "epoch": 0.35952899014463074,
      "grad_norm": 0.84375,
      "learning_rate": 0.0014826688975173084,
      "loss": 1.9314,
      "step": 2809
    },
    {
      "epoch": 0.359656981953155,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0014823057766119845,
      "loss": 1.8584,
      "step": 2810
    },
    {
      "epoch": 0.3597849737616792,
      "grad_norm": 0.796875,
      "learning_rate": 0.0014819425728145874,
      "loss": 1.916,
      "step": 2811
    },
    {
      "epoch": 0.3599129655702035,
      "grad_norm": 0.703125,
      "learning_rate": 0.0014815792861875398,
      "loss": 1.9193,
      "step": 2812
    },
    {
      "epoch": 0.36004095737872777,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0014812159167932784,
      "loss": 1.492,
      "step": 2813
    },
    {
      "epoch": 0.360168949187252,
      "grad_norm": 0.671875,
      "learning_rate": 0.0014808524646942537,
      "loss": 1.7099,
      "step": 2814
    },
    {
      "epoch": 0.36029694099577625,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0014804889299529318,
      "loss": 1.3755,
      "step": 2815
    },
    {
      "epoch": 0.36042493280430055,
      "grad_norm": 0.671875,
      "learning_rate": 0.0014801253126317912,
      "loss": 1.3531,
      "step": 2816
    },
    {
      "epoch": 0.3605529246128248,
      "grad_norm": 0.59375,
      "learning_rate": 0.0014797616127933259,
      "loss": 1.3387,
      "step": 2817
    },
    {
      "epoch": 0.36068091642134903,
      "grad_norm": 0.640625,
      "learning_rate": 0.001479397830500043,
      "loss": 1.8097,
      "step": 2818
    },
    {
      "epoch": 0.36080890822987327,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0014790339658144651,
      "loss": 2.1758,
      "step": 2819
    },
    {
      "epoch": 0.36093690003839757,
      "grad_norm": 0.65625,
      "learning_rate": 0.001478670018799128,
      "loss": 1.6263,
      "step": 2820
    },
    {
      "epoch": 0.3610648918469218,
      "grad_norm": 0.6875,
      "learning_rate": 0.0014783059895165818,
      "loss": 1.3112,
      "step": 2821
    },
    {
      "epoch": 0.36119288365544605,
      "grad_norm": 0.625,
      "learning_rate": 0.001477941878029391,
      "loss": 1.6563,
      "step": 2822
    },
    {
      "epoch": 0.3613208754639703,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0014775776844001339,
      "loss": 1.6563,
      "step": 2823
    },
    {
      "epoch": 0.36144886727249453,
      "grad_norm": 0.59375,
      "learning_rate": 0.001477213408691403,
      "loss": 1.5769,
      "step": 2824
    },
    {
      "epoch": 0.36157685908101883,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0014768490509658053,
      "loss": 1.706,
      "step": 2825
    },
    {
      "epoch": 0.3617048508895431,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0014764846112859614,
      "loss": 1.8487,
      "step": 2826
    },
    {
      "epoch": 0.3618328426980673,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0014761200897145063,
      "loss": 1.7212,
      "step": 2827
    },
    {
      "epoch": 0.36196083450659156,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0014757554863140884,
      "loss": 1.3699,
      "step": 2828
    },
    {
      "epoch": 0.36208882631511585,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0014753908011473716,
      "loss": 1.4182,
      "step": 2829
    },
    {
      "epoch": 0.3622168181236401,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0014750260342770328,
      "loss": 1.0392,
      "step": 2830
    },
    {
      "epoch": 0.36234480993216434,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0014746611857657623,
      "loss": 1.8387,
      "step": 2831
    },
    {
      "epoch": 0.3624728017406886,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0014742962556762657,
      "loss": 1.6684,
      "step": 2832
    },
    {
      "epoch": 0.3626007935492129,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0014739312440712627,
      "loss": 1.6874,
      "step": 2833
    },
    {
      "epoch": 0.3627287853577371,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0014735661510134857,
      "loss": 1.5708,
      "step": 2834
    },
    {
      "epoch": 0.36285677716626136,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0014732009765656823,
      "loss": 1.3741,
      "step": 2835
    },
    {
      "epoch": 0.3629847689747856,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0014728357207906132,
      "loss": 2.1666,
      "step": 2836
    },
    {
      "epoch": 0.3631127607833099,
      "grad_norm": 0.68359375,
      "learning_rate": 0.001472470383751054,
      "loss": 1.8469,
      "step": 2837
    },
    {
      "epoch": 0.36324075259183414,
      "grad_norm": 0.640625,
      "learning_rate": 0.0014721049655097935,
      "loss": 1.5226,
      "step": 2838
    },
    {
      "epoch": 0.3633687444003584,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0014717394661296352,
      "loss": 1.5178,
      "step": 2839
    },
    {
      "epoch": 0.3634967362088826,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0014713738856733955,
      "loss": 2.08,
      "step": 2840
    },
    {
      "epoch": 0.36362472801740686,
      "grad_norm": 0.5546875,
      "learning_rate": 0.001471008224203906,
      "loss": 1.4106,
      "step": 2841
    },
    {
      "epoch": 0.36375271982593116,
      "grad_norm": 0.85546875,
      "learning_rate": 0.001470642481784011,
      "loss": 1.6534,
      "step": 2842
    },
    {
      "epoch": 0.3638807116344554,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0014702766584765692,
      "loss": 2.2105,
      "step": 2843
    },
    {
      "epoch": 0.36400870344297964,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0014699107543444537,
      "loss": 1.7547,
      "step": 2844
    },
    {
      "epoch": 0.3641366952515039,
      "grad_norm": 0.7890625,
      "learning_rate": 0.001469544769450551,
      "loss": 2.2604,
      "step": 2845
    },
    {
      "epoch": 0.3642646870600282,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0014691787038577614,
      "loss": 1.2612,
      "step": 2846
    },
    {
      "epoch": 0.3643926788685524,
      "grad_norm": 0.76953125,
      "learning_rate": 0.001468812557628999,
      "loss": 1.4258,
      "step": 2847
    },
    {
      "epoch": 0.36452067067707666,
      "grad_norm": 0.546875,
      "learning_rate": 0.001468446330827192,
      "loss": 1.2685,
      "step": 2848
    },
    {
      "epoch": 0.3646486624856009,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0014680800235152831,
      "loss": 1.8224,
      "step": 2849
    },
    {
      "epoch": 0.3647766542941252,
      "grad_norm": 0.703125,
      "learning_rate": 0.0014677136357562274,
      "loss": 2.077,
      "step": 2850
    },
    {
      "epoch": 0.36490464610264944,
      "grad_norm": 0.61328125,
      "learning_rate": 0.001467347167612995,
      "loss": 1.6233,
      "step": 2851
    },
    {
      "epoch": 0.3650326379111737,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0014669806191485695,
      "loss": 1.3085,
      "step": 2852
    },
    {
      "epoch": 0.3651606297196979,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0014666139904259477,
      "loss": 1.4167,
      "step": 2853
    },
    {
      "epoch": 0.36528862152822217,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0014662472815081407,
      "loss": 2.3168,
      "step": 2854
    },
    {
      "epoch": 0.36541661333674647,
      "grad_norm": 0.90625,
      "learning_rate": 0.0014658804924581738,
      "loss": 1.956,
      "step": 2855
    },
    {
      "epoch": 0.3655446051452707,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0014655136233390857,
      "loss": 1.8016,
      "step": 2856
    },
    {
      "epoch": 0.36567259695379495,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0014651466742139282,
      "loss": 1.9128,
      "step": 2857
    },
    {
      "epoch": 0.3658005887623192,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0014647796451457682,
      "loss": 3.0111,
      "step": 2858
    },
    {
      "epoch": 0.3659285805708435,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0014644125361976848,
      "loss": 1.6864,
      "step": 2859
    },
    {
      "epoch": 0.36605657237936773,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0014640453474327724,
      "loss": 2.0294,
      "step": 2860
    },
    {
      "epoch": 0.36618456418789197,
      "grad_norm": 0.78125,
      "learning_rate": 0.0014636780789141382,
      "loss": 2.0022,
      "step": 2861
    },
    {
      "epoch": 0.3663125559964162,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0014633107307049026,
      "loss": 1.7574,
      "step": 2862
    },
    {
      "epoch": 0.3664405478049405,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0014629433028682014,
      "loss": 2.0888,
      "step": 2863
    },
    {
      "epoch": 0.36656853961346475,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0014625757954671822,
      "loss": 1.6688,
      "step": 2864
    },
    {
      "epoch": 0.366696531421989,
      "grad_norm": 0.6875,
      "learning_rate": 0.001462208208565007,
      "loss": 1.8058,
      "step": 2865
    },
    {
      "epoch": 0.36682452323051323,
      "grad_norm": 0.7265625,
      "learning_rate": 0.001461840542224852,
      "loss": 2.2845,
      "step": 2866
    },
    {
      "epoch": 0.3669525150390375,
      "grad_norm": 0.640625,
      "learning_rate": 0.0014614727965099072,
      "loss": 1.5955,
      "step": 2867
    },
    {
      "epoch": 0.36708050684756177,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0014611049714833746,
      "loss": 2.1679,
      "step": 2868
    },
    {
      "epoch": 0.367208498656086,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0014607370672084716,
      "loss": 2.7993,
      "step": 2869
    },
    {
      "epoch": 0.36733649046461025,
      "grad_norm": 0.65625,
      "learning_rate": 0.0014603690837484279,
      "loss": 1.5055,
      "step": 2870
    },
    {
      "epoch": 0.3674644822731345,
      "grad_norm": 0.59375,
      "learning_rate": 0.0014600010211664877,
      "loss": 0.8357,
      "step": 2871
    },
    {
      "epoch": 0.3675924740816588,
      "grad_norm": 0.64453125,
      "learning_rate": 0.001459632879525909,
      "loss": 1.3974,
      "step": 2872
    },
    {
      "epoch": 0.36772046589018303,
      "grad_norm": 0.703125,
      "learning_rate": 0.0014592646588899624,
      "loss": 2.0441,
      "step": 2873
    },
    {
      "epoch": 0.3678484576987073,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0014588963593219326,
      "loss": 1.48,
      "step": 2874
    },
    {
      "epoch": 0.3679764495072315,
      "grad_norm": 0.73046875,
      "learning_rate": 0.001458527980885118,
      "loss": 1.8019,
      "step": 2875
    },
    {
      "epoch": 0.3681044413157558,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0014581595236428302,
      "loss": 1.6273,
      "step": 2876
    },
    {
      "epoch": 0.36823243312428006,
      "grad_norm": 0.9609375,
      "learning_rate": 0.001457790987658395,
      "loss": 1.6814,
      "step": 2877
    },
    {
      "epoch": 0.3683604249328043,
      "grad_norm": 0.5703125,
      "learning_rate": 0.001457422372995151,
      "loss": 1.3315,
      "step": 2878
    },
    {
      "epoch": 0.36848841674132854,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0014570536797164507,
      "loss": 2.1558,
      "step": 2879
    },
    {
      "epoch": 0.36861640854985284,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0014566849078856596,
      "loss": 1.4502,
      "step": 2880
    },
    {
      "epoch": 0.3687444003583771,
      "grad_norm": 0.7109375,
      "learning_rate": 0.001456316057566158,
      "loss": 1.873,
      "step": 2881
    },
    {
      "epoch": 0.3688723921669013,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0014559471288213377,
      "loss": 1.4422,
      "step": 2882
    },
    {
      "epoch": 0.36900038397542556,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0014555781217146062,
      "loss": 1.4231,
      "step": 2883
    },
    {
      "epoch": 0.3691283757839498,
      "grad_norm": 0.734375,
      "learning_rate": 0.0014552090363093826,
      "loss": 1.6406,
      "step": 2884
    },
    {
      "epoch": 0.3692563675924741,
      "grad_norm": 0.578125,
      "learning_rate": 0.0014548398726691005,
      "loss": 0.7916,
      "step": 2885
    },
    {
      "epoch": 0.36938435940099834,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0014544706308572066,
      "loss": 1.3991,
      "step": 2886
    },
    {
      "epoch": 0.3695123512095226,
      "grad_norm": 0.5078125,
      "learning_rate": 0.001454101310937161,
      "loss": 1.1445,
      "step": 2887
    },
    {
      "epoch": 0.3696403430180468,
      "grad_norm": 0.5625,
      "learning_rate": 0.0014537319129724377,
      "loss": 1.302,
      "step": 2888
    },
    {
      "epoch": 0.3697683348265711,
      "grad_norm": 0.65625,
      "learning_rate": 0.0014533624370265232,
      "loss": 1.5098,
      "step": 2889
    },
    {
      "epoch": 0.36989632663509536,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0014529928831629184,
      "loss": 1.7258,
      "step": 2890
    },
    {
      "epoch": 0.3700243184436196,
      "grad_norm": 0.734375,
      "learning_rate": 0.0014526232514451367,
      "loss": 2.0742,
      "step": 2891
    },
    {
      "epoch": 0.37015231025214385,
      "grad_norm": 0.71875,
      "learning_rate": 0.001452253541936706,
      "loss": 2.8149,
      "step": 2892
    },
    {
      "epoch": 0.37028030206066814,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0014518837547011657,
      "loss": 2.0584,
      "step": 2893
    },
    {
      "epoch": 0.3704082938691924,
      "grad_norm": 0.6796875,
      "learning_rate": 0.001451513889802071,
      "loss": 1.9078,
      "step": 2894
    },
    {
      "epoch": 0.3705362856777166,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0014511439473029884,
      "loss": 2.7068,
      "step": 2895
    },
    {
      "epoch": 0.37066427748624087,
      "grad_norm": 0.65234375,
      "learning_rate": 0.001450773927267499,
      "loss": 1.5382,
      "step": 2896
    },
    {
      "epoch": 0.3707922692947651,
      "grad_norm": 0.69140625,
      "learning_rate": 0.001450403829759196,
      "loss": 2.0546,
      "step": 2897
    },
    {
      "epoch": 0.3709202611032894,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0014500336548416877,
      "loss": 1.9956,
      "step": 2898
    },
    {
      "epoch": 0.37104825291181365,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0014496634025785938,
      "loss": 2.467,
      "step": 2899
    },
    {
      "epoch": 0.3711762447203379,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0014492930730335488,
      "loss": 2.2398,
      "step": 2900
    },
    {
      "epoch": 0.37130423652886213,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0014489226662701992,
      "loss": 2.3485,
      "step": 2901
    },
    {
      "epoch": 0.3714322283373864,
      "grad_norm": 0.55859375,
      "learning_rate": 0.001448552182352206,
      "loss": 1.0926,
      "step": 2902
    },
    {
      "epoch": 0.37156022014591067,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0014481816213432427,
      "loss": 1.554,
      "step": 2903
    },
    {
      "epoch": 0.3716882119544349,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0014478109833069954,
      "loss": 1.4934,
      "step": 2904
    },
    {
      "epoch": 0.37181620376295915,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0014474402683071657,
      "loss": 1.3643,
      "step": 2905
    },
    {
      "epoch": 0.37194419557148345,
      "grad_norm": 0.5703125,
      "learning_rate": 0.001447069476407466,
      "loss": 1.323,
      "step": 2906
    },
    {
      "epoch": 0.3720721873800077,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0014466986076716235,
      "loss": 1.9439,
      "step": 2907
    },
    {
      "epoch": 0.37220017918853193,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0014463276621633774,
      "loss": 1.6974,
      "step": 2908
    },
    {
      "epoch": 0.3723281709970562,
      "grad_norm": 0.65625,
      "learning_rate": 0.001445956639946481,
      "loss": 1.4826,
      "step": 2909
    },
    {
      "epoch": 0.3724561628055804,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0014455855410847006,
      "loss": 1.4693,
      "step": 2910
    },
    {
      "epoch": 0.3725841546141047,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0014452143656418154,
      "loss": 1.7158,
      "step": 2911
    },
    {
      "epoch": 0.37271214642262895,
      "grad_norm": 0.66796875,
      "learning_rate": 0.001444843113681618,
      "loss": 1.3769,
      "step": 2912
    },
    {
      "epoch": 0.3728401382311532,
      "grad_norm": 0.78125,
      "learning_rate": 0.001444471785267914,
      "loss": 2.5381,
      "step": 2913
    },
    {
      "epoch": 0.37296813003967744,
      "grad_norm": 1.15625,
      "learning_rate": 0.0014441003804645223,
      "loss": 1.5985,
      "step": 2914
    },
    {
      "epoch": 0.37309612184820173,
      "grad_norm": 0.6640625,
      "learning_rate": 0.001443728899335275,
      "loss": 1.583,
      "step": 2915
    },
    {
      "epoch": 0.373224113656726,
      "grad_norm": 0.65625,
      "learning_rate": 0.0014433573419440164,
      "loss": 1.6741,
      "step": 2916
    },
    {
      "epoch": 0.3733521054652502,
      "grad_norm": 0.828125,
      "learning_rate": 0.0014429857083546054,
      "loss": 2.0903,
      "step": 2917
    },
    {
      "epoch": 0.37348009727377446,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0014426139986309136,
      "loss": 1.37,
      "step": 2918
    },
    {
      "epoch": 0.37360808908229876,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0014422422128368242,
      "loss": 1.8175,
      "step": 2919
    },
    {
      "epoch": 0.373736080890823,
      "grad_norm": 0.734375,
      "learning_rate": 0.0014418703510362355,
      "loss": 2.6955,
      "step": 2920
    },
    {
      "epoch": 0.37386407269934724,
      "grad_norm": 0.625,
      "learning_rate": 0.0014414984132930578,
      "loss": 1.3688,
      "step": 2921
    },
    {
      "epoch": 0.3739920645078715,
      "grad_norm": 0.609375,
      "learning_rate": 0.0014411263996712143,
      "loss": 1.7466,
      "step": 2922
    },
    {
      "epoch": 0.3741200563163958,
      "grad_norm": 0.671875,
      "learning_rate": 0.0014407543102346423,
      "loss": 1.7944,
      "step": 2923
    },
    {
      "epoch": 0.37424804812492,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0014403821450472906,
      "loss": 2.0558,
      "step": 2924
    },
    {
      "epoch": 0.37437603993344426,
      "grad_norm": 0.5625,
      "learning_rate": 0.0014400099041731226,
      "loss": 1.7137,
      "step": 2925
    },
    {
      "epoch": 0.3745040317419685,
      "grad_norm": 0.6015625,
      "learning_rate": 0.001439637587676113,
      "loss": 1.8885,
      "step": 2926
    },
    {
      "epoch": 0.37463202355049274,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0014392651956202512,
      "loss": 1.5503,
      "step": 2927
    },
    {
      "epoch": 0.37476001535901704,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0014388927280695382,
      "loss": 2.5328,
      "step": 2928
    },
    {
      "epoch": 0.3748880071675413,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0014385201850879894,
      "loss": 1.7202,
      "step": 2929
    },
    {
      "epoch": 0.3750159989760655,
      "grad_norm": 1.0546875,
      "learning_rate": 0.001438147566739632,
      "loss": 1.4654,
      "step": 2930
    },
    {
      "epoch": 0.37514399078458976,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0014377748730885062,
      "loss": 1.861,
      "step": 2931
    },
    {
      "epoch": 0.37527198259311406,
      "grad_norm": 0.671875,
      "learning_rate": 0.0014374021041986653,
      "loss": 1.8985,
      "step": 2932
    },
    {
      "epoch": 0.3753999744016383,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0014370292601341764,
      "loss": 1.3305,
      "step": 2933
    },
    {
      "epoch": 0.37552796621016254,
      "grad_norm": 0.703125,
      "learning_rate": 0.0014366563409591187,
      "loss": 2.4771,
      "step": 2934
    },
    {
      "epoch": 0.3756559580186868,
      "grad_norm": 0.515625,
      "learning_rate": 0.0014362833467375837,
      "loss": 0.8978,
      "step": 2935
    },
    {
      "epoch": 0.3757839498272111,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0014359102775336773,
      "loss": 1.4376,
      "step": 2936
    },
    {
      "epoch": 0.3759119416357353,
      "grad_norm": 0.70703125,
      "learning_rate": 0.001435537133411517,
      "loss": 1.6999,
      "step": 2937
    },
    {
      "epoch": 0.37603993344425957,
      "grad_norm": 0.6328125,
      "learning_rate": 0.001435163914435234,
      "loss": 1.6372,
      "step": 2938
    },
    {
      "epoch": 0.3761679252527838,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0014347906206689718,
      "loss": 1.9178,
      "step": 2939
    },
    {
      "epoch": 0.37629591706130805,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0014344172521768871,
      "loss": 1.8674,
      "step": 2940
    },
    {
      "epoch": 0.37642390886983235,
      "grad_norm": 0.78125,
      "learning_rate": 0.0014340438090231491,
      "loss": 1.9482,
      "step": 2941
    },
    {
      "epoch": 0.3765519006783566,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0014336702912719405,
      "loss": 1.5894,
      "step": 2942
    },
    {
      "epoch": 0.37667989248688083,
      "grad_norm": 0.6484375,
      "learning_rate": 0.001433296698987456,
      "loss": 1.8598,
      "step": 2943
    },
    {
      "epoch": 0.37680788429540507,
      "grad_norm": 0.5625,
      "learning_rate": 0.0014329230322339035,
      "loss": 1.3173,
      "step": 2944
    },
    {
      "epoch": 0.37693587610392937,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0014325492910755042,
      "loss": 1.9664,
      "step": 2945
    },
    {
      "epoch": 0.3770638679124536,
      "grad_norm": 0.6953125,
      "learning_rate": 0.001432175475576491,
      "loss": 1.7357,
      "step": 2946
    },
    {
      "epoch": 0.37719185972097785,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0014318015858011107,
      "loss": 1.349,
      "step": 2947
    },
    {
      "epoch": 0.3773198515295021,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0014314276218136217,
      "loss": 2.145,
      "step": 2948
    },
    {
      "epoch": 0.3774478433380264,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0014310535836782959,
      "loss": 1.2374,
      "step": 2949
    },
    {
      "epoch": 0.37757583514655063,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0014306794714594182,
      "loss": 1.8016,
      "step": 2950
    },
    {
      "epoch": 0.3777038269550749,
      "grad_norm": 0.625,
      "learning_rate": 0.0014303052852212858,
      "loss": 1.659,
      "step": 2951
    },
    {
      "epoch": 0.3778318187635991,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0014299310250282084,
      "loss": 1.0724,
      "step": 2952
    },
    {
      "epoch": 0.3779598105721234,
      "grad_norm": 0.62890625,
      "learning_rate": 0.001429556690944509,
      "loss": 1.6249,
      "step": 2953
    },
    {
      "epoch": 0.37808780238064765,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0014291822830345224,
      "loss": 1.0761,
      "step": 2954
    },
    {
      "epoch": 0.3782157941891719,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0014288078013625974,
      "loss": 1.9362,
      "step": 2955
    },
    {
      "epoch": 0.37834378599769614,
      "grad_norm": 0.671875,
      "learning_rate": 0.0014284332459930944,
      "loss": 1.3837,
      "step": 2956
    },
    {
      "epoch": 0.3784717778062204,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0014280586169903868,
      "loss": 1.6261,
      "step": 2957
    },
    {
      "epoch": 0.3785997696147447,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0014276839144188606,
      "loss": 2.0744,
      "step": 2958
    },
    {
      "epoch": 0.3787277614232689,
      "grad_norm": 0.640625,
      "learning_rate": 0.001427309138342915,
      "loss": 2.1608,
      "step": 2959
    },
    {
      "epoch": 0.37885575323179316,
      "grad_norm": 0.625,
      "learning_rate": 0.0014269342888269608,
      "loss": 1.1603,
      "step": 2960
    },
    {
      "epoch": 0.3789837450403174,
      "grad_norm": 0.8203125,
      "learning_rate": 0.001426559365935422,
      "loss": 2.6154,
      "step": 2961
    },
    {
      "epoch": 0.3791117368488417,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0014261843697327354,
      "loss": 1.3291,
      "step": 2962
    },
    {
      "epoch": 0.37923972865736594,
      "grad_norm": 0.734375,
      "learning_rate": 0.0014258093002833504,
      "loss": 2.1215,
      "step": 2963
    },
    {
      "epoch": 0.3793677204658902,
      "grad_norm": 0.66796875,
      "learning_rate": 0.001425434157651728,
      "loss": 1.321,
      "step": 2964
    },
    {
      "epoch": 0.3794957122744144,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0014250589419023433,
      "loss": 2.0127,
      "step": 2965
    },
    {
      "epoch": 0.3796237040829387,
      "grad_norm": 0.6796875,
      "learning_rate": 0.001424683653099683,
      "loss": 1.7933,
      "step": 2966
    },
    {
      "epoch": 0.37975169589146296,
      "grad_norm": 0.6875,
      "learning_rate": 0.0014243082913082466,
      "loss": 1.6117,
      "step": 2967
    },
    {
      "epoch": 0.3798796876999872,
      "grad_norm": 0.671875,
      "learning_rate": 0.001423932856592546,
      "loss": 1.567,
      "step": 2968
    },
    {
      "epoch": 0.38000767950851144,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0014235573490171056,
      "loss": 1.6992,
      "step": 2969
    },
    {
      "epoch": 0.3801356713170357,
      "grad_norm": 0.8203125,
      "learning_rate": 0.001423181768646463,
      "loss": 2.1282,
      "step": 2970
    },
    {
      "epoch": 0.38026366312556,
      "grad_norm": 0.62109375,
      "learning_rate": 0.001422806115545167,
      "loss": 1.7544,
      "step": 2971
    },
    {
      "epoch": 0.3803916549340842,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0014224303897777803,
      "loss": 1.7695,
      "step": 2972
    },
    {
      "epoch": 0.38051964674260846,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0014220545914088777,
      "loss": 1.8036,
      "step": 2973
    },
    {
      "epoch": 0.3806476385511327,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0014216787205030453,
      "loss": 1.162,
      "step": 2974
    },
    {
      "epoch": 0.380775630359657,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0014213027771248833,
      "loss": 1.5009,
      "step": 2975
    },
    {
      "epoch": 0.38090362216818124,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0014209267613390036,
      "loss": 1.5192,
      "step": 2976
    },
    {
      "epoch": 0.3810316139767055,
      "grad_norm": 0.625,
      "learning_rate": 0.0014205506732100303,
      "loss": 1.6141,
      "step": 2977
    },
    {
      "epoch": 0.3811596057852297,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0014201745128026007,
      "loss": 1.2063,
      "step": 2978
    },
    {
      "epoch": 0.381287597593754,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0014197982801813637,
      "loss": 1.96,
      "step": 2979
    },
    {
      "epoch": 0.38141558940227827,
      "grad_norm": 0.625,
      "learning_rate": 0.001419421975410981,
      "loss": 1.5785,
      "step": 2980
    },
    {
      "epoch": 0.3815435812108025,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0014190455985561273,
      "loss": 1.7526,
      "step": 2981
    },
    {
      "epoch": 0.38167157301932675,
      "grad_norm": 0.84375,
      "learning_rate": 0.0014186691496814881,
      "loss": 1.6122,
      "step": 2982
    },
    {
      "epoch": 0.381799564827851,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0014182926288517628,
      "loss": 1.6044,
      "step": 2983
    },
    {
      "epoch": 0.3819275566363753,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0014179160361316627,
      "loss": 1.9817,
      "step": 2984
    },
    {
      "epoch": 0.38205554844489953,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0014175393715859111,
      "loss": 1.5642,
      "step": 2985
    },
    {
      "epoch": 0.38218354025342377,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0014171626352792442,
      "loss": 1.2387,
      "step": 2986
    },
    {
      "epoch": 0.382311532061948,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00141678582727641,
      "loss": 1.3917,
      "step": 2987
    },
    {
      "epoch": 0.3824395238704723,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0014164089476421686,
      "loss": 1.5371,
      "step": 2988
    },
    {
      "epoch": 0.38256751567899655,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0014160319964412942,
      "loss": 1.9051,
      "step": 2989
    },
    {
      "epoch": 0.3826955074875208,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0014156549737385706,
      "loss": 1.8493,
      "step": 2990
    },
    {
      "epoch": 0.38282349929604503,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0014152778795987963,
      "loss": 1.4732,
      "step": 2991
    },
    {
      "epoch": 0.38295149110456933,
      "grad_norm": 0.75,
      "learning_rate": 0.0014149007140867803,
      "loss": 1.7962,
      "step": 2992
    },
    {
      "epoch": 0.38307948291309357,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0014145234772673452,
      "loss": 1.7035,
      "step": 2993
    },
    {
      "epoch": 0.3832074747216178,
      "grad_norm": 0.671875,
      "learning_rate": 0.0014141461692053245,
      "loss": 1.876,
      "step": 2994
    },
    {
      "epoch": 0.38333546653014205,
      "grad_norm": 0.84765625,
      "learning_rate": 0.001413768789965566,
      "loss": 2.1518,
      "step": 2995
    },
    {
      "epoch": 0.38346345833866635,
      "grad_norm": 0.71484375,
      "learning_rate": 0.001413391339612927,
      "loss": 1.6498,
      "step": 2996
    },
    {
      "epoch": 0.3835914501471906,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0014130138182122793,
      "loss": 1.7787,
      "step": 2997
    },
    {
      "epoch": 0.38371944195571484,
      "grad_norm": 0.7890625,
      "learning_rate": 0.001412636225828506,
      "loss": 2.2648,
      "step": 2998
    },
    {
      "epoch": 0.3838474337642391,
      "grad_norm": 0.61328125,
      "learning_rate": 0.001412258562526502,
      "loss": 1.281,
      "step": 2999
    },
    {
      "epoch": 0.3839754255727633,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0014118808283711758,
      "loss": 1.8461,
      "step": 3000
    },
    {
      "epoch": 0.3841034173812876,
      "grad_norm": 0.73828125,
      "learning_rate": 0.001411503023427446,
      "loss": 1.7706,
      "step": 3001
    },
    {
      "epoch": 0.38423140918981186,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0014111251477602455,
      "loss": 1.4304,
      "step": 3002
    },
    {
      "epoch": 0.3843594009983361,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0014107472014345177,
      "loss": 1.6504,
      "step": 3003
    },
    {
      "epoch": 0.38448739280686034,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0014103691845152192,
      "loss": 1.3872,
      "step": 3004
    },
    {
      "epoch": 0.38461538461538464,
      "grad_norm": 0.578125,
      "learning_rate": 0.001409991097067318,
      "loss": 1.3533,
      "step": 3005
    },
    {
      "epoch": 0.3847433764239089,
      "grad_norm": 0.6953125,
      "learning_rate": 0.001409612939155795,
      "loss": 2.1768,
      "step": 3006
    },
    {
      "epoch": 0.3848713682324331,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0014092347108456424,
      "loss": 1.5256,
      "step": 3007
    },
    {
      "epoch": 0.38499936004095736,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0014088564122018653,
      "loss": 2.3395,
      "step": 3008
    },
    {
      "epoch": 0.38512735184948166,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00140847804328948,
      "loss": 1.9746,
      "step": 3009
    },
    {
      "epoch": 0.3852553436580059,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0014080996041735151,
      "loss": 1.584,
      "step": 3010
    },
    {
      "epoch": 0.38538333546653014,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0014077210949190123,
      "loss": 1.8887,
      "step": 3011
    },
    {
      "epoch": 0.3855113272750544,
      "grad_norm": 0.640625,
      "learning_rate": 0.001407342515591024,
      "loss": 1.5695,
      "step": 3012
    },
    {
      "epoch": 0.3856393190835786,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0014069638662546158,
      "loss": 1.6608,
      "step": 3013
    },
    {
      "epoch": 0.3857673108921029,
      "grad_norm": 0.734375,
      "learning_rate": 0.0014065851469748638,
      "loss": 2.0418,
      "step": 3014
    },
    {
      "epoch": 0.38589530270062716,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0014062063578168576,
      "loss": 1.4212,
      "step": 3015
    },
    {
      "epoch": 0.3860232945091514,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0014058274988456984,
      "loss": 1.9445,
      "step": 3016
    },
    {
      "epoch": 0.38615128631767565,
      "grad_norm": 0.625,
      "learning_rate": 0.0014054485701264986,
      "loss": 1.466,
      "step": 3017
    },
    {
      "epoch": 0.38627927812619994,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0014050695717243845,
      "loss": 2.076,
      "step": 3018
    },
    {
      "epoch": 0.3864072699347242,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0014046905037044919,
      "loss": 2.3681,
      "step": 3019
    },
    {
      "epoch": 0.3865352617432484,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0014043113661319703,
      "loss": 1.9839,
      "step": 3020
    },
    {
      "epoch": 0.38666325355177267,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0014039321590719807,
      "loss": 1.9767,
      "step": 3021
    },
    {
      "epoch": 0.38679124536029696,
      "grad_norm": 0.6875,
      "learning_rate": 0.0014035528825896955,
      "loss": 1.2785,
      "step": 3022
    },
    {
      "epoch": 0.3869192371688212,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0014031735367503,
      "loss": 1.5653,
      "step": 3023
    },
    {
      "epoch": 0.38704722897734545,
      "grad_norm": 0.64453125,
      "learning_rate": 0.001402794121618991,
      "loss": 1.5398,
      "step": 3024
    },
    {
      "epoch": 0.3871752207858697,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0014024146372609769,
      "loss": 1.643,
      "step": 3025
    },
    {
      "epoch": 0.387303212594394,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0014020350837414784,
      "loss": 1.8591,
      "step": 3026
    },
    {
      "epoch": 0.38743120440291823,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0014016554611257276,
      "loss": 2.0616,
      "step": 3027
    },
    {
      "epoch": 0.38755919621144247,
      "grad_norm": 0.80859375,
      "learning_rate": 0.001401275769478969,
      "loss": 1.9311,
      "step": 3028
    },
    {
      "epoch": 0.3876871880199667,
      "grad_norm": 0.64453125,
      "learning_rate": 0.001400896008866459,
      "loss": 1.9572,
      "step": 3029
    },
    {
      "epoch": 0.38781517982849095,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0014005161793534655,
      "loss": 1.3475,
      "step": 3030
    },
    {
      "epoch": 0.38794317163701525,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0014001362810052683,
      "loss": 1.3152,
      "step": 3031
    },
    {
      "epoch": 0.3880711634455395,
      "grad_norm": 0.94921875,
      "learning_rate": 0.001399756313887159,
      "loss": 1.4649,
      "step": 3032
    },
    {
      "epoch": 0.38819915525406373,
      "grad_norm": 0.84375,
      "learning_rate": 0.0013993762780644412,
      "loss": 1.718,
      "step": 3033
    },
    {
      "epoch": 0.388327147062588,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0013989961736024302,
      "loss": 1.7172,
      "step": 3034
    },
    {
      "epoch": 0.38845513887111227,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0013986160005664532,
      "loss": 1.4262,
      "step": 3035
    },
    {
      "epoch": 0.3885831306796365,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0013982357590218491,
      "loss": 1.585,
      "step": 3036
    },
    {
      "epoch": 0.38871112248816075,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0013978554490339686,
      "loss": 1.7206,
      "step": 3037
    },
    {
      "epoch": 0.388839114296685,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0013974750706681736,
      "loss": 2.1836,
      "step": 3038
    },
    {
      "epoch": 0.3889671061052093,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0013970946239898392,
      "loss": 1.7603,
      "step": 3039
    },
    {
      "epoch": 0.38909509791373353,
      "grad_norm": 0.71875,
      "learning_rate": 0.001396714109064351,
      "loss": 1.8834,
      "step": 3040
    },
    {
      "epoch": 0.3892230897222578,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0013963335259571062,
      "loss": 1.9161,
      "step": 3041
    },
    {
      "epoch": 0.389351081530782,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0013959528747335148,
      "loss": 1.6412,
      "step": 3042
    },
    {
      "epoch": 0.38947907333930626,
      "grad_norm": 0.625,
      "learning_rate": 0.0013955721554589978,
      "loss": 1.5279,
      "step": 3043
    },
    {
      "epoch": 0.38960706514783056,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0013951913681989874,
      "loss": 1.7993,
      "step": 3044
    },
    {
      "epoch": 0.3897350569563548,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0013948105130189286,
      "loss": 2.1543,
      "step": 3045
    },
    {
      "epoch": 0.38986304876487904,
      "grad_norm": 1.1875,
      "learning_rate": 0.0013944295899842776,
      "loss": 1.6158,
      "step": 3046
    },
    {
      "epoch": 0.3899910405734033,
      "grad_norm": 0.640625,
      "learning_rate": 0.0013940485991605022,
      "loss": 1.7264,
      "step": 3047
    },
    {
      "epoch": 0.3901190323819276,
      "grad_norm": 0.578125,
      "learning_rate": 0.0013936675406130811,
      "loss": 1.6409,
      "step": 3048
    },
    {
      "epoch": 0.3902470241904518,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0013932864144075064,
      "loss": 1.6642,
      "step": 3049
    },
    {
      "epoch": 0.39037501599897606,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0013929052206092802,
      "loss": 2.2272,
      "step": 3050
    },
    {
      "epoch": 0.3905030078075003,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0013925239592839174,
      "loss": 2.0149,
      "step": 3051
    },
    {
      "epoch": 0.3906309996160246,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0013921426304969434,
      "loss": 1.4573,
      "step": 3052
    },
    {
      "epoch": 0.39075899142454884,
      "grad_norm": 1.046875,
      "learning_rate": 0.001391761234313896,
      "loss": 0.9883,
      "step": 3053
    },
    {
      "epoch": 0.3908869832330731,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0013913797708003243,
      "loss": 1.3383,
      "step": 3054
    },
    {
      "epoch": 0.3910149750415973,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0013909982400217885,
      "loss": 1.6537,
      "step": 3055
    },
    {
      "epoch": 0.39114296685012157,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0013906166420438618,
      "loss": 1.5555,
      "step": 3056
    },
    {
      "epoch": 0.39127095865864586,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0013902349769321272,
      "loss": 1.4126,
      "step": 3057
    },
    {
      "epoch": 0.3913989504671701,
      "grad_norm": 0.6875,
      "learning_rate": 0.0013898532447521803,
      "loss": 1.7207,
      "step": 3058
    },
    {
      "epoch": 0.39152694227569435,
      "grad_norm": 0.484375,
      "learning_rate": 0.0013894714455696283,
      "loss": 1.0625,
      "step": 3059
    },
    {
      "epoch": 0.3916549340842186,
      "grad_norm": 0.765625,
      "learning_rate": 0.001389089579450089,
      "loss": 1.6192,
      "step": 3060
    },
    {
      "epoch": 0.3917829258927429,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0013887076464591927,
      "loss": 1.8028,
      "step": 3061
    },
    {
      "epoch": 0.3919109177012671,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0013883256466625808,
      "loss": 2.1126,
      "step": 3062
    },
    {
      "epoch": 0.39203890950979137,
      "grad_norm": 0.71875,
      "learning_rate": 0.0013879435801259059,
      "loss": 1.549,
      "step": 3063
    },
    {
      "epoch": 0.3921669013183156,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0013875614469148329,
      "loss": 1.2345,
      "step": 3064
    },
    {
      "epoch": 0.3922948931268399,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0013871792470950366,
      "loss": 1.8556,
      "step": 3065
    },
    {
      "epoch": 0.39242288493536415,
      "grad_norm": 0.7265625,
      "learning_rate": 0.001386796980732205,
      "loss": 1.9797,
      "step": 3066
    },
    {
      "epoch": 0.3925508767438884,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0013864146478920366,
      "loss": 1.9219,
      "step": 3067
    },
    {
      "epoch": 0.39267886855241263,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0013860322486402416,
      "loss": 1.3075,
      "step": 3068
    },
    {
      "epoch": 0.3928068603609369,
      "grad_norm": 0.68359375,
      "learning_rate": 0.001385649783042541,
      "loss": 1.4742,
      "step": 3069
    },
    {
      "epoch": 0.39293485216946117,
      "grad_norm": 0.65625,
      "learning_rate": 0.0013852672511646682,
      "loss": 1.7269,
      "step": 3070
    },
    {
      "epoch": 0.3930628439779854,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0013848846530723677,
      "loss": 1.5405,
      "step": 3071
    },
    {
      "epoch": 0.39319083578650965,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0013845019888313946,
      "loss": 1.8432,
      "step": 3072
    },
    {
      "epoch": 0.3933188275950339,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0013841192585075165,
      "loss": 1.5151,
      "step": 3073
    },
    {
      "epoch": 0.3934468194035582,
      "grad_norm": 0.59375,
      "learning_rate": 0.0013837364621665114,
      "loss": 1.1654,
      "step": 3074
    },
    {
      "epoch": 0.39357481121208243,
      "grad_norm": 0.6875,
      "learning_rate": 0.0013833535998741696,
      "loss": 1.2094,
      "step": 3075
    },
    {
      "epoch": 0.3937028030206067,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0013829706716962915,
      "loss": 2.284,
      "step": 3076
    },
    {
      "epoch": 0.3938307948291309,
      "grad_norm": 0.625,
      "learning_rate": 0.00138258767769869,
      "loss": 1.673,
      "step": 3077
    },
    {
      "epoch": 0.3939587866376552,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0013822046179471886,
      "loss": 1.4047,
      "step": 3078
    },
    {
      "epoch": 0.39408677844617945,
      "grad_norm": 0.65625,
      "learning_rate": 0.0013818214925076224,
      "loss": 2.181,
      "step": 3079
    },
    {
      "epoch": 0.3942147702547037,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0013814383014458377,
      "loss": 1.772,
      "step": 3080
    },
    {
      "epoch": 0.39434276206322794,
      "grad_norm": 0.65625,
      "learning_rate": 0.0013810550448276923,
      "loss": 1.6046,
      "step": 3081
    },
    {
      "epoch": 0.39447075387175223,
      "grad_norm": 0.78125,
      "learning_rate": 0.0013806717227190554,
      "loss": 1.7823,
      "step": 3082
    },
    {
      "epoch": 0.3945987456802765,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0013802883351858055,
      "loss": 2.2111,
      "step": 3083
    },
    {
      "epoch": 0.3947267374888007,
      "grad_norm": 0.6015625,
      "learning_rate": 0.001379904882293836,
      "loss": 1.6126,
      "step": 3084
    },
    {
      "epoch": 0.39485472929732496,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0013795213641090483,
      "loss": 1.3325,
      "step": 3085
    },
    {
      "epoch": 0.3949827211058492,
      "grad_norm": 0.67578125,
      "learning_rate": 0.001379137780697357,
      "loss": 1.4418,
      "step": 3086
    },
    {
      "epoch": 0.3951107129143735,
      "grad_norm": 0.703125,
      "learning_rate": 0.0013787541321246862,
      "loss": 1.6134,
      "step": 3087
    },
    {
      "epoch": 0.39523870472289774,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0013783704184569725,
      "loss": 1.845,
      "step": 3088
    },
    {
      "epoch": 0.395366696531422,
      "grad_norm": 0.625,
      "learning_rate": 0.0013779866397601636,
      "loss": 1.8878,
      "step": 3089
    },
    {
      "epoch": 0.3954946883399462,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0013776027961002178,
      "loss": 1.6038,
      "step": 3090
    },
    {
      "epoch": 0.3956226801484705,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0013772188875431053,
      "loss": 1.7348,
      "step": 3091
    },
    {
      "epoch": 0.39575067195699476,
      "grad_norm": 0.890625,
      "learning_rate": 0.0013768349141548063,
      "loss": 1.5436,
      "step": 3092
    },
    {
      "epoch": 0.395878663765519,
      "grad_norm": 0.703125,
      "learning_rate": 0.0013764508760013136,
      "loss": 1.7762,
      "step": 3093
    },
    {
      "epoch": 0.39600665557404324,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0013760667731486295,
      "loss": 1.3375,
      "step": 3094
    },
    {
      "epoch": 0.39613464738256754,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0013756826056627693,
      "loss": 1.3371,
      "step": 3095
    },
    {
      "epoch": 0.3962626391910918,
      "grad_norm": 0.61328125,
      "learning_rate": 0.001375298373609758,
      "loss": 1.8017,
      "step": 3096
    },
    {
      "epoch": 0.396390630999616,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0013749140770556321,
      "loss": 1.2794,
      "step": 3097
    },
    {
      "epoch": 0.39651862280814026,
      "grad_norm": 0.62890625,
      "learning_rate": 0.001374529716066439,
      "loss": 1.604,
      "step": 3098
    },
    {
      "epoch": 0.39664661461666456,
      "grad_norm": 0.546875,
      "learning_rate": 0.0013741452907082372,
      "loss": 1.1431,
      "step": 3099
    },
    {
      "epoch": 0.3967746064251888,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0013737608010470973,
      "loss": 1.9227,
      "step": 3100
    },
    {
      "epoch": 0.39690259823371304,
      "grad_norm": 0.625,
      "learning_rate": 0.001373376247149099,
      "loss": 1.6141,
      "step": 3101
    },
    {
      "epoch": 0.3970305900422373,
      "grad_norm": 0.765625,
      "learning_rate": 0.0013729916290803352,
      "loss": 1.6846,
      "step": 3102
    },
    {
      "epoch": 0.3971585818507615,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0013726069469069083,
      "loss": 1.3175,
      "step": 3103
    },
    {
      "epoch": 0.3972865736592858,
      "grad_norm": 0.625,
      "learning_rate": 0.001372222200694932,
      "loss": 1.5514,
      "step": 3104
    },
    {
      "epoch": 0.39741456546781007,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0013718373905105313,
      "loss": 1.9435,
      "step": 3105
    },
    {
      "epoch": 0.3975425572763343,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0013714525164198418,
      "loss": 1.3087,
      "step": 3106
    },
    {
      "epoch": 0.39767054908485855,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0013710675784890107,
      "loss": 1.4216,
      "step": 3107
    },
    {
      "epoch": 0.39779854089338285,
      "grad_norm": 0.82421875,
      "learning_rate": 0.001370682576784196,
      "loss": 1.8052,
      "step": 3108
    },
    {
      "epoch": 0.3979265327019071,
      "grad_norm": 0.6796875,
      "learning_rate": 0.001370297511371566,
      "loss": 1.1608,
      "step": 3109
    },
    {
      "epoch": 0.39805452451043133,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0013699123823173007,
      "loss": 1.7608,
      "step": 3110
    },
    {
      "epoch": 0.39818251631895557,
      "grad_norm": 0.71875,
      "learning_rate": 0.0013695271896875909,
      "loss": 1.5223,
      "step": 3111
    },
    {
      "epoch": 0.39831050812747987,
      "grad_norm": 0.765625,
      "learning_rate": 0.0013691419335486379,
      "loss": 1.4166,
      "step": 3112
    },
    {
      "epoch": 0.3984384999360041,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0013687566139666547,
      "loss": 1.4842,
      "step": 3113
    },
    {
      "epoch": 0.39856649174452835,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0013683712310078643,
      "loss": 2.0265,
      "step": 3114
    },
    {
      "epoch": 0.3986944835530526,
      "grad_norm": 0.6640625,
      "learning_rate": 0.001367985784738501,
      "loss": 1.9539,
      "step": 3115
    },
    {
      "epoch": 0.39882247536157683,
      "grad_norm": 0.859375,
      "learning_rate": 0.00136760027522481,
      "loss": 2.2464,
      "step": 3116
    },
    {
      "epoch": 0.39895046717010113,
      "grad_norm": 1.0625,
      "learning_rate": 0.0013672147025330476,
      "loss": 1.5925,
      "step": 3117
    },
    {
      "epoch": 0.39907845897862537,
      "grad_norm": 0.671875,
      "learning_rate": 0.0013668290667294804,
      "loss": 1.6132,
      "step": 3118
    },
    {
      "epoch": 0.3992064507871496,
      "grad_norm": 0.65625,
      "learning_rate": 0.001366443367880387,
      "loss": 1.739,
      "step": 3119
    },
    {
      "epoch": 0.39933444259567386,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0013660576060520548,
      "loss": 1.5918,
      "step": 3120
    },
    {
      "epoch": 0.39946243440419815,
      "grad_norm": 1.078125,
      "learning_rate": 0.001365671781310784,
      "loss": 2.024,
      "step": 3121
    },
    {
      "epoch": 0.3995904262127224,
      "grad_norm": 0.875,
      "learning_rate": 0.0013652858937228845,
      "loss": 1.1176,
      "step": 3122
    },
    {
      "epoch": 0.39971841802124664,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0013648999433546773,
      "loss": 1.3741,
      "step": 3123
    },
    {
      "epoch": 0.3998464098297709,
      "grad_norm": 0.7265625,
      "learning_rate": 0.001364513930272495,
      "loss": 1.3982,
      "step": 3124
    },
    {
      "epoch": 0.3999744016382952,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0013641278545426792,
      "loss": 1.8582,
      "step": 3125
    },
    {
      "epoch": 0.4001023934468194,
      "grad_norm": 0.703125,
      "learning_rate": 0.001363741716231584,
      "loss": 2.2079,
      "step": 3126
    },
    {
      "epoch": 0.40023038525534366,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0013633555154055732,
      "loss": 1.4708,
      "step": 3127
    },
    {
      "epoch": 0.4003583770638679,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0013629692521310215,
      "loss": 1.5478,
      "step": 3128
    },
    {
      "epoch": 0.40048636887239214,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0013625829264743148,
      "loss": 1.714,
      "step": 3129
    },
    {
      "epoch": 0.40061436068091644,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0013621965385018496,
      "loss": 1.4313,
      "step": 3130
    },
    {
      "epoch": 0.4007423524894407,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0013618100882800323,
      "loss": 1.2326,
      "step": 3131
    },
    {
      "epoch": 0.4008703442979649,
      "grad_norm": 0.65625,
      "learning_rate": 0.0013614235758752813,
      "loss": 1.5607,
      "step": 3132
    },
    {
      "epoch": 0.40099833610648916,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0013610370013540248,
      "loss": 1.3011,
      "step": 3133
    },
    {
      "epoch": 0.40112632791501346,
      "grad_norm": 0.703125,
      "learning_rate": 0.0013606503647827016,
      "loss": 1.3873,
      "step": 3134
    },
    {
      "epoch": 0.4012543197235377,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0013602636662277622,
      "loss": 1.0389,
      "step": 3135
    },
    {
      "epoch": 0.40138231153206194,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0013598769057556663,
      "loss": 1.3943,
      "step": 3136
    },
    {
      "epoch": 0.4015103033405862,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0013594900834328855,
      "loss": 1.4725,
      "step": 3137
    },
    {
      "epoch": 0.4016382951491105,
      "grad_norm": 0.921875,
      "learning_rate": 0.0013591031993259008,
      "loss": 2.0203,
      "step": 3138
    },
    {
      "epoch": 0.4017662869576347,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0013587162535012049,
      "loss": 1.5781,
      "step": 3139
    },
    {
      "epoch": 0.40189427876615896,
      "grad_norm": 0.91796875,
      "learning_rate": 0.001358329246025301,
      "loss": 2.157,
      "step": 3140
    },
    {
      "epoch": 0.4020222705746832,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0013579421769647024,
      "loss": 2.5679,
      "step": 3141
    },
    {
      "epoch": 0.4021502623832075,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0013575550463859332,
      "loss": 1.968,
      "step": 3142
    },
    {
      "epoch": 0.40227825419173174,
      "grad_norm": 0.6796875,
      "learning_rate": 0.001357167854355528,
      "loss": 1.5567,
      "step": 3143
    },
    {
      "epoch": 0.402406246000256,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0013567806009400318,
      "loss": 0.9411,
      "step": 3144
    },
    {
      "epoch": 0.4025342378087802,
      "grad_norm": 0.87109375,
      "learning_rate": 0.001356393286206001,
      "loss": 2.3813,
      "step": 3145
    },
    {
      "epoch": 0.40266222961730447,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0013560059102200013,
      "loss": 1.7677,
      "step": 3146
    },
    {
      "epoch": 0.40279022142582876,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0013556184730486099,
      "loss": 1.4357,
      "step": 3147
    },
    {
      "epoch": 0.402918213234353,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0013552309747584144,
      "loss": 1.5339,
      "step": 3148
    },
    {
      "epoch": 0.40304620504287725,
      "grad_norm": 0.6640625,
      "learning_rate": 0.001354843415416012,
      "loss": 1.5793,
      "step": 3149
    },
    {
      "epoch": 0.4031741968514015,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0013544557950880113,
      "loss": 1.8711,
      "step": 3150
    },
    {
      "epoch": 0.4033021886599258,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0013540681138410316,
      "loss": 1.6158,
      "step": 3151
    },
    {
      "epoch": 0.40343018046845003,
      "grad_norm": 0.5625,
      "learning_rate": 0.0013536803717417016,
      "loss": 1.5156,
      "step": 3152
    },
    {
      "epoch": 0.40355817227697427,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0013532925688566615,
      "loss": 1.9058,
      "step": 3153
    },
    {
      "epoch": 0.4036861640854985,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0013529047052525615,
      "loss": 2.0544,
      "step": 3154
    },
    {
      "epoch": 0.4038141558940228,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0013525167809960617,
      "loss": 2.1876,
      "step": 3155
    },
    {
      "epoch": 0.40394214770254705,
      "grad_norm": 0.734375,
      "learning_rate": 0.0013521287961538338,
      "loss": 1.4873,
      "step": 3156
    },
    {
      "epoch": 0.4040701395110713,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0013517407507925591,
      "loss": 1.5106,
      "step": 3157
    },
    {
      "epoch": 0.40419813131959553,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0013513526449789293,
      "loss": 1.6863,
      "step": 3158
    },
    {
      "epoch": 0.4043261231281198,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0013509644787796473,
      "loss": 1.256,
      "step": 3159
    },
    {
      "epoch": 0.40445411493664407,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0013505762522614249,
      "loss": 1.9898,
      "step": 3160
    },
    {
      "epoch": 0.4045821067451683,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0013501879654909854,
      "loss": 1.3081,
      "step": 3161
    },
    {
      "epoch": 0.40471009855369255,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0013497996185350626,
      "loss": 1.3448,
      "step": 3162
    },
    {
      "epoch": 0.4048380903622168,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0013494112114604,
      "loss": 2.2031,
      "step": 3163
    },
    {
      "epoch": 0.4049660821707411,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0013490227443337514,
      "loss": 1.6086,
      "step": 3164
    },
    {
      "epoch": 0.40509407397926533,
      "grad_norm": 0.703125,
      "learning_rate": 0.0013486342172218816,
      "loss": 1.9751,
      "step": 3165
    },
    {
      "epoch": 0.4052220657877896,
      "grad_norm": 0.63671875,
      "learning_rate": 0.001348245630191565,
      "loss": 1.5674,
      "step": 3166
    },
    {
      "epoch": 0.4053500575963138,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0013478569833095865,
      "loss": 1.3691,
      "step": 3167
    },
    {
      "epoch": 0.4054780494048381,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0013474682766427416,
      "loss": 1.1611,
      "step": 3168
    },
    {
      "epoch": 0.40560604121336236,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0013470795102578357,
      "loss": 1.3798,
      "step": 3169
    },
    {
      "epoch": 0.4057340330218866,
      "grad_norm": 0.59375,
      "learning_rate": 0.0013466906842216847,
      "loss": 1.5548,
      "step": 3170
    },
    {
      "epoch": 0.40586202483041084,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0013463017986011148,
      "loss": 1.4671,
      "step": 3171
    },
    {
      "epoch": 0.4059900166389351,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0013459128534629619,
      "loss": 1.8885,
      "step": 3172
    },
    {
      "epoch": 0.4061180084474594,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0013455238488740728,
      "loss": 1.5053,
      "step": 3173
    },
    {
      "epoch": 0.4062460002559836,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0013451347849013045,
      "loss": 0.9789,
      "step": 3174
    },
    {
      "epoch": 0.40637399206450786,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0013447456616115237,
      "loss": 1.322,
      "step": 3175
    },
    {
      "epoch": 0.4065019838730321,
      "grad_norm": 0.65625,
      "learning_rate": 0.0013443564790716071,
      "loss": 1.5251,
      "step": 3176
    },
    {
      "epoch": 0.4066299756815564,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0013439672373484431,
      "loss": 1.4047,
      "step": 3177
    },
    {
      "epoch": 0.40675796749008064,
      "grad_norm": 0.828125,
      "learning_rate": 0.001343577936508928,
      "loss": 1.9439,
      "step": 3178
    },
    {
      "epoch": 0.4068859592986049,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0013431885766199702,
      "loss": 1.1518,
      "step": 3179
    },
    {
      "epoch": 0.4070139511071291,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0013427991577484874,
      "loss": 1.745,
      "step": 3180
    },
    {
      "epoch": 0.4071419429156534,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0013424096799614078,
      "loss": 1.4438,
      "step": 3181
    },
    {
      "epoch": 0.40726993472417766,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0013420201433256688,
      "loss": 1.1982,
      "step": 3182
    },
    {
      "epoch": 0.4073979265327019,
      "grad_norm": 0.65625,
      "learning_rate": 0.0013416305479082191,
      "loss": 1.2943,
      "step": 3183
    },
    {
      "epoch": 0.40752591834122615,
      "grad_norm": 0.75,
      "learning_rate": 0.0013412408937760168,
      "loss": 1.6744,
      "step": 3184
    },
    {
      "epoch": 0.40765391014975044,
      "grad_norm": 0.65625,
      "learning_rate": 0.0013408511809960303,
      "loss": 1.5203,
      "step": 3185
    },
    {
      "epoch": 0.4077819019582747,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0013404614096352385,
      "loss": 1.6436,
      "step": 3186
    },
    {
      "epoch": 0.4079098937667989,
      "grad_norm": 0.609375,
      "learning_rate": 0.001340071579760629,
      "loss": 1.3551,
      "step": 3187
    },
    {
      "epoch": 0.40803788557532317,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0013396816914392016,
      "loss": 1.4585,
      "step": 3188
    },
    {
      "epoch": 0.4081658773838474,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0013392917447379636,
      "loss": 1.5675,
      "step": 3189
    },
    {
      "epoch": 0.4082938691923717,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0013389017397239346,
      "loss": 1.3445,
      "step": 3190
    },
    {
      "epoch": 0.40842186100089595,
      "grad_norm": 0.6640625,
      "learning_rate": 0.001338511676464143,
      "loss": 1.5531,
      "step": 3191
    },
    {
      "epoch": 0.4085498528094202,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0013381215550256277,
      "loss": 1.0272,
      "step": 3192
    },
    {
      "epoch": 0.40867784461794443,
      "grad_norm": 0.671875,
      "learning_rate": 0.0013377313754754372,
      "loss": 1.6199,
      "step": 3193
    },
    {
      "epoch": 0.4088058364264687,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0013373411378806302,
      "loss": 1.4394,
      "step": 3194
    },
    {
      "epoch": 0.40893382823499297,
      "grad_norm": 0.84765625,
      "learning_rate": 0.001336950842308275,
      "loss": 1.6912,
      "step": 3195
    },
    {
      "epoch": 0.4090618200435172,
      "grad_norm": 0.9375,
      "learning_rate": 0.001336560488825451,
      "loss": 1.6453,
      "step": 3196
    },
    {
      "epoch": 0.40918981185204145,
      "grad_norm": 0.71875,
      "learning_rate": 0.0013361700774992464,
      "loss": 1.6802,
      "step": 3197
    },
    {
      "epoch": 0.40931780366056575,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0013357796083967597,
      "loss": 1.3206,
      "step": 3198
    },
    {
      "epoch": 0.40944579546909,
      "grad_norm": 0.75,
      "learning_rate": 0.0013353890815850993,
      "loss": 1.8972,
      "step": 3199
    },
    {
      "epoch": 0.40957378727761423,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0013349984971313836,
      "loss": 1.5101,
      "step": 3200
    },
    {
      "epoch": 0.4097017790861385,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0013346078551027408,
      "loss": 1.6004,
      "step": 3201
    },
    {
      "epoch": 0.4098297708946627,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0013342171555663091,
      "loss": 2.0847,
      "step": 3202
    },
    {
      "epoch": 0.409957762703187,
      "grad_norm": 0.67578125,
      "learning_rate": 0.001333826398589237,
      "loss": 2.0405,
      "step": 3203
    },
    {
      "epoch": 0.41008575451171125,
      "grad_norm": 0.625,
      "learning_rate": 0.0013334355842386815,
      "loss": 1.6316,
      "step": 3204
    },
    {
      "epoch": 0.4102137463202355,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0013330447125818115,
      "loss": 1.9201,
      "step": 3205
    },
    {
      "epoch": 0.41034173812875974,
      "grad_norm": 0.78125,
      "learning_rate": 0.0013326537836858038,
      "loss": 2.2348,
      "step": 3206
    },
    {
      "epoch": 0.41046972993728403,
      "grad_norm": 0.609375,
      "learning_rate": 0.001332262797617846,
      "loss": 1.029,
      "step": 3207
    },
    {
      "epoch": 0.4105977217458083,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0013318717544451358,
      "loss": 1.7886,
      "step": 3208
    },
    {
      "epoch": 0.4107257135543325,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0013314806542348801,
      "loss": 1.6665,
      "step": 3209
    },
    {
      "epoch": 0.41085370536285676,
      "grad_norm": 0.5625,
      "learning_rate": 0.0013310894970542957,
      "loss": 1.2825,
      "step": 3210
    },
    {
      "epoch": 0.41098169717138106,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0013306982829706091,
      "loss": 2.0388,
      "step": 3211
    },
    {
      "epoch": 0.4111096889799053,
      "grad_norm": 1.125,
      "learning_rate": 0.0013303070120510574,
      "loss": 1.8298,
      "step": 3212
    },
    {
      "epoch": 0.41123768078842954,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0013299156843628863,
      "loss": 1.5872,
      "step": 3213
    },
    {
      "epoch": 0.4113656725969538,
      "grad_norm": 1.03125,
      "learning_rate": 0.0013295242999733523,
      "loss": 1.8088,
      "step": 3214
    },
    {
      "epoch": 0.4114936644054781,
      "grad_norm": 0.765625,
      "learning_rate": 0.0013291328589497208,
      "loss": 2.1707,
      "step": 3215
    },
    {
      "epoch": 0.4116216562140023,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0013287413613592674,
      "loss": 1.5454,
      "step": 3216
    },
    {
      "epoch": 0.41174964802252656,
      "grad_norm": 1.3359375,
      "learning_rate": 0.001328349807269277,
      "loss": 1.8875,
      "step": 3217
    },
    {
      "epoch": 0.4118776398310508,
      "grad_norm": 0.9140625,
      "learning_rate": 0.001327958196747045,
      "loss": 1.5711,
      "step": 3218
    },
    {
      "epoch": 0.41200563163957504,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0013275665298598763,
      "loss": 1.8541,
      "step": 3219
    },
    {
      "epoch": 0.41213362344809934,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0013271748066750843,
      "loss": 2.0757,
      "step": 3220
    },
    {
      "epoch": 0.4122616152566236,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0013267830272599937,
      "loss": 1.7863,
      "step": 3221
    },
    {
      "epoch": 0.4123896070651478,
      "grad_norm": 0.640625,
      "learning_rate": 0.0013263911916819376,
      "loss": 1.4199,
      "step": 3222
    },
    {
      "epoch": 0.41251759887367206,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0013259993000082598,
      "loss": 2.1541,
      "step": 3223
    },
    {
      "epoch": 0.41264559068219636,
      "grad_norm": 0.625,
      "learning_rate": 0.0013256073523063127,
      "loss": 1.6949,
      "step": 3224
    },
    {
      "epoch": 0.4127735824907206,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0013252153486434596,
      "loss": 1.4258,
      "step": 3225
    },
    {
      "epoch": 0.41290157429924484,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0013248232890870721,
      "loss": 1.4577,
      "step": 3226
    },
    {
      "epoch": 0.4130295661077691,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0013244311737045324,
      "loss": 1.944,
      "step": 3227
    },
    {
      "epoch": 0.4131575579162934,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0013240390025632315,
      "loss": 1.4302,
      "step": 3228
    },
    {
      "epoch": 0.4132855497248176,
      "grad_norm": 0.75,
      "learning_rate": 0.0013236467757305704,
      "loss": 1.8283,
      "step": 3229
    },
    {
      "epoch": 0.41341354153334187,
      "grad_norm": 0.703125,
      "learning_rate": 0.0013232544932739601,
      "loss": 1.659,
      "step": 3230
    },
    {
      "epoch": 0.4135415333418661,
      "grad_norm": 0.703125,
      "learning_rate": 0.0013228621552608202,
      "loss": 1.3654,
      "step": 3231
    },
    {
      "epoch": 0.41366952515039035,
      "grad_norm": 0.78125,
      "learning_rate": 0.0013224697617585809,
      "loss": 1.8382,
      "step": 3232
    },
    {
      "epoch": 0.41379751695891465,
      "grad_norm": 0.5625,
      "learning_rate": 0.0013220773128346805,
      "loss": 1.1239,
      "step": 3233
    },
    {
      "epoch": 0.4139255087674389,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0013216848085565688,
      "loss": 1.76,
      "step": 3234
    },
    {
      "epoch": 0.41405350057596313,
      "grad_norm": 0.6796875,
      "learning_rate": 0.001321292248991703,
      "loss": 1.4854,
      "step": 3235
    },
    {
      "epoch": 0.41418149238448737,
      "grad_norm": 0.6875,
      "learning_rate": 0.0013208996342075517,
      "loss": 1.6808,
      "step": 3236
    },
    {
      "epoch": 0.41430948419301167,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0013205069642715915,
      "loss": 1.4843,
      "step": 3237
    },
    {
      "epoch": 0.4144374760015359,
      "grad_norm": 0.65625,
      "learning_rate": 0.0013201142392513096,
      "loss": 1.3534,
      "step": 3238
    },
    {
      "epoch": 0.41456546781006015,
      "grad_norm": 0.65625,
      "learning_rate": 0.0013197214592142014,
      "loss": 1.8532,
      "step": 3239
    },
    {
      "epoch": 0.4146934596185844,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0013193286242277735,
      "loss": 1.599,
      "step": 3240
    },
    {
      "epoch": 0.4148214514271087,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0013189357343595404,
      "loss": 1.7203,
      "step": 3241
    },
    {
      "epoch": 0.41494944323563293,
      "grad_norm": 0.65625,
      "learning_rate": 0.0013185427896770267,
      "loss": 1.4103,
      "step": 3242
    },
    {
      "epoch": 0.4150774350441572,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0013181497902477663,
      "loss": 1.7425,
      "step": 3243
    },
    {
      "epoch": 0.4152054268526814,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0013177567361393025,
      "loss": 1.4384,
      "step": 3244
    },
    {
      "epoch": 0.41533341866120566,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0013173636274191882,
      "loss": 1.2346,
      "step": 3245
    },
    {
      "epoch": 0.41546141046972995,
      "grad_norm": 0.59765625,
      "learning_rate": 0.001316970464154985,
      "loss": 1.413,
      "step": 3246
    },
    {
      "epoch": 0.4155894022782542,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0013165772464142653,
      "loss": 1.8685,
      "step": 3247
    },
    {
      "epoch": 0.41571739408677844,
      "grad_norm": 0.8125,
      "learning_rate": 0.0013161839742646092,
      "loss": 2.1633,
      "step": 3248
    },
    {
      "epoch": 0.4158453858953027,
      "grad_norm": 0.84375,
      "learning_rate": 0.001315790647773607,
      "loss": 1.9804,
      "step": 3249
    },
    {
      "epoch": 0.415973377703827,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0013153972670088582,
      "loss": 1.8009,
      "step": 3250
    },
    {
      "epoch": 0.4161013695123512,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0013150038320379723,
      "loss": 1.7432,
      "step": 3251
    },
    {
      "epoch": 0.41622936132087546,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0013146103429285667,
      "loss": 1.5213,
      "step": 3252
    },
    {
      "epoch": 0.4163573531293997,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0013142167997482693,
      "loss": 1.4661,
      "step": 3253
    },
    {
      "epoch": 0.416485344937924,
      "grad_norm": 0.671875,
      "learning_rate": 0.0013138232025647167,
      "loss": 1.7972,
      "step": 3254
    },
    {
      "epoch": 0.41661333674644824,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0013134295514455553,
      "loss": 1.2969,
      "step": 3255
    },
    {
      "epoch": 0.4167413285549725,
      "grad_norm": 0.609375,
      "learning_rate": 0.0013130358464584397,
      "loss": 1.4629,
      "step": 3256
    },
    {
      "epoch": 0.4168693203634967,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0013126420876710353,
      "loss": 2.2046,
      "step": 3257
    },
    {
      "epoch": 0.416997312172021,
      "grad_norm": 0.609375,
      "learning_rate": 0.0013122482751510153,
      "loss": 1.4503,
      "step": 3258
    },
    {
      "epoch": 0.41712530398054526,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0013118544089660634,
      "loss": 1.7137,
      "step": 3259
    },
    {
      "epoch": 0.4172532957890695,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0013114604891838711,
      "loss": 1.4641,
      "step": 3260
    },
    {
      "epoch": 0.41738128759759374,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0013110665158721406,
      "loss": 1.9427,
      "step": 3261
    },
    {
      "epoch": 0.417509279406118,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0013106724890985823,
      "loss": 1.3499,
      "step": 3262
    },
    {
      "epoch": 0.4176372712146423,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0013102784089309165,
      "loss": 1.3175,
      "step": 3263
    },
    {
      "epoch": 0.4177652630231665,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0013098842754368718,
      "loss": 1.3827,
      "step": 3264
    },
    {
      "epoch": 0.41789325483169076,
      "grad_norm": 0.703125,
      "learning_rate": 0.0013094900886841865,
      "loss": 1.3803,
      "step": 3265
    },
    {
      "epoch": 0.418021246640215,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0013090958487406079,
      "loss": 2.0383,
      "step": 3266
    },
    {
      "epoch": 0.4181492384487393,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0013087015556738928,
      "loss": 1.2524,
      "step": 3267
    },
    {
      "epoch": 0.41827723025726354,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0013083072095518068,
      "loss": 1.8613,
      "step": 3268
    },
    {
      "epoch": 0.4184052220657878,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0013079128104421245,
      "loss": 1.8132,
      "step": 3269
    },
    {
      "epoch": 0.418533213874312,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0013075183584126302,
      "loss": 1.6926,
      "step": 3270
    },
    {
      "epoch": 0.4186612056828363,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0013071238535311163,
      "loss": 1.5211,
      "step": 3271
    },
    {
      "epoch": 0.41878919749136057,
      "grad_norm": 0.609375,
      "learning_rate": 0.0013067292958653853,
      "loss": 1.3111,
      "step": 3272
    },
    {
      "epoch": 0.4189171892998848,
      "grad_norm": 0.76953125,
      "learning_rate": 0.001306334685483248,
      "loss": 1.2039,
      "step": 3273
    },
    {
      "epoch": 0.41904518110840905,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0013059400224525253,
      "loss": 1.5959,
      "step": 3274
    },
    {
      "epoch": 0.4191731729169333,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0013055453068410454,
      "loss": 1.8884,
      "step": 3275
    },
    {
      "epoch": 0.4193011647254576,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0013051505387166476,
      "loss": 1.2878,
      "step": 3276
    },
    {
      "epoch": 0.41942915653398183,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0013047557181471783,
      "loss": 1.4471,
      "step": 3277
    },
    {
      "epoch": 0.41955714834250607,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0013043608452004943,
      "loss": 1.0831,
      "step": 3278
    },
    {
      "epoch": 0.4196851401510303,
      "grad_norm": 0.6875,
      "learning_rate": 0.001303965919944461,
      "loss": 1.3966,
      "step": 3279
    },
    {
      "epoch": 0.4198131319595546,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0013035709424469525,
      "loss": 1.1363,
      "step": 3280
    },
    {
      "epoch": 0.41994112376807885,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0013031759127758522,
      "loss": 1.4959,
      "step": 3281
    },
    {
      "epoch": 0.4200691155766031,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0013027808309990522,
      "loss": 1.9854,
      "step": 3282
    },
    {
      "epoch": 0.42019710738512733,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0013023856971844544,
      "loss": 2.1157,
      "step": 3283
    },
    {
      "epoch": 0.42032509919365163,
      "grad_norm": 0.65625,
      "learning_rate": 0.0013019905113999678,
      "loss": 2.095,
      "step": 3284
    },
    {
      "epoch": 0.42045309100217587,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0013015952737135124,
      "loss": 2.0794,
      "step": 3285
    },
    {
      "epoch": 0.4205810828107001,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0013011999841930158,
      "loss": 1.5729,
      "step": 3286
    },
    {
      "epoch": 0.42070907461922435,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0013008046429064152,
      "loss": 1.9995,
      "step": 3287
    },
    {
      "epoch": 0.42083706642774865,
      "grad_norm": 0.67578125,
      "learning_rate": 0.001300409249921656,
      "loss": 1.8075,
      "step": 3288
    },
    {
      "epoch": 0.4209650582362729,
      "grad_norm": 0.6640625,
      "learning_rate": 0.001300013805306693,
      "loss": 1.7081,
      "step": 3289
    },
    {
      "epoch": 0.42109305004479713,
      "grad_norm": 0.828125,
      "learning_rate": 0.00129961830912949,
      "loss": 1.9376,
      "step": 3290
    },
    {
      "epoch": 0.4212210418533214,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0012992227614580195,
      "loss": 1.9304,
      "step": 3291
    },
    {
      "epoch": 0.4213490336618456,
      "grad_norm": 0.96875,
      "learning_rate": 0.0012988271623602628,
      "loss": 2.1461,
      "step": 3292
    },
    {
      "epoch": 0.4214770254703699,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0012984315119042094,
      "loss": 1.7212,
      "step": 3293
    },
    {
      "epoch": 0.42160501727889416,
      "grad_norm": 0.72265625,
      "learning_rate": 0.001298035810157859,
      "loss": 1.5468,
      "step": 3294
    },
    {
      "epoch": 0.4217330090874184,
      "grad_norm": 0.65234375,
      "learning_rate": 0.001297640057189219,
      "loss": 1.3799,
      "step": 3295
    },
    {
      "epoch": 0.42186100089594264,
      "grad_norm": 0.8671875,
      "learning_rate": 0.001297244253066306,
      "loss": 1.5107,
      "step": 3296
    },
    {
      "epoch": 0.42198899270446694,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0012968483978571452,
      "loss": 1.5036,
      "step": 3297
    },
    {
      "epoch": 0.4221169845129912,
      "grad_norm": 0.578125,
      "learning_rate": 0.0012964524916297714,
      "loss": 1.2885,
      "step": 3298
    },
    {
      "epoch": 0.4222449763215154,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0012960565344522267,
      "loss": 1.2853,
      "step": 3299
    },
    {
      "epoch": 0.42237296813003966,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0012956605263925631,
      "loss": 2.4334,
      "step": 3300
    },
    {
      "epoch": 0.42250095993856396,
      "grad_norm": 0.65234375,
      "learning_rate": 0.001295264467518841,
      "loss": 1.3032,
      "step": 3301
    },
    {
      "epoch": 0.4226289517470882,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0012948683578991298,
      "loss": 1.5153,
      "step": 3302
    },
    {
      "epoch": 0.42275694355561244,
      "grad_norm": 0.6796875,
      "learning_rate": 0.001294472197601507,
      "loss": 1.792,
      "step": 3303
    },
    {
      "epoch": 0.4228849353641367,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0012940759866940591,
      "loss": 1.1729,
      "step": 3304
    },
    {
      "epoch": 0.4230129271726609,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0012936797252448822,
      "loss": 1.3599,
      "step": 3305
    },
    {
      "epoch": 0.4231409189811852,
      "grad_norm": 0.828125,
      "learning_rate": 0.0012932834133220791,
      "loss": 1.6,
      "step": 3306
    },
    {
      "epoch": 0.42326891078970946,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0012928870509937632,
      "loss": 1.3691,
      "step": 3307
    },
    {
      "epoch": 0.4233969025982337,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0012924906383280552,
      "loss": 2.5634,
      "step": 3308
    },
    {
      "epoch": 0.42352489440675795,
      "grad_norm": 0.75,
      "learning_rate": 0.0012920941753930858,
      "loss": 1.4948,
      "step": 3309
    },
    {
      "epoch": 0.42365288621528224,
      "grad_norm": 0.6875,
      "learning_rate": 0.0012916976622569932,
      "loss": 1.922,
      "step": 3310
    },
    {
      "epoch": 0.4237808780238065,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0012913010989879243,
      "loss": 1.4629,
      "step": 3311
    },
    {
      "epoch": 0.4239088698323307,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0012909044856540355,
      "loss": 1.5908,
      "step": 3312
    },
    {
      "epoch": 0.42403686164085497,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0012905078223234908,
      "loss": 2.1848,
      "step": 3313
    },
    {
      "epoch": 0.42416485344937926,
      "grad_norm": 0.703125,
      "learning_rate": 0.0012901111090644637,
      "loss": 1.8984,
      "step": 3314
    },
    {
      "epoch": 0.4242928452579035,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0012897143459451354,
      "loss": 1.7739,
      "step": 3315
    },
    {
      "epoch": 0.42442083706642775,
      "grad_norm": 0.54296875,
      "learning_rate": 0.001289317533033696,
      "loss": 1.4739,
      "step": 3316
    },
    {
      "epoch": 0.424548828874952,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0012889206703983447,
      "loss": 1.3993,
      "step": 3317
    },
    {
      "epoch": 0.42467682068347623,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0012885237581072883,
      "loss": 1.4416,
      "step": 3318
    },
    {
      "epoch": 0.4248048124920005,
      "grad_norm": 0.8125,
      "learning_rate": 0.001288126796228743,
      "loss": 2.4509,
      "step": 3319
    },
    {
      "epoch": 0.42493280430052477,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0012877297848309331,
      "loss": 1.6117,
      "step": 3320
    },
    {
      "epoch": 0.425060796109049,
      "grad_norm": 0.609375,
      "learning_rate": 0.001287332723982091,
      "loss": 1.5292,
      "step": 3321
    },
    {
      "epoch": 0.42518878791757325,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0012869356137504586,
      "loss": 1.409,
      "step": 3322
    },
    {
      "epoch": 0.42531677972609755,
      "grad_norm": 0.6875,
      "learning_rate": 0.0012865384542042856,
      "loss": 1.6209,
      "step": 3323
    },
    {
      "epoch": 0.4254447715346218,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0012861412454118297,
      "loss": 1.7499,
      "step": 3324
    },
    {
      "epoch": 0.42557276334314603,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0012857439874413587,
      "loss": 1.6339,
      "step": 3325
    },
    {
      "epoch": 0.4257007551516703,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0012853466803611473,
      "loss": 2.2414,
      "step": 3326
    },
    {
      "epoch": 0.42582874696019457,
      "grad_norm": 0.796875,
      "learning_rate": 0.0012849493242394793,
      "loss": 2.5823,
      "step": 3327
    },
    {
      "epoch": 0.4259567387687188,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0012845519191446463,
      "loss": 1.678,
      "step": 3328
    },
    {
      "epoch": 0.42608473057724305,
      "grad_norm": 0.671875,
      "learning_rate": 0.0012841544651449497,
      "loss": 1.9392,
      "step": 3329
    },
    {
      "epoch": 0.4262127223857673,
      "grad_norm": 0.671875,
      "learning_rate": 0.0012837569623086976,
      "loss": 1.5999,
      "step": 3330
    },
    {
      "epoch": 0.4263407141942916,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0012833594107042076,
      "loss": 1.8916,
      "step": 3331
    },
    {
      "epoch": 0.42646870600281583,
      "grad_norm": 0.578125,
      "learning_rate": 0.0012829618103998059,
      "loss": 1.1979,
      "step": 3332
    },
    {
      "epoch": 0.4265966978113401,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0012825641614638253,
      "loss": 1.1127,
      "step": 3333
    },
    {
      "epoch": 0.4267246896198643,
      "grad_norm": 0.75,
      "learning_rate": 0.0012821664639646095,
      "loss": 2.5159,
      "step": 3334
    },
    {
      "epoch": 0.42685268142838856,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0012817687179705088,
      "loss": 2.1429,
      "step": 3335
    },
    {
      "epoch": 0.42698067323691286,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0012813709235498822,
      "loss": 1.7954,
      "step": 3336
    },
    {
      "epoch": 0.4271086650454371,
      "grad_norm": 0.703125,
      "learning_rate": 0.0012809730807710974,
      "loss": 1.7712,
      "step": 3337
    },
    {
      "epoch": 0.42723665685396134,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0012805751897025297,
      "loss": 1.4451,
      "step": 3338
    },
    {
      "epoch": 0.4273646486624856,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0012801772504125635,
      "loss": 1.4498,
      "step": 3339
    },
    {
      "epoch": 0.4274926404710099,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0012797792629695908,
      "loss": 1.6222,
      "step": 3340
    },
    {
      "epoch": 0.4276206322795341,
      "grad_norm": 0.640625,
      "learning_rate": 0.0012793812274420127,
      "loss": 1.067,
      "step": 3341
    },
    {
      "epoch": 0.42774862408805836,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0012789831438982375,
      "loss": 1.3138,
      "step": 3342
    },
    {
      "epoch": 0.4278766158965826,
      "grad_norm": 0.8359375,
      "learning_rate": 0.001278585012406683,
      "loss": 1.7521,
      "step": 3343
    },
    {
      "epoch": 0.4280046077051069,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0012781868330357738,
      "loss": 1.5049,
      "step": 3344
    },
    {
      "epoch": 0.42813259951363114,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0012777886058539438,
      "loss": 1.4626,
      "step": 3345
    },
    {
      "epoch": 0.4282605913221554,
      "grad_norm": 0.65625,
      "learning_rate": 0.0012773903309296348,
      "loss": 1.4665,
      "step": 3346
    },
    {
      "epoch": 0.4283885831306796,
      "grad_norm": 0.58203125,
      "learning_rate": 0.001276992008331297,
      "loss": 1.376,
      "step": 3347
    },
    {
      "epoch": 0.42851657493920386,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0012765936381273885,
      "loss": 2.123,
      "step": 3348
    },
    {
      "epoch": 0.42864456674772816,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0012761952203863757,
      "loss": 1.2421,
      "step": 3349
    },
    {
      "epoch": 0.4287725585562524,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0012757967551767328,
      "loss": 1.7098,
      "step": 3350
    },
    {
      "epoch": 0.42890055036477664,
      "grad_norm": 0.69140625,
      "learning_rate": 0.001275398242566943,
      "loss": 1.6417,
      "step": 3351
    },
    {
      "epoch": 0.4290285421733009,
      "grad_norm": 0.60546875,
      "learning_rate": 0.001274999682625497,
      "loss": 1.3395,
      "step": 3352
    },
    {
      "epoch": 0.4291565339818252,
      "grad_norm": 0.578125,
      "learning_rate": 0.0012746010754208933,
      "loss": 1.3381,
      "step": 3353
    },
    {
      "epoch": 0.4292845257903494,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00127420242102164,
      "loss": 1.6025,
      "step": 3354
    },
    {
      "epoch": 0.42941251759887367,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0012738037194962514,
      "loss": 1.6224,
      "step": 3355
    },
    {
      "epoch": 0.4295405094073979,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0012734049709132514,
      "loss": 1.5415,
      "step": 3356
    },
    {
      "epoch": 0.4296685012159222,
      "grad_norm": 0.65625,
      "learning_rate": 0.001273006175341171,
      "loss": 1.7885,
      "step": 3357
    },
    {
      "epoch": 0.42979649302444645,
      "grad_norm": 0.6875,
      "learning_rate": 0.0012726073328485502,
      "loss": 1.3686,
      "step": 3358
    },
    {
      "epoch": 0.4299244848329707,
      "grad_norm": 0.609375,
      "learning_rate": 0.001272208443503936,
      "loss": 0.9184,
      "step": 3359
    },
    {
      "epoch": 0.43005247664149493,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0012718095073758846,
      "loss": 1.7434,
      "step": 3360
    },
    {
      "epoch": 0.43018046845001917,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0012714105245329591,
      "loss": 1.1876,
      "step": 3361
    },
    {
      "epoch": 0.43030846025854347,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0012710114950437313,
      "loss": 1.4127,
      "step": 3362
    },
    {
      "epoch": 0.4304364520670677,
      "grad_norm": 0.8984375,
      "learning_rate": 0.001270612418976781,
      "loss": 2.0362,
      "step": 3363
    },
    {
      "epoch": 0.43056444387559195,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0012702132964006957,
      "loss": 1.8393,
      "step": 3364
    },
    {
      "epoch": 0.4306924356841162,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0012698141273840718,
      "loss": 1.1262,
      "step": 3365
    },
    {
      "epoch": 0.4308204274926405,
      "grad_norm": 0.59375,
      "learning_rate": 0.001269414911995512,
      "loss": 1.5473,
      "step": 3366
    },
    {
      "epoch": 0.43094841930116473,
      "grad_norm": 0.765625,
      "learning_rate": 0.0012690156503036287,
      "loss": 2.0267,
      "step": 3367
    },
    {
      "epoch": 0.431076411109689,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0012686163423770408,
      "loss": 1.5672,
      "step": 3368
    },
    {
      "epoch": 0.4312044029182132,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0012682169882843766,
      "loss": 1.6958,
      "step": 3369
    },
    {
      "epoch": 0.4313323947267375,
      "grad_norm": 0.94921875,
      "learning_rate": 0.001267817588094271,
      "loss": 1.9166,
      "step": 3370
    },
    {
      "epoch": 0.43146038653526175,
      "grad_norm": 0.5625,
      "learning_rate": 0.0012674181418753678,
      "loss": 1.5383,
      "step": 3371
    },
    {
      "epoch": 0.431588378343786,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0012670186496963178,
      "loss": 1.6633,
      "step": 3372
    },
    {
      "epoch": 0.43171637015231024,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0012666191116257807,
      "loss": 1.9452,
      "step": 3373
    },
    {
      "epoch": 0.43184436196083453,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0012662195277324236,
      "loss": 1.6633,
      "step": 3374
    },
    {
      "epoch": 0.4319723537693588,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0012658198980849212,
      "loss": 1.6215,
      "step": 3375
    },
    {
      "epoch": 0.432100345577883,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0012654202227519564,
      "loss": 1.2936,
      "step": 3376
    },
    {
      "epoch": 0.43222833738640726,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0012650205018022202,
      "loss": 1.5654,
      "step": 3377
    },
    {
      "epoch": 0.4323563291949315,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0012646207353044106,
      "loss": 1.6376,
      "step": 3378
    },
    {
      "epoch": 0.4324843210034558,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0012642209233272338,
      "loss": 1.2867,
      "step": 3379
    },
    {
      "epoch": 0.43261231281198004,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0012638210659394052,
      "loss": 1.6056,
      "step": 3380
    },
    {
      "epoch": 0.4327403046205043,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0012634211632096455,
      "loss": 1.2171,
      "step": 3381
    },
    {
      "epoch": 0.4328682964290285,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0012630212152066854,
      "loss": 1.0596,
      "step": 3382
    },
    {
      "epoch": 0.4329962882375528,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0012626212219992617,
      "loss": 1.6454,
      "step": 3383
    },
    {
      "epoch": 0.43312428004607706,
      "grad_norm": 0.78125,
      "learning_rate": 0.0012622211836561198,
      "loss": 1.7808,
      "step": 3384
    },
    {
      "epoch": 0.4332522718546013,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0012618211002460133,
      "loss": 1.3427,
      "step": 3385
    },
    {
      "epoch": 0.43338026366312554,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0012614209718377026,
      "loss": 1.2118,
      "step": 3386
    },
    {
      "epoch": 0.43350825547164984,
      "grad_norm": 0.640625,
      "learning_rate": 0.0012610207984999568,
      "loss": 1.7984,
      "step": 3387
    },
    {
      "epoch": 0.4336362472801741,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0012606205803015514,
      "loss": 2.2219,
      "step": 3388
    },
    {
      "epoch": 0.4337642390886983,
      "grad_norm": 0.66796875,
      "learning_rate": 0.001260220317311271,
      "loss": 1.3326,
      "step": 3389
    },
    {
      "epoch": 0.43389223089722256,
      "grad_norm": 0.65625,
      "learning_rate": 0.0012598200095979067,
      "loss": 1.8403,
      "step": 3390
    },
    {
      "epoch": 0.4340202227057468,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0012594196572302586,
      "loss": 1.0855,
      "step": 3391
    },
    {
      "epoch": 0.4341482145142711,
      "grad_norm": 0.6484375,
      "learning_rate": 0.001259019260277133,
      "loss": 1.4709,
      "step": 3392
    },
    {
      "epoch": 0.43427620632279534,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0012586188188073455,
      "loss": 1.7369,
      "step": 3393
    },
    {
      "epoch": 0.4344041981313196,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0012582183328897177,
      "loss": 1.7697,
      "step": 3394
    },
    {
      "epoch": 0.4345321899398438,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0012578178025930797,
      "loss": 1.5611,
      "step": 3395
    },
    {
      "epoch": 0.4346601817483681,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0012574172279862694,
      "loss": 2.0513,
      "step": 3396
    },
    {
      "epoch": 0.43478817355689237,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0012570166091381318,
      "loss": 1.2961,
      "step": 3397
    },
    {
      "epoch": 0.4349161653654166,
      "grad_norm": 0.5625,
      "learning_rate": 0.0012566159461175205,
      "loss": 1.3058,
      "step": 3398
    },
    {
      "epoch": 0.43504415717394085,
      "grad_norm": 0.6875,
      "learning_rate": 0.0012562152389932945,
      "loss": 1.6982,
      "step": 3399
    },
    {
      "epoch": 0.43517214898246515,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0012558144878343233,
      "loss": 2.0853,
      "step": 3400
    },
    {
      "epoch": 0.4353001407909894,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0012554136927094817,
      "loss": 1.3152,
      "step": 3401
    },
    {
      "epoch": 0.43542813259951363,
      "grad_norm": 0.6953125,
      "learning_rate": 0.001255012853687653,
      "loss": 1.5895,
      "step": 3402
    },
    {
      "epoch": 0.43555612440803787,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0012546119708377274,
      "loss": 1.8828,
      "step": 3403
    },
    {
      "epoch": 0.43568411621656217,
      "grad_norm": 0.765625,
      "learning_rate": 0.0012542110442286042,
      "loss": 1.761,
      "step": 3404
    },
    {
      "epoch": 0.4358121080250864,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0012538100739291882,
      "loss": 2.2002,
      "step": 3405
    },
    {
      "epoch": 0.43594009983361065,
      "grad_norm": 0.6796875,
      "learning_rate": 0.001253409060008393,
      "loss": 1.0123,
      "step": 3406
    },
    {
      "epoch": 0.4360680916421349,
      "grad_norm": 0.62890625,
      "learning_rate": 0.001253008002535139,
      "loss": 1.2089,
      "step": 3407
    },
    {
      "epoch": 0.43619608345065913,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0012526069015783552,
      "loss": 1.3318,
      "step": 3408
    },
    {
      "epoch": 0.43632407525918343,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0012522057572069772,
      "loss": 1.8096,
      "step": 3409
    },
    {
      "epoch": 0.43645206706770767,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0012518045694899474,
      "loss": 1.354,
      "step": 3410
    },
    {
      "epoch": 0.4365800588762319,
      "grad_norm": 0.7265625,
      "learning_rate": 0.001251403338496217,
      "loss": 1.5007,
      "step": 3411
    },
    {
      "epoch": 0.43670805068475615,
      "grad_norm": 0.6171875,
      "learning_rate": 0.001251002064294744,
      "loss": 1.9215,
      "step": 3412
    },
    {
      "epoch": 0.43683604249328045,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0012506007469544936,
      "loss": 2.1161,
      "step": 3413
    },
    {
      "epoch": 0.4369640343018047,
      "grad_norm": 0.734375,
      "learning_rate": 0.0012501993865444388,
      "loss": 1.6174,
      "step": 3414
    },
    {
      "epoch": 0.43709202611032894,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0012497979831335604,
      "loss": 0.9407,
      "step": 3415
    },
    {
      "epoch": 0.4372200179188532,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0012493965367908453,
      "loss": 2.3396,
      "step": 3416
    },
    {
      "epoch": 0.4373480097273775,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0012489950475852891,
      "loss": 1.4765,
      "step": 3417
    },
    {
      "epoch": 0.4374760015359017,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0012485935155858942,
      "loss": 1.8933,
      "step": 3418
    },
    {
      "epoch": 0.43760399334442596,
      "grad_norm": 0.671875,
      "learning_rate": 0.0012481919408616699,
      "loss": 1.6328,
      "step": 3419
    },
    {
      "epoch": 0.4377319851529502,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0012477903234816342,
      "loss": 2.0547,
      "step": 3420
    },
    {
      "epoch": 0.43785997696147444,
      "grad_norm": 0.9375,
      "learning_rate": 0.001247388663514811,
      "loss": 2.0868,
      "step": 3421
    },
    {
      "epoch": 0.43798796876999874,
      "grad_norm": 0.71484375,
      "learning_rate": 0.001246986961030232,
      "loss": 2.133,
      "step": 3422
    },
    {
      "epoch": 0.438115960578523,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0012465852160969365,
      "loss": 1.7411,
      "step": 3423
    },
    {
      "epoch": 0.4382439523870472,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0012461834287839707,
      "loss": 1.8637,
      "step": 3424
    },
    {
      "epoch": 0.43837194419557146,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0012457815991603885,
      "loss": 1.3558,
      "step": 3425
    },
    {
      "epoch": 0.43849993600409576,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0012453797272952505,
      "loss": 1.4696,
      "step": 3426
    },
    {
      "epoch": 0.43862792781262,
      "grad_norm": 0.578125,
      "learning_rate": 0.0012449778132576255,
      "loss": 1.3054,
      "step": 3427
    },
    {
      "epoch": 0.43875591962114424,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0012445758571165884,
      "loss": 1.2032,
      "step": 3428
    },
    {
      "epoch": 0.4388839114296685,
      "grad_norm": 0.85546875,
      "learning_rate": 0.001244173858941222,
      "loss": 1.6725,
      "step": 3429
    },
    {
      "epoch": 0.4390119032381928,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0012437718188006164,
      "loss": 1.4044,
      "step": 3430
    },
    {
      "epoch": 0.439139895046717,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0012433697367638687,
      "loss": 1.5909,
      "step": 3431
    },
    {
      "epoch": 0.43926788685524126,
      "grad_norm": 1.015625,
      "learning_rate": 0.001242967612900083,
      "loss": 2.3239,
      "step": 3432
    },
    {
      "epoch": 0.4393958786637655,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0012425654472783712,
      "loss": 0.9398,
      "step": 3433
    },
    {
      "epoch": 0.43952387047228975,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0012421632399678513,
      "loss": 1.5226,
      "step": 3434
    },
    {
      "epoch": 0.43965186228081404,
      "grad_norm": 0.828125,
      "learning_rate": 0.0012417609910376497,
      "loss": 1.9862,
      "step": 3435
    },
    {
      "epoch": 0.4397798540893383,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0012413587005568993,
      "loss": 0.8549,
      "step": 3436
    },
    {
      "epoch": 0.4399078458978625,
      "grad_norm": 0.59375,
      "learning_rate": 0.0012409563685947403,
      "loss": 1.5461,
      "step": 3437
    },
    {
      "epoch": 0.44003583770638677,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0012405539952203197,
      "loss": 2.0183,
      "step": 3438
    },
    {
      "epoch": 0.44016382951491106,
      "grad_norm": 0.71875,
      "learning_rate": 0.0012401515805027922,
      "loss": 1.8735,
      "step": 3439
    },
    {
      "epoch": 0.4402918213234353,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0012397491245113193,
      "loss": 1.5493,
      "step": 3440
    },
    {
      "epoch": 0.44041981313195955,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0012393466273150694,
      "loss": 1.789,
      "step": 3441
    },
    {
      "epoch": 0.4405478049404838,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0012389440889832182,
      "loss": 1.6762,
      "step": 3442
    },
    {
      "epoch": 0.4406757967490081,
      "grad_norm": 0.65625,
      "learning_rate": 0.0012385415095849486,
      "loss": 1.8088,
      "step": 3443
    },
    {
      "epoch": 0.44080378855753233,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0012381388891894505,
      "loss": 1.1798,
      "step": 3444
    },
    {
      "epoch": 0.44093178036605657,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00123773622786592,
      "loss": 1.8993,
      "step": 3445
    },
    {
      "epoch": 0.4410597721745808,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0012373335256835618,
      "loss": 1.4883,
      "step": 3446
    },
    {
      "epoch": 0.4411877639831051,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0012369307827115865,
      "loss": 1.2718,
      "step": 3447
    },
    {
      "epoch": 0.44131575579162935,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0012365279990192123,
      "loss": 2.244,
      "step": 3448
    },
    {
      "epoch": 0.4414437476001536,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0012361251746756637,
      "loss": 1.3261,
      "step": 3449
    },
    {
      "epoch": 0.44157173940867783,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0012357223097501726,
      "loss": 1.6683,
      "step": 3450
    },
    {
      "epoch": 0.4416997312172021,
      "grad_norm": 0.71875,
      "learning_rate": 0.0012353194043119781,
      "loss": 1.5258,
      "step": 3451
    },
    {
      "epoch": 0.44182772302572637,
      "grad_norm": 0.6484375,
      "learning_rate": 0.001234916458430326,
      "loss": 1.6429,
      "step": 3452
    },
    {
      "epoch": 0.4419557148342506,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0012345134721744692,
      "loss": 1.5034,
      "step": 3453
    },
    {
      "epoch": 0.44208370664277485,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0012341104456136671,
      "loss": 2.0101,
      "step": 3454
    },
    {
      "epoch": 0.4422116984512991,
      "grad_norm": 0.6484375,
      "learning_rate": 0.001233707378817187,
      "loss": 1.659,
      "step": 3455
    },
    {
      "epoch": 0.4423396902598234,
      "grad_norm": 0.71875,
      "learning_rate": 0.0012333042718543012,
      "loss": 2.1481,
      "step": 3456
    },
    {
      "epoch": 0.44246768206834763,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0012329011247942914,
      "loss": 1.4528,
      "step": 3457
    },
    {
      "epoch": 0.4425956738768719,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0012324979377064445,
      "loss": 1.6377,
      "step": 3458
    },
    {
      "epoch": 0.4427236656853961,
      "grad_norm": 0.6875,
      "learning_rate": 0.0012320947106600549,
      "loss": 1.9401,
      "step": 3459
    },
    {
      "epoch": 0.4428516574939204,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0012316914437244236,
      "loss": 1.4429,
      "step": 3460
    },
    {
      "epoch": 0.44297964930244466,
      "grad_norm": 0.78125,
      "learning_rate": 0.0012312881369688582,
      "loss": 1.6428,
      "step": 3461
    },
    {
      "epoch": 0.4431076411109689,
      "grad_norm": 0.609375,
      "learning_rate": 0.0012308847904626743,
      "loss": 1.2404,
      "step": 3462
    },
    {
      "epoch": 0.44323563291949314,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0012304814042751926,
      "loss": 1.4913,
      "step": 3463
    },
    {
      "epoch": 0.4433636247280174,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0012300779784757427,
      "loss": 1.3613,
      "step": 3464
    },
    {
      "epoch": 0.4434916165365417,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0012296745131336589,
      "loss": 1.1595,
      "step": 3465
    },
    {
      "epoch": 0.4436196083450659,
      "grad_norm": 0.78125,
      "learning_rate": 0.0012292710083182837,
      "loss": 1.3368,
      "step": 3466
    },
    {
      "epoch": 0.44374760015359016,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0012288674640989657,
      "loss": 1.6415,
      "step": 3467
    },
    {
      "epoch": 0.4438755919621144,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0012284638805450606,
      "loss": 1.0915,
      "step": 3468
    },
    {
      "epoch": 0.4440035837706387,
      "grad_norm": 0.703125,
      "learning_rate": 0.0012280602577259313,
      "loss": 1.8176,
      "step": 3469
    },
    {
      "epoch": 0.44413157557916294,
      "grad_norm": 0.578125,
      "learning_rate": 0.001227656595710946,
      "loss": 1.3079,
      "step": 3470
    },
    {
      "epoch": 0.4442595673876872,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0012272528945694815,
      "loss": 2.5854,
      "step": 3471
    },
    {
      "epoch": 0.4443875591962114,
      "grad_norm": 0.71875,
      "learning_rate": 0.0012268491543709194,
      "loss": 1.7278,
      "step": 3472
    },
    {
      "epoch": 0.4445155510047357,
      "grad_norm": 0.71875,
      "learning_rate": 0.0012264453751846503,
      "loss": 1.6548,
      "step": 3473
    },
    {
      "epoch": 0.44464354281325996,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0012260415570800687,
      "loss": 1.6534,
      "step": 3474
    },
    {
      "epoch": 0.4447715346217842,
      "grad_norm": 0.640625,
      "learning_rate": 0.0012256377001265784,
      "loss": 1.5329,
      "step": 3475
    },
    {
      "epoch": 0.44489952643030845,
      "grad_norm": 0.609375,
      "learning_rate": 0.0012252338043935885,
      "loss": 1.2831,
      "step": 3476
    },
    {
      "epoch": 0.44502751823883274,
      "grad_norm": 0.640625,
      "learning_rate": 0.001224829869950515,
      "loss": 1.3505,
      "step": 3477
    },
    {
      "epoch": 0.445155510047357,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0012244258968667802,
      "loss": 1.8848,
      "step": 3478
    },
    {
      "epoch": 0.4452835018558812,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0012240218852118138,
      "loss": 1.4611,
      "step": 3479
    },
    {
      "epoch": 0.44541149366440547,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0012236178350550518,
      "loss": 1.6044,
      "step": 3480
    },
    {
      "epoch": 0.4455394854729297,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0012232137464659367,
      "loss": 1.357,
      "step": 3481
    },
    {
      "epoch": 0.445667477281454,
      "grad_norm": 0.61328125,
      "learning_rate": 0.001222809619513918,
      "loss": 1.166,
      "step": 3482
    },
    {
      "epoch": 0.44579546908997825,
      "grad_norm": 0.640625,
      "learning_rate": 0.0012224054542684505,
      "loss": 1.6669,
      "step": 3483
    },
    {
      "epoch": 0.4459234608985025,
      "grad_norm": 0.59765625,
      "learning_rate": 0.001222001250798998,
      "loss": 1.0813,
      "step": 3484
    },
    {
      "epoch": 0.44605145270702673,
      "grad_norm": 0.65625,
      "learning_rate": 0.001221597009175028,
      "loss": 1.5182,
      "step": 3485
    },
    {
      "epoch": 0.446179444515551,
      "grad_norm": 0.6796875,
      "learning_rate": 0.001221192729466017,
      "loss": 1.0802,
      "step": 3486
    },
    {
      "epoch": 0.44630743632407527,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0012207884117414465,
      "loss": 1.0509,
      "step": 3487
    },
    {
      "epoch": 0.4464354281325995,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0012203840560708053,
      "loss": 1.4823,
      "step": 3488
    },
    {
      "epoch": 0.44656341994112375,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0012199796625235882,
      "loss": 1.7934,
      "step": 3489
    },
    {
      "epoch": 0.44669141174964805,
      "grad_norm": 0.66015625,
      "learning_rate": 0.001219575231169297,
      "loss": 1.4526,
      "step": 3490
    },
    {
      "epoch": 0.4468194035581723,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0012191707620774397,
      "loss": 1.5027,
      "step": 3491
    },
    {
      "epoch": 0.44694739536669653,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0012187662553175313,
      "loss": 1.7341,
      "step": 3492
    },
    {
      "epoch": 0.4470753871752208,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0012183617109590922,
      "loss": 1.7456,
      "step": 3493
    },
    {
      "epoch": 0.447203378983745,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0012179571290716503,
      "loss": 1.935,
      "step": 3494
    },
    {
      "epoch": 0.4473313707922693,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0012175525097247394,
      "loss": 1.4964,
      "step": 3495
    },
    {
      "epoch": 0.44745936260079355,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0012171478529879002,
      "loss": 1.9062,
      "step": 3496
    },
    {
      "epoch": 0.4475873544093178,
      "grad_norm": 0.77734375,
      "learning_rate": 0.001216743158930679,
      "loss": 1.205,
      "step": 3497
    },
    {
      "epoch": 0.44771534621784204,
      "grad_norm": 0.546875,
      "learning_rate": 0.0012163384276226293,
      "loss": 1.1237,
      "step": 3498
    },
    {
      "epoch": 0.44784333802636633,
      "grad_norm": 0.6875,
      "learning_rate": 0.001215933659133311,
      "loss": 1.6805,
      "step": 3499
    },
    {
      "epoch": 0.4479713298348906,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0012155288535322898,
      "loss": 1.7234,
      "step": 3500
    },
    {
      "epoch": 0.4480993216434148,
      "grad_norm": 0.55859375,
      "learning_rate": 0.001215124010889138,
      "loss": 1.0521,
      "step": 3501
    },
    {
      "epoch": 0.44822731345193906,
      "grad_norm": 0.56640625,
      "learning_rate": 0.001214719131273435,
      "loss": 1.2816,
      "step": 3502
    },
    {
      "epoch": 0.44835530526046335,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0012143142147547652,
      "loss": 1.6553,
      "step": 3503
    },
    {
      "epoch": 0.4484832970689876,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0012139092614027206,
      "loss": 2.029,
      "step": 3504
    },
    {
      "epoch": 0.44861128887751184,
      "grad_norm": 0.72265625,
      "learning_rate": 0.001213504271286899,
      "loss": 1.294,
      "step": 3505
    },
    {
      "epoch": 0.4487392806860361,
      "grad_norm": 0.6015625,
      "learning_rate": 0.001213099244476904,
      "loss": 1.4037,
      "step": 3506
    },
    {
      "epoch": 0.4488672724945603,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0012126941810423463,
      "loss": 1.3263,
      "step": 3507
    },
    {
      "epoch": 0.4489952643030846,
      "grad_norm": 0.703125,
      "learning_rate": 0.001212289081052843,
      "loss": 1.7489,
      "step": 3508
    },
    {
      "epoch": 0.44912325611160886,
      "grad_norm": 0.625,
      "learning_rate": 0.0012118839445780163,
      "loss": 1.7369,
      "step": 3509
    },
    {
      "epoch": 0.4492512479201331,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0012114787716874962,
      "loss": 1.3274,
      "step": 3510
    },
    {
      "epoch": 0.44937923972865734,
      "grad_norm": 0.59375,
      "learning_rate": 0.0012110735624509183,
      "loss": 1.7466,
      "step": 3511
    },
    {
      "epoch": 0.44950723153718164,
      "grad_norm": 0.75,
      "learning_rate": 0.0012106683169379239,
      "loss": 1.9582,
      "step": 3512
    },
    {
      "epoch": 0.4496352233457059,
      "grad_norm": 0.8125,
      "learning_rate": 0.001210263035218161,
      "loss": 2.1898,
      "step": 3513
    },
    {
      "epoch": 0.4497632151542301,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0012098577173612838,
      "loss": 1.1226,
      "step": 3514
    },
    {
      "epoch": 0.44989120696275436,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0012094523634369537,
      "loss": 1.4075,
      "step": 3515
    },
    {
      "epoch": 0.45001919877127866,
      "grad_norm": 0.55078125,
      "learning_rate": 0.001209046973514836,
      "loss": 1.0474,
      "step": 3516
    },
    {
      "epoch": 0.4501471905798029,
      "grad_norm": 0.5625,
      "learning_rate": 0.0012086415476646047,
      "loss": 1.2965,
      "step": 3517
    },
    {
      "epoch": 0.45027518238832714,
      "grad_norm": 0.640625,
      "learning_rate": 0.0012082360859559378,
      "loss": 1.8216,
      "step": 3518
    },
    {
      "epoch": 0.4504031741968514,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0012078305884585213,
      "loss": 1.8167,
      "step": 3519
    },
    {
      "epoch": 0.4505311660053757,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0012074250552420458,
      "loss": 1.4403,
      "step": 3520
    },
    {
      "epoch": 0.4506591578138999,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0012070194863762094,
      "loss": 1.5835,
      "step": 3521
    },
    {
      "epoch": 0.45078714962242417,
      "grad_norm": 0.90625,
      "learning_rate": 0.0012066138819307156,
      "loss": 1.6572,
      "step": 3522
    },
    {
      "epoch": 0.4509151414309484,
      "grad_norm": 0.703125,
      "learning_rate": 0.0012062082419752736,
      "loss": 2.1088,
      "step": 3523
    },
    {
      "epoch": 0.45104313323947265,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0012058025665795996,
      "loss": 1.8605,
      "step": 3524
    },
    {
      "epoch": 0.45117112504799695,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0012053968558134152,
      "loss": 1.293,
      "step": 3525
    },
    {
      "epoch": 0.4512991168565212,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0012049911097464492,
      "loss": 1.1194,
      "step": 3526
    },
    {
      "epoch": 0.45142710866504543,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0012045853284484345,
      "loss": 1.234,
      "step": 3527
    },
    {
      "epoch": 0.45155510047356967,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0012041795119891124,
      "loss": 0.9583,
      "step": 3528
    },
    {
      "epoch": 0.45168309228209397,
      "grad_norm": 0.71875,
      "learning_rate": 0.001203773660438228,
      "loss": 1.2495,
      "step": 3529
    },
    {
      "epoch": 0.4518110840906182,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0012033677738655336,
      "loss": 1.8333,
      "step": 3530
    },
    {
      "epoch": 0.45193907589914245,
      "grad_norm": 0.59765625,
      "learning_rate": 0.001202961852340788,
      "loss": 1.3328,
      "step": 3531
    },
    {
      "epoch": 0.4520670677076667,
      "grad_norm": 0.796875,
      "learning_rate": 0.0012025558959337548,
      "loss": 1.6723,
      "step": 3532
    },
    {
      "epoch": 0.452195059516191,
      "grad_norm": 0.59375,
      "learning_rate": 0.0012021499047142048,
      "loss": 1.6393,
      "step": 3533
    },
    {
      "epoch": 0.45232305132471523,
      "grad_norm": 0.703125,
      "learning_rate": 0.0012017438787519138,
      "loss": 1.009,
      "step": 3534
    },
    {
      "epoch": 0.45245104313323947,
      "grad_norm": 0.578125,
      "learning_rate": 0.0012013378181166638,
      "loss": 1.1971,
      "step": 3535
    },
    {
      "epoch": 0.4525790349417637,
      "grad_norm": 1.015625,
      "learning_rate": 0.0012009317228782433,
      "loss": 2.1447,
      "step": 3536
    },
    {
      "epoch": 0.45270702675028796,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0012005255931064461,
      "loss": 1.7181,
      "step": 3537
    },
    {
      "epoch": 0.45283501855881225,
      "grad_norm": 0.65625,
      "learning_rate": 0.0012001194288710723,
      "loss": 1.555,
      "step": 3538
    },
    {
      "epoch": 0.4529630103673365,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0011997132302419276,
      "loss": 1.7814,
      "step": 3539
    },
    {
      "epoch": 0.45309100217586074,
      "grad_norm": 0.76953125,
      "learning_rate": 0.001199306997288824,
      "loss": 1.8416,
      "step": 3540
    },
    {
      "epoch": 0.453218993984385,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0011989007300815793,
      "loss": 1.5658,
      "step": 3541
    },
    {
      "epoch": 0.4533469857929093,
      "grad_norm": 0.5078125,
      "learning_rate": 0.001198494428690017,
      "loss": 1.267,
      "step": 3542
    },
    {
      "epoch": 0.4534749776014335,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0011980880931839663,
      "loss": 1.3355,
      "step": 3543
    },
    {
      "epoch": 0.45360296940995776,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0011976817236332633,
      "loss": 2.3463,
      "step": 3544
    },
    {
      "epoch": 0.453730961218482,
      "grad_norm": 0.546875,
      "learning_rate": 0.0011972753201077487,
      "loss": 1.0164,
      "step": 3545
    },
    {
      "epoch": 0.4538589530270063,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0011968688826772694,
      "loss": 1.5796,
      "step": 3546
    },
    {
      "epoch": 0.45398694483553054,
      "grad_norm": 0.671875,
      "learning_rate": 0.0011964624114116785,
      "loss": 1.6686,
      "step": 3547
    },
    {
      "epoch": 0.4541149366440548,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0011960559063808347,
      "loss": 1.5594,
      "step": 3548
    },
    {
      "epoch": 0.454242928452579,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0011956493676546025,
      "loss": 1.0623,
      "step": 3549
    },
    {
      "epoch": 0.4543709202611033,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0011952427953028522,
      "loss": 1.8125,
      "step": 3550
    },
    {
      "epoch": 0.45449891206962756,
      "grad_norm": 0.65625,
      "learning_rate": 0.0011948361893954595,
      "loss": 1.9313,
      "step": 3551
    },
    {
      "epoch": 0.4546269038781518,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0011944295500023066,
      "loss": 1.2702,
      "step": 3552
    },
    {
      "epoch": 0.45475489568667604,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0011940228771932808,
      "loss": 1.1651,
      "step": 3553
    },
    {
      "epoch": 0.4548828874952003,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0011936161710382755,
      "loss": 1.4476,
      "step": 3554
    },
    {
      "epoch": 0.4550108793037246,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0011932094316071905,
      "loss": 1.622,
      "step": 3555
    },
    {
      "epoch": 0.4551388711122488,
      "grad_norm": 1.203125,
      "learning_rate": 0.0011928026589699292,
      "loss": 2.0715,
      "step": 3556
    },
    {
      "epoch": 0.45526686292077306,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0011923958531964032,
      "loss": 1.5558,
      "step": 3557
    },
    {
      "epoch": 0.4553948547292973,
      "grad_norm": 0.796875,
      "learning_rate": 0.0011919890143565278,
      "loss": 1.6278,
      "step": 3558
    },
    {
      "epoch": 0.4555228465378216,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0011915821425202257,
      "loss": 1.6173,
      "step": 3559
    },
    {
      "epoch": 0.45565083834634584,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0011911752377574238,
      "loss": 1.3943,
      "step": 3560
    },
    {
      "epoch": 0.4557788301548701,
      "grad_norm": 0.6328125,
      "learning_rate": 0.001190768300138056,
      "loss": 1.3418,
      "step": 3561
    },
    {
      "epoch": 0.4559068219633943,
      "grad_norm": 0.703125,
      "learning_rate": 0.00119036132973206,
      "loss": 1.6023,
      "step": 3562
    },
    {
      "epoch": 0.4560348137719186,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0011899543266093813,
      "loss": 0.8315,
      "step": 3563
    },
    {
      "epoch": 0.45616280558044286,
      "grad_norm": 0.671875,
      "learning_rate": 0.0011895472908399693,
      "loss": 1.3758,
      "step": 3564
    },
    {
      "epoch": 0.4562907973889671,
      "grad_norm": 0.8125,
      "learning_rate": 0.0011891402224937803,
      "loss": 1.831,
      "step": 3565
    },
    {
      "epoch": 0.45641878919749135,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0011887331216407756,
      "loss": 1.137,
      "step": 3566
    },
    {
      "epoch": 0.4565467810060156,
      "grad_norm": 0.578125,
      "learning_rate": 0.0011883259883509211,
      "loss": 1.1637,
      "step": 3567
    },
    {
      "epoch": 0.4566747728145399,
      "grad_norm": 0.671875,
      "learning_rate": 0.0011879188226941906,
      "loss": 1.4455,
      "step": 3568
    },
    {
      "epoch": 0.45680276462306413,
      "grad_norm": 0.6875,
      "learning_rate": 0.0011875116247405608,
      "loss": 1.8221,
      "step": 3569
    },
    {
      "epoch": 0.45693075643158837,
      "grad_norm": 0.640625,
      "learning_rate": 0.0011871043945600166,
      "loss": 1.6705,
      "step": 3570
    },
    {
      "epoch": 0.4570587482401126,
      "grad_norm": 0.66015625,
      "learning_rate": 0.001186697132222546,
      "loss": 1.1495,
      "step": 3571
    },
    {
      "epoch": 0.4571867400486369,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0011862898377981445,
      "loss": 1.5596,
      "step": 3572
    },
    {
      "epoch": 0.45731473185716115,
      "grad_norm": 0.671875,
      "learning_rate": 0.0011858825113568118,
      "loss": 1.5285,
      "step": 3573
    },
    {
      "epoch": 0.4574427236656854,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0011854751529685533,
      "loss": 1.53,
      "step": 3574
    },
    {
      "epoch": 0.45757071547420963,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0011850677627033806,
      "loss": 1.114,
      "step": 3575
    },
    {
      "epoch": 0.45769870728273393,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0011846603406313104,
      "loss": 1.3046,
      "step": 3576
    },
    {
      "epoch": 0.45782669909125817,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0011842528868223645,
      "loss": 1.5211,
      "step": 3577
    },
    {
      "epoch": 0.4579546908997824,
      "grad_norm": 0.859375,
      "learning_rate": 0.0011838454013465706,
      "loss": 2.0366,
      "step": 3578
    },
    {
      "epoch": 0.45808268270830665,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0011834378842739614,
      "loss": 0.8788,
      "step": 3579
    },
    {
      "epoch": 0.4582106745168309,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0011830303356745754,
      "loss": 1.7071,
      "step": 3580
    },
    {
      "epoch": 0.4583386663253552,
      "grad_norm": 0.71875,
      "learning_rate": 0.0011826227556184573,
      "loss": 2.0921,
      "step": 3581
    },
    {
      "epoch": 0.45846665813387943,
      "grad_norm": 0.703125,
      "learning_rate": 0.0011822151441756553,
      "loss": 1.5676,
      "step": 3582
    },
    {
      "epoch": 0.4585946499424037,
      "grad_norm": 0.67578125,
      "learning_rate": 0.001181807501416224,
      "loss": 2.1792,
      "step": 3583
    },
    {
      "epoch": 0.4587226417509279,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0011813998274102242,
      "loss": 1.1661,
      "step": 3584
    },
    {
      "epoch": 0.4588506335594522,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0011809921222277208,
      "loss": 2.1099,
      "step": 3585
    },
    {
      "epoch": 0.45897862536797646,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0011805843859387846,
      "loss": 1.3516,
      "step": 3586
    },
    {
      "epoch": 0.4591066171765007,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0011801766186134916,
      "loss": 2.0213,
      "step": 3587
    },
    {
      "epoch": 0.45923460898502494,
      "grad_norm": 0.5546875,
      "learning_rate": 0.001179768820321924,
      "loss": 1.3172,
      "step": 3588
    },
    {
      "epoch": 0.45936260079354924,
      "grad_norm": 0.65625,
      "learning_rate": 0.001179360991134167,
      "loss": 1.6249,
      "step": 3589
    },
    {
      "epoch": 0.4594905926020735,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0011789531311203146,
      "loss": 1.9561,
      "step": 3590
    },
    {
      "epoch": 0.4596185844105977,
      "grad_norm": 0.65625,
      "learning_rate": 0.0011785452403504624,
      "loss": 1.4122,
      "step": 3591
    },
    {
      "epoch": 0.45974657621912196,
      "grad_norm": 0.64453125,
      "learning_rate": 0.001178137318894714,
      "loss": 1.6869,
      "step": 3592
    },
    {
      "epoch": 0.45987456802764626,
      "grad_norm": 0.76171875,
      "learning_rate": 0.001177729366823177,
      "loss": 1.6937,
      "step": 3593
    },
    {
      "epoch": 0.4600025598361705,
      "grad_norm": 0.609375,
      "learning_rate": 0.0011773213842059646,
      "loss": 1.5811,
      "step": 3594
    },
    {
      "epoch": 0.46013055164469474,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0011769133711131959,
      "loss": 1.527,
      "step": 3595
    },
    {
      "epoch": 0.460258543453219,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0011765053276149935,
      "loss": 1.4385,
      "step": 3596
    },
    {
      "epoch": 0.4603865352617432,
      "grad_norm": 0.6328125,
      "learning_rate": 0.001176097253781487,
      "loss": 1.2286,
      "step": 3597
    },
    {
      "epoch": 0.4605145270702675,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00117568914968281,
      "loss": 1.2018,
      "step": 3598
    },
    {
      "epoch": 0.46064251887879176,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0011752810153891026,
      "loss": 1.8283,
      "step": 3599
    },
    {
      "epoch": 0.460770510687316,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0011748728509705082,
      "loss": 1.2967,
      "step": 3600
    },
    {
      "epoch": 0.46089850249584025,
      "grad_norm": 1.078125,
      "learning_rate": 0.0011744646564971776,
      "loss": 2.1083,
      "step": 3601
    },
    {
      "epoch": 0.46102649430436454,
      "grad_norm": 0.75390625,
      "learning_rate": 0.001174056432039265,
      "loss": 1.7734,
      "step": 3602
    },
    {
      "epoch": 0.4611544861128888,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0011736481776669307,
      "loss": 1.3838,
      "step": 3603
    },
    {
      "epoch": 0.461282477921413,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0011732398934503393,
      "loss": 1.6607,
      "step": 3604
    },
    {
      "epoch": 0.46141046972993727,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0011728315794596613,
      "loss": 1.2788,
      "step": 3605
    },
    {
      "epoch": 0.46153846153846156,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0011724232357650727,
      "loss": 1.7325,
      "step": 3606
    },
    {
      "epoch": 0.4616664533469858,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0011720148624367532,
      "loss": 1.4086,
      "step": 3607
    },
    {
      "epoch": 0.46179444515551005,
      "grad_norm": 0.6796875,
      "learning_rate": 0.001171606459544889,
      "loss": 1.1551,
      "step": 3608
    },
    {
      "epoch": 0.4619224369640343,
      "grad_norm": 0.578125,
      "learning_rate": 0.0011711980271596702,
      "loss": 1.4289,
      "step": 3609
    },
    {
      "epoch": 0.46205042877255853,
      "grad_norm": 1.0859375,
      "learning_rate": 0.001170789565351293,
      "loss": 1.2418,
      "step": 3610
    },
    {
      "epoch": 0.4621784205810828,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0011703810741899576,
      "loss": 1.426,
      "step": 3611
    },
    {
      "epoch": 0.46230641238960707,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0011699725537458708,
      "loss": 1.5447,
      "step": 3612
    },
    {
      "epoch": 0.4624344041981313,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0011695640040892423,
      "loss": 2.0317,
      "step": 3613
    },
    {
      "epoch": 0.46256239600665555,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0011691554252902887,
      "loss": 1.3231,
      "step": 3614
    },
    {
      "epoch": 0.46269038781517985,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0011687468174192308,
      "loss": 1.7783,
      "step": 3615
    },
    {
      "epoch": 0.4628183796237041,
      "grad_norm": 0.640625,
      "learning_rate": 0.0011683381805462944,
      "loss": 1.8521,
      "step": 3616
    },
    {
      "epoch": 0.46294637143222833,
      "grad_norm": 0.72265625,
      "learning_rate": 0.001167929514741711,
      "loss": 1.8957,
      "step": 3617
    },
    {
      "epoch": 0.4630743632407526,
      "grad_norm": 0.9296875,
      "learning_rate": 0.001167520820075715,
      "loss": 2.1175,
      "step": 3618
    },
    {
      "epoch": 0.46320235504927687,
      "grad_norm": 0.71875,
      "learning_rate": 0.0011671120966185485,
      "loss": 1.5379,
      "step": 3619
    },
    {
      "epoch": 0.4633303468578011,
      "grad_norm": 0.546875,
      "learning_rate": 0.0011667033444404567,
      "loss": 1.349,
      "step": 3620
    },
    {
      "epoch": 0.46345833866632535,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0011662945636116907,
      "loss": 1.2259,
      "step": 3621
    },
    {
      "epoch": 0.4635863304748496,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0011658857542025055,
      "loss": 1.4475,
      "step": 3622
    },
    {
      "epoch": 0.46371432228337384,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0011654769162831626,
      "loss": 1.4793,
      "step": 3623
    },
    {
      "epoch": 0.46384231409189813,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0011650680499239263,
      "loss": 1.5742,
      "step": 3624
    },
    {
      "epoch": 0.4639703059004224,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0011646591551950677,
      "loss": 1.482,
      "step": 3625
    },
    {
      "epoch": 0.4640982977089466,
      "grad_norm": 0.75,
      "learning_rate": 0.0011642502321668614,
      "loss": 1.889,
      "step": 3626
    },
    {
      "epoch": 0.46422628951747086,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0011638412809095878,
      "loss": 0.8742,
      "step": 3627
    },
    {
      "epoch": 0.46435428132599516,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0011634323014935322,
      "loss": 2.0855,
      "step": 3628
    },
    {
      "epoch": 0.4644822731345194,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0011630232939889836,
      "loss": 1.8454,
      "step": 3629
    },
    {
      "epoch": 0.46461026494304364,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0011626142584662369,
      "loss": 2.4242,
      "step": 3630
    },
    {
      "epoch": 0.4647382567515679,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0011622051949955914,
      "loss": 1.8019,
      "step": 3631
    },
    {
      "epoch": 0.4648662485600922,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0011617961036473519,
      "loss": 0.9078,
      "step": 3632
    },
    {
      "epoch": 0.4649942403686164,
      "grad_norm": 0.796875,
      "learning_rate": 0.0011613869844918265,
      "loss": 1.0928,
      "step": 3633
    },
    {
      "epoch": 0.46512223217714066,
      "grad_norm": 0.6875,
      "learning_rate": 0.0011609778375993296,
      "loss": 1.7201,
      "step": 3634
    },
    {
      "epoch": 0.4652502239856649,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0011605686630401792,
      "loss": 1.729,
      "step": 3635
    },
    {
      "epoch": 0.4653782157941892,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0011601594608846991,
      "loss": 1.2742,
      "step": 3636
    },
    {
      "epoch": 0.46550620760271344,
      "grad_norm": 0.87109375,
      "learning_rate": 0.001159750231203217,
      "loss": 2.0697,
      "step": 3637
    },
    {
      "epoch": 0.4656341994112377,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0011593409740660655,
      "loss": 1.6996,
      "step": 3638
    },
    {
      "epoch": 0.4657621912197619,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0011589316895435826,
      "loss": 1.1124,
      "step": 3639
    },
    {
      "epoch": 0.46589018302828616,
      "grad_norm": 0.640625,
      "learning_rate": 0.0011585223777061102,
      "loss": 1.3336,
      "step": 3640
    },
    {
      "epoch": 0.46601817483681046,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0011581130386239952,
      "loss": 1.9892,
      "step": 3641
    },
    {
      "epoch": 0.4661461666453347,
      "grad_norm": 0.765625,
      "learning_rate": 0.001157703672367589,
      "loss": 1.4224,
      "step": 3642
    },
    {
      "epoch": 0.46627415845385894,
      "grad_norm": 0.640625,
      "learning_rate": 0.0011572942790072484,
      "loss": 1.6858,
      "step": 3643
    },
    {
      "epoch": 0.4664021502623832,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0011568848586133336,
      "loss": 1.6775,
      "step": 3644
    },
    {
      "epoch": 0.4665301420709075,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0011564754112562106,
      "loss": 1.1986,
      "step": 3645
    },
    {
      "epoch": 0.4666581338794317,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0011560659370062492,
      "loss": 1.5059,
      "step": 3646
    },
    {
      "epoch": 0.46678612568795597,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0011556564359338243,
      "loss": 1.4367,
      "step": 3647
    },
    {
      "epoch": 0.4669141174964802,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0011552469081093158,
      "loss": 1.1936,
      "step": 3648
    },
    {
      "epoch": 0.4670421093050045,
      "grad_norm": 0.78515625,
      "learning_rate": 0.001154837353603107,
      "loss": 2.6349,
      "step": 3649
    },
    {
      "epoch": 0.46717010111352875,
      "grad_norm": 0.7265625,
      "learning_rate": 0.001154427772485587,
      "loss": 1.6054,
      "step": 3650
    },
    {
      "epoch": 0.467298092922053,
      "grad_norm": 0.6875,
      "learning_rate": 0.0011540181648271486,
      "loss": 1.7125,
      "step": 3651
    },
    {
      "epoch": 0.46742608473057723,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0011536085306981898,
      "loss": 1.881,
      "step": 3652
    },
    {
      "epoch": 0.46755407653910147,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0011531988701691124,
      "loss": 1.3769,
      "step": 3653
    },
    {
      "epoch": 0.46768206834762577,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0011527891833103237,
      "loss": 1.8845,
      "step": 3654
    },
    {
      "epoch": 0.46781006015615,
      "grad_norm": 0.65625,
      "learning_rate": 0.001152379470192235,
      "loss": 1.2916,
      "step": 3655
    },
    {
      "epoch": 0.46793805196467425,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0011519697308852621,
      "loss": 1.136,
      "step": 3656
    },
    {
      "epoch": 0.4680660437731985,
      "grad_norm": 0.6640625,
      "learning_rate": 0.001151559965459825,
      "loss": 1.356,
      "step": 3657
    },
    {
      "epoch": 0.4681940355817228,
      "grad_norm": 0.61328125,
      "learning_rate": 0.001151150173986349,
      "loss": 1.5294,
      "step": 3658
    },
    {
      "epoch": 0.46832202739024703,
      "grad_norm": 0.6640625,
      "learning_rate": 0.001150740356535263,
      "loss": 1.6881,
      "step": 3659
    },
    {
      "epoch": 0.4684500191987713,
      "grad_norm": 0.609375,
      "learning_rate": 0.0011503305131770013,
      "loss": 1.1755,
      "step": 3660
    },
    {
      "epoch": 0.4685780110072955,
      "grad_norm": 0.578125,
      "learning_rate": 0.0011499206439820019,
      "loss": 1.4778,
      "step": 3661
    },
    {
      "epoch": 0.4687060028158198,
      "grad_norm": 0.7890625,
      "learning_rate": 0.001149510749020707,
      "loss": 2.0204,
      "step": 3662
    },
    {
      "epoch": 0.46883399462434405,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0011491008283635647,
      "loss": 1.7821,
      "step": 3663
    },
    {
      "epoch": 0.4689619864328683,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0011486908820810251,
      "loss": 1.3973,
      "step": 3664
    },
    {
      "epoch": 0.46908997824139254,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0011482809102435457,
      "loss": 1.6127,
      "step": 3665
    },
    {
      "epoch": 0.46921797004991683,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0011478709129215856,
      "loss": 1.2009,
      "step": 3666
    },
    {
      "epoch": 0.4693459618584411,
      "grad_norm": 0.609375,
      "learning_rate": 0.0011474608901856102,
      "loss": 1.3941,
      "step": 3667
    },
    {
      "epoch": 0.4694739536669653,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0011470508421060881,
      "loss": 1.9559,
      "step": 3668
    },
    {
      "epoch": 0.46960194547548956,
      "grad_norm": 0.72265625,
      "learning_rate": 0.001146640768753493,
      "loss": 1.7697,
      "step": 3669
    },
    {
      "epoch": 0.4697299372840138,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0011462306701983025,
      "loss": 1.5683,
      "step": 3670
    },
    {
      "epoch": 0.4698579290925381,
      "grad_norm": 0.703125,
      "learning_rate": 0.0011458205465109989,
      "loss": 1.279,
      "step": 3671
    },
    {
      "epoch": 0.46998592090106234,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0011454103977620684,
      "loss": 1.5225,
      "step": 3672
    },
    {
      "epoch": 0.4701139127095866,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0011450002240220019,
      "loss": 1.936,
      "step": 3673
    },
    {
      "epoch": 0.4702419045181108,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0011445900253612945,
      "loss": 1.0406,
      "step": 3674
    },
    {
      "epoch": 0.4703698963266351,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0011441798018504445,
      "loss": 1.479,
      "step": 3675
    },
    {
      "epoch": 0.47049788813515936,
      "grad_norm": 0.6015625,
      "learning_rate": 0.001143769553559957,
      "loss": 1.1851,
      "step": 3676
    },
    {
      "epoch": 0.4706258799436836,
      "grad_norm": 0.75,
      "learning_rate": 0.001143359280560339,
      "loss": 1.4614,
      "step": 3677
    },
    {
      "epoch": 0.47075387175220784,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0011429489829221024,
      "loss": 1.4948,
      "step": 3678
    },
    {
      "epoch": 0.47088186356073214,
      "grad_norm": 0.671875,
      "learning_rate": 0.0011425386607157642,
      "loss": 1.5806,
      "step": 3679
    },
    {
      "epoch": 0.4710098553692564,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0011421283140118442,
      "loss": 1.2554,
      "step": 3680
    },
    {
      "epoch": 0.4711378471777806,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0011417179428808676,
      "loss": 1.5608,
      "step": 3681
    },
    {
      "epoch": 0.47126583898630486,
      "grad_norm": 0.65625,
      "learning_rate": 0.0011413075473933627,
      "loss": 1.6617,
      "step": 3682
    },
    {
      "epoch": 0.4713938307948291,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0011408971276198636,
      "loss": 1.4873,
      "step": 3683
    },
    {
      "epoch": 0.4715218226033534,
      "grad_norm": 0.609375,
      "learning_rate": 0.0011404866836309071,
      "loss": 1.064,
      "step": 3684
    },
    {
      "epoch": 0.47164981441187764,
      "grad_norm": 0.671875,
      "learning_rate": 0.0011400762154970346,
      "loss": 1.3172,
      "step": 3685
    },
    {
      "epoch": 0.4717778062204019,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0011396657232887917,
      "loss": 1.3606,
      "step": 3686
    },
    {
      "epoch": 0.4719057980289261,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0011392552070767282,
      "loss": 1.1649,
      "step": 3687
    },
    {
      "epoch": 0.4720337898374504,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0011388446669313979,
      "loss": 1.631,
      "step": 3688
    },
    {
      "epoch": 0.47216178164597467,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0011384341029233587,
      "loss": 3.2786,
      "step": 3689
    },
    {
      "epoch": 0.4722897734544989,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0011380235151231732,
      "loss": 1.8516,
      "step": 3690
    },
    {
      "epoch": 0.47241776526302315,
      "grad_norm": 0.75,
      "learning_rate": 0.001137612903601407,
      "loss": 2.0856,
      "step": 3691
    },
    {
      "epoch": 0.47254575707154745,
      "grad_norm": 0.703125,
      "learning_rate": 0.0011372022684286308,
      "loss": 1.8496,
      "step": 3692
    },
    {
      "epoch": 0.4726737488800717,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0011367916096754185,
      "loss": 1.7301,
      "step": 3693
    },
    {
      "epoch": 0.47280174068859593,
      "grad_norm": 0.71875,
      "learning_rate": 0.001136380927412349,
      "loss": 1.4864,
      "step": 3694
    },
    {
      "epoch": 0.47292973249712017,
      "grad_norm": 0.640625,
      "learning_rate": 0.0011359702217100041,
      "loss": 1.5395,
      "step": 3695
    },
    {
      "epoch": 0.4730577243056444,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0011355594926389708,
      "loss": 1.2392,
      "step": 3696
    },
    {
      "epoch": 0.4731857161141687,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0011351487402698388,
      "loss": 1.7836,
      "step": 3697
    },
    {
      "epoch": 0.47331370792269295,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0011347379646732033,
      "loss": 1.3885,
      "step": 3698
    },
    {
      "epoch": 0.4734416997312172,
      "grad_norm": 0.734375,
      "learning_rate": 0.0011343271659196624,
      "loss": 2.0586,
      "step": 3699
    },
    {
      "epoch": 0.47356969153974143,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0011339163440798187,
      "loss": 1.3707,
      "step": 3700
    },
    {
      "epoch": 0.47369768334826573,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0011335054992242784,
      "loss": 1.274,
      "step": 3701
    },
    {
      "epoch": 0.47382567515678997,
      "grad_norm": 0.70703125,
      "learning_rate": 0.001133094631423652,
      "loss": 1.6793,
      "step": 3702
    },
    {
      "epoch": 0.4739536669653142,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0011326837407485536,
      "loss": 2.303,
      "step": 3703
    },
    {
      "epoch": 0.47408165877383845,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0011322728272696017,
      "loss": 2.1251,
      "step": 3704
    },
    {
      "epoch": 0.47420965058236275,
      "grad_norm": 0.609375,
      "learning_rate": 0.0011318618910574185,
      "loss": 1.1609,
      "step": 3705
    },
    {
      "epoch": 0.474337642390887,
      "grad_norm": 0.65625,
      "learning_rate": 0.00113145093218263,
      "loss": 1.651,
      "step": 3706
    },
    {
      "epoch": 0.47446563419941123,
      "grad_norm": 0.6484375,
      "learning_rate": 0.001131039950715866,
      "loss": 1.5094,
      "step": 3707
    },
    {
      "epoch": 0.4745936260079355,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00113062894672776,
      "loss": 1.2045,
      "step": 3708
    },
    {
      "epoch": 0.4747216178164598,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0011302179202889505,
      "loss": 1.9101,
      "step": 3709
    },
    {
      "epoch": 0.474849609624984,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0011298068714700783,
      "loss": 1.7288,
      "step": 3710
    },
    {
      "epoch": 0.47497760143350826,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0011293958003417892,
      "loss": 1.7245,
      "step": 3711
    },
    {
      "epoch": 0.4751055932420325,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0011289847069747326,
      "loss": 1.8536,
      "step": 3712
    },
    {
      "epoch": 0.47523358505055674,
      "grad_norm": 0.71484375,
      "learning_rate": 0.001128573591439561,
      "loss": 1.5539,
      "step": 3713
    },
    {
      "epoch": 0.47536157685908104,
      "grad_norm": 0.625,
      "learning_rate": 0.0011281624538069322,
      "loss": 1.8019,
      "step": 3714
    },
    {
      "epoch": 0.4754895686676053,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0011277512941475058,
      "loss": 1.6224,
      "step": 3715
    },
    {
      "epoch": 0.4756175604761295,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0011273401125319472,
      "loss": 1.8874,
      "step": 3716
    },
    {
      "epoch": 0.47574555228465376,
      "grad_norm": 1.078125,
      "learning_rate": 0.0011269289090309238,
      "loss": 2.1388,
      "step": 3717
    },
    {
      "epoch": 0.47587354409317806,
      "grad_norm": 1.046875,
      "learning_rate": 0.0011265176837151082,
      "loss": 1.3203,
      "step": 3718
    },
    {
      "epoch": 0.4760015359017023,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0011261064366551762,
      "loss": 1.6939,
      "step": 3719
    },
    {
      "epoch": 0.47612952771022654,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0011256951679218064,
      "loss": 1.4087,
      "step": 3720
    },
    {
      "epoch": 0.4762575195187508,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0011252838775856829,
      "loss": 1.8989,
      "step": 3721
    },
    {
      "epoch": 0.4763855113272751,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0011248725657174922,
      "loss": 1.1949,
      "step": 3722
    },
    {
      "epoch": 0.4765135031357993,
      "grad_norm": 0.625,
      "learning_rate": 0.0011244612323879255,
      "loss": 1.3013,
      "step": 3723
    },
    {
      "epoch": 0.47664149494432356,
      "grad_norm": 0.609375,
      "learning_rate": 0.0011240498776676764,
      "loss": 1.7242,
      "step": 3724
    },
    {
      "epoch": 0.4767694867528478,
      "grad_norm": 0.56640625,
      "learning_rate": 0.001123638501627443,
      "loss": 1.3511,
      "step": 3725
    },
    {
      "epoch": 0.47689747856137205,
      "grad_norm": 1.046875,
      "learning_rate": 0.0011232271043379273,
      "loss": 1.1918,
      "step": 3726
    },
    {
      "epoch": 0.47702547036989634,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0011228156858698343,
      "loss": 2.1072,
      "step": 3727
    },
    {
      "epoch": 0.4771534621784206,
      "grad_norm": 0.703125,
      "learning_rate": 0.0011224042462938731,
      "loss": 1.4676,
      "step": 3728
    },
    {
      "epoch": 0.4772814539869448,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0011219927856807562,
      "loss": 2.0668,
      "step": 3729
    },
    {
      "epoch": 0.47740944579546907,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0011215813041011997,
      "loss": 1.2234,
      "step": 3730
    },
    {
      "epoch": 0.47753743760399336,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0011211698016259234,
      "loss": 1.5836,
      "step": 3731
    },
    {
      "epoch": 0.4776654294125176,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0011207582783256506,
      "loss": 1.3937,
      "step": 3732
    },
    {
      "epoch": 0.47779342122104185,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0011203467342711087,
      "loss": 1.961,
      "step": 3733
    },
    {
      "epoch": 0.4779214130295661,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0011199351695330279,
      "loss": 1.5636,
      "step": 3734
    },
    {
      "epoch": 0.4780494048380904,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0011195235841821422,
      "loss": 1.6219,
      "step": 3735
    },
    {
      "epoch": 0.4781773966466146,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0011191119782891891,
      "loss": 1.0168,
      "step": 3736
    },
    {
      "epoch": 0.47830538845513887,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0011187003519249105,
      "loss": 1.4966,
      "step": 3737
    },
    {
      "epoch": 0.4784333802636631,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0011182887051600502,
      "loss": 1.5385,
      "step": 3738
    },
    {
      "epoch": 0.4785613720721874,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0011178770380653566,
      "loss": 1.4995,
      "step": 3739
    },
    {
      "epoch": 0.47868936388071165,
      "grad_norm": 0.671875,
      "learning_rate": 0.001117465350711582,
      "loss": 1.6609,
      "step": 3740
    },
    {
      "epoch": 0.4788173556892359,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0011170536431694804,
      "loss": 1.4636,
      "step": 3741
    },
    {
      "epoch": 0.47894534749776013,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0011166419155098112,
      "loss": 1.2279,
      "step": 3742
    },
    {
      "epoch": 0.4790733393062844,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0011162301678033367,
      "loss": 1.4081,
      "step": 3743
    },
    {
      "epoch": 0.47920133111480867,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0011158184001208216,
      "loss": 2.1279,
      "step": 3744
    },
    {
      "epoch": 0.4793293229233329,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0011154066125330358,
      "loss": 1.5094,
      "step": 3745
    },
    {
      "epoch": 0.47945731473185715,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0011149948051107509,
      "loss": 1.4678,
      "step": 3746
    },
    {
      "epoch": 0.4795853065403814,
      "grad_norm": 0.9296875,
      "learning_rate": 0.001114582977924743,
      "loss": 2.1588,
      "step": 3747
    },
    {
      "epoch": 0.4797132983489057,
      "grad_norm": 0.84765625,
      "learning_rate": 0.001114171131045791,
      "loss": 1.4215,
      "step": 3748
    },
    {
      "epoch": 0.47984129015742993,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0011137592645446782,
      "loss": 2.2296,
      "step": 3749
    },
    {
      "epoch": 0.4799692819659542,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0011133473784921896,
      "loss": 1.5528,
      "step": 3750
    },
    {
      "epoch": 0.4800972737744784,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0011129354729591157,
      "loss": 1.7146,
      "step": 3751
    },
    {
      "epoch": 0.4802252655830027,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0011125235480162479,
      "loss": 1.2147,
      "step": 3752
    },
    {
      "epoch": 0.48035325739152696,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0011121116037343826,
      "loss": 1.4508,
      "step": 3753
    },
    {
      "epoch": 0.4804812492000512,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0011116996401843192,
      "loss": 1.6803,
      "step": 3754
    },
    {
      "epoch": 0.48060924100857544,
      "grad_norm": 0.578125,
      "learning_rate": 0.0011112876574368605,
      "loss": 1.3909,
      "step": 3755
    },
    {
      "epoch": 0.4807372328170997,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0011108756555628125,
      "loss": 1.5239,
      "step": 3756
    },
    {
      "epoch": 0.480865224625624,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0011104636346329837,
      "loss": 1.091,
      "step": 3757
    },
    {
      "epoch": 0.4809932164341482,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0011100515947181876,
      "loss": 2.1914,
      "step": 3758
    },
    {
      "epoch": 0.48112120824267246,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0011096395358892387,
      "loss": 1.6549,
      "step": 3759
    },
    {
      "epoch": 0.4812492000511967,
      "grad_norm": 0.796875,
      "learning_rate": 0.0011092274582169572,
      "loss": 2.5331,
      "step": 3760
    },
    {
      "epoch": 0.481377191859721,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0011088153617721645,
      "loss": 1.6781,
      "step": 3761
    },
    {
      "epoch": 0.48150518366824524,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0011084032466256867,
      "loss": 2.3348,
      "step": 3762
    },
    {
      "epoch": 0.4816331754767695,
      "grad_norm": 0.6640625,
      "learning_rate": 0.001107991112848352,
      "loss": 1.4134,
      "step": 3763
    },
    {
      "epoch": 0.4817611672852937,
      "grad_norm": 0.640625,
      "learning_rate": 0.0011075789605109925,
      "loss": 1.6416,
      "step": 3764
    },
    {
      "epoch": 0.481889159093818,
      "grad_norm": 0.69140625,
      "learning_rate": 0.001107166789684443,
      "loss": 1.6668,
      "step": 3765
    },
    {
      "epoch": 0.48201715090234226,
      "grad_norm": 0.55859375,
      "learning_rate": 0.001106754600439542,
      "loss": 1.4643,
      "step": 3766
    },
    {
      "epoch": 0.4821451427108665,
      "grad_norm": 0.85546875,
      "learning_rate": 0.001106342392847131,
      "loss": 1.3778,
      "step": 3767
    },
    {
      "epoch": 0.48227313451939074,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0011059301669780542,
      "loss": 0.9742,
      "step": 3768
    },
    {
      "epoch": 0.482401126327915,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0011055179229031598,
      "loss": 1.5225,
      "step": 3769
    },
    {
      "epoch": 0.4825291181364393,
      "grad_norm": 0.66015625,
      "learning_rate": 0.001105105660693298,
      "loss": 2.0289,
      "step": 3770
    },
    {
      "epoch": 0.4826571099449635,
      "grad_norm": 0.625,
      "learning_rate": 0.0011046933804193233,
      "loss": 2.0587,
      "step": 3771
    },
    {
      "epoch": 0.48278510175348777,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0011042810821520923,
      "loss": 1.5032,
      "step": 3772
    },
    {
      "epoch": 0.482913093562012,
      "grad_norm": 0.671875,
      "learning_rate": 0.0011038687659624658,
      "loss": 1.4424,
      "step": 3773
    },
    {
      "epoch": 0.4830410853705363,
      "grad_norm": 0.671875,
      "learning_rate": 0.0011034564319213065,
      "loss": 1.6551,
      "step": 3774
    },
    {
      "epoch": 0.48316907717906055,
      "grad_norm": 0.59375,
      "learning_rate": 0.0011030440800994805,
      "loss": 1.1529,
      "step": 3775
    },
    {
      "epoch": 0.4832970689875848,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0011026317105678578,
      "loss": 1.2561,
      "step": 3776
    },
    {
      "epoch": 0.48342506079610903,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0011022193233973104,
      "loss": 1.5169,
      "step": 3777
    },
    {
      "epoch": 0.4835530526046333,
      "grad_norm": 0.65625,
      "learning_rate": 0.0011018069186587139,
      "loss": 1.5597,
      "step": 3778
    },
    {
      "epoch": 0.48368104441315757,
      "grad_norm": 0.5625,
      "learning_rate": 0.0011013944964229463,
      "loss": 0.9717,
      "step": 3779
    },
    {
      "epoch": 0.4838090362216818,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00110098205676089,
      "loss": 1.6698,
      "step": 3780
    },
    {
      "epoch": 0.48393702803020605,
      "grad_norm": 0.77734375,
      "learning_rate": 0.001100569599743428,
      "loss": 1.61,
      "step": 3781
    },
    {
      "epoch": 0.48406501983873035,
      "grad_norm": 0.640625,
      "learning_rate": 0.0011001571254414492,
      "loss": 1.2466,
      "step": 3782
    },
    {
      "epoch": 0.4841930116472546,
      "grad_norm": 0.59375,
      "learning_rate": 0.001099744633925843,
      "loss": 1.3503,
      "step": 3783
    },
    {
      "epoch": 0.48432100345577883,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0010993321252675034,
      "loss": 1.4472,
      "step": 3784
    },
    {
      "epoch": 0.4844489952643031,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0010989195995373265,
      "loss": 1.1278,
      "step": 3785
    },
    {
      "epoch": 0.4845769870728273,
      "grad_norm": 0.6328125,
      "learning_rate": 0.001098507056806211,
      "loss": 1.2478,
      "step": 3786
    },
    {
      "epoch": 0.4847049788813516,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00109809449714506,
      "loss": 2.1651,
      "step": 3787
    },
    {
      "epoch": 0.48483297068987585,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0010976819206247777,
      "loss": 1.2023,
      "step": 3788
    },
    {
      "epoch": 0.4849609624984001,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0010972693273162727,
      "loss": 2.0888,
      "step": 3789
    },
    {
      "epoch": 0.48508895430692434,
      "grad_norm": 0.65625,
      "learning_rate": 0.0010968567172904558,
      "loss": 1.1907,
      "step": 3790
    },
    {
      "epoch": 0.48521694611544863,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0010964440906182404,
      "loss": 1.8563,
      "step": 3791
    },
    {
      "epoch": 0.4853449379239729,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0010960314473705433,
      "loss": 1.1354,
      "step": 3792
    },
    {
      "epoch": 0.4854729297324971,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0010956187876182838,
      "loss": 1.0686,
      "step": 3793
    },
    {
      "epoch": 0.48560092154102136,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0010952061114323842,
      "loss": 1.4601,
      "step": 3794
    },
    {
      "epoch": 0.48572891334954565,
      "grad_norm": 0.65625,
      "learning_rate": 0.0010947934188837696,
      "loss": 1.312,
      "step": 3795
    },
    {
      "epoch": 0.4858569051580699,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0010943807100433683,
      "loss": 1.1058,
      "step": 3796
    },
    {
      "epoch": 0.48598489696659414,
      "grad_norm": 1.015625,
      "learning_rate": 0.0010939679849821106,
      "loss": 2.0271,
      "step": 3797
    },
    {
      "epoch": 0.4861128887751184,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0010935552437709298,
      "loss": 1.5473,
      "step": 3798
    },
    {
      "epoch": 0.4862408805836426,
      "grad_norm": 0.703125,
      "learning_rate": 0.0010931424864807624,
      "loss": 1.4009,
      "step": 3799
    },
    {
      "epoch": 0.4863688723921669,
      "grad_norm": 0.71484375,
      "learning_rate": 0.001092729713182548,
      "loss": 1.6989,
      "step": 3800
    },
    {
      "epoch": 0.48649686420069116,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0010923169239472274,
      "loss": 1.2207,
      "step": 3801
    },
    {
      "epoch": 0.4866248560092154,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0010919041188457458,
      "loss": 1.7204,
      "step": 3802
    },
    {
      "epoch": 0.48675284781773964,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0010914912979490504,
      "loss": 1.7257,
      "step": 3803
    },
    {
      "epoch": 0.48688083962626394,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0010910784613280904,
      "loss": 1.3248,
      "step": 3804
    },
    {
      "epoch": 0.4870088314347882,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0010906656090538195,
      "loss": 1.2722,
      "step": 3805
    },
    {
      "epoch": 0.4871368232433124,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0010902527411971927,
      "loss": 2.2446,
      "step": 3806
    },
    {
      "epoch": 0.48726481505183666,
      "grad_norm": 0.625,
      "learning_rate": 0.0010898398578291682,
      "loss": 1.5673,
      "step": 3807
    },
    {
      "epoch": 0.48739280686036096,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0010894269590207063,
      "loss": 1.9282,
      "step": 3808
    },
    {
      "epoch": 0.4875207986688852,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0010890140448427702,
      "loss": 1.8628,
      "step": 3809
    },
    {
      "epoch": 0.48764879047740944,
      "grad_norm": 0.69140625,
      "learning_rate": 0.001088601115366327,
      "loss": 1.7521,
      "step": 3810
    },
    {
      "epoch": 0.4877767822859337,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0010881881706623442,
      "loss": 1.6367,
      "step": 3811
    },
    {
      "epoch": 0.4879047740944579,
      "grad_norm": 0.609375,
      "learning_rate": 0.0010877752108017934,
      "loss": 1.5187,
      "step": 3812
    },
    {
      "epoch": 0.4880327659029822,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0010873622358556485,
      "loss": 1.6453,
      "step": 3813
    },
    {
      "epoch": 0.48816075771150647,
      "grad_norm": 0.703125,
      "learning_rate": 0.001086949245894886,
      "loss": 1.7572,
      "step": 3814
    },
    {
      "epoch": 0.4882887495200307,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0010865362409904847,
      "loss": 1.2343,
      "step": 3815
    },
    {
      "epoch": 0.48841674132855495,
      "grad_norm": 0.546875,
      "learning_rate": 0.0010861232212134265,
      "loss": 1.3086,
      "step": 3816
    },
    {
      "epoch": 0.48854473313707925,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0010857101866346953,
      "loss": 1.0165,
      "step": 3817
    },
    {
      "epoch": 0.4886727249456035,
      "grad_norm": 0.609375,
      "learning_rate": 0.0010852971373252781,
      "loss": 1.6116,
      "step": 3818
    },
    {
      "epoch": 0.48880071675412773,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0010848840733561637,
      "loss": 0.9226,
      "step": 3819
    },
    {
      "epoch": 0.48892870856265197,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0010844709947983443,
      "loss": 1.1341,
      "step": 3820
    },
    {
      "epoch": 0.48905670037117627,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0010840579017228136,
      "loss": 2.2118,
      "step": 3821
    },
    {
      "epoch": 0.4891846921797005,
      "grad_norm": 0.703125,
      "learning_rate": 0.0010836447942005692,
      "loss": 1.125,
      "step": 3822
    },
    {
      "epoch": 0.48931268398822475,
      "grad_norm": 0.796875,
      "learning_rate": 0.0010832316723026092,
      "loss": 1.1425,
      "step": 3823
    },
    {
      "epoch": 0.489440675796749,
      "grad_norm": 0.71875,
      "learning_rate": 0.0010828185360999365,
      "loss": 1.3512,
      "step": 3824
    },
    {
      "epoch": 0.4895686676052733,
      "grad_norm": 0.578125,
      "learning_rate": 0.0010824053856635543,
      "loss": 1.1161,
      "step": 3825
    },
    {
      "epoch": 0.48969665941379753,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0010819922210644694,
      "loss": 1.0964,
      "step": 3826
    },
    {
      "epoch": 0.48982465122232177,
      "grad_norm": 0.546875,
      "learning_rate": 0.0010815790423736914,
      "loss": 1.1234,
      "step": 3827
    },
    {
      "epoch": 0.489952643030846,
      "grad_norm": 0.66015625,
      "learning_rate": 0.001081165849662231,
      "loss": 1.4927,
      "step": 3828
    },
    {
      "epoch": 0.49008063483937025,
      "grad_norm": 0.734375,
      "learning_rate": 0.0010807526430011028,
      "loss": 2.1187,
      "step": 3829
    },
    {
      "epoch": 0.49020862664789455,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0010803394224613223,
      "loss": 1.6346,
      "step": 3830
    },
    {
      "epoch": 0.4903366184564188,
      "grad_norm": 0.5625,
      "learning_rate": 0.001079926188113909,
      "loss": 0.8719,
      "step": 3831
    },
    {
      "epoch": 0.49046461026494304,
      "grad_norm": 0.578125,
      "learning_rate": 0.001079512940029883,
      "loss": 0.9436,
      "step": 3832
    },
    {
      "epoch": 0.4905926020734673,
      "grad_norm": 0.625,
      "learning_rate": 0.0010790996782802686,
      "loss": 1.5383,
      "step": 3833
    },
    {
      "epoch": 0.4907205938819916,
      "grad_norm": 0.59375,
      "learning_rate": 0.001078686402936091,
      "loss": 1.1728,
      "step": 3834
    },
    {
      "epoch": 0.4908485856905158,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0010782731140683785,
      "loss": 1.7422,
      "step": 3835
    },
    {
      "epoch": 0.49097657749904006,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0010778598117481611,
      "loss": 1.6897,
      "step": 3836
    },
    {
      "epoch": 0.4911045693075643,
      "grad_norm": 0.734375,
      "learning_rate": 0.001077446496046472,
      "loss": 1.429,
      "step": 3837
    },
    {
      "epoch": 0.4912325611160886,
      "grad_norm": 0.671875,
      "learning_rate": 0.0010770331670343458,
      "loss": 1.8426,
      "step": 3838
    },
    {
      "epoch": 0.49136055292461284,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00107661982478282,
      "loss": 1.2715,
      "step": 3839
    },
    {
      "epoch": 0.4914885447331371,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0010762064693629345,
      "loss": 1.1418,
      "step": 3840
    },
    {
      "epoch": 0.4916165365416613,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0010757931008457307,
      "loss": 1.6154,
      "step": 3841
    },
    {
      "epoch": 0.49174452835018556,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0010753797193022525,
      "loss": 2.0411,
      "step": 3842
    },
    {
      "epoch": 0.49187252015870986,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0010749663248035468,
      "loss": 1.2643,
      "step": 3843
    },
    {
      "epoch": 0.4920005119672341,
      "grad_norm": 0.69140625,
      "learning_rate": 0.001074552917420662,
      "loss": 1.6071,
      "step": 3844
    },
    {
      "epoch": 0.49212850377575834,
      "grad_norm": 0.65625,
      "learning_rate": 0.0010741394972246484,
      "loss": 1.8795,
      "step": 3845
    },
    {
      "epoch": 0.4922564955842826,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0010737260642865595,
      "loss": 1.1807,
      "step": 3846
    },
    {
      "epoch": 0.4923844873928069,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0010733126186774503,
      "loss": 1.3961,
      "step": 3847
    },
    {
      "epoch": 0.4925124792013311,
      "grad_norm": 0.5,
      "learning_rate": 0.001072899160468378,
      "loss": 1.1709,
      "step": 3848
    },
    {
      "epoch": 0.49264047100985536,
      "grad_norm": 0.703125,
      "learning_rate": 0.001072485689730402,
      "loss": 1.5396,
      "step": 3849
    },
    {
      "epoch": 0.4927684628183796,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0010720722065345844,
      "loss": 1.711,
      "step": 3850
    },
    {
      "epoch": 0.4928964546269039,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0010716587109519888,
      "loss": 1.7543,
      "step": 3851
    },
    {
      "epoch": 0.49302444643542814,
      "grad_norm": 0.58203125,
      "learning_rate": 0.001071245203053681,
      "loss": 1.1527,
      "step": 3852
    },
    {
      "epoch": 0.4931524382439524,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0010708316829107294,
      "loss": 1.7202,
      "step": 3853
    },
    {
      "epoch": 0.4932804300524766,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0010704181505942032,
      "loss": 0.8129,
      "step": 3854
    },
    {
      "epoch": 0.4934084218610009,
      "grad_norm": 0.78125,
      "learning_rate": 0.0010700046061751759,
      "loss": 1.7783,
      "step": 3855
    },
    {
      "epoch": 0.49353641366952516,
      "grad_norm": 0.66796875,
      "learning_rate": 0.001069591049724721,
      "loss": 1.5954,
      "step": 3856
    },
    {
      "epoch": 0.4936644054780494,
      "grad_norm": 0.609375,
      "learning_rate": 0.001069177481313915,
      "loss": 1.1819,
      "step": 3857
    },
    {
      "epoch": 0.49379239728657365,
      "grad_norm": 0.640625,
      "learning_rate": 0.0010687639010138368,
      "loss": 1.2262,
      "step": 3858
    },
    {
      "epoch": 0.4939203890950979,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0010683503088955663,
      "loss": 0.9776,
      "step": 3859
    },
    {
      "epoch": 0.4940483809036222,
      "grad_norm": 0.83984375,
      "learning_rate": 0.001067936705030186,
      "loss": 1.7527,
      "step": 3860
    },
    {
      "epoch": 0.49417637271214643,
      "grad_norm": 0.65625,
      "learning_rate": 0.001067523089488781,
      "loss": 1.5657,
      "step": 3861
    },
    {
      "epoch": 0.49430436452067067,
      "grad_norm": 0.609375,
      "learning_rate": 0.0010671094623424375,
      "loss": 1.6608,
      "step": 3862
    },
    {
      "epoch": 0.4944323563291949,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0010666958236622442,
      "loss": 1.5407,
      "step": 3863
    },
    {
      "epoch": 0.4945603481377192,
      "grad_norm": 0.703125,
      "learning_rate": 0.0010662821735192914,
      "loss": 1.4158,
      "step": 3864
    },
    {
      "epoch": 0.49468833994624345,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0010658685119846712,
      "loss": 1.3248,
      "step": 3865
    },
    {
      "epoch": 0.4948163317547677,
      "grad_norm": 0.765625,
      "learning_rate": 0.0010654548391294792,
      "loss": 1.7434,
      "step": 3866
    },
    {
      "epoch": 0.49494432356329193,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0010650411550248108,
      "loss": 1.7785,
      "step": 3867
    },
    {
      "epoch": 0.49507231537181623,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0010646274597417645,
      "loss": 1.4754,
      "step": 3868
    },
    {
      "epoch": 0.49520030718034047,
      "grad_norm": 0.703125,
      "learning_rate": 0.001064213753351441,
      "loss": 1.8486,
      "step": 3869
    },
    {
      "epoch": 0.4953282989888647,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0010638000359249418,
      "loss": 1.2117,
      "step": 3870
    },
    {
      "epoch": 0.49545629079738895,
      "grad_norm": 0.71484375,
      "learning_rate": 0.001063386307533371,
      "loss": 1.9305,
      "step": 3871
    },
    {
      "epoch": 0.4955842826059132,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0010629725682478354,
      "loss": 1.3997,
      "step": 3872
    },
    {
      "epoch": 0.4957122744144375,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0010625588181394424,
      "loss": 0.9641,
      "step": 3873
    },
    {
      "epoch": 0.49584026622296173,
      "grad_norm": 0.64453125,
      "learning_rate": 0.001062145057279301,
      "loss": 1.5378,
      "step": 3874
    },
    {
      "epoch": 0.495968258031486,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0010617312857385236,
      "loss": 1.5752,
      "step": 3875
    },
    {
      "epoch": 0.4960962498400102,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0010613175035882228,
      "loss": 2.0252,
      "step": 3876
    },
    {
      "epoch": 0.4962242416485345,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0010609037108995144,
      "loss": 1.1878,
      "step": 3877
    },
    {
      "epoch": 0.49635223345705876,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0010604899077435149,
      "loss": 1.6707,
      "step": 3878
    },
    {
      "epoch": 0.496480225265583,
      "grad_norm": 0.546875,
      "learning_rate": 0.0010600760941913434,
      "loss": 1.0217,
      "step": 3879
    },
    {
      "epoch": 0.49660821707410724,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0010596622703141208,
      "loss": 1.5655,
      "step": 3880
    },
    {
      "epoch": 0.49673620888263154,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0010592484361829687,
      "loss": 1.174,
      "step": 3881
    },
    {
      "epoch": 0.4968642006911558,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0010588345918690113,
      "loss": 1.5208,
      "step": 3882
    },
    {
      "epoch": 0.49699219249968,
      "grad_norm": 0.66796875,
      "learning_rate": 0.001058420737443375,
      "loss": 1.7663,
      "step": 3883
    },
    {
      "epoch": 0.49712018430820426,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0010580068729771873,
      "loss": 1.5014,
      "step": 3884
    },
    {
      "epoch": 0.4972481761167285,
      "grad_norm": 0.5703125,
      "learning_rate": 0.001057592998541577,
      "loss": 0.9018,
      "step": 3885
    },
    {
      "epoch": 0.4973761679252528,
      "grad_norm": 0.65625,
      "learning_rate": 0.0010571791142076758,
      "loss": 1.6119,
      "step": 3886
    },
    {
      "epoch": 0.49750415973377704,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0010567652200466156,
      "loss": 1.131,
      "step": 3887
    },
    {
      "epoch": 0.4976321515423013,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0010563513161295313,
      "loss": 1.8132,
      "step": 3888
    },
    {
      "epoch": 0.4977601433508255,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0010559374025275597,
      "loss": 1.2968,
      "step": 3889
    },
    {
      "epoch": 0.4978881351593498,
      "grad_norm": 0.765625,
      "learning_rate": 0.0010555234793118375,
      "loss": 1.7384,
      "step": 3890
    },
    {
      "epoch": 0.49801612696787406,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0010551095465535045,
      "loss": 1.2981,
      "step": 3891
    },
    {
      "epoch": 0.4981441187763983,
      "grad_norm": 0.66796875,
      "learning_rate": 0.001054695604323702,
      "loss": 1.5123,
      "step": 3892
    },
    {
      "epoch": 0.49827211058492255,
      "grad_norm": 0.703125,
      "learning_rate": 0.0010542816526935722,
      "loss": 1.8546,
      "step": 3893
    },
    {
      "epoch": 0.49840010239344684,
      "grad_norm": 0.953125,
      "learning_rate": 0.0010538676917342597,
      "loss": 2.4513,
      "step": 3894
    },
    {
      "epoch": 0.4985280942019711,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0010534537215169107,
      "loss": 1.8315,
      "step": 3895
    },
    {
      "epoch": 0.4986560860104953,
      "grad_norm": 0.74609375,
      "learning_rate": 0.001053039742112672,
      "loss": 1.6972,
      "step": 3896
    },
    {
      "epoch": 0.49878407781901957,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0010526257535926936,
      "loss": 1.8721,
      "step": 3897
    },
    {
      "epoch": 0.49891206962754386,
      "grad_norm": 0.6796875,
      "learning_rate": 0.001052211756028125,
      "loss": 1.5659,
      "step": 3898
    },
    {
      "epoch": 0.4990400614360681,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0010517977494901192,
      "loss": 1.3848,
      "step": 3899
    },
    {
      "epoch": 0.49916805324459235,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0010513837340498297,
      "loss": 1.2277,
      "step": 3900
    },
    {
      "epoch": 0.4992960450531166,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0010509697097784122,
      "loss": 1.5398,
      "step": 3901
    },
    {
      "epoch": 0.49942403686164083,
      "grad_norm": 0.6484375,
      "learning_rate": 0.001050555676747023,
      "loss": 1.6108,
      "step": 3902
    },
    {
      "epoch": 0.4995520286701651,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00105014163502682,
      "loss": 1.4297,
      "step": 3903
    },
    {
      "epoch": 0.49968002047868937,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0010497275846889638,
      "loss": 1.5052,
      "step": 3904
    },
    {
      "epoch": 0.4998080122872136,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0010493135258046154,
      "loss": 1.0537,
      "step": 3905
    },
    {
      "epoch": 0.49993600409573785,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0010488994584449374,
      "loss": 1.4475,
      "step": 3906
    },
    {
      "epoch": 0.5000639959042621,
      "grad_norm": 1.0078125,
      "learning_rate": 0.001048485382681094,
      "loss": 1.6393,
      "step": 3907
    },
    {
      "epoch": 0.5001919877127864,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0010480712985842513,
      "loss": 0.9537,
      "step": 3908
    },
    {
      "epoch": 0.5003199795213107,
      "grad_norm": 0.703125,
      "learning_rate": 0.001047657206225576,
      "loss": 1.8019,
      "step": 3909
    },
    {
      "epoch": 0.5004479713298349,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0010472431056762358,
      "loss": 1.7795,
      "step": 3910
    },
    {
      "epoch": 0.5005759631383592,
      "grad_norm": 0.8359375,
      "learning_rate": 0.001046828997007402,
      "loss": 1.2779,
      "step": 3911
    },
    {
      "epoch": 0.5007039549468834,
      "grad_norm": 0.734375,
      "learning_rate": 0.001046414880290245,
      "loss": 1.6997,
      "step": 3912
    },
    {
      "epoch": 0.5008319467554077,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0010460007555959383,
      "loss": 1.6984,
      "step": 3913
    },
    {
      "epoch": 0.500959938563932,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0010455866229956549,
      "loss": 1.6141,
      "step": 3914
    },
    {
      "epoch": 0.5010879303724561,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0010451724825605707,
      "loss": 1.6885,
      "step": 3915
    },
    {
      "epoch": 0.5012159221809804,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0010447583343618626,
      "loss": 1.7968,
      "step": 3916
    },
    {
      "epoch": 0.5013439139895046,
      "grad_norm": 0.640625,
      "learning_rate": 0.001044344178470709,
      "loss": 1.2313,
      "step": 3917
    },
    {
      "epoch": 0.5014719057980289,
      "grad_norm": 0.9140625,
      "learning_rate": 0.001043930014958288,
      "loss": 1.6876,
      "step": 3918
    },
    {
      "epoch": 0.5015998976065532,
      "grad_norm": 0.75,
      "learning_rate": 0.0010435158438957819,
      "loss": 1.9363,
      "step": 3919
    },
    {
      "epoch": 0.5017278894150774,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0010431016653543715,
      "loss": 1.3795,
      "step": 3920
    },
    {
      "epoch": 0.5018558812236017,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0010426874794052408,
      "loss": 1.4119,
      "step": 3921
    },
    {
      "epoch": 0.501983873032126,
      "grad_norm": 0.640625,
      "learning_rate": 0.001042273286119574,
      "loss": 1.4926,
      "step": 3922
    },
    {
      "epoch": 0.5021118648406502,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0010418590855685572,
      "loss": 2.0563,
      "step": 3923
    },
    {
      "epoch": 0.5022398566491745,
      "grad_norm": 0.75,
      "learning_rate": 0.0010414448778233777,
      "loss": 2.3923,
      "step": 3924
    },
    {
      "epoch": 0.5023678484576987,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0010410306629552231,
      "loss": 1.4498,
      "step": 3925
    },
    {
      "epoch": 0.502495840266223,
      "grad_norm": 0.625,
      "learning_rate": 0.0010406164410352834,
      "loss": 1.0972,
      "step": 3926
    },
    {
      "epoch": 0.5026238320747473,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0010402022121347495,
      "loss": 0.9747,
      "step": 3927
    },
    {
      "epoch": 0.5027518238832714,
      "grad_norm": 0.5859375,
      "learning_rate": 0.001039787976324813,
      "loss": 1.2369,
      "step": 3928
    },
    {
      "epoch": 0.5028798156917957,
      "grad_norm": 0.671875,
      "learning_rate": 0.0010393737336766673,
      "loss": 1.7741,
      "step": 3929
    },
    {
      "epoch": 0.5030078075003199,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0010389594842615067,
      "loss": 1.4652,
      "step": 3930
    },
    {
      "epoch": 0.5031357993088442,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0010385452281505268,
      "loss": 1.6996,
      "step": 3931
    },
    {
      "epoch": 0.5032637911173685,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0010381309654149235,
      "loss": 1.4017,
      "step": 3932
    },
    {
      "epoch": 0.5033917829258927,
      "grad_norm": 0.65625,
      "learning_rate": 0.0010377166961258954,
      "loss": 1.7911,
      "step": 3933
    },
    {
      "epoch": 0.503519774734417,
      "grad_norm": 0.640625,
      "learning_rate": 0.0010373024203546412,
      "loss": 1.293,
      "step": 3934
    },
    {
      "epoch": 0.5036477665429413,
      "grad_norm": 0.5859375,
      "learning_rate": 0.001036888138172361,
      "loss": 0.8172,
      "step": 3935
    },
    {
      "epoch": 0.5037757583514655,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0010364738496502555,
      "loss": 1.3114,
      "step": 3936
    },
    {
      "epoch": 0.5039037501599898,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0010360595548595272,
      "loss": 1.3741,
      "step": 3937
    },
    {
      "epoch": 0.504031741968514,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0010356452538713794,
      "loss": 1.9196,
      "step": 3938
    },
    {
      "epoch": 0.5041597337770383,
      "grad_norm": 0.671875,
      "learning_rate": 0.0010352309467570164,
      "loss": 1.334,
      "step": 3939
    },
    {
      "epoch": 0.5042877255855626,
      "grad_norm": 0.6875,
      "learning_rate": 0.0010348166335876438,
      "loss": 1.4792,
      "step": 3940
    },
    {
      "epoch": 0.5044157173940867,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0010344023144344672,
      "loss": 0.7615,
      "step": 3941
    },
    {
      "epoch": 0.504543709202611,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0010339879893686954,
      "loss": 1.3423,
      "step": 3942
    },
    {
      "epoch": 0.5046717010111352,
      "grad_norm": 0.71875,
      "learning_rate": 0.0010335736584615356,
      "loss": 1.6923,
      "step": 3943
    },
    {
      "epoch": 0.5047996928196595,
      "grad_norm": 0.5,
      "learning_rate": 0.001033159321784198,
      "loss": 1.0903,
      "step": 3944
    },
    {
      "epoch": 0.5049276846281838,
      "grad_norm": 0.76171875,
      "learning_rate": 0.001032744979407893,
      "loss": 1.5427,
      "step": 3945
    },
    {
      "epoch": 0.505055676436708,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0010323306314038324,
      "loss": 1.0237,
      "step": 3946
    },
    {
      "epoch": 0.5051836682452323,
      "grad_norm": 0.75,
      "learning_rate": 0.001031916277843228,
      "loss": 1.2268,
      "step": 3947
    },
    {
      "epoch": 0.5053116600537566,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0010315019187972934,
      "loss": 1.4482,
      "step": 3948
    },
    {
      "epoch": 0.5054396518622808,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0010310875543372425,
      "loss": 1.5664,
      "step": 3949
    },
    {
      "epoch": 0.5055676436708051,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0010306731845342917,
      "loss": 0.8626,
      "step": 3950
    },
    {
      "epoch": 0.5056956354793293,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0010302588094596558,
      "loss": 1.5475,
      "step": 3951
    },
    {
      "epoch": 0.5058236272878536,
      "grad_norm": 0.625,
      "learning_rate": 0.001029844429184553,
      "loss": 1.0782,
      "step": 3952
    },
    {
      "epoch": 0.5059516190963779,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0010294300437802008,
      "loss": 1.7669,
      "step": 3953
    },
    {
      "epoch": 0.506079610904902,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0010290156533178181,
      "loss": 1.0951,
      "step": 3954
    },
    {
      "epoch": 0.5062076027134264,
      "grad_norm": 0.65625,
      "learning_rate": 0.0010286012578686243,
      "loss": 1.14,
      "step": 3955
    },
    {
      "epoch": 0.5063355945219506,
      "grad_norm": 0.71875,
      "learning_rate": 0.0010281868575038404,
      "loss": 1.1679,
      "step": 3956
    },
    {
      "epoch": 0.5064635863304748,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0010277724522946882,
      "loss": 1.5888,
      "step": 3957
    },
    {
      "epoch": 0.5065915781389991,
      "grad_norm": 0.8359375,
      "learning_rate": 0.001027358042312389,
      "loss": 2.2238,
      "step": 3958
    },
    {
      "epoch": 0.5067195699475233,
      "grad_norm": 0.640625,
      "learning_rate": 0.0010269436276281664,
      "loss": 1.2349,
      "step": 3959
    },
    {
      "epoch": 0.5068475617560476,
      "grad_norm": 0.65234375,
      "learning_rate": 0.001026529208313244,
      "loss": 1.5128,
      "step": 3960
    },
    {
      "epoch": 0.5069755535645719,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0010261147844388473,
      "loss": 2.1263,
      "step": 3961
    },
    {
      "epoch": 0.5071035453730961,
      "grad_norm": 0.7421875,
      "learning_rate": 0.001025700356076201,
      "loss": 1.6227,
      "step": 3962
    },
    {
      "epoch": 0.5072315371816204,
      "grad_norm": 0.6640625,
      "learning_rate": 0.001025285923296531,
      "loss": 0.9586,
      "step": 3963
    },
    {
      "epoch": 0.5073595289901446,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0010248714861710655,
      "loss": 1.9431,
      "step": 3964
    },
    {
      "epoch": 0.5074875207986689,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0010244570447710314,
      "loss": 1.2959,
      "step": 3965
    },
    {
      "epoch": 0.5076155126071932,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0010240425991676569,
      "loss": 1.4285,
      "step": 3966
    },
    {
      "epoch": 0.5077435044157174,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0010236281494321719,
      "loss": 1.4839,
      "step": 3967
    },
    {
      "epoch": 0.5078714962242417,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0010232136956358064,
      "loss": 1.3947,
      "step": 3968
    },
    {
      "epoch": 0.507999488032766,
      "grad_norm": 0.59375,
      "learning_rate": 0.0010227992378497904,
      "loss": 1.0504,
      "step": 3969
    },
    {
      "epoch": 0.5081274798412901,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0010223847761453558,
      "loss": 1.2146,
      "step": 3970
    },
    {
      "epoch": 0.5082554716498144,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0010219703105937335,
      "loss": 1.5681,
      "step": 3971
    },
    {
      "epoch": 0.5083834634583386,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0010215558412661572,
      "loss": 0.6115,
      "step": 3972
    },
    {
      "epoch": 0.5085114552668629,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0010211413682338602,
      "loss": 1.4061,
      "step": 3973
    },
    {
      "epoch": 0.5086394470753872,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0010207268915680757,
      "loss": 1.5206,
      "step": 3974
    },
    {
      "epoch": 0.5087674388839114,
      "grad_norm": 0.734375,
      "learning_rate": 0.0010203124113400389,
      "loss": 1.5486,
      "step": 3975
    },
    {
      "epoch": 0.5088954306924357,
      "grad_norm": 0.66796875,
      "learning_rate": 0.001019897927620984,
      "loss": 1.6695,
      "step": 3976
    },
    {
      "epoch": 0.5090234225009599,
      "grad_norm": 0.578125,
      "learning_rate": 0.001019483440482148,
      "loss": 1.0495,
      "step": 3977
    },
    {
      "epoch": 0.5091514143094842,
      "grad_norm": 0.703125,
      "learning_rate": 0.0010190689499947664,
      "loss": 1.9144,
      "step": 3978
    },
    {
      "epoch": 0.5092794061180085,
      "grad_norm": 0.546875,
      "learning_rate": 0.0010186544562300766,
      "loss": 1.1555,
      "step": 3979
    },
    {
      "epoch": 0.5094073979265327,
      "grad_norm": 0.625,
      "learning_rate": 0.0010182399592593154,
      "loss": 1.2608,
      "step": 3980
    },
    {
      "epoch": 0.509535389735057,
      "grad_norm": 0.578125,
      "learning_rate": 0.0010178254591537218,
      "loss": 1.2789,
      "step": 3981
    },
    {
      "epoch": 0.5096633815435813,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0010174109559845336,
      "loss": 1.7769,
      "step": 3982
    },
    {
      "epoch": 0.5097913733521054,
      "grad_norm": 0.59375,
      "learning_rate": 0.0010169964498229894,
      "loss": 1.3682,
      "step": 3983
    },
    {
      "epoch": 0.5099193651606297,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0010165819407403306,
      "loss": 2.0932,
      "step": 3984
    },
    {
      "epoch": 0.5100473569691539,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0010161674288077958,
      "loss": 1.3771,
      "step": 3985
    },
    {
      "epoch": 0.5101753487776782,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0010157529140966263,
      "loss": 1.892,
      "step": 3986
    },
    {
      "epoch": 0.5103033405862025,
      "grad_norm": 0.59375,
      "learning_rate": 0.0010153383966780625,
      "loss": 1.4063,
      "step": 3987
    },
    {
      "epoch": 0.5104313323947267,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0010149238766233467,
      "loss": 1.505,
      "step": 3988
    },
    {
      "epoch": 0.510559324203251,
      "grad_norm": 0.640625,
      "learning_rate": 0.0010145093540037202,
      "loss": 1.6485,
      "step": 3989
    },
    {
      "epoch": 0.5106873160117752,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0010140948288904265,
      "loss": 1.9461,
      "step": 3990
    },
    {
      "epoch": 0.5108153078202995,
      "grad_norm": 0.640625,
      "learning_rate": 0.0010136803013547071,
      "loss": 1.2003,
      "step": 3991
    },
    {
      "epoch": 0.5109432996288238,
      "grad_norm": 0.625,
      "learning_rate": 0.0010132657714678063,
      "loss": 1.4739,
      "step": 3992
    },
    {
      "epoch": 0.511071291437348,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0010128512393009672,
      "loss": 1.3088,
      "step": 3993
    },
    {
      "epoch": 0.5111992832458723,
      "grad_norm": 0.78125,
      "learning_rate": 0.001012436704925434,
      "loss": 1.3081,
      "step": 3994
    },
    {
      "epoch": 0.5113272750543966,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0010120221684124515,
      "loss": 1.2492,
      "step": 3995
    },
    {
      "epoch": 0.5114552668629208,
      "grad_norm": 0.5625,
      "learning_rate": 0.0010116076298332644,
      "loss": 1.2274,
      "step": 3996
    },
    {
      "epoch": 0.511583258671445,
      "grad_norm": 0.76953125,
      "learning_rate": 0.001011193089259118,
      "loss": 1.5074,
      "step": 3997
    },
    {
      "epoch": 0.5117112504799692,
      "grad_norm": 0.6796875,
      "learning_rate": 0.001010778546761257,
      "loss": 1.9034,
      "step": 3998
    },
    {
      "epoch": 0.5118392422884935,
      "grad_norm": 0.625,
      "learning_rate": 0.0010103640024109282,
      "loss": 1.5582,
      "step": 3999
    },
    {
      "epoch": 0.5119672340970178,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0010099494562793771,
      "loss": 1.3784,
      "step": 4000
    },
    {
      "epoch": 0.512095225905542,
      "grad_norm": 1.578125,
      "learning_rate": 0.001009534908437851,
      "loss": 1.387,
      "step": 4001
    },
    {
      "epoch": 0.5122232177140663,
      "grad_norm": 0.71875,
      "learning_rate": 0.0010091203589575961,
      "loss": 1.8561,
      "step": 4002
    },
    {
      "epoch": 0.5123512095225905,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0010087058079098594,
      "loss": 1.3315,
      "step": 4003
    },
    {
      "epoch": 0.5124792013311148,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0010082912553658885,
      "loss": 2.037,
      "step": 4004
    },
    {
      "epoch": 0.5126071931396391,
      "grad_norm": 0.671875,
      "learning_rate": 0.0010078767013969303,
      "loss": 1.45,
      "step": 4005
    },
    {
      "epoch": 0.5127351849481633,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0010074621460742339,
      "loss": 1.8623,
      "step": 4006
    },
    {
      "epoch": 0.5128631767566876,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0010070475894690463,
      "loss": 1.41,
      "step": 4007
    },
    {
      "epoch": 0.5129911685652119,
      "grad_norm": 0.65625,
      "learning_rate": 0.0010066330316526161,
      "loss": 1.1012,
      "step": 4008
    },
    {
      "epoch": 0.5131191603737361,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0010062184726961918,
      "loss": 0.8734,
      "step": 4009
    },
    {
      "epoch": 0.5132471521822604,
      "grad_norm": 0.65234375,
      "learning_rate": 0.001005803912671022,
      "loss": 1.1599,
      "step": 4010
    },
    {
      "epoch": 0.5133751439907845,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0010053893516483556,
      "loss": 1.6462,
      "step": 4011
    },
    {
      "epoch": 0.5135031357993088,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0010049747896994422,
      "loss": 1.6703,
      "step": 4012
    },
    {
      "epoch": 0.5136311276078331,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0010045602268955298,
      "loss": 1.5783,
      "step": 4013
    },
    {
      "epoch": 0.5137591194163573,
      "grad_norm": 0.75,
      "learning_rate": 0.0010041456633078688,
      "loss": 1.1486,
      "step": 4014
    },
    {
      "epoch": 0.5138871112248816,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0010037310990077082,
      "loss": 1.2451,
      "step": 4015
    },
    {
      "epoch": 0.5140151030334058,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0010033165340662977,
      "loss": 1.6737,
      "step": 4016
    },
    {
      "epoch": 0.5141430948419301,
      "grad_norm": 0.61328125,
      "learning_rate": 0.001002901968554887,
      "loss": 1.2031,
      "step": 4017
    },
    {
      "epoch": 0.5142710866504544,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0010024874025447259,
      "loss": 1.3048,
      "step": 4018
    },
    {
      "epoch": 0.5143990784589786,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0010020728361070643,
      "loss": 1.0953,
      "step": 4019
    },
    {
      "epoch": 0.5145270702675029,
      "grad_norm": 1.25,
      "learning_rate": 0.0010016582693131524,
      "loss": 1.4212,
      "step": 4020
    },
    {
      "epoch": 0.5146550620760272,
      "grad_norm": 0.640625,
      "learning_rate": 0.0010012437022342397,
      "loss": 1.1844,
      "step": 4021
    },
    {
      "epoch": 0.5147830538845514,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0010008291349415765,
      "loss": 1.3277,
      "step": 4022
    },
    {
      "epoch": 0.5149110456930757,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0010004145675064136,
      "loss": 1.5768,
      "step": 4023
    },
    {
      "epoch": 0.5150390375015999,
      "grad_norm": 0.6484375,
      "learning_rate": 0.001,
      "loss": 1.0423,
      "step": 4024
    },
    {
      "epoch": 0.5151670293101241,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0009995854324935866,
      "loss": 1.7394,
      "step": 4025
    },
    {
      "epoch": 0.5152950211186484,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0009991708650584233,
      "loss": 1.4498,
      "step": 4026
    },
    {
      "epoch": 0.5154230129271726,
      "grad_norm": 0.625,
      "learning_rate": 0.0009987562977657606,
      "loss": 1.1581,
      "step": 4027
    },
    {
      "epoch": 0.5155510047356969,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0009983417306868479,
      "loss": 1.4062,
      "step": 4028
    },
    {
      "epoch": 0.5156789965442212,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0009979271638929357,
      "loss": 1.3091,
      "step": 4029
    },
    {
      "epoch": 0.5158069883527454,
      "grad_norm": 0.765625,
      "learning_rate": 0.0009975125974552742,
      "loss": 2.0614,
      "step": 4030
    },
    {
      "epoch": 0.5159349801612697,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0009970980314451135,
      "loss": 1.4728,
      "step": 4031
    },
    {
      "epoch": 0.5160629719697939,
      "grad_norm": 0.96875,
      "learning_rate": 0.0009966834659337024,
      "loss": 1.56,
      "step": 4032
    },
    {
      "epoch": 0.5161909637783182,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0009962689009922919,
      "loss": 1.4569,
      "step": 4033
    },
    {
      "epoch": 0.5163189555868425,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0009958543366921312,
      "loss": 1.4094,
      "step": 4034
    },
    {
      "epoch": 0.5164469473953667,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00099543977310447,
      "loss": 1.3107,
      "step": 4035
    },
    {
      "epoch": 0.516574939203891,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009950252103005583,
      "loss": 1.2362,
      "step": 4036
    },
    {
      "epoch": 0.5167029310124152,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0009946106483516444,
      "loss": 2.0884,
      "step": 4037
    },
    {
      "epoch": 0.5168309228209395,
      "grad_norm": 0.578125,
      "learning_rate": 0.0009941960873289782,
      "loss": 1.0722,
      "step": 4038
    },
    {
      "epoch": 0.5169589146294638,
      "grad_norm": 0.78515625,
      "learning_rate": 0.000993781527303808,
      "loss": 1.1221,
      "step": 4039
    },
    {
      "epoch": 0.5170869064379879,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0009933669683473841,
      "loss": 1.6631,
      "step": 4040
    },
    {
      "epoch": 0.5172148982465122,
      "grad_norm": 0.68359375,
      "learning_rate": 0.000992952410530954,
      "loss": 1.5704,
      "step": 4041
    },
    {
      "epoch": 0.5173428900550365,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0009925378539257664,
      "loss": 1.3625,
      "step": 4042
    },
    {
      "epoch": 0.5174708818635607,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0009921232986030695,
      "loss": 1.1639,
      "step": 4043
    },
    {
      "epoch": 0.517598873672085,
      "grad_norm": 0.64453125,
      "learning_rate": 0.000991708744634112,
      "loss": 0.9606,
      "step": 4044
    },
    {
      "epoch": 0.5177268654806092,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0009912941920901409,
      "loss": 1.1759,
      "step": 4045
    },
    {
      "epoch": 0.5178548572891335,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0009908796410424041,
      "loss": 0.8484,
      "step": 4046
    },
    {
      "epoch": 0.5179828490976578,
      "grad_norm": 0.53515625,
      "learning_rate": 0.000990465091562149,
      "loss": 1.1101,
      "step": 4047
    },
    {
      "epoch": 0.518110840906182,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0009900505437206227,
      "loss": 1.5603,
      "step": 4048
    },
    {
      "epoch": 0.5182388327147063,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0009896359975890719,
      "loss": 1.1434,
      "step": 4049
    },
    {
      "epoch": 0.5183668245232305,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0009892214532387432,
      "loss": 1.5629,
      "step": 4050
    },
    {
      "epoch": 0.5184948163317548,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0009888069107408823,
      "loss": 1.4095,
      "step": 4051
    },
    {
      "epoch": 0.5186228081402791,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0009883923701667356,
      "loss": 2.0055,
      "step": 4052
    },
    {
      "epoch": 0.5187507999488032,
      "grad_norm": 0.734375,
      "learning_rate": 0.0009879778315875487,
      "loss": 1.8771,
      "step": 4053
    },
    {
      "epoch": 0.5188787917573275,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0009875632950745662,
      "loss": 1.7181,
      "step": 4054
    },
    {
      "epoch": 0.5190067835658518,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0009871487606990328,
      "loss": 1.7509,
      "step": 4055
    },
    {
      "epoch": 0.519134775374376,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0009867342285321938,
      "loss": 2.2135,
      "step": 4056
    },
    {
      "epoch": 0.5192627671829003,
      "grad_norm": 0.59375,
      "learning_rate": 0.000986319698645293,
      "loss": 1.1546,
      "step": 4057
    },
    {
      "epoch": 0.5193907589914245,
      "grad_norm": 0.609375,
      "learning_rate": 0.000985905171109574,
      "loss": 1.4116,
      "step": 4058
    },
    {
      "epoch": 0.5195187507999488,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00098549064599628,
      "loss": 1.2887,
      "step": 4059
    },
    {
      "epoch": 0.5196467426084731,
      "grad_norm": 0.65625,
      "learning_rate": 0.0009850761233766536,
      "loss": 1.1935,
      "step": 4060
    },
    {
      "epoch": 0.5197747344169973,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0009846616033219377,
      "loss": 1.0899,
      "step": 4061
    },
    {
      "epoch": 0.5199027262255216,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0009842470859033742,
      "loss": 1.478,
      "step": 4062
    },
    {
      "epoch": 0.5200307180340458,
      "grad_norm": 0.609375,
      "learning_rate": 0.0009838325711922045,
      "loss": 1.2456,
      "step": 4063
    },
    {
      "epoch": 0.5201587098425701,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0009834180592596694,
      "loss": 1.2698,
      "step": 4064
    },
    {
      "epoch": 0.5202867016510944,
      "grad_norm": 0.6875,
      "learning_rate": 0.0009830035501770104,
      "loss": 1.4175,
      "step": 4065
    },
    {
      "epoch": 0.5204146934596185,
      "grad_norm": 0.59375,
      "learning_rate": 0.0009825890440154667,
      "loss": 1.4279,
      "step": 4066
    },
    {
      "epoch": 0.5205426852681428,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0009821745408462787,
      "loss": 1.2622,
      "step": 4067
    },
    {
      "epoch": 0.5206706770766671,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0009817600407406846,
      "loss": 1.2092,
      "step": 4068
    },
    {
      "epoch": 0.5207986688851913,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0009813455437699236,
      "loss": 1.0571,
      "step": 4069
    },
    {
      "epoch": 0.5209266606937156,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0009809310500052337,
      "loss": 1.2428,
      "step": 4070
    },
    {
      "epoch": 0.5210546525022398,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0009805165595178523,
      "loss": 1.717,
      "step": 4071
    },
    {
      "epoch": 0.5211826443107641,
      "grad_norm": 0.85546875,
      "learning_rate": 0.000980102072379016,
      "loss": 1.9695,
      "step": 4072
    },
    {
      "epoch": 0.5213106361192884,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0009796875886599614,
      "loss": 1.2984,
      "step": 4073
    },
    {
      "epoch": 0.5214386279278126,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0009792731084319244,
      "loss": 1.1901,
      "step": 4074
    },
    {
      "epoch": 0.5215666197363369,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0009788586317661403,
      "loss": 1.8083,
      "step": 4075
    },
    {
      "epoch": 0.5216946115448611,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0009784441587338429,
      "loss": 2.1293,
      "step": 4076
    },
    {
      "epoch": 0.5218226033533854,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0009780296894062667,
      "loss": 1.3798,
      "step": 4077
    },
    {
      "epoch": 0.5219505951619097,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0009776152238546445,
      "loss": 1.4126,
      "step": 4078
    },
    {
      "epoch": 0.5220785869704339,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0009772007621502094,
      "loss": 1.1239,
      "step": 4079
    },
    {
      "epoch": 0.5222065787789582,
      "grad_norm": 0.734375,
      "learning_rate": 0.0009767863043641939,
      "loss": 1.9069,
      "step": 4080
    },
    {
      "epoch": 0.5223345705874824,
      "grad_norm": 0.66015625,
      "learning_rate": 0.000976371850567828,
      "loss": 1.4066,
      "step": 4081
    },
    {
      "epoch": 0.5224625623960066,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0009759574008323431,
      "loss": 1.3571,
      "step": 4082
    },
    {
      "epoch": 0.5225905542045309,
      "grad_norm": 0.6328125,
      "learning_rate": 0.000975542955228969,
      "loss": 1.6479,
      "step": 4083
    },
    {
      "epoch": 0.5227185460130551,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0009751285138289348,
      "loss": 1.0956,
      "step": 4084
    },
    {
      "epoch": 0.5228465378215794,
      "grad_norm": 0.75390625,
      "learning_rate": 0.000974714076703469,
      "loss": 1.4513,
      "step": 4085
    },
    {
      "epoch": 0.5229745296301037,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0009742996439237995,
      "loss": 1.028,
      "step": 4086
    },
    {
      "epoch": 0.5231025214386279,
      "grad_norm": 0.765625,
      "learning_rate": 0.000973885215561153,
      "loss": 1.6738,
      "step": 4087
    },
    {
      "epoch": 0.5232305132471522,
      "grad_norm": 0.6875,
      "learning_rate": 0.0009734707916867559,
      "loss": 1.2342,
      "step": 4088
    },
    {
      "epoch": 0.5233585050556764,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0009730563723718339,
      "loss": 1.4473,
      "step": 4089
    },
    {
      "epoch": 0.5234864968642007,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0009726419576876112,
      "loss": 1.8384,
      "step": 4090
    },
    {
      "epoch": 0.523614488672725,
      "grad_norm": 0.59375,
      "learning_rate": 0.0009722275477053121,
      "loss": 1.0617,
      "step": 4091
    },
    {
      "epoch": 0.5237424804812492,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0009718131424961594,
      "loss": 1.205,
      "step": 4092
    },
    {
      "epoch": 0.5238704722897735,
      "grad_norm": 0.8125,
      "learning_rate": 0.0009713987421313758,
      "loss": 1.7948,
      "step": 4093
    },
    {
      "epoch": 0.5239984640982978,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0009709843466821821,
      "loss": 1.6287,
      "step": 4094
    },
    {
      "epoch": 0.5241264559068219,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0009705699562197993,
      "loss": 1.7061,
      "step": 4095
    },
    {
      "epoch": 0.5242544477153462,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0009701555708154469,
      "loss": 1.9281,
      "step": 4096
    },
    {
      "epoch": 0.5243824395238704,
      "grad_norm": 1.03125,
      "learning_rate": 0.0009697411905403443,
      "loss": 2.1192,
      "step": 4097
    },
    {
      "epoch": 0.5245104313323947,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0009693268154657088,
      "loss": 1.51,
      "step": 4098
    },
    {
      "epoch": 0.524638423140919,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0009689124456627576,
      "loss": 1.4854,
      "step": 4099
    },
    {
      "epoch": 0.5247664149494432,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0009684980812027068,
      "loss": 2.0303,
      "step": 4100
    },
    {
      "epoch": 0.5248944067579675,
      "grad_norm": 0.859375,
      "learning_rate": 0.0009680837221567721,
      "loss": 1.8658,
      "step": 4101
    },
    {
      "epoch": 0.5250223985664918,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0009676693685961679,
      "loss": 0.8916,
      "step": 4102
    },
    {
      "epoch": 0.525150390375016,
      "grad_norm": 0.6484375,
      "learning_rate": 0.000967255020592107,
      "loss": 1.5684,
      "step": 4103
    },
    {
      "epoch": 0.5252783821835403,
      "grad_norm": 0.5,
      "learning_rate": 0.000966840678215802,
      "loss": 0.808,
      "step": 4104
    },
    {
      "epoch": 0.5254063739920645,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0009664263415384644,
      "loss": 1.0635,
      "step": 4105
    },
    {
      "epoch": 0.5255343658005888,
      "grad_norm": 0.6796875,
      "learning_rate": 0.000966012010631305,
      "loss": 1.494,
      "step": 4106
    },
    {
      "epoch": 0.5256623576091131,
      "grad_norm": 0.546875,
      "learning_rate": 0.0009655976855655328,
      "loss": 1.0793,
      "step": 4107
    },
    {
      "epoch": 0.5257903494176372,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0009651833664123566,
      "loss": 1.4969,
      "step": 4108
    },
    {
      "epoch": 0.5259183412261615,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0009647690532429837,
      "loss": 0.8243,
      "step": 4109
    },
    {
      "epoch": 0.5260463330346857,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0009643547461286206,
      "loss": 1.2972,
      "step": 4110
    },
    {
      "epoch": 0.52617432484321,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0009639404451404728,
      "loss": 1.1914,
      "step": 4111
    },
    {
      "epoch": 0.5263023166517343,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0009635261503497445,
      "loss": 1.3283,
      "step": 4112
    },
    {
      "epoch": 0.5264303084602585,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0009631118618276391,
      "loss": 1.7427,
      "step": 4113
    },
    {
      "epoch": 0.5265583002687828,
      "grad_norm": 0.78125,
      "learning_rate": 0.0009626975796453586,
      "loss": 1.7884,
      "step": 4114
    },
    {
      "epoch": 0.5266862920773071,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0009622833038741048,
      "loss": 2.08,
      "step": 4115
    },
    {
      "epoch": 0.5268142838858313,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0009618690345850764,
      "loss": 1.613,
      "step": 4116
    },
    {
      "epoch": 0.5269422756943556,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0009614547718494735,
      "loss": 1.2432,
      "step": 4117
    },
    {
      "epoch": 0.5270702675028798,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0009610405157384933,
      "loss": 1.4507,
      "step": 4118
    },
    {
      "epoch": 0.5271982593114041,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0009606262663233325,
      "loss": 1.0298,
      "step": 4119
    },
    {
      "epoch": 0.5273262511199284,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0009602120236751872,
      "loss": 1.2197,
      "step": 4120
    },
    {
      "epoch": 0.5274542429284526,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0009597977878652508,
      "loss": 1.0416,
      "step": 4121
    },
    {
      "epoch": 0.5275822347369769,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0009593835589647167,
      "loss": 1.5423,
      "step": 4122
    },
    {
      "epoch": 0.527710226545501,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0009589693370447768,
      "loss": 1.0568,
      "step": 4123
    },
    {
      "epoch": 0.5278382183540253,
      "grad_norm": 0.9375,
      "learning_rate": 0.0009585551221766228,
      "loss": 1.2903,
      "step": 4124
    },
    {
      "epoch": 0.5279662101625496,
      "grad_norm": 0.84375,
      "learning_rate": 0.000958140914431443,
      "loss": 1.519,
      "step": 4125
    },
    {
      "epoch": 0.5280942019710738,
      "grad_norm": 0.578125,
      "learning_rate": 0.0009577267138804262,
      "loss": 1.2483,
      "step": 4126
    },
    {
      "epoch": 0.5282221937795981,
      "grad_norm": 0.640625,
      "learning_rate": 0.0009573125205947593,
      "loss": 1.0377,
      "step": 4127
    },
    {
      "epoch": 0.5283501855881224,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0009568983346456288,
      "loss": 1.6639,
      "step": 4128
    },
    {
      "epoch": 0.5284781773966466,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0009564841561042186,
      "loss": 1.2322,
      "step": 4129
    },
    {
      "epoch": 0.5286061692051709,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0009560699850417121,
      "loss": 1.4687,
      "step": 4130
    },
    {
      "epoch": 0.5287341610136951,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0009556558215292915,
      "loss": 1.1696,
      "step": 4131
    },
    {
      "epoch": 0.5288621528222194,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0009552416656381373,
      "loss": 0.6933,
      "step": 4132
    },
    {
      "epoch": 0.5289901446307437,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0009548275174394294,
      "loss": 1.268,
      "step": 4133
    },
    {
      "epoch": 0.5291181364392679,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0009544133770043453,
      "loss": 1.4249,
      "step": 4134
    },
    {
      "epoch": 0.5292461282477922,
      "grad_norm": 0.65234375,
      "learning_rate": 0.000953999244404062,
      "loss": 1.2348,
      "step": 4135
    },
    {
      "epoch": 0.5293741200563163,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0009535851197097548,
      "loss": 1.3731,
      "step": 4136
    },
    {
      "epoch": 0.5295021118648406,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0009531710029925984,
      "loss": 1.7501,
      "step": 4137
    },
    {
      "epoch": 0.5296301036733649,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0009527568943237645,
      "loss": 1.2982,
      "step": 4138
    },
    {
      "epoch": 0.5297580954818891,
      "grad_norm": 0.546875,
      "learning_rate": 0.0009523427937744245,
      "loss": 1.1068,
      "step": 4139
    },
    {
      "epoch": 0.5298860872904134,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0009519287014157488,
      "loss": 1.8894,
      "step": 4140
    },
    {
      "epoch": 0.5300140790989377,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0009515146173189057,
      "loss": 1.3177,
      "step": 4141
    },
    {
      "epoch": 0.5301420709074619,
      "grad_norm": 0.625,
      "learning_rate": 0.0009511005415550629,
      "loss": 1.4394,
      "step": 4142
    },
    {
      "epoch": 0.5302700627159862,
      "grad_norm": 0.640625,
      "learning_rate": 0.0009506864741953848,
      "loss": 1.2318,
      "step": 4143
    },
    {
      "epoch": 0.5303980545245104,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0009502724153110364,
      "loss": 1.5223,
      "step": 4144
    },
    {
      "epoch": 0.5305260463330347,
      "grad_norm": 0.59375,
      "learning_rate": 0.0009498583649731801,
      "loss": 1.4218,
      "step": 4145
    },
    {
      "epoch": 0.530654038141559,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0009494443232529776,
      "loss": 0.9453,
      "step": 4146
    },
    {
      "epoch": 0.5307820299500832,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0009490302902215882,
      "loss": 1.3953,
      "step": 4147
    },
    {
      "epoch": 0.5309100217586075,
      "grad_norm": 0.71875,
      "learning_rate": 0.0009486162659501703,
      "loss": 1.9138,
      "step": 4148
    },
    {
      "epoch": 0.5310380135671317,
      "grad_norm": 0.609375,
      "learning_rate": 0.0009482022505098808,
      "loss": 1.2434,
      "step": 4149
    },
    {
      "epoch": 0.531166005375656,
      "grad_norm": 0.76171875,
      "learning_rate": 0.000947788243971875,
      "loss": 1.9496,
      "step": 4150
    },
    {
      "epoch": 0.5312939971841802,
      "grad_norm": 0.6875,
      "learning_rate": 0.0009473742464073068,
      "loss": 1.8374,
      "step": 4151
    },
    {
      "epoch": 0.5314219889927044,
      "grad_norm": 0.75,
      "learning_rate": 0.000946960257887328,
      "loss": 1.9386,
      "step": 4152
    },
    {
      "epoch": 0.5315499808012287,
      "grad_norm": 0.859375,
      "learning_rate": 0.0009465462784830895,
      "loss": 1.1859,
      "step": 4153
    },
    {
      "epoch": 0.531677972609753,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0009461323082657401,
      "loss": 1.4338,
      "step": 4154
    },
    {
      "epoch": 0.5318059644182772,
      "grad_norm": 0.734375,
      "learning_rate": 0.0009457183473064279,
      "loss": 1.4393,
      "step": 4155
    },
    {
      "epoch": 0.5319339562268015,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0009453043956762983,
      "loss": 1.6619,
      "step": 4156
    },
    {
      "epoch": 0.5320619480353257,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0009448904534464956,
      "loss": 0.7814,
      "step": 4157
    },
    {
      "epoch": 0.53218993984385,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0009444765206881626,
      "loss": 1.6256,
      "step": 4158
    },
    {
      "epoch": 0.5323179316523743,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0009440625974724408,
      "loss": 0.7259,
      "step": 4159
    },
    {
      "epoch": 0.5324459234608985,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0009436486838704688,
      "loss": 2.1991,
      "step": 4160
    },
    {
      "epoch": 0.5325739152694228,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0009432347799533845,
      "loss": 1.2727,
      "step": 4161
    },
    {
      "epoch": 0.532701907077947,
      "grad_norm": 0.671875,
      "learning_rate": 0.0009428208857923245,
      "loss": 1.1378,
      "step": 4162
    },
    {
      "epoch": 0.5328298988864713,
      "grad_norm": 0.60546875,
      "learning_rate": 0.000942407001458423,
      "loss": 0.9146,
      "step": 4163
    },
    {
      "epoch": 0.5329578906949956,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0009419931270228133,
      "loss": 1.24,
      "step": 4164
    },
    {
      "epoch": 0.5330858825035197,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0009415792625566253,
      "loss": 1.0764,
      "step": 4165
    },
    {
      "epoch": 0.533213874312044,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0009411654081309888,
      "loss": 2.4241,
      "step": 4166
    },
    {
      "epoch": 0.5333418661205683,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0009407515638170317,
      "loss": 2.3665,
      "step": 4167
    },
    {
      "epoch": 0.5334698579290925,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0009403377296858797,
      "loss": 1.3332,
      "step": 4168
    },
    {
      "epoch": 0.5335978497376168,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0009399239058086568,
      "loss": 1.4247,
      "step": 4169
    },
    {
      "epoch": 0.533725841546141,
      "grad_norm": 0.5625,
      "learning_rate": 0.0009395100922564852,
      "loss": 1.2996,
      "step": 4170
    },
    {
      "epoch": 0.5338538333546653,
      "grad_norm": 0.671875,
      "learning_rate": 0.0009390962891004858,
      "loss": 1.8806,
      "step": 4171
    },
    {
      "epoch": 0.5339818251631896,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0009386824964117773,
      "loss": 1.5017,
      "step": 4172
    },
    {
      "epoch": 0.5341098169717138,
      "grad_norm": 0.640625,
      "learning_rate": 0.0009382687142614767,
      "loss": 1.2444,
      "step": 4173
    },
    {
      "epoch": 0.5342378087802381,
      "grad_norm": 0.640625,
      "learning_rate": 0.0009378549427206991,
      "loss": 1.5432,
      "step": 4174
    },
    {
      "epoch": 0.5343658005887624,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0009374411818605579,
      "loss": 1.3327,
      "step": 4175
    },
    {
      "epoch": 0.5344937923972866,
      "grad_norm": 0.65625,
      "learning_rate": 0.0009370274317521644,
      "loss": 1.4451,
      "step": 4176
    },
    {
      "epoch": 0.5346217842058109,
      "grad_norm": 0.59375,
      "learning_rate": 0.0009366136924666288,
      "loss": 1.1007,
      "step": 4177
    },
    {
      "epoch": 0.534749776014335,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0009361999640750585,
      "loss": 1.5631,
      "step": 4178
    },
    {
      "epoch": 0.5348777678228593,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0009357862466485594,
      "loss": 1.5469,
      "step": 4179
    },
    {
      "epoch": 0.5350057596313836,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0009353725402582356,
      "loss": 1.623,
      "step": 4180
    },
    {
      "epoch": 0.5351337514399078,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0009349588449751898,
      "loss": 1.4546,
      "step": 4181
    },
    {
      "epoch": 0.5352617432484321,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0009345451608705213,
      "loss": 1.1013,
      "step": 4182
    },
    {
      "epoch": 0.5353897350569563,
      "grad_norm": 0.64453125,
      "learning_rate": 0.000934131488015329,
      "loss": 1.2902,
      "step": 4183
    },
    {
      "epoch": 0.5355177268654806,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0009337178264807088,
      "loss": 1.4758,
      "step": 4184
    },
    {
      "epoch": 0.5356457186740049,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0009333041763377558,
      "loss": 1.4345,
      "step": 4185
    },
    {
      "epoch": 0.5357737104825291,
      "grad_norm": 0.84375,
      "learning_rate": 0.0009328905376575628,
      "loss": 1.9367,
      "step": 4186
    },
    {
      "epoch": 0.5359017022910534,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0009324769105112192,
      "loss": 1.8046,
      "step": 4187
    },
    {
      "epoch": 0.5360296940995777,
      "grad_norm": 0.58984375,
      "learning_rate": 0.000932063294969814,
      "loss": 1.4072,
      "step": 4188
    },
    {
      "epoch": 0.5361576859081019,
      "grad_norm": 0.69140625,
      "learning_rate": 0.000931649691104434,
      "loss": 1.194,
      "step": 4189
    },
    {
      "epoch": 0.5362856777166262,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0009312360989861637,
      "loss": 0.9666,
      "step": 4190
    },
    {
      "epoch": 0.5364136695251503,
      "grad_norm": 0.65625,
      "learning_rate": 0.0009308225186860852,
      "loss": 1.0125,
      "step": 4191
    },
    {
      "epoch": 0.5365416613336746,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0009304089502752791,
      "loss": 1.2579,
      "step": 4192
    },
    {
      "epoch": 0.5366696531421989,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0009299953938248242,
      "loss": 1.6354,
      "step": 4193
    },
    {
      "epoch": 0.5367976449507231,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0009295818494057967,
      "loss": 1.4725,
      "step": 4194
    },
    {
      "epoch": 0.5369256367592474,
      "grad_norm": 0.63671875,
      "learning_rate": 0.000929168317089271,
      "loss": 1.2696,
      "step": 4195
    },
    {
      "epoch": 0.5370536285677716,
      "grad_norm": 0.640625,
      "learning_rate": 0.0009287547969463191,
      "loss": 1.3346,
      "step": 4196
    },
    {
      "epoch": 0.5371816203762959,
      "grad_norm": 0.671875,
      "learning_rate": 0.0009283412890480113,
      "loss": 1.5821,
      "step": 4197
    },
    {
      "epoch": 0.5373096121848202,
      "grad_norm": 0.65625,
      "learning_rate": 0.0009279277934654156,
      "loss": 1.2,
      "step": 4198
    },
    {
      "epoch": 0.5374376039933444,
      "grad_norm": 0.75,
      "learning_rate": 0.0009275143102695984,
      "loss": 1.4555,
      "step": 4199
    },
    {
      "epoch": 0.5375655958018687,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0009271008395316223,
      "loss": 1.1416,
      "step": 4200
    },
    {
      "epoch": 0.537693587610393,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009266873813225499,
      "loss": 1.0973,
      "step": 4201
    },
    {
      "epoch": 0.5378215794189172,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0009262739357134406,
      "loss": 1.1359,
      "step": 4202
    },
    {
      "epoch": 0.5379495712274415,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0009258605027753515,
      "loss": 1.659,
      "step": 4203
    },
    {
      "epoch": 0.5380775630359657,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0009254470825793384,
      "loss": 1.0914,
      "step": 4204
    },
    {
      "epoch": 0.53820555484449,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0009250336751964535,
      "loss": 1.2921,
      "step": 4205
    },
    {
      "epoch": 0.5383335466530142,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0009246202806977476,
      "loss": 1.1683,
      "step": 4206
    },
    {
      "epoch": 0.5384615384615384,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0009242068991542694,
      "loss": 1.9805,
      "step": 4207
    },
    {
      "epoch": 0.5385895302700627,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0009237935306370659,
      "loss": 1.3031,
      "step": 4208
    },
    {
      "epoch": 0.5387175220785869,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0009233801752171801,
      "loss": 1.2648,
      "step": 4209
    },
    {
      "epoch": 0.5388455138871112,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0009229668329656543,
      "loss": 1.4744,
      "step": 4210
    },
    {
      "epoch": 0.5389735056956355,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0009225535039535282,
      "loss": 1.1753,
      "step": 4211
    },
    {
      "epoch": 0.5391014975041597,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0009221401882518393,
      "loss": 1.1226,
      "step": 4212
    },
    {
      "epoch": 0.539229489312684,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0009217268859316219,
      "loss": 1.5472,
      "step": 4213
    },
    {
      "epoch": 0.5393574811212083,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0009213135970639092,
      "loss": 1.181,
      "step": 4214
    },
    {
      "epoch": 0.5394854729297325,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0009209003217197315,
      "loss": 1.6803,
      "step": 4215
    },
    {
      "epoch": 0.5396134647382568,
      "grad_norm": 0.72265625,
      "learning_rate": 0.000920487059970117,
      "loss": 1.4109,
      "step": 4216
    },
    {
      "epoch": 0.539741456546781,
      "grad_norm": 0.59375,
      "learning_rate": 0.0009200738118860913,
      "loss": 1.1871,
      "step": 4217
    },
    {
      "epoch": 0.5398694483553053,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0009196605775386777,
      "loss": 1.0023,
      "step": 4218
    },
    {
      "epoch": 0.5399974401638296,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0009192473569988973,
      "loss": 1.3049,
      "step": 4219
    },
    {
      "epoch": 0.5401254319723537,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0009188341503377689,
      "loss": 1.3564,
      "step": 4220
    },
    {
      "epoch": 0.540253423780878,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0009184209576263092,
      "loss": 1.1808,
      "step": 4221
    },
    {
      "epoch": 0.5403814155894022,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0009180077789355306,
      "loss": 1.9248,
      "step": 4222
    },
    {
      "epoch": 0.5405094073979265,
      "grad_norm": 0.70703125,
      "learning_rate": 0.000917594614336446,
      "loss": 1.7733,
      "step": 4223
    },
    {
      "epoch": 0.5406373992064508,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0009171814639000637,
      "loss": 1.7673,
      "step": 4224
    },
    {
      "epoch": 0.540765391014975,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0009167683276973906,
      "loss": 1.4037,
      "step": 4225
    },
    {
      "epoch": 0.5408933828234993,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0009163552057994313,
      "loss": 0.7341,
      "step": 4226
    },
    {
      "epoch": 0.5410213746320236,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0009159420982771864,
      "loss": 1.0691,
      "step": 4227
    },
    {
      "epoch": 0.5411493664405478,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0009155290052016559,
      "loss": 1.263,
      "step": 4228
    },
    {
      "epoch": 0.5412773582490721,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0009151159266438362,
      "loss": 2.0575,
      "step": 4229
    },
    {
      "epoch": 0.5414053500575963,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0009147028626747222,
      "loss": 1.1804,
      "step": 4230
    },
    {
      "epoch": 0.5415333418661206,
      "grad_norm": 0.765625,
      "learning_rate": 0.0009142898133653048,
      "loss": 2.2662,
      "step": 4231
    },
    {
      "epoch": 0.5416613336746449,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0009138767787865737,
      "loss": 1.3173,
      "step": 4232
    },
    {
      "epoch": 0.541789325483169,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0009134637590095154,
      "loss": 1.2609,
      "step": 4233
    },
    {
      "epoch": 0.5419173172916933,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0009130507541051142,
      "loss": 1.0688,
      "step": 4234
    },
    {
      "epoch": 0.5420453091002175,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0009126377641443517,
      "loss": 1.471,
      "step": 4235
    },
    {
      "epoch": 0.5421733009087418,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0009122247891982069,
      "loss": 1.4828,
      "step": 4236
    },
    {
      "epoch": 0.5423012927172661,
      "grad_norm": 0.65625,
      "learning_rate": 0.0009118118293376561,
      "loss": 1.2028,
      "step": 4237
    },
    {
      "epoch": 0.5424292845257903,
      "grad_norm": 0.640625,
      "learning_rate": 0.0009113988846336732,
      "loss": 1.2975,
      "step": 4238
    },
    {
      "epoch": 0.5425572763343146,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0009109859551572297,
      "loss": 1.2243,
      "step": 4239
    },
    {
      "epoch": 0.5426852681428389,
      "grad_norm": 0.67578125,
      "learning_rate": 0.000910573040979294,
      "loss": 1.3711,
      "step": 4240
    },
    {
      "epoch": 0.5428132599513631,
      "grad_norm": 0.6875,
      "learning_rate": 0.0009101601421708319,
      "loss": 1.1919,
      "step": 4241
    },
    {
      "epoch": 0.5429412517598874,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0009097472588028071,
      "loss": 1.1624,
      "step": 4242
    },
    {
      "epoch": 0.5430692435684116,
      "grad_norm": 0.59375,
      "learning_rate": 0.0009093343909461806,
      "loss": 1.2973,
      "step": 4243
    },
    {
      "epoch": 0.5431972353769359,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0009089215386719097,
      "loss": 1.1945,
      "step": 4244
    },
    {
      "epoch": 0.5433252271854602,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00090850870205095,
      "loss": 1.6441,
      "step": 4245
    },
    {
      "epoch": 0.5434532189939844,
      "grad_norm": 0.90625,
      "learning_rate": 0.0009080958811542542,
      "loss": 1.5184,
      "step": 4246
    },
    {
      "epoch": 0.5435812108025087,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0009076830760527725,
      "loss": 1.4538,
      "step": 4247
    },
    {
      "epoch": 0.543709202611033,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0009072702868174524,
      "loss": 1.172,
      "step": 4248
    },
    {
      "epoch": 0.5438371944195571,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0009068575135192377,
      "loss": 1.2365,
      "step": 4249
    },
    {
      "epoch": 0.5439651862280814,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0009064447562290704,
      "loss": 1.3475,
      "step": 4250
    },
    {
      "epoch": 0.5440931780366056,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0009060320150178897,
      "loss": 1.0337,
      "step": 4251
    },
    {
      "epoch": 0.5442211698451299,
      "grad_norm": 0.6171875,
      "learning_rate": 0.000905619289956632,
      "loss": 1.1218,
      "step": 4252
    },
    {
      "epoch": 0.5443491616536542,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0009052065811162304,
      "loss": 1.9926,
      "step": 4253
    },
    {
      "epoch": 0.5444771534621784,
      "grad_norm": 0.59765625,
      "learning_rate": 0.000904793888567616,
      "loss": 1.1876,
      "step": 4254
    },
    {
      "epoch": 0.5446051452707027,
      "grad_norm": 0.625,
      "learning_rate": 0.0009043812123817164,
      "loss": 1.3979,
      "step": 4255
    },
    {
      "epoch": 0.5447331370792269,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0009039685526294568,
      "loss": 1.4995,
      "step": 4256
    },
    {
      "epoch": 0.5448611288877512,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0009035559093817598,
      "loss": 1.4829,
      "step": 4257
    },
    {
      "epoch": 0.5449891206962755,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0009031432827095444,
      "loss": 1.4966,
      "step": 4258
    },
    {
      "epoch": 0.5451171125047997,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0009027306726837272,
      "loss": 1.926,
      "step": 4259
    },
    {
      "epoch": 0.545245104313324,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0009023180793752223,
      "loss": 1.6835,
      "step": 4260
    },
    {
      "epoch": 0.5453730961218483,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0009019055028549404,
      "loss": 1.1597,
      "step": 4261
    },
    {
      "epoch": 0.5455010879303724,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0009014929431937891,
      "loss": 1.3332,
      "step": 4262
    },
    {
      "epoch": 0.5456290797388967,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0009010804004626738,
      "loss": 1.682,
      "step": 4263
    },
    {
      "epoch": 0.5457570715474209,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0009006678747324967,
      "loss": 1.3329,
      "step": 4264
    },
    {
      "epoch": 0.5458850633559452,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0009002553660741574,
      "loss": 1.1675,
      "step": 4265
    },
    {
      "epoch": 0.5460130551644695,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008998428745585513,
      "loss": 1.0613,
      "step": 4266
    },
    {
      "epoch": 0.5461410469729937,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0008994304002565722,
      "loss": 2.1313,
      "step": 4267
    },
    {
      "epoch": 0.546269038781518,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0008990179432391103,
      "loss": 1.6178,
      "step": 4268
    },
    {
      "epoch": 0.5463970305900422,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0008986055035770536,
      "loss": 1.51,
      "step": 4269
    },
    {
      "epoch": 0.5465250223985665,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0008981930813412865,
      "loss": 1.6904,
      "step": 4270
    },
    {
      "epoch": 0.5466530142070908,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0008977806766026899,
      "loss": 1.5791,
      "step": 4271
    },
    {
      "epoch": 0.546781006015615,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0008973682894321425,
      "loss": 2.3671,
      "step": 4272
    },
    {
      "epoch": 0.5469089978241393,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008969559199005195,
      "loss": 1.3767,
      "step": 4273
    },
    {
      "epoch": 0.5470369896326636,
      "grad_norm": 0.546875,
      "learning_rate": 0.000896543568078694,
      "loss": 1.0,
      "step": 4274
    },
    {
      "epoch": 0.5471649814411877,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0008961312340375345,
      "loss": 1.4543,
      "step": 4275
    },
    {
      "epoch": 0.547292973249712,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0008957189178479077,
      "loss": 2.1927,
      "step": 4276
    },
    {
      "epoch": 0.5474209650582362,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008953066195806769,
      "loss": 1.4581,
      "step": 4277
    },
    {
      "epoch": 0.5475489568667605,
      "grad_norm": 0.58984375,
      "learning_rate": 0.000894894339306702,
      "loss": 1.5452,
      "step": 4278
    },
    {
      "epoch": 0.5476769486752848,
      "grad_norm": 0.515625,
      "learning_rate": 0.0008944820770968406,
      "loss": 1.1578,
      "step": 4279
    },
    {
      "epoch": 0.547804940483809,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0008940698330219459,
      "loss": 1.5027,
      "step": 4280
    },
    {
      "epoch": 0.5479329322923333,
      "grad_norm": 0.671875,
      "learning_rate": 0.0008936576071528692,
      "loss": 1.6731,
      "step": 4281
    },
    {
      "epoch": 0.5480609241008575,
      "grad_norm": 0.5625,
      "learning_rate": 0.000893245399560458,
      "loss": 1.1323,
      "step": 4282
    },
    {
      "epoch": 0.5481889159093818,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0008928332103155572,
      "loss": 1.0187,
      "step": 4283
    },
    {
      "epoch": 0.5483169077179061,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008924210394890078,
      "loss": 1.3356,
      "step": 4284
    },
    {
      "epoch": 0.5484448995264303,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0008920088871516481,
      "loss": 1.6728,
      "step": 4285
    },
    {
      "epoch": 0.5485728913349546,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0008915967533743134,
      "loss": 1.1037,
      "step": 4286
    },
    {
      "epoch": 0.5487008831434789,
      "grad_norm": 0.703125,
      "learning_rate": 0.0008911846382278352,
      "loss": 1.4628,
      "step": 4287
    },
    {
      "epoch": 0.548828874952003,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0008907725417830431,
      "loss": 1.3463,
      "step": 4288
    },
    {
      "epoch": 0.5489568667605274,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008903604641107614,
      "loss": 0.884,
      "step": 4289
    },
    {
      "epoch": 0.5490848585690515,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008899484052818126,
      "loss": 0.9491,
      "step": 4290
    },
    {
      "epoch": 0.5492128503775758,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0008895363653670161,
      "loss": 2.0905,
      "step": 4291
    },
    {
      "epoch": 0.5493408421861001,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0008891243444371879,
      "loss": 1.7515,
      "step": 4292
    },
    {
      "epoch": 0.5494688339946243,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0008887123425631396,
      "loss": 1.1553,
      "step": 4293
    },
    {
      "epoch": 0.5495968258031486,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0008883003598156809,
      "loss": 0.97,
      "step": 4294
    },
    {
      "epoch": 0.5497248176116728,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0008878883962656175,
      "loss": 2.0329,
      "step": 4295
    },
    {
      "epoch": 0.5498528094201971,
      "grad_norm": 0.640625,
      "learning_rate": 0.0008874764519837526,
      "loss": 1.518,
      "step": 4296
    },
    {
      "epoch": 0.5499808012287214,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008870645270408848,
      "loss": 1.2498,
      "step": 4297
    },
    {
      "epoch": 0.5501087930372456,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008866526215078103,
      "loss": 1.0313,
      "step": 4298
    },
    {
      "epoch": 0.5502367848457699,
      "grad_norm": 0.60546875,
      "learning_rate": 0.000886240735455322,
      "loss": 1.4513,
      "step": 4299
    },
    {
      "epoch": 0.5503647766542942,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0008858288689542089,
      "loss": 1.6362,
      "step": 4300
    },
    {
      "epoch": 0.5504927684628184,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0008854170220752573,
      "loss": 1.2716,
      "step": 4301
    },
    {
      "epoch": 0.5506207602713427,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0008850051948892494,
      "loss": 1.9504,
      "step": 4302
    },
    {
      "epoch": 0.5507487520798668,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0008845933874669645,
      "loss": 0.8233,
      "step": 4303
    },
    {
      "epoch": 0.5508767438883911,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0008841815998791784,
      "loss": 1.3901,
      "step": 4304
    },
    {
      "epoch": 0.5510047356969154,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0008837698321966639,
      "loss": 1.0866,
      "step": 4305
    },
    {
      "epoch": 0.5511327275054396,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008833580844901887,
      "loss": 0.9383,
      "step": 4306
    },
    {
      "epoch": 0.5512607193139639,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0008829463568305196,
      "loss": 1.4528,
      "step": 4307
    },
    {
      "epoch": 0.5513887111224881,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0008825346492884183,
      "loss": 1.3723,
      "step": 4308
    },
    {
      "epoch": 0.5515167029310124,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008821229619346432,
      "loss": 1.2352,
      "step": 4309
    },
    {
      "epoch": 0.5516446947395367,
      "grad_norm": 0.609375,
      "learning_rate": 0.0008817112948399502,
      "loss": 1.2808,
      "step": 4310
    },
    {
      "epoch": 0.5517726865480609,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008812996480750899,
      "loss": 0.986,
      "step": 4311
    },
    {
      "epoch": 0.5519006783565852,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0008808880217108109,
      "loss": 1.3825,
      "step": 4312
    },
    {
      "epoch": 0.5520286701651095,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0008804764158178577,
      "loss": 1.1918,
      "step": 4313
    },
    {
      "epoch": 0.5521566619736337,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008800648304669725,
      "loss": 1.1634,
      "step": 4314
    },
    {
      "epoch": 0.552284653782158,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0008796532657288915,
      "loss": 0.7287,
      "step": 4315
    },
    {
      "epoch": 0.5524126455906821,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008792417216743494,
      "loss": 1.1964,
      "step": 4316
    },
    {
      "epoch": 0.5525406373992064,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0008788301983740769,
      "loss": 1.2879,
      "step": 4317
    },
    {
      "epoch": 0.5526686292077307,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0008784186958988006,
      "loss": 1.4235,
      "step": 4318
    },
    {
      "epoch": 0.5527966210162549,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0008780072143192442,
      "loss": 1.7344,
      "step": 4319
    },
    {
      "epoch": 0.5529246128247792,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0008775957537061272,
      "loss": 1.9381,
      "step": 4320
    },
    {
      "epoch": 0.5530526046333035,
      "grad_norm": 0.546875,
      "learning_rate": 0.0008771843141301658,
      "loss": 0.8636,
      "step": 4321
    },
    {
      "epoch": 0.5531805964418277,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0008767728956620728,
      "loss": 1.3933,
      "step": 4322
    },
    {
      "epoch": 0.553308588250352,
      "grad_norm": 0.671875,
      "learning_rate": 0.0008763614983725572,
      "loss": 1.5717,
      "step": 4323
    },
    {
      "epoch": 0.5534365800588762,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0008759501223323239,
      "loss": 1.3145,
      "step": 4324
    },
    {
      "epoch": 0.5535645718674005,
      "grad_norm": 0.703125,
      "learning_rate": 0.0008755387676120746,
      "loss": 1.3863,
      "step": 4325
    },
    {
      "epoch": 0.5536925636759248,
      "grad_norm": 0.734375,
      "learning_rate": 0.0008751274342825076,
      "loss": 1.9962,
      "step": 4326
    },
    {
      "epoch": 0.553820555484449,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008747161224143174,
      "loss": 1.0187,
      "step": 4327
    },
    {
      "epoch": 0.5539485472929733,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0008743048320781937,
      "loss": 1.4983,
      "step": 4328
    },
    {
      "epoch": 0.5540765391014975,
      "grad_norm": 0.6875,
      "learning_rate": 0.0008738935633448242,
      "loss": 1.8387,
      "step": 4329
    },
    {
      "epoch": 0.5542045309100218,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008734823162848918,
      "loss": 0.7988,
      "step": 4330
    },
    {
      "epoch": 0.554332522718546,
      "grad_norm": 0.625,
      "learning_rate": 0.0008730710909690761,
      "loss": 1.3431,
      "step": 4331
    },
    {
      "epoch": 0.5544605145270702,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0008726598874680533,
      "loss": 1.7849,
      "step": 4332
    },
    {
      "epoch": 0.5545885063355945,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0008722487058524943,
      "loss": 0.6922,
      "step": 4333
    },
    {
      "epoch": 0.5547164981441188,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008718375461930681,
      "loss": 1.0125,
      "step": 4334
    },
    {
      "epoch": 0.554844489952643,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008714264085604388,
      "loss": 1.0779,
      "step": 4335
    },
    {
      "epoch": 0.5549724817611673,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0008710152930252677,
      "loss": 1.4712,
      "step": 4336
    },
    {
      "epoch": 0.5551004735696915,
      "grad_norm": 0.7109375,
      "learning_rate": 0.000870604199658211,
      "loss": 1.4511,
      "step": 4337
    },
    {
      "epoch": 0.5552284653782158,
      "grad_norm": 0.9375,
      "learning_rate": 0.000870193128529922,
      "loss": 1.413,
      "step": 4338
    },
    {
      "epoch": 0.5553564571867401,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0008697820797110499,
      "loss": 1.914,
      "step": 4339
    },
    {
      "epoch": 0.5554844489952643,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008693710532722401,
      "loss": 0.9238,
      "step": 4340
    },
    {
      "epoch": 0.5556124408037886,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0008689600492841345,
      "loss": 1.7752,
      "step": 4341
    },
    {
      "epoch": 0.5557404326123128,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0008685490678173704,
      "loss": 1.708,
      "step": 4342
    },
    {
      "epoch": 0.5558684244208371,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0008681381089425815,
      "loss": 1.1525,
      "step": 4343
    },
    {
      "epoch": 0.5559964162293614,
      "grad_norm": 0.796875,
      "learning_rate": 0.0008677271727303981,
      "loss": 1.8011,
      "step": 4344
    },
    {
      "epoch": 0.5561244080378855,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0008673162592514465,
      "loss": 1.6868,
      "step": 4345
    },
    {
      "epoch": 0.5562523998464098,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008669053685763482,
      "loss": 1.1865,
      "step": 4346
    },
    {
      "epoch": 0.5563803916549341,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0008664945007757218,
      "loss": 1.8071,
      "step": 4347
    },
    {
      "epoch": 0.5565083834634583,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0008660836559201815,
      "loss": 0.9648,
      "step": 4348
    },
    {
      "epoch": 0.5566363752719826,
      "grad_norm": 0.71875,
      "learning_rate": 0.0008656728340803382,
      "loss": 1.1396,
      "step": 4349
    },
    {
      "epoch": 0.5567643670805068,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0008652620353267971,
      "loss": 0.838,
      "step": 4350
    },
    {
      "epoch": 0.5568923588890311,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0008648512597301613,
      "loss": 1.5257,
      "step": 4351
    },
    {
      "epoch": 0.5570203506975554,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0008644405073610296,
      "loss": 1.5273,
      "step": 4352
    },
    {
      "epoch": 0.5571483425060796,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0008640297782899959,
      "loss": 1.331,
      "step": 4353
    },
    {
      "epoch": 0.5572763343146039,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0008636190725876514,
      "loss": 1.6568,
      "step": 4354
    },
    {
      "epoch": 0.5574043261231281,
      "grad_norm": 0.6875,
      "learning_rate": 0.0008632083903245816,
      "loss": 1.4478,
      "step": 4355
    },
    {
      "epoch": 0.5575323179316524,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0008627977315713693,
      "loss": 1.513,
      "step": 4356
    },
    {
      "epoch": 0.5576603097401767,
      "grad_norm": 0.6484375,
      "learning_rate": 0.000862387096398593,
      "loss": 1.4821,
      "step": 4357
    },
    {
      "epoch": 0.5577883015487008,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0008619764848768271,
      "loss": 2.5756,
      "step": 4358
    },
    {
      "epoch": 0.5579162933572251,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0008615658970766414,
      "loss": 1.5676,
      "step": 4359
    },
    {
      "epoch": 0.5580442851657494,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0008611553330686024,
      "loss": 1.0942,
      "step": 4360
    },
    {
      "epoch": 0.5581722769742736,
      "grad_norm": 0.828125,
      "learning_rate": 0.0008607447929232722,
      "loss": 1.7723,
      "step": 4361
    },
    {
      "epoch": 0.5583002687827979,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0008603342767112086,
      "loss": 1.1241,
      "step": 4362
    },
    {
      "epoch": 0.5584282605913221,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0008599237845029657,
      "loss": 1.461,
      "step": 4363
    },
    {
      "epoch": 0.5585562523998464,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0008595133163690931,
      "loss": 1.5682,
      "step": 4364
    },
    {
      "epoch": 0.5586842442083707,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008591028723801364,
      "loss": 1.2932,
      "step": 4365
    },
    {
      "epoch": 0.5588122360168949,
      "grad_norm": 0.79296875,
      "learning_rate": 0.000858692452606637,
      "loss": 1.8981,
      "step": 4366
    },
    {
      "epoch": 0.5589402278254192,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0008582820571191328,
      "loss": 1.0333,
      "step": 4367
    },
    {
      "epoch": 0.5590682196339434,
      "grad_norm": 0.56640625,
      "learning_rate": 0.000857871685988156,
      "loss": 1.2186,
      "step": 4368
    },
    {
      "epoch": 0.5591962114424677,
      "grad_norm": 0.75,
      "learning_rate": 0.000857461339284236,
      "loss": 1.3539,
      "step": 4369
    },
    {
      "epoch": 0.559324203250992,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008570510170778974,
      "loss": 1.2119,
      "step": 4370
    },
    {
      "epoch": 0.5594521950595162,
      "grad_norm": 0.490234375,
      "learning_rate": 0.000856640719439661,
      "loss": 0.8532,
      "step": 4371
    },
    {
      "epoch": 0.5595801868680405,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0008562304464400432,
      "loss": 1.4385,
      "step": 4372
    },
    {
      "epoch": 0.5597081786765647,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0008558201981495554,
      "loss": 1.87,
      "step": 4373
    },
    {
      "epoch": 0.5598361704850889,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0008554099746387058,
      "loss": 1.352,
      "step": 4374
    },
    {
      "epoch": 0.5599641622936132,
      "grad_norm": 0.609375,
      "learning_rate": 0.0008549997759779981,
      "loss": 1.0808,
      "step": 4375
    },
    {
      "epoch": 0.5600921541021374,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0008545896022379319,
      "loss": 1.636,
      "step": 4376
    },
    {
      "epoch": 0.5602201459106617,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0008541794534890014,
      "loss": 1.6119,
      "step": 4377
    },
    {
      "epoch": 0.560348137719186,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0008537693298016977,
      "loss": 1.1804,
      "step": 4378
    },
    {
      "epoch": 0.5604761295277102,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0008533592312465071,
      "loss": 1.0555,
      "step": 4379
    },
    {
      "epoch": 0.5606041213362345,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008529491578939122,
      "loss": 1.2443,
      "step": 4380
    },
    {
      "epoch": 0.5607321131447587,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0008525391098143903,
      "loss": 1.3695,
      "step": 4381
    },
    {
      "epoch": 0.560860104953283,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0008521290870784147,
      "loss": 1.5425,
      "step": 4382
    },
    {
      "epoch": 0.5609880967618073,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0008517190897564546,
      "loss": 1.5795,
      "step": 4383
    },
    {
      "epoch": 0.5611160885703315,
      "grad_norm": 0.5625,
      "learning_rate": 0.0008513091179189748,
      "loss": 1.0565,
      "step": 4384
    },
    {
      "epoch": 0.5612440803788558,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0008508991716364359,
      "loss": 1.1234,
      "step": 4385
    },
    {
      "epoch": 0.5613720721873801,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0008504892509792931,
      "loss": 1.2966,
      "step": 4386
    },
    {
      "epoch": 0.5615000639959042,
      "grad_norm": 0.65625,
      "learning_rate": 0.0008500793560179984,
      "loss": 1.485,
      "step": 4387
    },
    {
      "epoch": 0.5616280558044285,
      "grad_norm": 0.765625,
      "learning_rate": 0.0008496694868229987,
      "loss": 1.4752,
      "step": 4388
    },
    {
      "epoch": 0.5617560476129527,
      "grad_norm": 0.83203125,
      "learning_rate": 0.000849259643464737,
      "loss": 1.3528,
      "step": 4389
    },
    {
      "epoch": 0.561884039421477,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008488498260136512,
      "loss": 1.287,
      "step": 4390
    },
    {
      "epoch": 0.5620120312300013,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0008484400345401751,
      "loss": 1.7458,
      "step": 4391
    },
    {
      "epoch": 0.5621400230385255,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0008480302691147381,
      "loss": 1.3392,
      "step": 4392
    },
    {
      "epoch": 0.5622680148470498,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008476205298077649,
      "loss": 1.1461,
      "step": 4393
    },
    {
      "epoch": 0.562396006655574,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0008472108166896765,
      "loss": 1.1251,
      "step": 4394
    },
    {
      "epoch": 0.5625239984640983,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0008468011298308879,
      "loss": 1.1454,
      "step": 4395
    },
    {
      "epoch": 0.5626519902726226,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0008463914693018105,
      "loss": 0.8711,
      "step": 4396
    },
    {
      "epoch": 0.5627799820811468,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0008459818351728514,
      "loss": 1.2614,
      "step": 4397
    },
    {
      "epoch": 0.5629079738896711,
      "grad_norm": 0.609375,
      "learning_rate": 0.0008455722275144134,
      "loss": 1.4918,
      "step": 4398
    },
    {
      "epoch": 0.5630359656981954,
      "grad_norm": 0.671875,
      "learning_rate": 0.0008451626463968933,
      "loss": 1.4231,
      "step": 4399
    },
    {
      "epoch": 0.5631639575067195,
      "grad_norm": 0.609375,
      "learning_rate": 0.0008447530918906845,
      "loss": 1.5839,
      "step": 4400
    },
    {
      "epoch": 0.5632919493152438,
      "grad_norm": 0.671875,
      "learning_rate": 0.0008443435640661756,
      "loss": 1.1573,
      "step": 4401
    },
    {
      "epoch": 0.563419941123768,
      "grad_norm": 0.78125,
      "learning_rate": 0.0008439340629937509,
      "loss": 1.8678,
      "step": 4402
    },
    {
      "epoch": 0.5635479329322923,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008435245887437896,
      "loss": 0.9931,
      "step": 4403
    },
    {
      "epoch": 0.5636759247408166,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0008431151413866666,
      "loss": 1.3119,
      "step": 4404
    },
    {
      "epoch": 0.5638039165493408,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008427057209927517,
      "loss": 1.0753,
      "step": 4405
    },
    {
      "epoch": 0.5639319083578651,
      "grad_norm": 0.63671875,
      "learning_rate": 0.000842296327632411,
      "loss": 1.16,
      "step": 4406
    },
    {
      "epoch": 0.5640599001663894,
      "grad_norm": 0.984375,
      "learning_rate": 0.000841886961376005,
      "loss": 1.3551,
      "step": 4407
    },
    {
      "epoch": 0.5641878919749136,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00084147762229389,
      "loss": 1.9919,
      "step": 4408
    },
    {
      "epoch": 0.5643158837834379,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0008410683104564176,
      "loss": 1.0035,
      "step": 4409
    },
    {
      "epoch": 0.5644438755919621,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0008406590259339345,
      "loss": 0.8481,
      "step": 4410
    },
    {
      "epoch": 0.5645718674004864,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008402497687967836,
      "loss": 1.5049,
      "step": 4411
    },
    {
      "epoch": 0.5646998592090107,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008398405391153012,
      "loss": 1.4768,
      "step": 4412
    },
    {
      "epoch": 0.5648278510175349,
      "grad_norm": 0.70703125,
      "learning_rate": 0.000839431336959821,
      "loss": 1.7466,
      "step": 4413
    },
    {
      "epoch": 0.5649558428260592,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008390221624006706,
      "loss": 0.657,
      "step": 4414
    },
    {
      "epoch": 0.5650838346345833,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008386130155081735,
      "loss": 0.9652,
      "step": 4415
    },
    {
      "epoch": 0.5652118264431076,
      "grad_norm": 0.71875,
      "learning_rate": 0.0008382038963526486,
      "loss": 1.2852,
      "step": 4416
    },
    {
      "epoch": 0.5653398182516319,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0008377948050044086,
      "loss": 1.4829,
      "step": 4417
    },
    {
      "epoch": 0.5654678100601561,
      "grad_norm": 0.671875,
      "learning_rate": 0.0008373857415337634,
      "loss": 1.5093,
      "step": 4418
    },
    {
      "epoch": 0.5655958018686804,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008369767060110164,
      "loss": 1.2521,
      "step": 4419
    },
    {
      "epoch": 0.5657237936772047,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0008365676985064684,
      "loss": 1.5782,
      "step": 4420
    },
    {
      "epoch": 0.5658517854857289,
      "grad_norm": 0.765625,
      "learning_rate": 0.0008361587190904123,
      "loss": 1.8736,
      "step": 4421
    },
    {
      "epoch": 0.5659797772942532,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0008357497678331389,
      "loss": 1.6473,
      "step": 4422
    },
    {
      "epoch": 0.5661077691027774,
      "grad_norm": 0.6875,
      "learning_rate": 0.0008353408448049327,
      "loss": 1.3583,
      "step": 4423
    },
    {
      "epoch": 0.5662357609113017,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0008349319500760738,
      "loss": 1.2999,
      "step": 4424
    },
    {
      "epoch": 0.566363752719826,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0008345230837168378,
      "loss": 1.6895,
      "step": 4425
    },
    {
      "epoch": 0.5664917445283502,
      "grad_norm": 0.953125,
      "learning_rate": 0.0008341142457974944,
      "loss": 1.9241,
      "step": 4426
    },
    {
      "epoch": 0.5666197363368745,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0008337054363883094,
      "loss": 1.1482,
      "step": 4427
    },
    {
      "epoch": 0.5667477281453986,
      "grad_norm": 0.640625,
      "learning_rate": 0.0008332966555595431,
      "loss": 1.5623,
      "step": 4428
    },
    {
      "epoch": 0.5668757199539229,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0008328879033814516,
      "loss": 1.1108,
      "step": 4429
    },
    {
      "epoch": 0.5670037117624472,
      "grad_norm": 0.62109375,
      "learning_rate": 0.000832479179924285,
      "loss": 1.2429,
      "step": 4430
    },
    {
      "epoch": 0.5671317035709714,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008320704852582895,
      "loss": 1.0244,
      "step": 4431
    },
    {
      "epoch": 0.5672596953794957,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008316618194537054,
      "loss": 1.1599,
      "step": 4432
    },
    {
      "epoch": 0.56738768718802,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0008312531825807695,
      "loss": 1.4514,
      "step": 4433
    },
    {
      "epoch": 0.5675156789965442,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0008308445747097115,
      "loss": 1.5856,
      "step": 4434
    },
    {
      "epoch": 0.5676436708050685,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0008304359959107578,
      "loss": 1.8614,
      "step": 4435
    },
    {
      "epoch": 0.5677716626135927,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0008300274462541295,
      "loss": 1.1103,
      "step": 4436
    },
    {
      "epoch": 0.567899654422117,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0008296189258100422,
      "loss": 0.9999,
      "step": 4437
    },
    {
      "epoch": 0.5680276462306413,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0008292104346487075,
      "loss": 1.562,
      "step": 4438
    },
    {
      "epoch": 0.5681556380391655,
      "grad_norm": 0.640625,
      "learning_rate": 0.00082880197284033,
      "loss": 1.3044,
      "step": 4439
    },
    {
      "epoch": 0.5682836298476898,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0008283935404551113,
      "loss": 1.7831,
      "step": 4440
    },
    {
      "epoch": 0.568411621656214,
      "grad_norm": 0.609375,
      "learning_rate": 0.0008279851375632468,
      "loss": 1.2123,
      "step": 4441
    },
    {
      "epoch": 0.5685396134647382,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0008275767642349276,
      "loss": 1.3365,
      "step": 4442
    },
    {
      "epoch": 0.5686676052732625,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0008271684205403386,
      "loss": 1.1675,
      "step": 4443
    },
    {
      "epoch": 0.5687955970817867,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0008267601065496611,
      "loss": 1.3143,
      "step": 4444
    },
    {
      "epoch": 0.568923588890311,
      "grad_norm": 0.71875,
      "learning_rate": 0.0008263518223330697,
      "loss": 1.735,
      "step": 4445
    },
    {
      "epoch": 0.5690515806988353,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008259435679607351,
      "loss": 1.312,
      "step": 4446
    },
    {
      "epoch": 0.5691795725073595,
      "grad_norm": 0.640625,
      "learning_rate": 0.0008255353435028227,
      "loss": 1.4377,
      "step": 4447
    },
    {
      "epoch": 0.5693075643158838,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0008251271490294918,
      "loss": 1.2474,
      "step": 4448
    },
    {
      "epoch": 0.569435556124408,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0008247189846108978,
      "loss": 1.9107,
      "step": 4449
    },
    {
      "epoch": 0.5695635479329323,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00082431085031719,
      "loss": 1.4296,
      "step": 4450
    },
    {
      "epoch": 0.5696915397414566,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0008239027462185134,
      "loss": 1.5815,
      "step": 4451
    },
    {
      "epoch": 0.5698195315499808,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0008234946723850067,
      "loss": 1.0631,
      "step": 4452
    },
    {
      "epoch": 0.5699475233585051,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0008230866288868044,
      "loss": 1.3246,
      "step": 4453
    },
    {
      "epoch": 0.5700755151670293,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008226786157940352,
      "loss": 0.7762,
      "step": 4454
    },
    {
      "epoch": 0.5702035069755536,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008222706331768231,
      "loss": 1.1439,
      "step": 4455
    },
    {
      "epoch": 0.5703314987840779,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0008218626811052865,
      "loss": 1.2006,
      "step": 4456
    },
    {
      "epoch": 0.570459490592602,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0008214547596495377,
      "loss": 1.515,
      "step": 4457
    },
    {
      "epoch": 0.5705874824011263,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0008210468688796857,
      "loss": 1.4144,
      "step": 4458
    },
    {
      "epoch": 0.5707154742096506,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0008206390088658326,
      "loss": 1.4718,
      "step": 4459
    },
    {
      "epoch": 0.5708434660181748,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0008202311796780766,
      "loss": 1.1659,
      "step": 4460
    },
    {
      "epoch": 0.5709714578266991,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0008198233813865083,
      "loss": 1.9878,
      "step": 4461
    },
    {
      "epoch": 0.5710994496352233,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0008194156140612156,
      "loss": 1.8078,
      "step": 4462
    },
    {
      "epoch": 0.5712274414437476,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0008190078777722794,
      "loss": 1.7769,
      "step": 4463
    },
    {
      "epoch": 0.5713554332522719,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0008186001725897761,
      "loss": 1.6706,
      "step": 4464
    },
    {
      "epoch": 0.5714834250607961,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0008181924985837761,
      "loss": 1.6283,
      "step": 4465
    },
    {
      "epoch": 0.5716114168693204,
      "grad_norm": 0.640625,
      "learning_rate": 0.000817784855824345,
      "loss": 1.098,
      "step": 4466
    },
    {
      "epoch": 0.5717394086778446,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008173772443815429,
      "loss": 1.3043,
      "step": 4467
    },
    {
      "epoch": 0.5718674004863689,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0008169696643254243,
      "loss": 1.5847,
      "step": 4468
    },
    {
      "epoch": 0.5719953922948932,
      "grad_norm": 0.625,
      "learning_rate": 0.0008165621157260387,
      "loss": 1.5264,
      "step": 4469
    },
    {
      "epoch": 0.5721233841034173,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0008161545986534297,
      "loss": 1.3202,
      "step": 4470
    },
    {
      "epoch": 0.5722513759119416,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0008157471131776356,
      "loss": 1.1569,
      "step": 4471
    },
    {
      "epoch": 0.5723793677204659,
      "grad_norm": 0.609375,
      "learning_rate": 0.0008153396593686897,
      "loss": 1.3465,
      "step": 4472
    },
    {
      "epoch": 0.5725073595289901,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0008149322372966195,
      "loss": 1.4418,
      "step": 4473
    },
    {
      "epoch": 0.5726353513375144,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0008145248470314468,
      "loss": 1.5695,
      "step": 4474
    },
    {
      "epoch": 0.5727633431460386,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0008141174886431885,
      "loss": 1.3151,
      "step": 4475
    },
    {
      "epoch": 0.5728913349545629,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0008137101622018556,
      "loss": 1.2453,
      "step": 4476
    },
    {
      "epoch": 0.5730193267630872,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008133028677774539,
      "loss": 1.0484,
      "step": 4477
    },
    {
      "epoch": 0.5731473185716114,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0008128956054399838,
      "loss": 0.6781,
      "step": 4478
    },
    {
      "epoch": 0.5732753103801357,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0008124883752594393,
      "loss": 1.2821,
      "step": 4479
    },
    {
      "epoch": 0.57340330218866,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0008120811773058097,
      "loss": 1.7967,
      "step": 4480
    },
    {
      "epoch": 0.5735312939971842,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0008116740116490787,
      "loss": 1.5003,
      "step": 4481
    },
    {
      "epoch": 0.5736592858057085,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0008112668783592249,
      "loss": 1.7481,
      "step": 4482
    },
    {
      "epoch": 0.5737872776142326,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0008108597775062198,
      "loss": 1.7487,
      "step": 4483
    },
    {
      "epoch": 0.573915269422757,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0008104527091600306,
      "loss": 1.5298,
      "step": 4484
    },
    {
      "epoch": 0.5740432612312812,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0008100456733906189,
      "loss": 1.0491,
      "step": 4485
    },
    {
      "epoch": 0.5741712530398054,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00080963867026794,
      "loss": 1.5736,
      "step": 4486
    },
    {
      "epoch": 0.5742992448483297,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0008092316998619444,
      "loss": 1.8488,
      "step": 4487
    },
    {
      "epoch": 0.5744272366568539,
      "grad_norm": 0.71875,
      "learning_rate": 0.0008088247622425762,
      "loss": 1.6848,
      "step": 4488
    },
    {
      "epoch": 0.5745552284653782,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0008084178574797743,
      "loss": 1.6393,
      "step": 4489
    },
    {
      "epoch": 0.5746832202739025,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0008080109856434721,
      "loss": 1.6666,
      "step": 4490
    },
    {
      "epoch": 0.5748112120824267,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0008076041468035971,
      "loss": 1.2944,
      "step": 4491
    },
    {
      "epoch": 0.574939203890951,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0008071973410300709,
      "loss": 1.832,
      "step": 4492
    },
    {
      "epoch": 0.5750671956994753,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0008067905683928099,
      "loss": 1.16,
      "step": 4493
    },
    {
      "epoch": 0.5751951875079995,
      "grad_norm": 0.78125,
      "learning_rate": 0.0008063838289617243,
      "loss": 1.7755,
      "step": 4494
    },
    {
      "epoch": 0.5753231793165238,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0008059771228067197,
      "loss": 1.2727,
      "step": 4495
    },
    {
      "epoch": 0.575451171125048,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0008055704499976936,
      "loss": 1.4275,
      "step": 4496
    },
    {
      "epoch": 0.5755791629335723,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0008051638106045406,
      "loss": 1.3017,
      "step": 4497
    },
    {
      "epoch": 0.5757071547420965,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0008047572046971479,
      "loss": 1.251,
      "step": 4498
    },
    {
      "epoch": 0.5758351465506207,
      "grad_norm": 0.59375,
      "learning_rate": 0.0008043506323453975,
      "loss": 1.2082,
      "step": 4499
    },
    {
      "epoch": 0.575963138359145,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0008039440936191655,
      "loss": 0.8828,
      "step": 4500
    },
    {
      "epoch": 0.5760911301676692,
      "grad_norm": 0.671875,
      "learning_rate": 0.0008035375885883217,
      "loss": 1.2343,
      "step": 4501
    },
    {
      "epoch": 0.5762191219761935,
      "grad_norm": 0.578125,
      "learning_rate": 0.0008031311173227309,
      "loss": 0.8972,
      "step": 4502
    },
    {
      "epoch": 0.5763471137847178,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0008027246798922513,
      "loss": 1.3374,
      "step": 4503
    },
    {
      "epoch": 0.576475105593242,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0008023182763667369,
      "loss": 1.0575,
      "step": 4504
    },
    {
      "epoch": 0.5766030974017663,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0008019119068160338,
      "loss": 1.202,
      "step": 4505
    },
    {
      "epoch": 0.5767310892102906,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0008015055713099833,
      "loss": 1.2835,
      "step": 4506
    },
    {
      "epoch": 0.5768590810188148,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0008010992699184208,
      "loss": 1.0384,
      "step": 4507
    },
    {
      "epoch": 0.5769870728273391,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0008006930027111761,
      "loss": 1.6066,
      "step": 4508
    },
    {
      "epoch": 0.5771150646358633,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0008002867697580727,
      "loss": 1.1585,
      "step": 4509
    },
    {
      "epoch": 0.5772430564443876,
      "grad_norm": 0.65234375,
      "learning_rate": 0.000799880571128928,
      "loss": 1.1551,
      "step": 4510
    },
    {
      "epoch": 0.5773710482529119,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0007994744068935541,
      "loss": 1.3523,
      "step": 4511
    },
    {
      "epoch": 0.577499040061436,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0007990682771217567,
      "loss": 1.4698,
      "step": 4512
    },
    {
      "epoch": 0.5776270318699603,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0007986621818833364,
      "loss": 1.22,
      "step": 4513
    },
    {
      "epoch": 0.5777550236784845,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0007982561212480863,
      "loss": 1.5497,
      "step": 4514
    },
    {
      "epoch": 0.5778830154870088,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0007978500952857952,
      "loss": 1.1446,
      "step": 4515
    },
    {
      "epoch": 0.5780110072955331,
      "grad_norm": 0.67578125,
      "learning_rate": 0.000797444104066245,
      "loss": 1.9027,
      "step": 4516
    },
    {
      "epoch": 0.5781389991040573,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0007970381476592125,
      "loss": 1.3971,
      "step": 4517
    },
    {
      "epoch": 0.5782669909125816,
      "grad_norm": 0.609375,
      "learning_rate": 0.0007966322261344665,
      "loss": 1.2502,
      "step": 4518
    },
    {
      "epoch": 0.5783949827211059,
      "grad_norm": 0.90625,
      "learning_rate": 0.0007962263395617724,
      "loss": 1.7286,
      "step": 4519
    },
    {
      "epoch": 0.5785229745296301,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0007958204880108879,
      "loss": 1.2071,
      "step": 4520
    },
    {
      "epoch": 0.5786509663381544,
      "grad_norm": 0.671875,
      "learning_rate": 0.0007954146715515654,
      "loss": 1.4566,
      "step": 4521
    },
    {
      "epoch": 0.5787789581466786,
      "grad_norm": 0.671875,
      "learning_rate": 0.0007950088902535513,
      "loss": 1.4658,
      "step": 4522
    },
    {
      "epoch": 0.5789069499552029,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0007946031441865848,
      "loss": 1.4755,
      "step": 4523
    },
    {
      "epoch": 0.5790349417637272,
      "grad_norm": 0.59375,
      "learning_rate": 0.0007941974334204008,
      "loss": 0.8621,
      "step": 4524
    },
    {
      "epoch": 0.5791629335722513,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0007937917580247265,
      "loss": 1.0185,
      "step": 4525
    },
    {
      "epoch": 0.5792909253807756,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0007933861180692849,
      "loss": 1.6932,
      "step": 4526
    },
    {
      "epoch": 0.5794189171892998,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0007929805136237908,
      "loss": 1.27,
      "step": 4527
    },
    {
      "epoch": 0.5795469089978241,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0007925749447579544,
      "loss": 1.5213,
      "step": 4528
    },
    {
      "epoch": 0.5796749008063484,
      "grad_norm": 0.66015625,
      "learning_rate": 0.000792169411541479,
      "loss": 1.3332,
      "step": 4529
    },
    {
      "epoch": 0.5798028926148726,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0007917639140440622,
      "loss": 1.2427,
      "step": 4530
    },
    {
      "epoch": 0.5799308844233969,
      "grad_norm": 0.640625,
      "learning_rate": 0.0007913584523353957,
      "loss": 1.2353,
      "step": 4531
    },
    {
      "epoch": 0.5800588762319212,
      "grad_norm": 0.91796875,
      "learning_rate": 0.000790953026485164,
      "loss": 1.9407,
      "step": 4532
    },
    {
      "epoch": 0.5801868680404454,
      "grad_norm": 1.390625,
      "learning_rate": 0.0007905476365630465,
      "loss": 1.4423,
      "step": 4533
    },
    {
      "epoch": 0.5803148598489697,
      "grad_norm": 1.015625,
      "learning_rate": 0.000790142282638716,
      "loss": 1.1971,
      "step": 4534
    },
    {
      "epoch": 0.5804428516574939,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0007897369647818394,
      "loss": 0.9681,
      "step": 4535
    },
    {
      "epoch": 0.5805708434660182,
      "grad_norm": 0.609375,
      "learning_rate": 0.0007893316830620765,
      "loss": 0.99,
      "step": 4536
    },
    {
      "epoch": 0.5806988352745425,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0007889264375490818,
      "loss": 1.693,
      "step": 4537
    },
    {
      "epoch": 0.5808268270830667,
      "grad_norm": 0.59375,
      "learning_rate": 0.0007885212283125036,
      "loss": 1.1069,
      "step": 4538
    },
    {
      "epoch": 0.580954818891591,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0007881160554219835,
      "loss": 1.0052,
      "step": 4539
    },
    {
      "epoch": 0.5810828107001151,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0007877109189471575,
      "loss": 1.3596,
      "step": 4540
    },
    {
      "epoch": 0.5812108025086394,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0007873058189576537,
      "loss": 2.118,
      "step": 4541
    },
    {
      "epoch": 0.5813387943171637,
      "grad_norm": 0.66796875,
      "learning_rate": 0.000786900755523096,
      "loss": 1.2682,
      "step": 4542
    },
    {
      "epoch": 0.5814667861256879,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007864957287131012,
      "loss": 2.0278,
      "step": 4543
    },
    {
      "epoch": 0.5815947779342122,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0007860907385972796,
      "loss": 1.5356,
      "step": 4544
    },
    {
      "epoch": 0.5817227697427365,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0007856857852452349,
      "loss": 1.3216,
      "step": 4545
    },
    {
      "epoch": 0.5818507615512607,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0007852808687265653,
      "loss": 1.246,
      "step": 4546
    },
    {
      "epoch": 0.581978753359785,
      "grad_norm": 0.6953125,
      "learning_rate": 0.000784875989110862,
      "loss": 1.3242,
      "step": 4547
    },
    {
      "epoch": 0.5821067451683092,
      "grad_norm": 0.609375,
      "learning_rate": 0.0007844711464677106,
      "loss": 1.3714,
      "step": 4548
    },
    {
      "epoch": 0.5822347369768335,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007840663408666893,
      "loss": 2.1842,
      "step": 4549
    },
    {
      "epoch": 0.5823627287853578,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0007836615723773709,
      "loss": 1.0917,
      "step": 4550
    },
    {
      "epoch": 0.582490720593882,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0007832568410693213,
      "loss": 0.6619,
      "step": 4551
    },
    {
      "epoch": 0.5826187124024063,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0007828521470121001,
      "loss": 2.2204,
      "step": 4552
    },
    {
      "epoch": 0.5827467042109306,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0007824474902752608,
      "loss": 0.9002,
      "step": 4553
    },
    {
      "epoch": 0.5828746960194547,
      "grad_norm": 0.671875,
      "learning_rate": 0.0007820428709283499,
      "loss": 1.4881,
      "step": 4554
    },
    {
      "epoch": 0.583002687827979,
      "grad_norm": 0.578125,
      "learning_rate": 0.0007816382890409079,
      "loss": 0.8103,
      "step": 4555
    },
    {
      "epoch": 0.5831306796365032,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0007812337446824689,
      "loss": 1.372,
      "step": 4556
    },
    {
      "epoch": 0.5832586714450275,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0007808292379225604,
      "loss": 1.2082,
      "step": 4557
    },
    {
      "epoch": 0.5833866632535518,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0007804247688307031,
      "loss": 1.1683,
      "step": 4558
    },
    {
      "epoch": 0.583514655062076,
      "grad_norm": 0.64453125,
      "learning_rate": 0.000780020337476412,
      "loss": 0.9364,
      "step": 4559
    },
    {
      "epoch": 0.5836426468706003,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0007796159439291949,
      "loss": 1.1179,
      "step": 4560
    },
    {
      "epoch": 0.5837706386791245,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0007792115882585536,
      "loss": 0.9364,
      "step": 4561
    },
    {
      "epoch": 0.5838986304876488,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0007788072705339834,
      "loss": 1.8486,
      "step": 4562
    },
    {
      "epoch": 0.5840266222961731,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0007784029908249723,
      "loss": 1.9848,
      "step": 4563
    },
    {
      "epoch": 0.5841546141046973,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0007779987492010021,
      "loss": 1.1959,
      "step": 4564
    },
    {
      "epoch": 0.5842826059132216,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0007775945457315492,
      "loss": 1.0815,
      "step": 4565
    },
    {
      "epoch": 0.5844105977217459,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0007771903804860825,
      "loss": 1.2552,
      "step": 4566
    },
    {
      "epoch": 0.58453858953027,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0007767862535340635,
      "loss": 1.1754,
      "step": 4567
    },
    {
      "epoch": 0.5846665813387943,
      "grad_norm": 0.671875,
      "learning_rate": 0.0007763821649449483,
      "loss": 1.2125,
      "step": 4568
    },
    {
      "epoch": 0.5847945731473185,
      "grad_norm": 0.671875,
      "learning_rate": 0.0007759781147881863,
      "loss": 1.37,
      "step": 4569
    },
    {
      "epoch": 0.5849225649558428,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00077557410313322,
      "loss": 1.2623,
      "step": 4570
    },
    {
      "epoch": 0.5850505567643671,
      "grad_norm": 0.71875,
      "learning_rate": 0.0007751701300494856,
      "loss": 1.487,
      "step": 4571
    },
    {
      "epoch": 0.5851785485728913,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0007747661956064118,
      "loss": 0.8685,
      "step": 4572
    },
    {
      "epoch": 0.5853065403814156,
      "grad_norm": 0.765625,
      "learning_rate": 0.0007743622998734217,
      "loss": 1.3459,
      "step": 4573
    },
    {
      "epoch": 0.5854345321899398,
      "grad_norm": 0.546875,
      "learning_rate": 0.0007739584429199314,
      "loss": 1.0747,
      "step": 4574
    },
    {
      "epoch": 0.5855625239984641,
      "grad_norm": 0.65625,
      "learning_rate": 0.0007735546248153503,
      "loss": 1.208,
      "step": 4575
    },
    {
      "epoch": 0.5856905158069884,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0007731508456290805,
      "loss": 1.0512,
      "step": 4576
    },
    {
      "epoch": 0.5858185076155126,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0007727471054305188,
      "loss": 1.3662,
      "step": 4577
    },
    {
      "epoch": 0.5859464994240369,
      "grad_norm": 0.52734375,
      "learning_rate": 0.000772343404289054,
      "loss": 0.9988,
      "step": 4578
    },
    {
      "epoch": 0.5860744912325612,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0007719397422740691,
      "loss": 0.8572,
      "step": 4579
    },
    {
      "epoch": 0.5862024830410854,
      "grad_norm": 0.828125,
      "learning_rate": 0.0007715361194549394,
      "loss": 1.4125,
      "step": 4580
    },
    {
      "epoch": 0.5863304748496097,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0007711325359010344,
      "loss": 1.1526,
      "step": 4581
    },
    {
      "epoch": 0.5864584666581338,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0007707289916817164,
      "loss": 1.7563,
      "step": 4582
    },
    {
      "epoch": 0.5865864584666581,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0007703254868663412,
      "loss": 1.9295,
      "step": 4583
    },
    {
      "epoch": 0.5867144502751824,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0007699220215242577,
      "loss": 1.5186,
      "step": 4584
    },
    {
      "epoch": 0.5868424420837066,
      "grad_norm": 0.609375,
      "learning_rate": 0.0007695185957248074,
      "loss": 1.263,
      "step": 4585
    },
    {
      "epoch": 0.5869704338922309,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0007691152095373259,
      "loss": 0.9177,
      "step": 4586
    },
    {
      "epoch": 0.5870984257007551,
      "grad_norm": 0.625,
      "learning_rate": 0.0007687118630311417,
      "loss": 1.3851,
      "step": 4587
    },
    {
      "epoch": 0.5872264175092794,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0007683085562755769,
      "loss": 1.2397,
      "step": 4588
    },
    {
      "epoch": 0.5873544093178037,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0007679052893399454,
      "loss": 1.3353,
      "step": 4589
    },
    {
      "epoch": 0.5874824011263279,
      "grad_norm": 0.671875,
      "learning_rate": 0.0007675020622935557,
      "loss": 1.7398,
      "step": 4590
    },
    {
      "epoch": 0.5876103929348522,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0007670988752057087,
      "loss": 1.2832,
      "step": 4591
    },
    {
      "epoch": 0.5877383847433765,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0007666957281456989,
      "loss": 1.3132,
      "step": 4592
    },
    {
      "epoch": 0.5878663765519007,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0007662926211828136,
      "loss": 1.7802,
      "step": 4593
    },
    {
      "epoch": 0.587994368360425,
      "grad_norm": 0.69140625,
      "learning_rate": 0.000765889554386333,
      "loss": 1.0574,
      "step": 4594
    },
    {
      "epoch": 0.5881223601689491,
      "grad_norm": 0.75390625,
      "learning_rate": 0.000765486527825531,
      "loss": 2.0212,
      "step": 4595
    },
    {
      "epoch": 0.5882503519774734,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0007650835415696739,
      "loss": 1.1435,
      "step": 4596
    },
    {
      "epoch": 0.5883783437859977,
      "grad_norm": 0.58984375,
      "learning_rate": 0.000764680595688022,
      "loss": 1.0434,
      "step": 4597
    },
    {
      "epoch": 0.5885063355945219,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0007642776902498275,
      "loss": 1.4035,
      "step": 4598
    },
    {
      "epoch": 0.5886343274030462,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0007638748253243365,
      "loss": 1.237,
      "step": 4599
    },
    {
      "epoch": 0.5887623192115704,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0007634720009807879,
      "loss": 1.3477,
      "step": 4600
    },
    {
      "epoch": 0.5888903110200947,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0007630692172884137,
      "loss": 1.0663,
      "step": 4601
    },
    {
      "epoch": 0.589018302828619,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0007626664743164382,
      "loss": 1.2084,
      "step": 4602
    },
    {
      "epoch": 0.5891462946371432,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0007622637721340801,
      "loss": 1.2349,
      "step": 4603
    },
    {
      "epoch": 0.5892742864456675,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0007618611108105497,
      "loss": 1.2576,
      "step": 4604
    },
    {
      "epoch": 0.5894022782541918,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0007614584904150514,
      "loss": 1.5495,
      "step": 4605
    },
    {
      "epoch": 0.589530270062716,
      "grad_norm": 0.640625,
      "learning_rate": 0.000761055911016782,
      "loss": 1.0249,
      "step": 4606
    },
    {
      "epoch": 0.5896582618712403,
      "grad_norm": 0.6875,
      "learning_rate": 0.0007606533726849308,
      "loss": 1.4756,
      "step": 4607
    },
    {
      "epoch": 0.5897862536797644,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0007602508754886809,
      "loss": 0.9497,
      "step": 4608
    },
    {
      "epoch": 0.5899142454882887,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0007598484194972076,
      "loss": 0.8174,
      "step": 4609
    },
    {
      "epoch": 0.590042237296813,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0007594460047796806,
      "loss": 1.2791,
      "step": 4610
    },
    {
      "epoch": 0.5901702291053372,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00075904363140526,
      "loss": 1.8278,
      "step": 4611
    },
    {
      "epoch": 0.5902982209138615,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0007586412994431009,
      "loss": 1.0455,
      "step": 4612
    },
    {
      "epoch": 0.5904262127223857,
      "grad_norm": 0.734375,
      "learning_rate": 0.0007582390089623505,
      "loss": 1.6248,
      "step": 4613
    },
    {
      "epoch": 0.59055420453091,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0007578367600321489,
      "loss": 1.4511,
      "step": 4614
    },
    {
      "epoch": 0.5906821963394343,
      "grad_norm": 0.578125,
      "learning_rate": 0.0007574345527216293,
      "loss": 1.1509,
      "step": 4615
    },
    {
      "epoch": 0.5908101881479585,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0007570323870999172,
      "loss": 1.0214,
      "step": 4616
    },
    {
      "epoch": 0.5909381799564828,
      "grad_norm": 0.625,
      "learning_rate": 0.0007566302632361313,
      "loss": 1.3928,
      "step": 4617
    },
    {
      "epoch": 0.5910661717650071,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0007562281811993835,
      "loss": 1.1588,
      "step": 4618
    },
    {
      "epoch": 0.5911941635735313,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0007558261410587781,
      "loss": 1.121,
      "step": 4619
    },
    {
      "epoch": 0.5913221553820556,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0007554241428834118,
      "loss": 0.8999,
      "step": 4620
    },
    {
      "epoch": 0.5914501471905798,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0007550221867423745,
      "loss": 0.95,
      "step": 4621
    },
    {
      "epoch": 0.591578138999104,
      "grad_norm": 0.765625,
      "learning_rate": 0.0007546202727047494,
      "loss": 1.5063,
      "step": 4622
    },
    {
      "epoch": 0.5917061308076283,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0007542184008396116,
      "loss": 1.8455,
      "step": 4623
    },
    {
      "epoch": 0.5918341226161525,
      "grad_norm": 0.734375,
      "learning_rate": 0.0007538165712160298,
      "loss": 1.6098,
      "step": 4624
    },
    {
      "epoch": 0.5919621144246768,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0007534147839030638,
      "loss": 1.454,
      "step": 4625
    },
    {
      "epoch": 0.5920901062332011,
      "grad_norm": 0.609375,
      "learning_rate": 0.0007530130389697681,
      "loss": 1.1808,
      "step": 4626
    },
    {
      "epoch": 0.5922180980417253,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0007526113364851891,
      "loss": 1.485,
      "step": 4627
    },
    {
      "epoch": 0.5923460898502496,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0007522096765183662,
      "loss": 1.4046,
      "step": 4628
    },
    {
      "epoch": 0.5924740816587738,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0007518080591383302,
      "loss": 1.8756,
      "step": 4629
    },
    {
      "epoch": 0.5926020734672981,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0007514064844141061,
      "loss": 1.8065,
      "step": 4630
    },
    {
      "epoch": 0.5927300652758224,
      "grad_norm": 0.63671875,
      "learning_rate": 0.000751004952414711,
      "loss": 1.4318,
      "step": 4631
    },
    {
      "epoch": 0.5928580570843466,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0007506034632091549,
      "loss": 1.6261,
      "step": 4632
    },
    {
      "epoch": 0.5929860488928709,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0007502020168664399,
      "loss": 1.6221,
      "step": 4633
    },
    {
      "epoch": 0.5931140407013951,
      "grad_norm": 0.703125,
      "learning_rate": 0.0007498006134555613,
      "loss": 1.2176,
      "step": 4634
    },
    {
      "epoch": 0.5932420325099194,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0007493992530455067,
      "loss": 1.1739,
      "step": 4635
    },
    {
      "epoch": 0.5933700243184437,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0007489979357052563,
      "loss": 1.2085,
      "step": 4636
    },
    {
      "epoch": 0.5934980161269678,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0007485966615037833,
      "loss": 1.3527,
      "step": 4637
    },
    {
      "epoch": 0.5936260079354921,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0007481954305100527,
      "loss": 1.0749,
      "step": 4638
    },
    {
      "epoch": 0.5937539997440164,
      "grad_norm": 0.56640625,
      "learning_rate": 0.000747794242793023,
      "loss": 1.3163,
      "step": 4639
    },
    {
      "epoch": 0.5938819915525406,
      "grad_norm": 0.78125,
      "learning_rate": 0.0007473930984216446,
      "loss": 1.5381,
      "step": 4640
    },
    {
      "epoch": 0.5940099833610649,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0007469919974648609,
      "loss": 0.8487,
      "step": 4641
    },
    {
      "epoch": 0.5941379751695891,
      "grad_norm": 0.5625,
      "learning_rate": 0.0007465909399916073,
      "loss": 1.1553,
      "step": 4642
    },
    {
      "epoch": 0.5942659669781134,
      "grad_norm": 0.765625,
      "learning_rate": 0.000746189926070812,
      "loss": 1.677,
      "step": 4643
    },
    {
      "epoch": 0.5943939587866377,
      "grad_norm": 0.6328125,
      "learning_rate": 0.000745788955771396,
      "loss": 1.3615,
      "step": 4644
    },
    {
      "epoch": 0.5945219505951619,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0007453880291622726,
      "loss": 2.0083,
      "step": 4645
    },
    {
      "epoch": 0.5946499424036862,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0007449871463123477,
      "loss": 1.1726,
      "step": 4646
    },
    {
      "epoch": 0.5947779342122104,
      "grad_norm": 0.625,
      "learning_rate": 0.0007445863072905185,
      "loss": 0.8827,
      "step": 4647
    },
    {
      "epoch": 0.5949059260207347,
      "grad_norm": 0.640625,
      "learning_rate": 0.0007441855121656768,
      "loss": 1.1548,
      "step": 4648
    },
    {
      "epoch": 0.595033917829259,
      "grad_norm": 0.703125,
      "learning_rate": 0.0007437847610067052,
      "loss": 1.7901,
      "step": 4649
    },
    {
      "epoch": 0.5951619096377831,
      "grad_norm": 0.625,
      "learning_rate": 0.0007433840538824801,
      "loss": 1.0021,
      "step": 4650
    },
    {
      "epoch": 0.5952899014463074,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0007429833908618682,
      "loss": 1.3999,
      "step": 4651
    },
    {
      "epoch": 0.5954178932548317,
      "grad_norm": 0.609375,
      "learning_rate": 0.0007425827720137308,
      "loss": 1.0789,
      "step": 4652
    },
    {
      "epoch": 0.5955458850633559,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0007421821974069205,
      "loss": 1.7312,
      "step": 4653
    },
    {
      "epoch": 0.5956738768718802,
      "grad_norm": 0.671875,
      "learning_rate": 0.0007417816671102824,
      "loss": 1.2687,
      "step": 4654
    },
    {
      "epoch": 0.5958018686804044,
      "grad_norm": 0.78125,
      "learning_rate": 0.000741381181192655,
      "loss": 1.5126,
      "step": 4655
    },
    {
      "epoch": 0.5959298604889287,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0007409807397228671,
      "loss": 1.0053,
      "step": 4656
    },
    {
      "epoch": 0.596057852297453,
      "grad_norm": 0.65625,
      "learning_rate": 0.0007405803427697416,
      "loss": 1.4998,
      "step": 4657
    },
    {
      "epoch": 0.5961858441059772,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0007401799904020933,
      "loss": 1.8025,
      "step": 4658
    },
    {
      "epoch": 0.5963138359145015,
      "grad_norm": 0.703125,
      "learning_rate": 0.0007397796826887294,
      "loss": 1.0622,
      "step": 4659
    },
    {
      "epoch": 0.5964418277230257,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0007393794196984488,
      "loss": 1.1864,
      "step": 4660
    },
    {
      "epoch": 0.59656981953155,
      "grad_norm": 0.578125,
      "learning_rate": 0.0007389792015000434,
      "loss": 1.4319,
      "step": 4661
    },
    {
      "epoch": 0.5966978113400743,
      "grad_norm": 0.6875,
      "learning_rate": 0.0007385790281622973,
      "loss": 1.3739,
      "step": 4662
    },
    {
      "epoch": 0.5968258031485985,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0007381788997539868,
      "loss": 1.5197,
      "step": 4663
    },
    {
      "epoch": 0.5969537949571228,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0007377788163438803,
      "loss": 1.3252,
      "step": 4664
    },
    {
      "epoch": 0.597081786765647,
      "grad_norm": 0.625,
      "learning_rate": 0.0007373787780007386,
      "loss": 1.2227,
      "step": 4665
    },
    {
      "epoch": 0.5972097785741712,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0007369787847933149,
      "loss": 1.3285,
      "step": 4666
    },
    {
      "epoch": 0.5973377703826955,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0007365788367903543,
      "loss": 1.4279,
      "step": 4667
    },
    {
      "epoch": 0.5974657621912197,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0007361789340605952,
      "loss": 1.3899,
      "step": 4668
    },
    {
      "epoch": 0.597593753999744,
      "grad_norm": 0.609375,
      "learning_rate": 0.0007357790766727662,
      "loss": 1.389,
      "step": 4669
    },
    {
      "epoch": 0.5977217458082683,
      "grad_norm": 0.75,
      "learning_rate": 0.0007353792646955896,
      "loss": 1.1124,
      "step": 4670
    },
    {
      "epoch": 0.5978497376167925,
      "grad_norm": 0.609375,
      "learning_rate": 0.00073497949819778,
      "loss": 1.2151,
      "step": 4671
    },
    {
      "epoch": 0.5979777294253168,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0007345797772480439,
      "loss": 1.1354,
      "step": 4672
    },
    {
      "epoch": 0.598105721233841,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0007341801019150791,
      "loss": 1.2121,
      "step": 4673
    },
    {
      "epoch": 0.5982337130423653,
      "grad_norm": 0.5625,
      "learning_rate": 0.0007337804722675766,
      "loss": 0.9902,
      "step": 4674
    },
    {
      "epoch": 0.5983617048508896,
      "grad_norm": 0.65625,
      "learning_rate": 0.0007333808883742192,
      "loss": 1.476,
      "step": 4675
    },
    {
      "epoch": 0.5984896966594138,
      "grad_norm": 0.78125,
      "learning_rate": 0.0007329813503036822,
      "loss": 1.6588,
      "step": 4676
    },
    {
      "epoch": 0.5986176884679381,
      "grad_norm": 0.75,
      "learning_rate": 0.0007325818581246325,
      "loss": 2.3395,
      "step": 4677
    },
    {
      "epoch": 0.5987456802764624,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0007321824119057292,
      "loss": 1.1687,
      "step": 4678
    },
    {
      "epoch": 0.5988736720849865,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0007317830117156234,
      "loss": 1.2539,
      "step": 4679
    },
    {
      "epoch": 0.5990016638935108,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0007313836576229591,
      "loss": 1.6219,
      "step": 4680
    },
    {
      "epoch": 0.599129655702035,
      "grad_norm": 0.8125,
      "learning_rate": 0.0007309843496963715,
      "loss": 1.3939,
      "step": 4681
    },
    {
      "epoch": 0.5992576475105593,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0007305850880044881,
      "loss": 1.2223,
      "step": 4682
    },
    {
      "epoch": 0.5993856393190836,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0007301858726159282,
      "loss": 1.7117,
      "step": 4683
    },
    {
      "epoch": 0.5995136311276078,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0007297867035993041,
      "loss": 1.6053,
      "step": 4684
    },
    {
      "epoch": 0.5996416229361321,
      "grad_norm": 0.59375,
      "learning_rate": 0.0007293875810232194,
      "loss": 1.0262,
      "step": 4685
    },
    {
      "epoch": 0.5997696147446563,
      "grad_norm": 0.5859375,
      "learning_rate": 0.000728988504956269,
      "loss": 1.3633,
      "step": 4686
    },
    {
      "epoch": 0.5998976065531806,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007285894754670413,
      "loss": 1.8994,
      "step": 4687
    },
    {
      "epoch": 0.6000255983617049,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0007281904926241156,
      "loss": 1.1981,
      "step": 4688
    },
    {
      "epoch": 0.6001535901702291,
      "grad_norm": 0.640625,
      "learning_rate": 0.000727791556496064,
      "loss": 1.1131,
      "step": 4689
    },
    {
      "epoch": 0.6002815819787534,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0007273926671514503,
      "loss": 0.9762,
      "step": 4690
    },
    {
      "epoch": 0.6004095737872777,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0007269938246588293,
      "loss": 1.2084,
      "step": 4691
    },
    {
      "epoch": 0.6005375655958018,
      "grad_norm": 0.953125,
      "learning_rate": 0.0007265950290867489,
      "loss": 1.1265,
      "step": 4692
    },
    {
      "epoch": 0.6006655574043261,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0007261962805037485,
      "loss": 1.2312,
      "step": 4693
    },
    {
      "epoch": 0.6007935492128503,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0007257975789783604,
      "loss": 1.5783,
      "step": 4694
    },
    {
      "epoch": 0.6009215410213746,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0007253989245791068,
      "loss": 1.6508,
      "step": 4695
    },
    {
      "epoch": 0.6010495328298989,
      "grad_norm": 0.671875,
      "learning_rate": 0.0007250003173745035,
      "loss": 1.4562,
      "step": 4696
    },
    {
      "epoch": 0.6011775246384231,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0007246017574330572,
      "loss": 1.5864,
      "step": 4697
    },
    {
      "epoch": 0.6013055164469474,
      "grad_norm": 0.671875,
      "learning_rate": 0.0007242032448232671,
      "loss": 1.8514,
      "step": 4698
    },
    {
      "epoch": 0.6014335082554717,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0007238047796136247,
      "loss": 1.5123,
      "step": 4699
    },
    {
      "epoch": 0.6015615000639959,
      "grad_norm": 0.609375,
      "learning_rate": 0.0007234063618726116,
      "loss": 1.167,
      "step": 4700
    },
    {
      "epoch": 0.6016894918725202,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0007230079916687029,
      "loss": 1.2442,
      "step": 4701
    },
    {
      "epoch": 0.6018174836810444,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0007226096690703651,
      "loss": 0.8747,
      "step": 4702
    },
    {
      "epoch": 0.6019454754895687,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0007222113941460565,
      "loss": 1.2378,
      "step": 4703
    },
    {
      "epoch": 0.602073467298093,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0007218131669642265,
      "loss": 1.0564,
      "step": 4704
    },
    {
      "epoch": 0.6022014591066172,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0007214149875933172,
      "loss": 0.8584,
      "step": 4705
    },
    {
      "epoch": 0.6023294509151415,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0007210168561017625,
      "loss": 1.7615,
      "step": 4706
    },
    {
      "epoch": 0.6024574427236656,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0007206187725579873,
      "loss": 1.1488,
      "step": 4707
    },
    {
      "epoch": 0.6025854345321899,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0007202207370304093,
      "loss": 1.3649,
      "step": 4708
    },
    {
      "epoch": 0.6027134263407142,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0007198227495874368,
      "loss": 1.4799,
      "step": 4709
    },
    {
      "epoch": 0.6028414181492384,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0007194248102974704,
      "loss": 1.4204,
      "step": 4710
    },
    {
      "epoch": 0.6029694099577627,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0007190269192289027,
      "loss": 1.425,
      "step": 4711
    },
    {
      "epoch": 0.603097401766287,
      "grad_norm": 0.65625,
      "learning_rate": 0.0007186290764501182,
      "loss": 1.6658,
      "step": 4712
    },
    {
      "epoch": 0.6032253935748112,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0007182312820294915,
      "loss": 1.2663,
      "step": 4713
    },
    {
      "epoch": 0.6033533853833355,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0007178335360353907,
      "loss": 1.3327,
      "step": 4714
    },
    {
      "epoch": 0.6034813771918597,
      "grad_norm": 0.671875,
      "learning_rate": 0.0007174358385361745,
      "loss": 1.0643,
      "step": 4715
    },
    {
      "epoch": 0.603609369000384,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0007170381896001947,
      "loss": 0.9233,
      "step": 4716
    },
    {
      "epoch": 0.6037373608089083,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0007166405892957926,
      "loss": 1.702,
      "step": 4717
    },
    {
      "epoch": 0.6038653526174325,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0007162430376913027,
      "loss": 1.157,
      "step": 4718
    },
    {
      "epoch": 0.6039933444259568,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0007158455348550505,
      "loss": 1.1133,
      "step": 4719
    },
    {
      "epoch": 0.6041213362344809,
      "grad_norm": 0.640625,
      "learning_rate": 0.0007154480808553536,
      "loss": 1.2952,
      "step": 4720
    },
    {
      "epoch": 0.6042493280430052,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0007150506757605211,
      "loss": 0.8525,
      "step": 4721
    },
    {
      "epoch": 0.6043773198515295,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0007146533196388528,
      "loss": 1.5574,
      "step": 4722
    },
    {
      "epoch": 0.6045053116600537,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0007142560125586412,
      "loss": 1.1473,
      "step": 4723
    },
    {
      "epoch": 0.604633303468578,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0007138587545881701,
      "loss": 1.6452,
      "step": 4724
    },
    {
      "epoch": 0.6047612952771023,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0007134615457957148,
      "loss": 1.2599,
      "step": 4725
    },
    {
      "epoch": 0.6048892870856265,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0007130643862495416,
      "loss": 1.4838,
      "step": 4726
    },
    {
      "epoch": 0.6050172788941508,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0007126672760179091,
      "loss": 1.0988,
      "step": 4727
    },
    {
      "epoch": 0.605145270702675,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0007122702151690671,
      "loss": 1.23,
      "step": 4728
    },
    {
      "epoch": 0.6052732625111993,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0007118732037712569,
      "loss": 1.2082,
      "step": 4729
    },
    {
      "epoch": 0.6054012543197236,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0007114762418927121,
      "loss": 1.028,
      "step": 4730
    },
    {
      "epoch": 0.6055292461282478,
      "grad_norm": 1.03125,
      "learning_rate": 0.0007110793296016555,
      "loss": 1.515,
      "step": 4731
    },
    {
      "epoch": 0.6056572379367721,
      "grad_norm": 0.6484375,
      "learning_rate": 0.000710682466966304,
      "loss": 1.4934,
      "step": 4732
    },
    {
      "epoch": 0.6057852297452962,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0007102856540548648,
      "loss": 2.0649,
      "step": 4733
    },
    {
      "epoch": 0.6059132215538205,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0007098888909355368,
      "loss": 1.1106,
      "step": 4734
    },
    {
      "epoch": 0.6060412133623448,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0007094921776765094,
      "loss": 0.9666,
      "step": 4735
    },
    {
      "epoch": 0.606169205170869,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0007090955143459648,
      "loss": 1.327,
      "step": 4736
    },
    {
      "epoch": 0.6062971969793933,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0007086989010120759,
      "loss": 1.2004,
      "step": 4737
    },
    {
      "epoch": 0.6064251887879176,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0007083023377430069,
      "loss": 1.7891,
      "step": 4738
    },
    {
      "epoch": 0.6065531805964418,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0007079058246069145,
      "loss": 1.8124,
      "step": 4739
    },
    {
      "epoch": 0.6066811724049661,
      "grad_norm": 0.765625,
      "learning_rate": 0.0007075093616719449,
      "loss": 1.3497,
      "step": 4740
    },
    {
      "epoch": 0.6068091642134903,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0007071129490062372,
      "loss": 1.4755,
      "step": 4741
    },
    {
      "epoch": 0.6069371560220146,
      "grad_norm": 0.67578125,
      "learning_rate": 0.000706716586677921,
      "loss": 1.8759,
      "step": 4742
    },
    {
      "epoch": 0.6070651478305389,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0007063202747551182,
      "loss": 1.9106,
      "step": 4743
    },
    {
      "epoch": 0.6071931396390631,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0007059240133059408,
      "loss": 1.7444,
      "step": 4744
    },
    {
      "epoch": 0.6073211314475874,
      "grad_norm": 0.69921875,
      "learning_rate": 0.000705527802398493,
      "loss": 1.827,
      "step": 4745
    },
    {
      "epoch": 0.6074491232561116,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0007051316421008702,
      "loss": 1.2123,
      "step": 4746
    },
    {
      "epoch": 0.6075771150646359,
      "grad_norm": 0.55078125,
      "learning_rate": 0.000704735532481159,
      "loss": 1.326,
      "step": 4747
    },
    {
      "epoch": 0.6077051068731602,
      "grad_norm": 0.5625,
      "learning_rate": 0.000704339473607437,
      "loss": 1.22,
      "step": 4748
    },
    {
      "epoch": 0.6078330986816843,
      "grad_norm": 0.6875,
      "learning_rate": 0.0007039434655477735,
      "loss": 1.554,
      "step": 4749
    },
    {
      "epoch": 0.6079610904902086,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0007035475083702287,
      "loss": 1.3969,
      "step": 4750
    },
    {
      "epoch": 0.6080890822987329,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0007031516021428545,
      "loss": 2.0431,
      "step": 4751
    },
    {
      "epoch": 0.6082170741072571,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0007027557469336945,
      "loss": 1.1029,
      "step": 4752
    },
    {
      "epoch": 0.6083450659157814,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0007023599428107814,
      "loss": 1.6144,
      "step": 4753
    },
    {
      "epoch": 0.6084730577243056,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0007019641898421412,
      "loss": 1.083,
      "step": 4754
    },
    {
      "epoch": 0.6086010495328299,
      "grad_norm": 0.609375,
      "learning_rate": 0.0007015684880957905,
      "loss": 1.558,
      "step": 4755
    },
    {
      "epoch": 0.6087290413413542,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0007011728376397378,
      "loss": 1.3267,
      "step": 4756
    },
    {
      "epoch": 0.6088570331498784,
      "grad_norm": 0.8125,
      "learning_rate": 0.0007007772385419807,
      "loss": 2.1946,
      "step": 4757
    },
    {
      "epoch": 0.6089850249584027,
      "grad_norm": 0.640625,
      "learning_rate": 0.00070038169087051,
      "loss": 1.242,
      "step": 4758
    },
    {
      "epoch": 0.6091130167669269,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0006999861946933071,
      "loss": 1.0725,
      "step": 4759
    },
    {
      "epoch": 0.6092410085754512,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0006995907500783443,
      "loss": 1.162,
      "step": 4760
    },
    {
      "epoch": 0.6093690003839755,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0006991953570935853,
      "loss": 1.0569,
      "step": 4761
    },
    {
      "epoch": 0.6094969921924996,
      "grad_norm": 0.640625,
      "learning_rate": 0.0006988000158069845,
      "loss": 1.2618,
      "step": 4762
    },
    {
      "epoch": 0.6096249840010239,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0006984047262864879,
      "loss": 1.5448,
      "step": 4763
    },
    {
      "epoch": 0.6097529758095482,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0006980094886000324,
      "loss": 1.4866,
      "step": 4764
    },
    {
      "epoch": 0.6098809676180724,
      "grad_norm": 0.671875,
      "learning_rate": 0.0006976143028155459,
      "loss": 2.0439,
      "step": 4765
    },
    {
      "epoch": 0.6100089594265967,
      "grad_norm": 0.65625,
      "learning_rate": 0.0006972191690009477,
      "loss": 1.3675,
      "step": 4766
    },
    {
      "epoch": 0.6101369512351209,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0006968240872241478,
      "loss": 1.6122,
      "step": 4767
    },
    {
      "epoch": 0.6102649430436452,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0006964290575530475,
      "loss": 1.2167,
      "step": 4768
    },
    {
      "epoch": 0.6103929348521695,
      "grad_norm": 0.609375,
      "learning_rate": 0.0006960340800555392,
      "loss": 1.2139,
      "step": 4769
    },
    {
      "epoch": 0.6105209266606937,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0006956391547995057,
      "loss": 0.7637,
      "step": 4770
    },
    {
      "epoch": 0.610648918469218,
      "grad_norm": 0.6875,
      "learning_rate": 0.0006952442818528219,
      "loss": 1.6221,
      "step": 4771
    },
    {
      "epoch": 0.6107769102777423,
      "grad_norm": 0.65625,
      "learning_rate": 0.0006948494612833526,
      "loss": 1.2529,
      "step": 4772
    },
    {
      "epoch": 0.6109049020862665,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0006944546931589546,
      "loss": 1.1695,
      "step": 4773
    },
    {
      "epoch": 0.6110328938947908,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0006940599775474752,
      "loss": 1.494,
      "step": 4774
    },
    {
      "epoch": 0.611160885703315,
      "grad_norm": 0.59375,
      "learning_rate": 0.000693665314516752,
      "loss": 1.3506,
      "step": 4775
    },
    {
      "epoch": 0.6112888775118392,
      "grad_norm": 0.546875,
      "learning_rate": 0.0006932707041346147,
      "loss": 1.0163,
      "step": 4776
    },
    {
      "epoch": 0.6114168693203635,
      "grad_norm": 0.828125,
      "learning_rate": 0.0006928761464688835,
      "loss": 1.7052,
      "step": 4777
    },
    {
      "epoch": 0.6115448611288877,
      "grad_norm": 0.578125,
      "learning_rate": 0.0006924816415873703,
      "loss": 0.9351,
      "step": 4778
    },
    {
      "epoch": 0.611672852937412,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0006920871895578757,
      "loss": 1.4286,
      "step": 4779
    },
    {
      "epoch": 0.6118008447459362,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0006916927904481934,
      "loss": 1.443,
      "step": 4780
    },
    {
      "epoch": 0.6119288365544605,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0006912984443261072,
      "loss": 1.4222,
      "step": 4781
    },
    {
      "epoch": 0.6120568283629848,
      "grad_norm": 0.59375,
      "learning_rate": 0.000690904151259392,
      "loss": 1.2301,
      "step": 4782
    },
    {
      "epoch": 0.612184820171509,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0006905099113158138,
      "loss": 1.3626,
      "step": 4783
    },
    {
      "epoch": 0.6123128119800333,
      "grad_norm": 0.625,
      "learning_rate": 0.0006901157245631284,
      "loss": 1.4507,
      "step": 4784
    },
    {
      "epoch": 0.6124408037885576,
      "grad_norm": 0.546875,
      "learning_rate": 0.0006897215910690836,
      "loss": 0.947,
      "step": 4785
    },
    {
      "epoch": 0.6125687955970818,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0006893275109014174,
      "loss": 1.3153,
      "step": 4786
    },
    {
      "epoch": 0.6126967874056061,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0006889334841278595,
      "loss": 1.4223,
      "step": 4787
    },
    {
      "epoch": 0.6128247792141303,
      "grad_norm": 0.6875,
      "learning_rate": 0.0006885395108161289,
      "loss": 1.5116,
      "step": 4788
    },
    {
      "epoch": 0.6129527710226546,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0006881455910339369,
      "loss": 1.3433,
      "step": 4789
    },
    {
      "epoch": 0.6130807628311788,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0006877517248489848,
      "loss": 2.3247,
      "step": 4790
    },
    {
      "epoch": 0.613208754639703,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0006873579123289649,
      "loss": 0.8419,
      "step": 4791
    },
    {
      "epoch": 0.6133367464482273,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0006869641535415605,
      "loss": 1.9199,
      "step": 4792
    },
    {
      "epoch": 0.6134647382567515,
      "grad_norm": 0.71875,
      "learning_rate": 0.0006865704485544451,
      "loss": 1.027,
      "step": 4793
    },
    {
      "epoch": 0.6135927300652758,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0006861767974352833,
      "loss": 1.8671,
      "step": 4794
    },
    {
      "epoch": 0.6137207218738001,
      "grad_norm": 0.671875,
      "learning_rate": 0.0006857832002517307,
      "loss": 1.1517,
      "step": 4795
    },
    {
      "epoch": 0.6138487136823243,
      "grad_norm": 0.6875,
      "learning_rate": 0.0006853896570714335,
      "loss": 1.6693,
      "step": 4796
    },
    {
      "epoch": 0.6139767054908486,
      "grad_norm": 0.58984375,
      "learning_rate": 0.000684996167962028,
      "loss": 1.5,
      "step": 4797
    },
    {
      "epoch": 0.6141046972993729,
      "grad_norm": 0.796875,
      "learning_rate": 0.0006846027329911417,
      "loss": 1.4504,
      "step": 4798
    },
    {
      "epoch": 0.6142326891078971,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0006842093522263929,
      "loss": 1.525,
      "step": 4799
    },
    {
      "epoch": 0.6143606809164214,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0006838160257353911,
      "loss": 1.4458,
      "step": 4800
    },
    {
      "epoch": 0.6144886727249456,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0006834227535857349,
      "loss": 1.2935,
      "step": 4801
    },
    {
      "epoch": 0.6146166645334699,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0006830295358450149,
      "loss": 1.4761,
      "step": 4802
    },
    {
      "epoch": 0.6147446563419942,
      "grad_norm": 0.87109375,
      "learning_rate": 0.000682636372580812,
      "loss": 0.9584,
      "step": 4803
    },
    {
      "epoch": 0.6148726481505183,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0006822432638606974,
      "loss": 1.3589,
      "step": 4804
    },
    {
      "epoch": 0.6150006399590426,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0006818502097522339,
      "loss": 1.6068,
      "step": 4805
    },
    {
      "epoch": 0.6151286317675668,
      "grad_norm": 0.703125,
      "learning_rate": 0.0006814572103229735,
      "loss": 1.6604,
      "step": 4806
    },
    {
      "epoch": 0.6152566235760911,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0006810642656404596,
      "loss": 1.8452,
      "step": 4807
    },
    {
      "epoch": 0.6153846153846154,
      "grad_norm": 0.625,
      "learning_rate": 0.0006806713757722265,
      "loss": 0.9694,
      "step": 4808
    },
    {
      "epoch": 0.6155126071931396,
      "grad_norm": 0.84375,
      "learning_rate": 0.0006802785407857986,
      "loss": 1.6006,
      "step": 4809
    },
    {
      "epoch": 0.6156405990016639,
      "grad_norm": 0.609375,
      "learning_rate": 0.0006798857607486908,
      "loss": 1.2345,
      "step": 4810
    },
    {
      "epoch": 0.6157685908101882,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0006794930357284086,
      "loss": 1.3567,
      "step": 4811
    },
    {
      "epoch": 0.6158965826187124,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0006791003657924484,
      "loss": 1.6533,
      "step": 4812
    },
    {
      "epoch": 0.6160245744272367,
      "grad_norm": 0.62890625,
      "learning_rate": 0.000678707751008297,
      "loss": 1.2787,
      "step": 4813
    },
    {
      "epoch": 0.6161525662357609,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0006783151914434318,
      "loss": 0.876,
      "step": 4814
    },
    {
      "epoch": 0.6162805580442852,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0006779226871653195,
      "loss": 1.4385,
      "step": 4815
    },
    {
      "epoch": 0.6164085498528095,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0006775302382414194,
      "loss": 1.2716,
      "step": 4816
    },
    {
      "epoch": 0.6165365416613336,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0006771378447391798,
      "loss": 1.6251,
      "step": 4817
    },
    {
      "epoch": 0.616664533469858,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0006767455067260403,
      "loss": 1.2591,
      "step": 4818
    },
    {
      "epoch": 0.6167925252783821,
      "grad_norm": 0.625,
      "learning_rate": 0.0006763532242694298,
      "loss": 1.1345,
      "step": 4819
    },
    {
      "epoch": 0.6169205170869064,
      "grad_norm": 0.640625,
      "learning_rate": 0.0006759609974367689,
      "loss": 1.119,
      "step": 4820
    },
    {
      "epoch": 0.6170485088954307,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0006755688262954677,
      "loss": 0.8338,
      "step": 4821
    },
    {
      "epoch": 0.6171765007039549,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0006751767109129278,
      "loss": 1.5406,
      "step": 4822
    },
    {
      "epoch": 0.6173044925124792,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0006747846513565406,
      "loss": 1.8614,
      "step": 4823
    },
    {
      "epoch": 0.6174324843210035,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0006743926476936873,
      "loss": 1.2091,
      "step": 4824
    },
    {
      "epoch": 0.6175604761295277,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0006740006999917405,
      "loss": 1.6475,
      "step": 4825
    },
    {
      "epoch": 0.617688467938052,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0006736088083180625,
      "loss": 1.2553,
      "step": 4826
    },
    {
      "epoch": 0.6178164597465762,
      "grad_norm": 0.625,
      "learning_rate": 0.0006732169727400067,
      "loss": 0.8051,
      "step": 4827
    },
    {
      "epoch": 0.6179444515551005,
      "grad_norm": 0.6875,
      "learning_rate": 0.0006728251933249159,
      "loss": 1.1643,
      "step": 4828
    },
    {
      "epoch": 0.6180724433636248,
      "grad_norm": 0.59375,
      "learning_rate": 0.0006724334701401239,
      "loss": 1.4546,
      "step": 4829
    },
    {
      "epoch": 0.618200435172149,
      "grad_norm": 0.703125,
      "learning_rate": 0.0006720418032529547,
      "loss": 1.1787,
      "step": 4830
    },
    {
      "epoch": 0.6183284269806733,
      "grad_norm": 0.546875,
      "learning_rate": 0.0006716501927307231,
      "loss": 1.223,
      "step": 4831
    },
    {
      "epoch": 0.6184564187891974,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0006712586386407328,
      "loss": 1.5472,
      "step": 4832
    },
    {
      "epoch": 0.6185844105977217,
      "grad_norm": 0.703125,
      "learning_rate": 0.0006708671410502793,
      "loss": 1.8289,
      "step": 4833
    },
    {
      "epoch": 0.618712402406246,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0006704757000266477,
      "loss": 1.372,
      "step": 4834
    },
    {
      "epoch": 0.6188403942147702,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0006700843156371136,
      "loss": 1.2011,
      "step": 4835
    },
    {
      "epoch": 0.6189683860232945,
      "grad_norm": 0.5546875,
      "learning_rate": 0.000669692987948943,
      "loss": 0.9845,
      "step": 4836
    },
    {
      "epoch": 0.6190963778318188,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0006693017170293909,
      "loss": 1.6607,
      "step": 4837
    },
    {
      "epoch": 0.619224369640343,
      "grad_norm": 0.703125,
      "learning_rate": 0.0006689105029457045,
      "loss": 1.7245,
      "step": 4838
    },
    {
      "epoch": 0.6193523614488673,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00066851934576512,
      "loss": 1.5095,
      "step": 4839
    },
    {
      "epoch": 0.6194803532573915,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0006681282455548645,
      "loss": 1.9428,
      "step": 4840
    },
    {
      "epoch": 0.6196083450659158,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0006677372023821542,
      "loss": 1.1136,
      "step": 4841
    },
    {
      "epoch": 0.6197363368744401,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0006673462163141966,
      "loss": 1.6873,
      "step": 4842
    },
    {
      "epoch": 0.6198643286829643,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0006669552874181888,
      "loss": 1.5206,
      "step": 4843
    },
    {
      "epoch": 0.6199923204914886,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0006665644157613182,
      "loss": 0.9308,
      "step": 4844
    },
    {
      "epoch": 0.6201203123000127,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0006661736014107635,
      "loss": 1.5269,
      "step": 4845
    },
    {
      "epoch": 0.620248304108537,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0006657828444336911,
      "loss": 1.4726,
      "step": 4846
    },
    {
      "epoch": 0.6203762959170613,
      "grad_norm": 1.078125,
      "learning_rate": 0.0006653921448972594,
      "loss": 2.1705,
      "step": 4847
    },
    {
      "epoch": 0.6205042877255855,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0006650015028686167,
      "loss": 1.2939,
      "step": 4848
    },
    {
      "epoch": 0.6206322795341098,
      "grad_norm": 0.609375,
      "learning_rate": 0.000664610918414901,
      "loss": 1.4533,
      "step": 4849
    },
    {
      "epoch": 0.6207602713426341,
      "grad_norm": 0.59375,
      "learning_rate": 0.0006642203916032406,
      "loss": 1.0645,
      "step": 4850
    },
    {
      "epoch": 0.6208882631511583,
      "grad_norm": 0.703125,
      "learning_rate": 0.0006638299225007537,
      "loss": 1.2943,
      "step": 4851
    },
    {
      "epoch": 0.6210162549596826,
      "grad_norm": 0.703125,
      "learning_rate": 0.0006634395111745489,
      "loss": 1.2249,
      "step": 4852
    },
    {
      "epoch": 0.6211442467682068,
      "grad_norm": 0.71484375,
      "learning_rate": 0.000663049157691725,
      "loss": 1.403,
      "step": 4853
    },
    {
      "epoch": 0.6212722385767311,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0006626588621193701,
      "loss": 1.4503,
      "step": 4854
    },
    {
      "epoch": 0.6214002303852554,
      "grad_norm": 0.546875,
      "learning_rate": 0.000662268624524563,
      "loss": 0.8254,
      "step": 4855
    },
    {
      "epoch": 0.6215282221937796,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0006618784449743722,
      "loss": 1.1634,
      "step": 4856
    },
    {
      "epoch": 0.6216562140023039,
      "grad_norm": 0.609375,
      "learning_rate": 0.0006614883235358569,
      "loss": 1.0928,
      "step": 4857
    },
    {
      "epoch": 0.6217842058108282,
      "grad_norm": 0.796875,
      "learning_rate": 0.0006610982602760657,
      "loss": 1.5334,
      "step": 4858
    },
    {
      "epoch": 0.6219121976193523,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0006607082552620367,
      "loss": 1.4291,
      "step": 4859
    },
    {
      "epoch": 0.6220401894278766,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0006603183085607987,
      "loss": 1.1244,
      "step": 4860
    },
    {
      "epoch": 0.6221681812364008,
      "grad_norm": 0.578125,
      "learning_rate": 0.0006599284202393708,
      "loss": 1.0621,
      "step": 4861
    },
    {
      "epoch": 0.6222961730449251,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0006595385903647619,
      "loss": 0.977,
      "step": 4862
    },
    {
      "epoch": 0.6224241648534494,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0006591488190039698,
      "loss": 1.1656,
      "step": 4863
    },
    {
      "epoch": 0.6225521566619736,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0006587591062239834,
      "loss": 0.8212,
      "step": 4864
    },
    {
      "epoch": 0.6226801484704979,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0006583694520917811,
      "loss": 2.024,
      "step": 4865
    },
    {
      "epoch": 0.6228081402790221,
      "grad_norm": 0.609375,
      "learning_rate": 0.0006579798566743314,
      "loss": 0.8808,
      "step": 4866
    },
    {
      "epoch": 0.6229361320875464,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0006575903200385926,
      "loss": 1.2533,
      "step": 4867
    },
    {
      "epoch": 0.6230641238960707,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0006572008422515127,
      "loss": 1.1203,
      "step": 4868
    },
    {
      "epoch": 0.6231921157045949,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0006568114233800298,
      "loss": 1.3658,
      "step": 4869
    },
    {
      "epoch": 0.6233201075131192,
      "grad_norm": 0.734375,
      "learning_rate": 0.000656422063491072,
      "loss": 1.8813,
      "step": 4870
    },
    {
      "epoch": 0.6234480993216435,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0006560327626515573,
      "loss": 1.0848,
      "step": 4871
    },
    {
      "epoch": 0.6235760911301677,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0006556435209283929,
      "loss": 0.9864,
      "step": 4872
    },
    {
      "epoch": 0.623704082938692,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0006552543383884765,
      "loss": 0.8598,
      "step": 4873
    },
    {
      "epoch": 0.6238320747472161,
      "grad_norm": 0.703125,
      "learning_rate": 0.0006548652150986955,
      "loss": 1.3318,
      "step": 4874
    },
    {
      "epoch": 0.6239600665557404,
      "grad_norm": 0.765625,
      "learning_rate": 0.000654476151125927,
      "loss": 1.0499,
      "step": 4875
    },
    {
      "epoch": 0.6240880583642647,
      "grad_norm": 0.703125,
      "learning_rate": 0.0006540871465370382,
      "loss": 1.574,
      "step": 4876
    },
    {
      "epoch": 0.6242160501727889,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0006536982013988853,
      "loss": 1.0239,
      "step": 4877
    },
    {
      "epoch": 0.6243440419813132,
      "grad_norm": 0.78125,
      "learning_rate": 0.0006533093157783152,
      "loss": 1.2115,
      "step": 4878
    },
    {
      "epoch": 0.6244720337898374,
      "grad_norm": 0.578125,
      "learning_rate": 0.0006529204897421643,
      "loss": 1.1829,
      "step": 4879
    },
    {
      "epoch": 0.6246000255983617,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0006525317233572588,
      "loss": 1.714,
      "step": 4880
    },
    {
      "epoch": 0.624728017406886,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0006521430166904139,
      "loss": 1.2249,
      "step": 4881
    },
    {
      "epoch": 0.6248560092154102,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0006517543698084351,
      "loss": 1.1305,
      "step": 4882
    },
    {
      "epoch": 0.6249840010239345,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0006513657827781185,
      "loss": 1.7397,
      "step": 4883
    },
    {
      "epoch": 0.6251119928324588,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0006509772556662488,
      "loss": 1.5089,
      "step": 4884
    },
    {
      "epoch": 0.625239984640983,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0006505887885396003,
      "loss": 1.1685,
      "step": 4885
    },
    {
      "epoch": 0.6253679764495073,
      "grad_norm": 0.671875,
      "learning_rate": 0.0006502003814649375,
      "loss": 1.1113,
      "step": 4886
    },
    {
      "epoch": 0.6254959682580314,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0006498120345090146,
      "loss": 1.6866,
      "step": 4887
    },
    {
      "epoch": 0.6256239600665557,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0006494237477385753,
      "loss": 1.1366,
      "step": 4888
    },
    {
      "epoch": 0.62575195187508,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0006490355212203532,
      "loss": 1.6263,
      "step": 4889
    },
    {
      "epoch": 0.6258799436836042,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0006486473550210707,
      "loss": 0.8933,
      "step": 4890
    },
    {
      "epoch": 0.6260079354921285,
      "grad_norm": 0.6171875,
      "learning_rate": 0.000648259249207441,
      "loss": 1.4276,
      "step": 4891
    },
    {
      "epoch": 0.6261359273006527,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0006478712038461662,
      "loss": 1.1693,
      "step": 4892
    },
    {
      "epoch": 0.626263919109177,
      "grad_norm": 0.734375,
      "learning_rate": 0.0006474832190039386,
      "loss": 1.3703,
      "step": 4893
    },
    {
      "epoch": 0.6263919109177013,
      "grad_norm": 0.609375,
      "learning_rate": 0.0006470952947474389,
      "loss": 1.4274,
      "step": 4894
    },
    {
      "epoch": 0.6265199027262255,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0006467074311433385,
      "loss": 0.6944,
      "step": 4895
    },
    {
      "epoch": 0.6266478945347498,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0006463196282582984,
      "loss": 1.5647,
      "step": 4896
    },
    {
      "epoch": 0.6267758863432741,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0006459318861589684,
      "loss": 1.2763,
      "step": 4897
    },
    {
      "epoch": 0.6269038781517983,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0006455442049119887,
      "loss": 1.4294,
      "step": 4898
    },
    {
      "epoch": 0.6270318699603226,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0006451565845839882,
      "loss": 1.075,
      "step": 4899
    },
    {
      "epoch": 0.6271598617688467,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0006447690252415858,
      "loss": 1.3863,
      "step": 4900
    },
    {
      "epoch": 0.627287853577371,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0006443815269513899,
      "loss": 1.0408,
      "step": 4901
    },
    {
      "epoch": 0.6274158453858953,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0006439940897799989,
      "loss": 1.0629,
      "step": 4902
    },
    {
      "epoch": 0.6275438371944195,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0006436067137939994,
      "loss": 1.6001,
      "step": 4903
    },
    {
      "epoch": 0.6276718290029438,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0006432193990599684,
      "loss": 1.3539,
      "step": 4904
    },
    {
      "epoch": 0.627799820811468,
      "grad_norm": 0.71875,
      "learning_rate": 0.0006428321456444722,
      "loss": 1.9321,
      "step": 4905
    },
    {
      "epoch": 0.6279278126199923,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0006424449536140668,
      "loss": 1.6035,
      "step": 4906
    },
    {
      "epoch": 0.6280558044285166,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0006420578230352978,
      "loss": 1.146,
      "step": 4907
    },
    {
      "epoch": 0.6281837962370408,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0006416707539746991,
      "loss": 1.479,
      "step": 4908
    },
    {
      "epoch": 0.6283117880455651,
      "grad_norm": 0.69921875,
      "learning_rate": 0.000641283746498795,
      "loss": 1.6063,
      "step": 4909
    },
    {
      "epoch": 0.6284397798540894,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0006408968006740994,
      "loss": 1.6608,
      "step": 4910
    },
    {
      "epoch": 0.6285677716626136,
      "grad_norm": 0.6875,
      "learning_rate": 0.000640509916567115,
      "loss": 1.7442,
      "step": 4911
    },
    {
      "epoch": 0.6286957634711379,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0006401230942443339,
      "loss": 1.0778,
      "step": 4912
    },
    {
      "epoch": 0.6288237552796621,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0006397363337722379,
      "loss": 1.3228,
      "step": 4913
    },
    {
      "epoch": 0.6289517470881864,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0006393496352172982,
      "loss": 1.7837,
      "step": 4914
    },
    {
      "epoch": 0.6290797388967106,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0006389629986459756,
      "loss": 1.001,
      "step": 4915
    },
    {
      "epoch": 0.6292077307052348,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0006385764241247189,
      "loss": 0.7678,
      "step": 4916
    },
    {
      "epoch": 0.6293357225137591,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0006381899117199678,
      "loss": 1.1762,
      "step": 4917
    },
    {
      "epoch": 0.6294637143222833,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0006378034614981507,
      "loss": 1.016,
      "step": 4918
    },
    {
      "epoch": 0.6295917061308076,
      "grad_norm": 0.515625,
      "learning_rate": 0.0006374170735256853,
      "loss": 0.787,
      "step": 4919
    },
    {
      "epoch": 0.6297196979393319,
      "grad_norm": 0.59375,
      "learning_rate": 0.000637030747868979,
      "loss": 1.2652,
      "step": 4920
    },
    {
      "epoch": 0.6298476897478561,
      "grad_norm": 0.55078125,
      "learning_rate": 0.000636644484594427,
      "loss": 0.8987,
      "step": 4921
    },
    {
      "epoch": 0.6299756815563804,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0006362582837684161,
      "loss": 1.0087,
      "step": 4922
    },
    {
      "epoch": 0.6301036733649047,
      "grad_norm": 0.609375,
      "learning_rate": 0.0006358721454573207,
      "loss": 1.07,
      "step": 4923
    },
    {
      "epoch": 0.6302316651734289,
      "grad_norm": 0.734375,
      "learning_rate": 0.0006354860697275053,
      "loss": 1.6581,
      "step": 4924
    },
    {
      "epoch": 0.6303596569819532,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0006351000566453226,
      "loss": 1.2312,
      "step": 4925
    },
    {
      "epoch": 0.6304876487904774,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0006347141062771158,
      "loss": 0.9688,
      "step": 4926
    },
    {
      "epoch": 0.6306156405990017,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0006343282186892163,
      "loss": 1.2568,
      "step": 4927
    },
    {
      "epoch": 0.630743632407526,
      "grad_norm": 0.59375,
      "learning_rate": 0.0006339423939479452,
      "loss": 0.97,
      "step": 4928
    },
    {
      "epoch": 0.6308716242160501,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0006335566321196135,
      "loss": 1.2551,
      "step": 4929
    },
    {
      "epoch": 0.6309996160245744,
      "grad_norm": 0.828125,
      "learning_rate": 0.0006331709332705195,
      "loss": 1.8014,
      "step": 4930
    },
    {
      "epoch": 0.6311276078330987,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0006327852974669526,
      "loss": 0.8369,
      "step": 4931
    },
    {
      "epoch": 0.6312555996416229,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00063239972477519,
      "loss": 0.8759,
      "step": 4932
    },
    {
      "epoch": 0.6313835914501472,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0006320142152614994,
      "loss": 1.0278,
      "step": 4933
    },
    {
      "epoch": 0.6315115832586714,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0006316287689921361,
      "loss": 1.5576,
      "step": 4934
    },
    {
      "epoch": 0.6316395750671957,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0006312433860333456,
      "loss": 1.1302,
      "step": 4935
    },
    {
      "epoch": 0.63176756687572,
      "grad_norm": 0.66015625,
      "learning_rate": 0.000630858066451362,
      "loss": 1.0118,
      "step": 4936
    },
    {
      "epoch": 0.6318955586842442,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0006304728103124094,
      "loss": 1.3935,
      "step": 4937
    },
    {
      "epoch": 0.6320235504927685,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0006300876176826992,
      "loss": 1.6025,
      "step": 4938
    },
    {
      "epoch": 0.6321515423012927,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0006297024886284342,
      "loss": 0.9382,
      "step": 4939
    },
    {
      "epoch": 0.632279534109817,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0006293174232158041,
      "loss": 1.5135,
      "step": 4940
    },
    {
      "epoch": 0.6324075259183413,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0006289324215109893,
      "loss": 1.7593,
      "step": 4941
    },
    {
      "epoch": 0.6325355177268654,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0006285474835801586,
      "loss": 1.3229,
      "step": 4942
    },
    {
      "epoch": 0.6326635095353897,
      "grad_norm": 0.59375,
      "learning_rate": 0.000628162609489469,
      "loss": 1.0865,
      "step": 4943
    },
    {
      "epoch": 0.632791501343914,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0006277777993050681,
      "loss": 1.1507,
      "step": 4944
    },
    {
      "epoch": 0.6329194931524382,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0006273930530930918,
      "loss": 0.7403,
      "step": 4945
    },
    {
      "epoch": 0.6330474849609625,
      "grad_norm": 0.76171875,
      "learning_rate": 0.000627008370919665,
      "loss": 1.3479,
      "step": 4946
    },
    {
      "epoch": 0.6331754767694867,
      "grad_norm": 0.67578125,
      "learning_rate": 0.000626623752850901,
      "loss": 1.1847,
      "step": 4947
    },
    {
      "epoch": 0.633303468578011,
      "grad_norm": 0.75,
      "learning_rate": 0.0006262391989529031,
      "loss": 1.3102,
      "step": 4948
    },
    {
      "epoch": 0.6334314603865353,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0006258547092917628,
      "loss": 1.1099,
      "step": 4949
    },
    {
      "epoch": 0.6335594521950595,
      "grad_norm": 0.640625,
      "learning_rate": 0.0006254702839335611,
      "loss": 1.0875,
      "step": 4950
    },
    {
      "epoch": 0.6336874440035838,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0006250859229443684,
      "loss": 1.5115,
      "step": 4951
    },
    {
      "epoch": 0.633815435812108,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0006247016263902423,
      "loss": 1.523,
      "step": 4952
    },
    {
      "epoch": 0.6339434276206323,
      "grad_norm": 0.84375,
      "learning_rate": 0.0006243173943372307,
      "loss": 0.9869,
      "step": 4953
    },
    {
      "epoch": 0.6340714194291566,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0006239332268513704,
      "loss": 1.9566,
      "step": 4954
    },
    {
      "epoch": 0.6341994112376808,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0006235491239986868,
      "loss": 1.5691,
      "step": 4955
    },
    {
      "epoch": 0.634327403046205,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0006231650858451939,
      "loss": 1.2274,
      "step": 4956
    },
    {
      "epoch": 0.6344553948547293,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0006227811124568949,
      "loss": 1.1999,
      "step": 4957
    },
    {
      "epoch": 0.6345833866632535,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0006223972038997823,
      "loss": 1.3626,
      "step": 4958
    },
    {
      "epoch": 0.6347113784717778,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0006220133602398365,
      "loss": 1.5445,
      "step": 4959
    },
    {
      "epoch": 0.634839370280302,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0006216295815430278,
      "loss": 1.6048,
      "step": 4960
    },
    {
      "epoch": 0.6349673620888263,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0006212458678753141,
      "loss": 1.5734,
      "step": 4961
    },
    {
      "epoch": 0.6350953538973506,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0006208622193026434,
      "loss": 1.3465,
      "step": 4962
    },
    {
      "epoch": 0.6352233457058748,
      "grad_norm": 0.734375,
      "learning_rate": 0.0006204786358909516,
      "loss": 1.2818,
      "step": 4963
    },
    {
      "epoch": 0.6353513375143991,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0006200951177061642,
      "loss": 1.1779,
      "step": 4964
    },
    {
      "epoch": 0.6354793293229233,
      "grad_norm": 0.59375,
      "learning_rate": 0.0006197116648141945,
      "loss": 1.5599,
      "step": 4965
    },
    {
      "epoch": 0.6356073211314476,
      "grad_norm": 0.69140625,
      "learning_rate": 0.000619328277280945,
      "loss": 1.5154,
      "step": 4966
    },
    {
      "epoch": 0.6357353129399719,
      "grad_norm": 0.6875,
      "learning_rate": 0.0006189449551723075,
      "loss": 1.5694,
      "step": 4967
    },
    {
      "epoch": 0.6358633047484961,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0006185616985541624,
      "loss": 1.2526,
      "step": 4968
    },
    {
      "epoch": 0.6359912965570204,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0006181785074923778,
      "loss": 0.8735,
      "step": 4969
    },
    {
      "epoch": 0.6361192883655447,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0006177953820528117,
      "loss": 1.7162,
      "step": 4970
    },
    {
      "epoch": 0.6362472801740688,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0006174123223013102,
      "loss": 0.9307,
      "step": 4971
    },
    {
      "epoch": 0.6363752719825931,
      "grad_norm": 1.03125,
      "learning_rate": 0.0006170293283037087,
      "loss": 1.7504,
      "step": 4972
    },
    {
      "epoch": 0.6365032637911173,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0006166464001258307,
      "loss": 1.3911,
      "step": 4973
    },
    {
      "epoch": 0.6366312555996416,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0006162635378334886,
      "loss": 0.9654,
      "step": 4974
    },
    {
      "epoch": 0.6367592474081659,
      "grad_norm": 0.6875,
      "learning_rate": 0.0006158807414924836,
      "loss": 1.3407,
      "step": 4975
    },
    {
      "epoch": 0.6368872392166901,
      "grad_norm": 0.578125,
      "learning_rate": 0.0006154980111686053,
      "loss": 1.2524,
      "step": 4976
    },
    {
      "epoch": 0.6370152310252144,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0006151153469276326,
      "loss": 1.9194,
      "step": 4977
    },
    {
      "epoch": 0.6371432228337386,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0006147327488353318,
      "loss": 1.3834,
      "step": 4978
    },
    {
      "epoch": 0.6372712146422629,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0006143502169574591,
      "loss": 1.1503,
      "step": 4979
    },
    {
      "epoch": 0.6373992064507872,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0006139677513597587,
      "loss": 1.4305,
      "step": 4980
    },
    {
      "epoch": 0.6375271982593114,
      "grad_norm": 0.78125,
      "learning_rate": 0.0006135853521079635,
      "loss": 2.6308,
      "step": 4981
    },
    {
      "epoch": 0.6376551900678357,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0006132030192677952,
      "loss": 1.7531,
      "step": 4982
    },
    {
      "epoch": 0.63778318187636,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0006128207529049634,
      "loss": 1.7162,
      "step": 4983
    },
    {
      "epoch": 0.6379111736848841,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0006124385530851673,
      "loss": 2.0491,
      "step": 4984
    },
    {
      "epoch": 0.6380391654934084,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0006120564198740939,
      "loss": 1.4225,
      "step": 4985
    },
    {
      "epoch": 0.6381671573019326,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0006116743533374194,
      "loss": 1.1854,
      "step": 4986
    },
    {
      "epoch": 0.6382951491104569,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0006112923535408073,
      "loss": 1.09,
      "step": 4987
    },
    {
      "epoch": 0.6384231409189812,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0006109104205499111,
      "loss": 1.4868,
      "step": 4988
    },
    {
      "epoch": 0.6385511327275054,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0006105285544303717,
      "loss": 1.78,
      "step": 4989
    },
    {
      "epoch": 0.6386791245360297,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0006101467552478195,
      "loss": 1.5208,
      "step": 4990
    },
    {
      "epoch": 0.6388071163445539,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0006097650230678731,
      "loss": 0.8292,
      "step": 4991
    },
    {
      "epoch": 0.6389351081530782,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0006093833579561385,
      "loss": 1.2322,
      "step": 4992
    },
    {
      "epoch": 0.6390630999616025,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0006090017599782116,
      "loss": 1.713,
      "step": 4993
    },
    {
      "epoch": 0.6391910917701267,
      "grad_norm": 0.59375,
      "learning_rate": 0.0006086202291996761,
      "loss": 1.4056,
      "step": 4994
    },
    {
      "epoch": 0.639319083578651,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0006082387656861044,
      "loss": 1.4612,
      "step": 4995
    },
    {
      "epoch": 0.6394470753871753,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0006078573695030568,
      "loss": 1.0407,
      "step": 4996
    },
    {
      "epoch": 0.6395750671956995,
      "grad_norm": 0.734375,
      "learning_rate": 0.0006074760407160828,
      "loss": 1.0215,
      "step": 4997
    },
    {
      "epoch": 0.6397030590042238,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0006070947793907197,
      "loss": 1.4747,
      "step": 4998
    },
    {
      "epoch": 0.6398310508127479,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0006067135855924937,
      "loss": 1.282,
      "step": 4999
    },
    {
      "epoch": 0.6399590426212722,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0006063324593869189,
      "loss": 1.0313,
      "step": 5000
    },
    {
      "epoch": 0.6400870344297965,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0006059514008394982,
      "loss": 1.2856,
      "step": 5001
    },
    {
      "epoch": 0.6402150262383207,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0006055704100157225,
      "loss": 1.4022,
      "step": 5002
    },
    {
      "epoch": 0.640343018046845,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0006051894869810713,
      "loss": 1.2325,
      "step": 5003
    },
    {
      "epoch": 0.6404710098553693,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0006048086318010127,
      "loss": 0.7098,
      "step": 5004
    },
    {
      "epoch": 0.6405990016638935,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0006044278445410025,
      "loss": 1.8001,
      "step": 5005
    },
    {
      "epoch": 0.6407269934724178,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0006040471252664851,
      "loss": 1.133,
      "step": 5006
    },
    {
      "epoch": 0.640854985280942,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0006036664740428937,
      "loss": 1.4818,
      "step": 5007
    },
    {
      "epoch": 0.6409829770894663,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0006032858909356494,
      "loss": 1.7535,
      "step": 5008
    },
    {
      "epoch": 0.6411109688979906,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0006029053760101611,
      "loss": 1.5021,
      "step": 5009
    },
    {
      "epoch": 0.6412389607065148,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0006025249293318265,
      "loss": 1.033,
      "step": 5010
    },
    {
      "epoch": 0.6413669525150391,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0006021445509660315,
      "loss": 1.0947,
      "step": 5011
    },
    {
      "epoch": 0.6414949443235632,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0006017642409781508,
      "loss": 1.3554,
      "step": 5012
    },
    {
      "epoch": 0.6416229361320875,
      "grad_norm": 0.6328125,
      "learning_rate": 0.000601383999433547,
      "loss": 1.2178,
      "step": 5013
    },
    {
      "epoch": 0.6417509279406118,
      "grad_norm": 0.671875,
      "learning_rate": 0.00060100382639757,
      "loss": 1.8817,
      "step": 5014
    },
    {
      "epoch": 0.641878919749136,
      "grad_norm": 0.71875,
      "learning_rate": 0.0006006237219355591,
      "loss": 1.2399,
      "step": 5015
    },
    {
      "epoch": 0.6420069115576603,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0006002436861128411,
      "loss": 1.2032,
      "step": 5016
    },
    {
      "epoch": 0.6421349033661846,
      "grad_norm": 0.64453125,
      "learning_rate": 0.000599863718994732,
      "loss": 1.0483,
      "step": 5017
    },
    {
      "epoch": 0.6422628951747088,
      "grad_norm": 0.671875,
      "learning_rate": 0.0005994838206465346,
      "loss": 1.7032,
      "step": 5018
    },
    {
      "epoch": 0.6423908869832331,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0005991039911335409,
      "loss": 0.7853,
      "step": 5019
    },
    {
      "epoch": 0.6425188787917573,
      "grad_norm": 0.640625,
      "learning_rate": 0.000598724230521031,
      "loss": 1.3414,
      "step": 5020
    },
    {
      "epoch": 0.6426468706002816,
      "grad_norm": 0.65625,
      "learning_rate": 0.0005983445388742726,
      "loss": 1.7385,
      "step": 5021
    },
    {
      "epoch": 0.6427748624088059,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0005979649162585219,
      "loss": 1.6314,
      "step": 5022
    },
    {
      "epoch": 0.6429028542173301,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0005975853627390232,
      "loss": 1.4017,
      "step": 5023
    },
    {
      "epoch": 0.6430308460258544,
      "grad_norm": 0.609375,
      "learning_rate": 0.000597205878381009,
      "loss": 1.2564,
      "step": 5024
    },
    {
      "epoch": 0.6431588378343785,
      "grad_norm": 0.765625,
      "learning_rate": 0.0005968264632496998,
      "loss": 0.9936,
      "step": 5025
    },
    {
      "epoch": 0.6432868296429028,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0005964471174103047,
      "loss": 1.5215,
      "step": 5026
    },
    {
      "epoch": 0.6434148214514271,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0005960678409280195,
      "loss": 1.2219,
      "step": 5027
    },
    {
      "epoch": 0.6435428132599513,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0005956886338680297,
      "loss": 1.0766,
      "step": 5028
    },
    {
      "epoch": 0.6436708050684756,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0005953094962955081,
      "loss": 1.4005,
      "step": 5029
    },
    {
      "epoch": 0.6437987968769999,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0005949304282756158,
      "loss": 1.2517,
      "step": 5030
    },
    {
      "epoch": 0.6439267886855241,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0005945514298735013,
      "loss": 1.494,
      "step": 5031
    },
    {
      "epoch": 0.6440547804940484,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0005941725011543018,
      "loss": 1.2447,
      "step": 5032
    },
    {
      "epoch": 0.6441827723025726,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0005937936421831425,
      "loss": 1.1716,
      "step": 5033
    },
    {
      "epoch": 0.6443107641110969,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0005934148530251362,
      "loss": 1.0756,
      "step": 5034
    },
    {
      "epoch": 0.6444387559196212,
      "grad_norm": 0.671875,
      "learning_rate": 0.0005930361337453847,
      "loss": 1.6194,
      "step": 5035
    },
    {
      "epoch": 0.6445667477281454,
      "grad_norm": 0.546875,
      "learning_rate": 0.000592657484408976,
      "loss": 1.1681,
      "step": 5036
    },
    {
      "epoch": 0.6446947395366697,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0005922789050809878,
      "loss": 0.8457,
      "step": 5037
    },
    {
      "epoch": 0.6448227313451939,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0005919003958264848,
      "loss": 1.2229,
      "step": 5038
    },
    {
      "epoch": 0.6449507231537182,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0005915219567105205,
      "loss": 1.5029,
      "step": 5039
    },
    {
      "epoch": 0.6450787149622424,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0005911435877981351,
      "loss": 1.2128,
      "step": 5040
    },
    {
      "epoch": 0.6452067067707666,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0005907652891543576,
      "loss": 1.0483,
      "step": 5041
    },
    {
      "epoch": 0.6453346985792909,
      "grad_norm": 0.58203125,
      "learning_rate": 0.000590387060844205,
      "loss": 1.0554,
      "step": 5042
    },
    {
      "epoch": 0.6454626903878152,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0005900089029326818,
      "loss": 1.4797,
      "step": 5043
    },
    {
      "epoch": 0.6455906821963394,
      "grad_norm": 0.6484375,
      "learning_rate": 0.000589630815484781,
      "loss": 1.5789,
      "step": 5044
    },
    {
      "epoch": 0.6457186740048637,
      "grad_norm": 0.609375,
      "learning_rate": 0.0005892527985654824,
      "loss": 1.0829,
      "step": 5045
    },
    {
      "epoch": 0.6458466658133879,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0005888748522397547,
      "loss": 1.3243,
      "step": 5046
    },
    {
      "epoch": 0.6459746576219122,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0005884969765725539,
      "loss": 1.1866,
      "step": 5047
    },
    {
      "epoch": 0.6461026494304365,
      "grad_norm": 0.71875,
      "learning_rate": 0.0005881191716288248,
      "loss": 1.751,
      "step": 5048
    },
    {
      "epoch": 0.6462306412389607,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0005877414374734982,
      "loss": 1.1301,
      "step": 5049
    },
    {
      "epoch": 0.646358633047485,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0005873637741714941,
      "loss": 1.8711,
      "step": 5050
    },
    {
      "epoch": 0.6464866248560092,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0005869861817877207,
      "loss": 0.601,
      "step": 5051
    },
    {
      "epoch": 0.6466146166645335,
      "grad_norm": 0.546875,
      "learning_rate": 0.0005866086603870733,
      "loss": 1.0536,
      "step": 5052
    },
    {
      "epoch": 0.6467426084730578,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0005862312100344345,
      "loss": 1.0077,
      "step": 5053
    },
    {
      "epoch": 0.6468706002815819,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0005858538307946755,
      "loss": 0.9364,
      "step": 5054
    },
    {
      "epoch": 0.6469985920901062,
      "grad_norm": 0.7109375,
      "learning_rate": 0.000585476522732655,
      "loss": 1.417,
      "step": 5055
    },
    {
      "epoch": 0.6471265838986305,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0005850992859132197,
      "loss": 1.094,
      "step": 5056
    },
    {
      "epoch": 0.6472545757071547,
      "grad_norm": 0.60546875,
      "learning_rate": 0.000584722120401204,
      "loss": 1.1563,
      "step": 5057
    },
    {
      "epoch": 0.647382567515679,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0005843450262614293,
      "loss": 1.5279,
      "step": 5058
    },
    {
      "epoch": 0.6475105593242032,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0005839680035587061,
      "loss": 1.5112,
      "step": 5059
    },
    {
      "epoch": 0.6476385511327275,
      "grad_norm": 0.72265625,
      "learning_rate": 0.000583591052357831,
      "loss": 1.487,
      "step": 5060
    },
    {
      "epoch": 0.6477665429412518,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0005832141727235905,
      "loss": 1.3857,
      "step": 5061
    },
    {
      "epoch": 0.647894534749776,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0005828373647207561,
      "loss": 1.8734,
      "step": 5062
    },
    {
      "epoch": 0.6480225265583003,
      "grad_norm": 0.64453125,
      "learning_rate": 0.000582460628414089,
      "loss": 1.3269,
      "step": 5063
    },
    {
      "epoch": 0.6481505183668245,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0005820839638683373,
      "loss": 1.1869,
      "step": 5064
    },
    {
      "epoch": 0.6482785101753488,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0005817073711482371,
      "loss": 1.14,
      "step": 5065
    },
    {
      "epoch": 0.6484065019838731,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0005813308503185122,
      "loss": 2.0241,
      "step": 5066
    },
    {
      "epoch": 0.6485344937923972,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0005809544014438732,
      "loss": 1.3701,
      "step": 5067
    },
    {
      "epoch": 0.6486624856009215,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0005805780245890191,
      "loss": 1.3391,
      "step": 5068
    },
    {
      "epoch": 0.6487904774094458,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0005802017198186366,
      "loss": 1.1065,
      "step": 5069
    },
    {
      "epoch": 0.64891846921797,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0005798254871973996,
      "loss": 0.7584,
      "step": 5070
    },
    {
      "epoch": 0.6490464610264943,
      "grad_norm": 0.890625,
      "learning_rate": 0.0005794493267899699,
      "loss": 1.3103,
      "step": 5071
    },
    {
      "epoch": 0.6491744528350185,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0005790732386609966,
      "loss": 1.5109,
      "step": 5072
    },
    {
      "epoch": 0.6493024446435428,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0005786972228751169,
      "loss": 1.4972,
      "step": 5073
    },
    {
      "epoch": 0.6494304364520671,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0005783212794969548,
      "loss": 1.6149,
      "step": 5074
    },
    {
      "epoch": 0.6495584282605913,
      "grad_norm": 0.75,
      "learning_rate": 0.0005779454085911229,
      "loss": 1.5918,
      "step": 5075
    },
    {
      "epoch": 0.6496864200691156,
      "grad_norm": 0.625,
      "learning_rate": 0.0005775696102222199,
      "loss": 1.296,
      "step": 5076
    },
    {
      "epoch": 0.6498144118776399,
      "grad_norm": 0.69921875,
      "learning_rate": 0.000577193884454833,
      "loss": 1.0357,
      "step": 5077
    },
    {
      "epoch": 0.6499424036861641,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0005768182313535372,
      "loss": 1.042,
      "step": 5078
    },
    {
      "epoch": 0.6500703954946884,
      "grad_norm": 0.765625,
      "learning_rate": 0.0005764426509828948,
      "loss": 1.7256,
      "step": 5079
    },
    {
      "epoch": 0.6501983873032126,
      "grad_norm": 0.703125,
      "learning_rate": 0.0005760671434074541,
      "loss": 1.551,
      "step": 5080
    },
    {
      "epoch": 0.6503263791117369,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0005756917086917532,
      "loss": 1.2646,
      "step": 5081
    },
    {
      "epoch": 0.6504543709202611,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0005753163469003167,
      "loss": 1.3207,
      "step": 5082
    },
    {
      "epoch": 0.6505823627287853,
      "grad_norm": 0.6875,
      "learning_rate": 0.0005749410580976568,
      "loss": 1.1246,
      "step": 5083
    },
    {
      "epoch": 0.6507103545373096,
      "grad_norm": 0.5625,
      "learning_rate": 0.000574565842348272,
      "loss": 0.9948,
      "step": 5084
    },
    {
      "epoch": 0.6508383463458338,
      "grad_norm": 0.765625,
      "learning_rate": 0.00057419069971665,
      "loss": 1.2798,
      "step": 5085
    },
    {
      "epoch": 0.6509663381543581,
      "grad_norm": 0.6875,
      "learning_rate": 0.0005738156302672646,
      "loss": 1.9316,
      "step": 5086
    },
    {
      "epoch": 0.6510943299628824,
      "grad_norm": 0.55859375,
      "learning_rate": 0.000573440634064578,
      "loss": 1.0281,
      "step": 5087
    },
    {
      "epoch": 0.6512223217714066,
      "grad_norm": 0.65625,
      "learning_rate": 0.0005730657111730397,
      "loss": 1.3116,
      "step": 5088
    },
    {
      "epoch": 0.6513503135799309,
      "grad_norm": 0.515625,
      "learning_rate": 0.0005726908616570854,
      "loss": 0.9845,
      "step": 5089
    },
    {
      "epoch": 0.6514783053884552,
      "grad_norm": 0.8125,
      "learning_rate": 0.0005723160855811395,
      "loss": 0.8995,
      "step": 5090
    },
    {
      "epoch": 0.6516062971969794,
      "grad_norm": 0.796875,
      "learning_rate": 0.0005719413830096134,
      "loss": 1.545,
      "step": 5091
    },
    {
      "epoch": 0.6517342890055037,
      "grad_norm": 0.640625,
      "learning_rate": 0.0005715667540069059,
      "loss": 1.4223,
      "step": 5092
    },
    {
      "epoch": 0.6518622808140279,
      "grad_norm": 0.609375,
      "learning_rate": 0.0005711921986374026,
      "loss": 0.7426,
      "step": 5093
    },
    {
      "epoch": 0.6519902726225522,
      "grad_norm": 0.609375,
      "learning_rate": 0.0005708177169654776,
      "loss": 1.1203,
      "step": 5094
    },
    {
      "epoch": 0.6521182644310765,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0005704433090554912,
      "loss": 1.3863,
      "step": 5095
    },
    {
      "epoch": 0.6522462562396006,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0005700689749717915,
      "loss": 1.0289,
      "step": 5096
    },
    {
      "epoch": 0.6523742480481249,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0005696947147787144,
      "loss": 1.4358,
      "step": 5097
    },
    {
      "epoch": 0.6525022398566491,
      "grad_norm": 0.68359375,
      "learning_rate": 0.000569320528540582,
      "loss": 1.5146,
      "step": 5098
    },
    {
      "epoch": 0.6526302316651734,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0005689464163217042,
      "loss": 1.0689,
      "step": 5099
    },
    {
      "epoch": 0.6527582234736977,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0005685723781863785,
      "loss": 1.0983,
      "step": 5100
    },
    {
      "epoch": 0.6528862152822219,
      "grad_norm": 0.640625,
      "learning_rate": 0.0005681984141988899,
      "loss": 1.6144,
      "step": 5101
    },
    {
      "epoch": 0.6530142070907462,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0005678245244235093,
      "loss": 1.1565,
      "step": 5102
    },
    {
      "epoch": 0.6531421988992705,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0005674507089244957,
      "loss": 1.36,
      "step": 5103
    },
    {
      "epoch": 0.6532701907077947,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0005670769677660962,
      "loss": 1.5307,
      "step": 5104
    },
    {
      "epoch": 0.653398182516319,
      "grad_norm": 0.6875,
      "learning_rate": 0.0005667033010125442,
      "loss": 1.1854,
      "step": 5105
    },
    {
      "epoch": 0.6535261743248432,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0005663297087280597,
      "loss": 1.4473,
      "step": 5106
    },
    {
      "epoch": 0.6536541661333675,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0005659561909768509,
      "loss": 1.3884,
      "step": 5107
    },
    {
      "epoch": 0.6537821579418918,
      "grad_norm": 0.65234375,
      "learning_rate": 0.000565582747823113,
      "loss": 1.3487,
      "step": 5108
    },
    {
      "epoch": 0.653910149750416,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0005652093793310282,
      "loss": 1.1562,
      "step": 5109
    },
    {
      "epoch": 0.6540381415589402,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0005648360855647663,
      "loss": 1.027,
      "step": 5110
    },
    {
      "epoch": 0.6541661333674644,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0005644628665884832,
      "loss": 1.0925,
      "step": 5111
    },
    {
      "epoch": 0.6542941251759887,
      "grad_norm": 0.71875,
      "learning_rate": 0.0005640897224663228,
      "loss": 1.8154,
      "step": 5112
    },
    {
      "epoch": 0.654422116984513,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0005637166532624163,
      "loss": 1.3287,
      "step": 5113
    },
    {
      "epoch": 0.6545501087930372,
      "grad_norm": 0.671875,
      "learning_rate": 0.0005633436590408814,
      "loss": 1.4533,
      "step": 5114
    },
    {
      "epoch": 0.6546781006015615,
      "grad_norm": 0.796875,
      "learning_rate": 0.0005629707398658234,
      "loss": 1.4055,
      "step": 5115
    },
    {
      "epoch": 0.6548060924100858,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0005625978958013344,
      "loss": 0.8711,
      "step": 5116
    },
    {
      "epoch": 0.65493408421861,
      "grad_norm": 0.6484375,
      "learning_rate": 0.000562225126911494,
      "loss": 1.5439,
      "step": 5117
    },
    {
      "epoch": 0.6550620760271343,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0005618524332603681,
      "loss": 1.0745,
      "step": 5118
    },
    {
      "epoch": 0.6551900678356585,
      "grad_norm": 0.65625,
      "learning_rate": 0.0005614798149120108,
      "loss": 1.4413,
      "step": 5119
    },
    {
      "epoch": 0.6553180596441828,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0005611072719304617,
      "loss": 1.6468,
      "step": 5120
    },
    {
      "epoch": 0.6554460514527071,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0005607348043797491,
      "loss": 1.4039,
      "step": 5121
    },
    {
      "epoch": 0.6555740432612313,
      "grad_norm": 0.765625,
      "learning_rate": 0.0005603624123238871,
      "loss": 1.9516,
      "step": 5122
    },
    {
      "epoch": 0.6557020350697556,
      "grad_norm": 0.6875,
      "learning_rate": 0.0005599900958268781,
      "loss": 1.4599,
      "step": 5123
    },
    {
      "epoch": 0.6558300268782797,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0005596178549527097,
      "loss": 1.6002,
      "step": 5124
    },
    {
      "epoch": 0.655958018686804,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0005592456897653578,
      "loss": 1.4348,
      "step": 5125
    },
    {
      "epoch": 0.6560860104953283,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0005588736003287854,
      "loss": 1.1642,
      "step": 5126
    },
    {
      "epoch": 0.6562140023038525,
      "grad_norm": 0.6875,
      "learning_rate": 0.000558501586706942,
      "loss": 1.7344,
      "step": 5127
    },
    {
      "epoch": 0.6563419941123768,
      "grad_norm": 0.6875,
      "learning_rate": 0.0005581296489637647,
      "loss": 1.266,
      "step": 5128
    },
    {
      "epoch": 0.6564699859209011,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0005577577871631759,
      "loss": 1.6556,
      "step": 5129
    },
    {
      "epoch": 0.6565979777294253,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0005573860013690867,
      "loss": 1.313,
      "step": 5130
    },
    {
      "epoch": 0.6567259695379496,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0005570142916453945,
      "loss": 1.7377,
      "step": 5131
    },
    {
      "epoch": 0.6568539613464738,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0005566426580559839,
      "loss": 1.3629,
      "step": 5132
    },
    {
      "epoch": 0.6569819531549981,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0005562711006647257,
      "loss": 1.8414,
      "step": 5133
    },
    {
      "epoch": 0.6571099449635224,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0005558996195354779,
      "loss": 1.0037,
      "step": 5134
    },
    {
      "epoch": 0.6572379367720466,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0005555282147320862,
      "loss": 1.0892,
      "step": 5135
    },
    {
      "epoch": 0.6573659285805709,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0005551568863183822,
      "loss": 1.6798,
      "step": 5136
    },
    {
      "epoch": 0.657493920389095,
      "grad_norm": 0.8125,
      "learning_rate": 0.0005547856343581847,
      "loss": 1.4363,
      "step": 5137
    },
    {
      "epoch": 0.6576219121976193,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0005544144589152995,
      "loss": 1.3957,
      "step": 5138
    },
    {
      "epoch": 0.6577499040061436,
      "grad_norm": 0.65625,
      "learning_rate": 0.000554043360053519,
      "loss": 1.153,
      "step": 5139
    },
    {
      "epoch": 0.6578778958146678,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0005536723378366227,
      "loss": 1.515,
      "step": 5140
    },
    {
      "epoch": 0.6580058876231921,
      "grad_norm": 0.5859375,
      "learning_rate": 0.000553301392328377,
      "loss": 0.9713,
      "step": 5141
    },
    {
      "epoch": 0.6581338794317164,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0005529305235925341,
      "loss": 1.0887,
      "step": 5142
    },
    {
      "epoch": 0.6582618712402406,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0005525597316928344,
      "loss": 1.1021,
      "step": 5143
    },
    {
      "epoch": 0.6583898630487649,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0005521890166930044,
      "loss": 1.2916,
      "step": 5144
    },
    {
      "epoch": 0.6585178548572891,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0005518183786567581,
      "loss": 0.8911,
      "step": 5145
    },
    {
      "epoch": 0.6586458466658134,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0005514478176477944,
      "loss": 1.4778,
      "step": 5146
    },
    {
      "epoch": 0.6587738384743377,
      "grad_norm": 0.64453125,
      "learning_rate": 0.000551077333729801,
      "loss": 1.6903,
      "step": 5147
    },
    {
      "epoch": 0.6589018302828619,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0005507069269664512,
      "loss": 1.115,
      "step": 5148
    },
    {
      "epoch": 0.6590298220913862,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0005503365974214059,
      "loss": 1.1155,
      "step": 5149
    },
    {
      "epoch": 0.6591578138999105,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0005499663451583125,
      "loss": 1.2685,
      "step": 5150
    },
    {
      "epoch": 0.6592858057084346,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0005495961702408038,
      "loss": 1.5617,
      "step": 5151
    },
    {
      "epoch": 0.6594137975169589,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0005492260727325011,
      "loss": 1.7788,
      "step": 5152
    },
    {
      "epoch": 0.6595417893254831,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0005488560526970116,
      "loss": 1.0001,
      "step": 5153
    },
    {
      "epoch": 0.6596697811340074,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0005484861101979294,
      "loss": 1.8185,
      "step": 5154
    },
    {
      "epoch": 0.6597977729425317,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0005481162452988344,
      "loss": 1.4408,
      "step": 5155
    },
    {
      "epoch": 0.6599257647510559,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0005477464580632944,
      "loss": 1.3478,
      "step": 5156
    },
    {
      "epoch": 0.6600537565595802,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0005473767485548634,
      "loss": 1.3659,
      "step": 5157
    },
    {
      "epoch": 0.6601817483681044,
      "grad_norm": 0.6875,
      "learning_rate": 0.0005470071168370819,
      "loss": 1.5904,
      "step": 5158
    },
    {
      "epoch": 0.6603097401766287,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0005466375629734767,
      "loss": 1.3782,
      "step": 5159
    },
    {
      "epoch": 0.660437731985153,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0005462680870275624,
      "loss": 1.3528,
      "step": 5160
    },
    {
      "epoch": 0.6605657237936772,
      "grad_norm": 0.75,
      "learning_rate": 0.0005458986890628388,
      "loss": 1.6975,
      "step": 5161
    },
    {
      "epoch": 0.6606937156022015,
      "grad_norm": 0.703125,
      "learning_rate": 0.0005455293691427934,
      "loss": 1.4231,
      "step": 5162
    },
    {
      "epoch": 0.6608217074107258,
      "grad_norm": 0.6875,
      "learning_rate": 0.0005451601273308997,
      "loss": 1.2495,
      "step": 5163
    },
    {
      "epoch": 0.66094969921925,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0005447909636906176,
      "loss": 2.2551,
      "step": 5164
    },
    {
      "epoch": 0.6610776910277743,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0005444218782853941,
      "loss": 1.4076,
      "step": 5165
    },
    {
      "epoch": 0.6612056828362984,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0005440528711786623,
      "loss": 0.7835,
      "step": 5166
    },
    {
      "epoch": 0.6613336746448227,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0005436839424338425,
      "loss": 1.1845,
      "step": 5167
    },
    {
      "epoch": 0.661461666453347,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0005433150921143406,
      "loss": 1.8271,
      "step": 5168
    },
    {
      "epoch": 0.6615896582618712,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0005429463202835498,
      "loss": 1.5447,
      "step": 5169
    },
    {
      "epoch": 0.6617176500703955,
      "grad_norm": 0.625,
      "learning_rate": 0.0005425776270048492,
      "loss": 1.2679,
      "step": 5170
    },
    {
      "epoch": 0.6618456418789197,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0005422090123416048,
      "loss": 1.4374,
      "step": 5171
    },
    {
      "epoch": 0.661973633687444,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0005418404763571698,
      "loss": 1.0942,
      "step": 5172
    },
    {
      "epoch": 0.6621016254959683,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0005414720191148822,
      "loss": 1.3071,
      "step": 5173
    },
    {
      "epoch": 0.6622296173044925,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0005411036406780676,
      "loss": 1.7055,
      "step": 5174
    },
    {
      "epoch": 0.6623576091130168,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0005407353411100378,
      "loss": 1.3506,
      "step": 5175
    },
    {
      "epoch": 0.6624856009215411,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0005403671204740915,
      "loss": 1.8215,
      "step": 5176
    },
    {
      "epoch": 0.6626135927300653,
      "grad_norm": 0.6875,
      "learning_rate": 0.0005399989788335125,
      "loss": 1.555,
      "step": 5177
    },
    {
      "epoch": 0.6627415845385896,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0005396309162515725,
      "loss": 1.108,
      "step": 5178
    },
    {
      "epoch": 0.6628695763471137,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0005392629327915288,
      "loss": 1.2861,
      "step": 5179
    },
    {
      "epoch": 0.662997568155638,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0005388950285166255,
      "loss": 1.3633,
      "step": 5180
    },
    {
      "epoch": 0.6631255599641623,
      "grad_norm": 1.0625,
      "learning_rate": 0.000538527203490093,
      "loss": 1.0638,
      "step": 5181
    },
    {
      "epoch": 0.6632535517726865,
      "grad_norm": 1.4375,
      "learning_rate": 0.0005381594577751476,
      "loss": 1.2699,
      "step": 5182
    },
    {
      "epoch": 0.6633815435812108,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0005377917914349929,
      "loss": 1.4237,
      "step": 5183
    },
    {
      "epoch": 0.663509535389735,
      "grad_norm": 0.5859375,
      "learning_rate": 0.000537424204532818,
      "loss": 0.9474,
      "step": 5184
    },
    {
      "epoch": 0.6636375271982593,
      "grad_norm": 0.59375,
      "learning_rate": 0.000537056697131799,
      "loss": 1.3439,
      "step": 5185
    },
    {
      "epoch": 0.6637655190067836,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0005366892692950974,
      "loss": 1.3281,
      "step": 5186
    },
    {
      "epoch": 0.6638935108153078,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0005363219210858621,
      "loss": 1.4607,
      "step": 5187
    },
    {
      "epoch": 0.6640215026238321,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0005359546525672275,
      "loss": 1.9073,
      "step": 5188
    },
    {
      "epoch": 0.6641494944323564,
      "grad_norm": 0.765625,
      "learning_rate": 0.0005355874638023155,
      "loss": 1.6473,
      "step": 5189
    },
    {
      "epoch": 0.6642774862408806,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0005352203548542322,
      "loss": 1.6528,
      "step": 5190
    },
    {
      "epoch": 0.6644054780494049,
      "grad_norm": 0.62890625,
      "learning_rate": 0.000534853325786072,
      "loss": 1.56,
      "step": 5191
    },
    {
      "epoch": 0.664533469857929,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0005344863766609147,
      "loss": 1.5443,
      "step": 5192
    },
    {
      "epoch": 0.6646614616664533,
      "grad_norm": 0.921875,
      "learning_rate": 0.000534119507541826,
      "loss": 1.5247,
      "step": 5193
    },
    {
      "epoch": 0.6647894534749776,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0005337527184918594,
      "loss": 1.5099,
      "step": 5194
    },
    {
      "epoch": 0.6649174452835018,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0005333860095740525,
      "loss": 1.9678,
      "step": 5195
    },
    {
      "epoch": 0.6650454370920261,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0005330193808514307,
      "loss": 1.144,
      "step": 5196
    },
    {
      "epoch": 0.6651734289005503,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0005326528323870048,
      "loss": 2.119,
      "step": 5197
    },
    {
      "epoch": 0.6653014207090746,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0005322863642437727,
      "loss": 1.5036,
      "step": 5198
    },
    {
      "epoch": 0.6654294125175989,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0005319199764847171,
      "loss": 1.2921,
      "step": 5199
    },
    {
      "epoch": 0.6655574043261231,
      "grad_norm": 0.70703125,
      "learning_rate": 0.000531553669172808,
      "loss": 1.6066,
      "step": 5200
    },
    {
      "epoch": 0.6656853961346474,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0005311874423710013,
      "loss": 1.0669,
      "step": 5201
    },
    {
      "epoch": 0.6658133879431717,
      "grad_norm": 0.66796875,
      "learning_rate": 0.000530821296142239,
      "loss": 1.212,
      "step": 5202
    },
    {
      "epoch": 0.6659413797516959,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0005304552305494492,
      "loss": 1.1703,
      "step": 5203
    },
    {
      "epoch": 0.6660693715602202,
      "grad_norm": 0.765625,
      "learning_rate": 0.0005300892456555463,
      "loss": 2.0756,
      "step": 5204
    },
    {
      "epoch": 0.6661973633687444,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0005297233415234307,
      "loss": 1.316,
      "step": 5205
    },
    {
      "epoch": 0.6663253551772687,
      "grad_norm": 0.84375,
      "learning_rate": 0.000529357518215989,
      "loss": 1.6274,
      "step": 5206
    },
    {
      "epoch": 0.666453346985793,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0005289917757960943,
      "loss": 1.808,
      "step": 5207
    },
    {
      "epoch": 0.6665813387943171,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0005286261143266044,
      "loss": 1.1153,
      "step": 5208
    },
    {
      "epoch": 0.6667093306028414,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0005282605338703649,
      "loss": 1.7069,
      "step": 5209
    },
    {
      "epoch": 0.6668373224113656,
      "grad_norm": 0.578125,
      "learning_rate": 0.0005278950344902064,
      "loss": 1.0566,
      "step": 5210
    },
    {
      "epoch": 0.6669653142198899,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000527529616248946,
      "loss": 1.2708,
      "step": 5211
    },
    {
      "epoch": 0.6670933060284142,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0005271642792093872,
      "loss": 1.2833,
      "step": 5212
    },
    {
      "epoch": 0.6672212978369384,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0005267990234343182,
      "loss": 1.1994,
      "step": 5213
    },
    {
      "epoch": 0.6673492896454627,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0005264338489865145,
      "loss": 1.3932,
      "step": 5214
    },
    {
      "epoch": 0.667477281453987,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0005260687559287376,
      "loss": 1.6421,
      "step": 5215
    },
    {
      "epoch": 0.6676052732625112,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0005257037443237342,
      "loss": 1.5255,
      "step": 5216
    },
    {
      "epoch": 0.6677332650710355,
      "grad_norm": 0.57421875,
      "learning_rate": 0.000525338814234238,
      "loss": 1.1002,
      "step": 5217
    },
    {
      "epoch": 0.6678612568795597,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0005249739657229675,
      "loss": 1.1163,
      "step": 5218
    },
    {
      "epoch": 0.667989248688084,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0005246091988526283,
      "loss": 1.0708,
      "step": 5219
    },
    {
      "epoch": 0.6681172404966083,
      "grad_norm": 0.65625,
      "learning_rate": 0.0005242445136859117,
      "loss": 1.4233,
      "step": 5220
    },
    {
      "epoch": 0.6682452323051324,
      "grad_norm": 0.59375,
      "learning_rate": 0.0005238799102854941,
      "loss": 1.5655,
      "step": 5221
    },
    {
      "epoch": 0.6683732241136567,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0005235153887140389,
      "loss": 1.2544,
      "step": 5222
    },
    {
      "epoch": 0.668501215922181,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0005231509490341949,
      "loss": 1.2307,
      "step": 5223
    },
    {
      "epoch": 0.6686292077307052,
      "grad_norm": 0.703125,
      "learning_rate": 0.0005227865913085972,
      "loss": 1.0569,
      "step": 5224
    },
    {
      "epoch": 0.6687571995392295,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0005224223155998663,
      "loss": 1.3251,
      "step": 5225
    },
    {
      "epoch": 0.6688851913477537,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0005220581219706091,
      "loss": 0.8304,
      "step": 5226
    },
    {
      "epoch": 0.669013183156278,
      "grad_norm": 0.6875,
      "learning_rate": 0.0005216940104834181,
      "loss": 1.2095,
      "step": 5227
    },
    {
      "epoch": 0.6691411749648023,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0005213299812008718,
      "loss": 1.2238,
      "step": 5228
    },
    {
      "epoch": 0.6692691667733265,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0005209660341855351,
      "loss": 1.3511,
      "step": 5229
    },
    {
      "epoch": 0.6693971585818508,
      "grad_norm": 0.7578125,
      "learning_rate": 0.000520602169499957,
      "loss": 2.3763,
      "step": 5230
    },
    {
      "epoch": 0.669525150390375,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0005202383872066744,
      "loss": 1.2547,
      "step": 5231
    },
    {
      "epoch": 0.6696531421988993,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0005198746873682088,
      "loss": 1.1886,
      "step": 5232
    },
    {
      "epoch": 0.6697811340074236,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0005195110700470683,
      "loss": 1.4403,
      "step": 5233
    },
    {
      "epoch": 0.6699091258159477,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0005191475353057464,
      "loss": 1.2551,
      "step": 5234
    },
    {
      "epoch": 0.670037117624472,
      "grad_norm": 0.515625,
      "learning_rate": 0.0005187840832067221,
      "loss": 0.8575,
      "step": 5235
    },
    {
      "epoch": 0.6701651094329963,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0005184207138124605,
      "loss": 0.7625,
      "step": 5236
    },
    {
      "epoch": 0.6702931012415205,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0005180574271854128,
      "loss": 1.1778,
      "step": 5237
    },
    {
      "epoch": 0.6704210930500448,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0005176942233880157,
      "loss": 1.0791,
      "step": 5238
    },
    {
      "epoch": 0.670549084858569,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0005173311024826916,
      "loss": 1.6703,
      "step": 5239
    },
    {
      "epoch": 0.6706770766670933,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0005169680645318486,
      "loss": 1.8026,
      "step": 5240
    },
    {
      "epoch": 0.6708050684756176,
      "grad_norm": 0.68359375,
      "learning_rate": 0.000516605109597881,
      "loss": 1.1484,
      "step": 5241
    },
    {
      "epoch": 0.6709330602841418,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0005162422377431681,
      "loss": 1.7545,
      "step": 5242
    },
    {
      "epoch": 0.6710610520926661,
      "grad_norm": 0.625,
      "learning_rate": 0.000515879449030076,
      "loss": 1.1391,
      "step": 5243
    },
    {
      "epoch": 0.6711890439011903,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0005155167435209549,
      "loss": 1.1684,
      "step": 5244
    },
    {
      "epoch": 0.6713170357097146,
      "grad_norm": 0.6484375,
      "learning_rate": 0.000515154121278142,
      "loss": 0.9142,
      "step": 5245
    },
    {
      "epoch": 0.6714450275182389,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0005147915823639599,
      "loss": 1.3487,
      "step": 5246
    },
    {
      "epoch": 0.671573019326763,
      "grad_norm": 0.6875,
      "learning_rate": 0.0005144291268407167,
      "loss": 1.4144,
      "step": 5247
    },
    {
      "epoch": 0.6717010111352874,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0005140667547707064,
      "loss": 0.6348,
      "step": 5248
    },
    {
      "epoch": 0.6718290029438116,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0005137044662162085,
      "loss": 1.0744,
      "step": 5249
    },
    {
      "epoch": 0.6719569947523358,
      "grad_norm": 0.63671875,
      "learning_rate": 0.000513342261239488,
      "loss": 1.3106,
      "step": 5250
    },
    {
      "epoch": 0.6720849865608601,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0005129801399027963,
      "loss": 1.1352,
      "step": 5251
    },
    {
      "epoch": 0.6722129783693843,
      "grad_norm": 0.734375,
      "learning_rate": 0.0005126181022683689,
      "loss": 1.2877,
      "step": 5252
    },
    {
      "epoch": 0.6723409701779086,
      "grad_norm": 0.78125,
      "learning_rate": 0.0005122561483984282,
      "loss": 1.3454,
      "step": 5253
    },
    {
      "epoch": 0.6724689619864329,
      "grad_norm": 0.734375,
      "learning_rate": 0.0005118942783551818,
      "loss": 1.656,
      "step": 5254
    },
    {
      "epoch": 0.6725969537949571,
      "grad_norm": 0.703125,
      "learning_rate": 0.0005115324922008233,
      "loss": 1.6532,
      "step": 5255
    },
    {
      "epoch": 0.6727249456034814,
      "grad_norm": 0.65625,
      "learning_rate": 0.0005111707899975317,
      "loss": 1.2042,
      "step": 5256
    },
    {
      "epoch": 0.6728529374120056,
      "grad_norm": 0.546875,
      "learning_rate": 0.0005108091718074706,
      "loss": 0.9941,
      "step": 5257
    },
    {
      "epoch": 0.6729809292205299,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0005104476376927901,
      "loss": 1.4217,
      "step": 5258
    },
    {
      "epoch": 0.6731089210290542,
      "grad_norm": 0.6875,
      "learning_rate": 0.000510086187715626,
      "loss": 1.8042,
      "step": 5259
    },
    {
      "epoch": 0.6732369128375784,
      "grad_norm": 0.578125,
      "learning_rate": 0.0005097248219380994,
      "loss": 0.9559,
      "step": 5260
    },
    {
      "epoch": 0.6733649046461027,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0005093635404223167,
      "loss": 1.4691,
      "step": 5261
    },
    {
      "epoch": 0.673492896454627,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00050900234323037,
      "loss": 1.5334,
      "step": 5262
    },
    {
      "epoch": 0.6736208882631511,
      "grad_norm": 0.640625,
      "learning_rate": 0.0005086412304243369,
      "loss": 1.0055,
      "step": 5263
    },
    {
      "epoch": 0.6737488800716754,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0005082802020662807,
      "loss": 1.2939,
      "step": 5264
    },
    {
      "epoch": 0.6738768718801996,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0005079192582182501,
      "loss": 1.1867,
      "step": 5265
    },
    {
      "epoch": 0.6740048636887239,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0005075583989422783,
      "loss": 1.6785,
      "step": 5266
    },
    {
      "epoch": 0.6741328554972482,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0005071976243003855,
      "loss": 1.6552,
      "step": 5267
    },
    {
      "epoch": 0.6742608473057724,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0005068369343545767,
      "loss": 1.103,
      "step": 5268
    },
    {
      "epoch": 0.6743888391142967,
      "grad_norm": 0.703125,
      "learning_rate": 0.0005064763291668425,
      "loss": 1.381,
      "step": 5269
    },
    {
      "epoch": 0.6745168309228209,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0005061158087991577,
      "loss": 1.8188,
      "step": 5270
    },
    {
      "epoch": 0.6746448227313452,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0005057553733134845,
      "loss": 0.9878,
      "step": 5271
    },
    {
      "epoch": 0.6747728145398695,
      "grad_norm": 0.5625,
      "learning_rate": 0.0005053950227717697,
      "loss": 1.0572,
      "step": 5272
    },
    {
      "epoch": 0.6749008063483937,
      "grad_norm": 0.546875,
      "learning_rate": 0.0005050347572359452,
      "loss": 0.8947,
      "step": 5273
    },
    {
      "epoch": 0.675028798156918,
      "grad_norm": 0.640625,
      "learning_rate": 0.000504674576767928,
      "loss": 1.0357,
      "step": 5274
    },
    {
      "epoch": 0.6751567899654423,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0005043144814296215,
      "loss": 1.5175,
      "step": 5275
    },
    {
      "epoch": 0.6752847817739664,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0005039544712829136,
      "loss": 1.2048,
      "step": 5276
    },
    {
      "epoch": 0.6754127735824907,
      "grad_norm": 0.625,
      "learning_rate": 0.0005035945463896781,
      "loss": 1.3787,
      "step": 5277
    },
    {
      "epoch": 0.6755407653910149,
      "grad_norm": 0.734375,
      "learning_rate": 0.0005032347068117742,
      "loss": 1.9546,
      "step": 5278
    },
    {
      "epoch": 0.6756687571995392,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0005028749526110454,
      "loss": 2.0544,
      "step": 5279
    },
    {
      "epoch": 0.6757967490080635,
      "grad_norm": 0.625,
      "learning_rate": 0.0005025152838493219,
      "loss": 1.3904,
      "step": 5280
    },
    {
      "epoch": 0.6759247408165877,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005021557005884183,
      "loss": 1.6347,
      "step": 5281
    },
    {
      "epoch": 0.676052732625112,
      "grad_norm": 0.734375,
      "learning_rate": 0.0005017962028901348,
      "loss": 1.8253,
      "step": 5282
    },
    {
      "epoch": 0.6761807244336362,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0005014367908162574,
      "loss": 1.7767,
      "step": 5283
    },
    {
      "epoch": 0.6763087162421605,
      "grad_norm": 0.640625,
      "learning_rate": 0.0005010774644285564,
      "loss": 2.0672,
      "step": 5284
    },
    {
      "epoch": 0.6764367080506848,
      "grad_norm": 0.61328125,
      "learning_rate": 0.000500718223788788,
      "loss": 1.6539,
      "step": 5285
    },
    {
      "epoch": 0.676564699859209,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0005003590689586937,
      "loss": 0.747,
      "step": 5286
    },
    {
      "epoch": 0.6766926916677333,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0005000000000000002,
      "loss": 0.9819,
      "step": 5287
    },
    {
      "epoch": 0.6768206834762576,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0004996410169744187,
      "loss": 1.7672,
      "step": 5288
    },
    {
      "epoch": 0.6769486752847818,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0004992821199436468,
      "loss": 1.3313,
      "step": 5289
    },
    {
      "epoch": 0.677076667093306,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0004989233089693664,
      "loss": 0.9475,
      "step": 5290
    },
    {
      "epoch": 0.6772046589018302,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0004985645841132458,
      "loss": 1.4755,
      "step": 5291
    },
    {
      "epoch": 0.6773326507103545,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0004982059454369366,
      "loss": 0.8909,
      "step": 5292
    },
    {
      "epoch": 0.6774606425188788,
      "grad_norm": 0.703125,
      "learning_rate": 0.0004978473930020767,
      "loss": 1.6385,
      "step": 5293
    },
    {
      "epoch": 0.677588634327403,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0004974889268702901,
      "loss": 1.2777,
      "step": 5294
    },
    {
      "epoch": 0.6777166261359273,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0004971305471031845,
      "loss": 1.3129,
      "step": 5295
    },
    {
      "epoch": 0.6778446179444515,
      "grad_norm": 0.75,
      "learning_rate": 0.0004967722537623537,
      "loss": 1.5128,
      "step": 5296
    },
    {
      "epoch": 0.6779726097529758,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0004964140469093755,
      "loss": 1.4913,
      "step": 5297
    },
    {
      "epoch": 0.6781006015615001,
      "grad_norm": 0.65625,
      "learning_rate": 0.0004960559266058139,
      "loss": 1.3797,
      "step": 5298
    },
    {
      "epoch": 0.6782285933700243,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0004956978929132177,
      "loss": 1.0408,
      "step": 5299
    },
    {
      "epoch": 0.6783565851785486,
      "grad_norm": 0.72265625,
      "learning_rate": 0.000495339945893121,
      "loss": 1.0762,
      "step": 5300
    },
    {
      "epoch": 0.6784845769870729,
      "grad_norm": 0.640625,
      "learning_rate": 0.0004949820856070424,
      "loss": 1.5949,
      "step": 5301
    },
    {
      "epoch": 0.6786125687955971,
      "grad_norm": 0.6875,
      "learning_rate": 0.000494624312116486,
      "loss": 1.2899,
      "step": 5302
    },
    {
      "epoch": 0.6787405606041214,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0004942666254829413,
      "loss": 1.0439,
      "step": 5303
    },
    {
      "epoch": 0.6788685524126455,
      "grad_norm": 0.609375,
      "learning_rate": 0.0004939090257678824,
      "loss": 0.889,
      "step": 5304
    },
    {
      "epoch": 0.6789965442211698,
      "grad_norm": 0.640625,
      "learning_rate": 0.0004935515130327686,
      "loss": 1.2029,
      "step": 5305
    },
    {
      "epoch": 0.6791245360296941,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0004931940873390443,
      "loss": 0.8974,
      "step": 5306
    },
    {
      "epoch": 0.6792525278382183,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0004928367487481389,
      "loss": 1.4494,
      "step": 5307
    },
    {
      "epoch": 0.6793805196467426,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0004924794973214668,
      "loss": 1.3243,
      "step": 5308
    },
    {
      "epoch": 0.6795085114552669,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0004921223331204278,
      "loss": 1.2025,
      "step": 5309
    },
    {
      "epoch": 0.6796365032637911,
      "grad_norm": 0.71875,
      "learning_rate": 0.0004917652562064058,
      "loss": 1.483,
      "step": 5310
    },
    {
      "epoch": 0.6797644950723154,
      "grad_norm": 0.609375,
      "learning_rate": 0.0004914082666407705,
      "loss": 1.179,
      "step": 5311
    },
    {
      "epoch": 0.6798924868808396,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0004910513644848761,
      "loss": 1.5026,
      "step": 5312
    },
    {
      "epoch": 0.6800204786893639,
      "grad_norm": 1.328125,
      "learning_rate": 0.000490694549800063,
      "loss": 1.9998,
      "step": 5313
    },
    {
      "epoch": 0.6801484704978882,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0004903378226476544,
      "loss": 0.7294,
      "step": 5314
    },
    {
      "epoch": 0.6802764623064124,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0004899811830889597,
      "loss": 1.2398,
      "step": 5315
    },
    {
      "epoch": 0.6804044541149367,
      "grad_norm": 0.62890625,
      "learning_rate": 0.000489624631185274,
      "loss": 1.1768,
      "step": 5316
    },
    {
      "epoch": 0.6805324459234608,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0004892681669978761,
      "loss": 1.584,
      "step": 5317
    },
    {
      "epoch": 0.6806604377319851,
      "grad_norm": 0.703125,
      "learning_rate": 0.0004889117905880307,
      "loss": 1.8735,
      "step": 5318
    },
    {
      "epoch": 0.6807884295405094,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0004885555020169858,
      "loss": 1.5507,
      "step": 5319
    },
    {
      "epoch": 0.6809164213490336,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00048819930134597615,
      "loss": 1.3476,
      "step": 5320
    },
    {
      "epoch": 0.6810444131575579,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00048784318863622037,
      "loss": 1.2752,
      "step": 5321
    },
    {
      "epoch": 0.6811724049660822,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00048748716394892257,
      "loss": 1.447,
      "step": 5322
    },
    {
      "epoch": 0.6813003967746064,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0004871312273452706,
      "loss": 0.9485,
      "step": 5323
    },
    {
      "epoch": 0.6814283885831307,
      "grad_norm": 0.734375,
      "learning_rate": 0.0004867753788864385,
      "loss": 1.8318,
      "step": 5324
    },
    {
      "epoch": 0.6815563803916549,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0004864196186335844,
      "loss": 0.9194,
      "step": 5325
    },
    {
      "epoch": 0.6816843722001792,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0004860639466478516,
      "loss": 0.9975,
      "step": 5326
    },
    {
      "epoch": 0.6818123640087035,
      "grad_norm": 0.515625,
      "learning_rate": 0.000485708362990368,
      "loss": 0.8249,
      "step": 5327
    },
    {
      "epoch": 0.6819403558172277,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0004853528677222465,
      "loss": 1.4133,
      "step": 5328
    },
    {
      "epoch": 0.682068347625752,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0004849974609045849,
      "loss": 0.9676,
      "step": 5329
    },
    {
      "epoch": 0.6821963394342762,
      "grad_norm": 0.75,
      "learning_rate": 0.0004846421425984652,
      "loss": 1.8357,
      "step": 5330
    },
    {
      "epoch": 0.6823243312428005,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0004842869128649553,
      "loss": 1.7265,
      "step": 5331
    },
    {
      "epoch": 0.6824523230513247,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0004839317717651063,
      "loss": 1.1953,
      "step": 5332
    },
    {
      "epoch": 0.6825803148598489,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0004835767193599555,
      "loss": 0.9396,
      "step": 5333
    },
    {
      "epoch": 0.6827083066683732,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0004832217557105243,
      "loss": 1.3267,
      "step": 5334
    },
    {
      "epoch": 0.6828362984768975,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0004828668808778194,
      "loss": 1.5007,
      "step": 5335
    },
    {
      "epoch": 0.6829642902854217,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00048251209492283087,
      "loss": 1.1638,
      "step": 5336
    },
    {
      "epoch": 0.683092282093946,
      "grad_norm": 0.55078125,
      "learning_rate": 0.000482157397906535,
      "loss": 0.9997,
      "step": 5337
    },
    {
      "epoch": 0.6832202739024702,
      "grad_norm": 0.65625,
      "learning_rate": 0.0004818027898898919,
      "loss": 1.2109,
      "step": 5338
    },
    {
      "epoch": 0.6833482657109945,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00048144827093384723,
      "loss": 1.6108,
      "step": 5339
    },
    {
      "epoch": 0.6834762575195188,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00048109384109933095,
      "loss": 1.0177,
      "step": 5340
    },
    {
      "epoch": 0.683604249328043,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00048073950044725687,
      "loss": 1.537,
      "step": 5341
    },
    {
      "epoch": 0.6837322411365673,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00048038524903852456,
      "loss": 0.9097,
      "step": 5342
    },
    {
      "epoch": 0.6838602329450915,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0004800310869340178,
      "loss": 0.9789,
      "step": 5343
    },
    {
      "epoch": 0.6839882247536158,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00047967701419460553,
      "loss": 0.8388,
      "step": 5344
    },
    {
      "epoch": 0.6841162165621401,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00047932303088114,
      "loss": 1.1898,
      "step": 5345
    },
    {
      "epoch": 0.6842442083706642,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00047896913705445957,
      "loss": 1.062,
      "step": 5346
    },
    {
      "epoch": 0.6843722001791885,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0004786153327753865,
      "loss": 0.919,
      "step": 5347
    },
    {
      "epoch": 0.6845001919877128,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0004782616181047278,
      "loss": 1.2271,
      "step": 5348
    },
    {
      "epoch": 0.684628183796237,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00047790799310327514,
      "loss": 1.169,
      "step": 5349
    },
    {
      "epoch": 0.6847561756047613,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00047755445783180474,
      "loss": 2.1309,
      "step": 5350
    },
    {
      "epoch": 0.6848841674132855,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00047720101235107726,
      "loss": 1.1076,
      "step": 5351
    },
    {
      "epoch": 0.6850121592218098,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00047684765672183807,
      "loss": 1.1818,
      "step": 5352
    },
    {
      "epoch": 0.6851401510303341,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0004764943910048175,
      "loss": 1.4034,
      "step": 5353
    },
    {
      "epoch": 0.6852681428388583,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0004761412152607292,
      "loss": 0.94,
      "step": 5354
    },
    {
      "epoch": 0.6853961346473826,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0004757881295502725,
      "loss": 1.3163,
      "step": 5355
    },
    {
      "epoch": 0.6855241264559068,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00047543513393413106,
      "loss": 1.171,
      "step": 5356
    },
    {
      "epoch": 0.6856521182644311,
      "grad_norm": 0.625,
      "learning_rate": 0.0004750822284729727,
      "loss": 0.9642,
      "step": 5357
    },
    {
      "epoch": 0.6857801100729554,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00047472941322745043,
      "loss": 0.8478,
      "step": 5358
    },
    {
      "epoch": 0.6859081018814795,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0004743766882582006,
      "loss": 1.8073,
      "step": 5359
    },
    {
      "epoch": 0.6860360936900038,
      "grad_norm": 0.78125,
      "learning_rate": 0.0004740240536258449,
      "loss": 1.1491,
      "step": 5360
    },
    {
      "epoch": 0.6861640854985281,
      "grad_norm": 0.609375,
      "learning_rate": 0.000473671509390989,
      "loss": 1.4852,
      "step": 5361
    },
    {
      "epoch": 0.6862920773070523,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00047331905561422464,
      "loss": 1.1144,
      "step": 5362
    },
    {
      "epoch": 0.6864200691155766,
      "grad_norm": 0.828125,
      "learning_rate": 0.00047296669235612546,
      "loss": 1.2925,
      "step": 5363
    },
    {
      "epoch": 0.6865480609241008,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00047261441967725107,
      "loss": 1.5892,
      "step": 5364
    },
    {
      "epoch": 0.6866760527326251,
      "grad_norm": 0.71875,
      "learning_rate": 0.0004722622376381455,
      "loss": 1.278,
      "step": 5365
    },
    {
      "epoch": 0.6868040445411494,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00047191014629933704,
      "loss": 1.1191,
      "step": 5366
    },
    {
      "epoch": 0.6869320363496736,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0004715581457213377,
      "loss": 1.6018,
      "step": 5367
    },
    {
      "epoch": 0.6870600281581979,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00047120623596464496,
      "loss": 1.3193,
      "step": 5368
    },
    {
      "epoch": 0.6871880199667221,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00047085441708974007,
      "loss": 0.8773,
      "step": 5369
    },
    {
      "epoch": 0.6873160117752464,
      "grad_norm": 0.5625,
      "learning_rate": 0.00047050268915708896,
      "loss": 1.143,
      "step": 5370
    },
    {
      "epoch": 0.6874440035837707,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0004701510522271416,
      "loss": 1.4249,
      "step": 5371
    },
    {
      "epoch": 0.6875719953922949,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00046979950636033254,
      "loss": 0.952,
      "step": 5372
    },
    {
      "epoch": 0.6876999872008192,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0004694480516170807,
      "loss": 1.3743,
      "step": 5373
    },
    {
      "epoch": 0.6878279790093434,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00046909668805778936,
      "loss": 1.7114,
      "step": 5374
    },
    {
      "epoch": 0.6879559708178676,
      "grad_norm": 0.609375,
      "learning_rate": 0.00046874541574284625,
      "loss": 1.1846,
      "step": 5375
    },
    {
      "epoch": 0.6880839626263919,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00046839423473262257,
      "loss": 1.3133,
      "step": 5376
    },
    {
      "epoch": 0.6882119544349161,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0004680431450874748,
      "loss": 1.178,
      "step": 5377
    },
    {
      "epoch": 0.6883399462434404,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0004676921468677434,
      "loss": 1.357,
      "step": 5378
    },
    {
      "epoch": 0.6884679380519647,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00046734124013375324,
      "loss": 1.2325,
      "step": 5379
    },
    {
      "epoch": 0.6885959298604889,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0004669904249458136,
      "loss": 0.9748,
      "step": 5380
    },
    {
      "epoch": 0.6887239216690132,
      "grad_norm": 0.53125,
      "learning_rate": 0.00046663970136421684,
      "loss": 0.747,
      "step": 5381
    },
    {
      "epoch": 0.6888519134775375,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0004662890694492412,
      "loss": 1.6421,
      "step": 5382
    },
    {
      "epoch": 0.6889799052860617,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00046593852926114787,
      "loss": 0.8991,
      "step": 5383
    },
    {
      "epoch": 0.689107897094586,
      "grad_norm": 0.69921875,
      "learning_rate": 0.000465588080860184,
      "loss": 1.0704,
      "step": 5384
    },
    {
      "epoch": 0.6892358889031102,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00046523772430657887,
      "loss": 1.2781,
      "step": 5385
    },
    {
      "epoch": 0.6893638807116345,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00046488745966054713,
      "loss": 0.9693,
      "step": 5386
    },
    {
      "epoch": 0.6894918725201588,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0004645372869822876,
      "loss": 1.0416,
      "step": 5387
    },
    {
      "epoch": 0.6896198643286829,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0004641872063319833,
      "loss": 1.2591,
      "step": 5388
    },
    {
      "epoch": 0.6897478561372072,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0004638372177698007,
      "loss": 1.2081,
      "step": 5389
    },
    {
      "epoch": 0.6898758479457314,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00046348732135589144,
      "loss": 1.7203,
      "step": 5390
    },
    {
      "epoch": 0.6900038397542557,
      "grad_norm": 0.71875,
      "learning_rate": 0.0004631375171503907,
      "loss": 1.1161,
      "step": 5391
    },
    {
      "epoch": 0.69013183156278,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0004627878052134181,
      "loss": 0.9612,
      "step": 5392
    },
    {
      "epoch": 0.6902598233713042,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00046243818560507733,
      "loss": 0.9925,
      "step": 5393
    },
    {
      "epoch": 0.6903878151798285,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0004620886583854562,
      "loss": 1.1552,
      "step": 5394
    },
    {
      "epoch": 0.6905158069883528,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0004617392236146266,
      "loss": 1.5563,
      "step": 5395
    },
    {
      "epoch": 0.690643798796877,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0004613898813526445,
      "loss": 1.1805,
      "step": 5396
    },
    {
      "epoch": 0.6907717906054013,
      "grad_norm": 0.578125,
      "learning_rate": 0.0004610406316595506,
      "loss": 1.2188,
      "step": 5397
    },
    {
      "epoch": 0.6908997824139255,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0004606914745953682,
      "loss": 1.7372,
      "step": 5398
    },
    {
      "epoch": 0.6910277742224498,
      "grad_norm": 0.6015625,
      "learning_rate": 0.000460342410220106,
      "loss": 1.104,
      "step": 5399
    },
    {
      "epoch": 0.6911557660309741,
      "grad_norm": 0.75,
      "learning_rate": 0.00045999343859375643,
      "loss": 2.1693,
      "step": 5400
    },
    {
      "epoch": 0.6912837578394982,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00045964455977629595,
      "loss": 1.703,
      "step": 5401
    },
    {
      "epoch": 0.6914117496480225,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0004592957738276854,
      "loss": 0.9165,
      "step": 5402
    },
    {
      "epoch": 0.6915397414565467,
      "grad_norm": 0.609375,
      "learning_rate": 0.0004589470808078685,
      "loss": 0.8038,
      "step": 5403
    },
    {
      "epoch": 0.691667733265071,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00045859848077677424,
      "loss": 0.9878,
      "step": 5404
    },
    {
      "epoch": 0.6917957250735953,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0004582499737943151,
      "loss": 1.6525,
      "step": 5405
    },
    {
      "epoch": 0.6919237168821195,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00045790155992038787,
      "loss": 1.3526,
      "step": 5406
    },
    {
      "epoch": 0.6920517086906438,
      "grad_norm": 0.65625,
      "learning_rate": 0.00045755323921487293,
      "loss": 1.5359,
      "step": 5407
    },
    {
      "epoch": 0.6921797004991681,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00045720501173763487,
      "loss": 1.2661,
      "step": 5408
    },
    {
      "epoch": 0.6923076923076923,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00045685687754852224,
      "loss": 1.3323,
      "step": 5409
    },
    {
      "epoch": 0.6924356841162166,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00045650883670736756,
      "loss": 1.0182,
      "step": 5410
    },
    {
      "epoch": 0.6925636759247408,
      "grad_norm": 0.578125,
      "learning_rate": 0.00045616088927398767,
      "loss": 0.9394,
      "step": 5411
    },
    {
      "epoch": 0.6926916677332651,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0004558130353081823,
      "loss": 1.037,
      "step": 5412
    },
    {
      "epoch": 0.6928196595417894,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0004554652748697361,
      "loss": 1.2838,
      "step": 5413
    },
    {
      "epoch": 0.6929476513503136,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0004551176080184173,
      "loss": 1.6451,
      "step": 5414
    },
    {
      "epoch": 0.6930756431588379,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00045477003481397815,
      "loss": 1.3476,
      "step": 5415
    },
    {
      "epoch": 0.693203634967362,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00045442255531615483,
      "loss": 1.2118,
      "step": 5416
    },
    {
      "epoch": 0.6933316267758863,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00045407516958466724,
      "loss": 1.0358,
      "step": 5417
    },
    {
      "epoch": 0.6934596185844106,
      "grad_norm": 0.703125,
      "learning_rate": 0.0004537278776792192,
      "loss": 1.1848,
      "step": 5418
    },
    {
      "epoch": 0.6935876103929348,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0004533806796594989,
      "loss": 1.558,
      "step": 5419
    },
    {
      "epoch": 0.6937156022014591,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0004530335755851772,
      "loss": 1.1938,
      "step": 5420
    },
    {
      "epoch": 0.6938435940099834,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0004526865655159099,
      "loss": 1.2026,
      "step": 5421
    },
    {
      "epoch": 0.6939715858185076,
      "grad_norm": 0.59375,
      "learning_rate": 0.00045233964951133643,
      "loss": 1.0856,
      "step": 5422
    },
    {
      "epoch": 0.6940995776270319,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00045199282763107974,
      "loss": 1.0349,
      "step": 5423
    },
    {
      "epoch": 0.6942275694355561,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00045164609993474737,
      "loss": 1.1385,
      "step": 5424
    },
    {
      "epoch": 0.6943555612440804,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0004512994664819292,
      "loss": 0.9035,
      "step": 5425
    },
    {
      "epoch": 0.6944835530526047,
      "grad_norm": 0.65625,
      "learning_rate": 0.0004509529273322003,
      "loss": 1.3202,
      "step": 5426
    },
    {
      "epoch": 0.6946115448611289,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0004506064825451189,
      "loss": 1.4018,
      "step": 5427
    },
    {
      "epoch": 0.6947395366696532,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00045026013218022734,
      "loss": 1.5868,
      "step": 5428
    },
    {
      "epoch": 0.6948675284781773,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0004499138762970515,
      "loss": 1.0554,
      "step": 5429
    },
    {
      "epoch": 0.6949955202867016,
      "grad_norm": 0.55078125,
      "learning_rate": 0.000449567714955101,
      "loss": 0.9893,
      "step": 5430
    },
    {
      "epoch": 0.6951235120952259,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0004492216482138692,
      "loss": 0.9679,
      "step": 5431
    },
    {
      "epoch": 0.6952515039037501,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00044887567613283353,
      "loss": 1.3626,
      "step": 5432
    },
    {
      "epoch": 0.6953794957122744,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00044852979877145504,
      "loss": 1.0933,
      "step": 5433
    },
    {
      "epoch": 0.6955074875207987,
      "grad_norm": 0.640625,
      "learning_rate": 0.00044818401618917784,
      "loss": 1.3552,
      "step": 5434
    },
    {
      "epoch": 0.6956354793293229,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0004478383284454305,
      "loss": 1.1603,
      "step": 5435
    },
    {
      "epoch": 0.6957634711378472,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0004474927355996251,
      "loss": 0.8761,
      "step": 5436
    },
    {
      "epoch": 0.6958914629463714,
      "grad_norm": 0.71875,
      "learning_rate": 0.0004471472377111574,
      "loss": 1.0215,
      "step": 5437
    },
    {
      "epoch": 0.6960194547548957,
      "grad_norm": 0.53125,
      "learning_rate": 0.00044680183483940673,
      "loss": 0.7736,
      "step": 5438
    },
    {
      "epoch": 0.69614744656342,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0004464565270437363,
      "loss": 1.7273,
      "step": 5439
    },
    {
      "epoch": 0.6962754383719442,
      "grad_norm": 0.671875,
      "learning_rate": 0.00044611131438349283,
      "loss": 1.4319,
      "step": 5440
    },
    {
      "epoch": 0.6964034301804685,
      "grad_norm": 0.734375,
      "learning_rate": 0.0004457661969180067,
      "loss": 1.7506,
      "step": 5441
    },
    {
      "epoch": 0.6965314219889926,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0004454211747065923,
      "loss": 1.6409,
      "step": 5442
    },
    {
      "epoch": 0.696659413797517,
      "grad_norm": 0.71875,
      "learning_rate": 0.0004450762478085465,
      "loss": 1.1067,
      "step": 5443
    },
    {
      "epoch": 0.6967874056060412,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00044473141628315104,
      "loss": 1.6947,
      "step": 5444
    },
    {
      "epoch": 0.6969153974145654,
      "grad_norm": 0.796875,
      "learning_rate": 0.0004443866801896709,
      "loss": 1.4037,
      "step": 5445
    },
    {
      "epoch": 0.6970433892230897,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0004440420395873547,
      "loss": 1.0186,
      "step": 5446
    },
    {
      "epoch": 0.697171381031614,
      "grad_norm": 0.578125,
      "learning_rate": 0.00044369749453543393,
      "loss": 1.0325,
      "step": 5447
    },
    {
      "epoch": 0.6972993728401382,
      "grad_norm": 0.671875,
      "learning_rate": 0.00044335304509312466,
      "loss": 1.4335,
      "step": 5448
    },
    {
      "epoch": 0.6974273646486625,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00044300869131962595,
      "loss": 1.4833,
      "step": 5449
    },
    {
      "epoch": 0.6975553564571867,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00044266443327412085,
      "loss": 1.3632,
      "step": 5450
    },
    {
      "epoch": 0.697683348265711,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0004423202710157754,
      "loss": 0.8878,
      "step": 5451
    },
    {
      "epoch": 0.6978113400742353,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00044197620460373967,
      "loss": 0.8826,
      "step": 5452
    },
    {
      "epoch": 0.6979393318827595,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0004416322340971469,
      "loss": 1.2079,
      "step": 5453
    },
    {
      "epoch": 0.6980673236912838,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00044128835955511416,
      "loss": 1.0911,
      "step": 5454
    },
    {
      "epoch": 0.6981953154998081,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0004409445810367421,
      "loss": 1.5344,
      "step": 5455
    },
    {
      "epoch": 0.6983233073083323,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00044060089860111396,
      "loss": 0.8415,
      "step": 5456
    },
    {
      "epoch": 0.6984512991168565,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0004402573123072975,
      "loss": 1.2915,
      "step": 5457
    },
    {
      "epoch": 0.6985792909253807,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0004399138222143436,
      "loss": 1.1171,
      "step": 5458
    },
    {
      "epoch": 0.698707282733905,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0004395704283812871,
      "loss": 1.1954,
      "step": 5459
    },
    {
      "epoch": 0.6988352745424293,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0004392271308671445,
      "loss": 1.8424,
      "step": 5460
    },
    {
      "epoch": 0.6989632663509535,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00043888392973091817,
      "loss": 1.7688,
      "step": 5461
    },
    {
      "epoch": 0.6990912581594778,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00043854082503159264,
      "loss": 1.1134,
      "step": 5462
    },
    {
      "epoch": 0.699219249968002,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00043819781682813576,
      "loss": 1.3505,
      "step": 5463
    },
    {
      "epoch": 0.6993472417765263,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00043785490517949955,
      "loss": 0.8827,
      "step": 5464
    },
    {
      "epoch": 0.6994752335850506,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00043751209014461826,
      "loss": 0.9663,
      "step": 5465
    },
    {
      "epoch": 0.6996032253935748,
      "grad_norm": 0.65625,
      "learning_rate": 0.00043716937178241036,
      "loss": 1.6808,
      "step": 5466
    },
    {
      "epoch": 0.6997312172020991,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00043682675015177774,
      "loss": 1.4897,
      "step": 5467
    },
    {
      "epoch": 0.6998592090106234,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0004364842253116059,
      "loss": 1.2283,
      "step": 5468
    },
    {
      "epoch": 0.6999872008191476,
      "grad_norm": 0.609375,
      "learning_rate": 0.00043614179732076244,
      "loss": 1.2354,
      "step": 5469
    },
    {
      "epoch": 0.7001151926276719,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0004357994662380996,
      "loss": 1.0873,
      "step": 5470
    },
    {
      "epoch": 0.700243184436196,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0004354572321224525,
      "loss": 0.9833,
      "step": 5471
    },
    {
      "epoch": 0.7003711762447203,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00043511509503263955,
      "loss": 1.9097,
      "step": 5472
    },
    {
      "epoch": 0.7004991680532446,
      "grad_norm": 0.703125,
      "learning_rate": 0.0004347730550274628,
      "loss": 1.4709,
      "step": 5473
    },
    {
      "epoch": 0.7006271598617688,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0004344311121657071,
      "loss": 0.7995,
      "step": 5474
    },
    {
      "epoch": 0.7007551516702931,
      "grad_norm": 0.75,
      "learning_rate": 0.000434089266506141,
      "loss": 1.3459,
      "step": 5475
    },
    {
      "epoch": 0.7008831434788173,
      "grad_norm": 0.6875,
      "learning_rate": 0.0004337475181075162,
      "loss": 1.302,
      "step": 5476
    },
    {
      "epoch": 0.7010111352873416,
      "grad_norm": 0.703125,
      "learning_rate": 0.00043340586702856823,
      "loss": 1.5572,
      "step": 5477
    },
    {
      "epoch": 0.7011391270958659,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00043306431332801434,
      "loss": 1.4949,
      "step": 5478
    },
    {
      "epoch": 0.7012671189043901,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0004327228570645566,
      "loss": 1.6552,
      "step": 5479
    },
    {
      "epoch": 0.7013951107129144,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0004323814982968799,
      "loss": 1.0335,
      "step": 5480
    },
    {
      "epoch": 0.7015231025214387,
      "grad_norm": 0.59375,
      "learning_rate": 0.0004320402370836525,
      "loss": 1.18,
      "step": 5481
    },
    {
      "epoch": 0.7016510943299629,
      "grad_norm": 0.67578125,
      "learning_rate": 0.000431699073483525,
      "loss": 1.3708,
      "step": 5482
    },
    {
      "epoch": 0.7017790861384872,
      "grad_norm": 0.75390625,
      "learning_rate": 0.000431358007555132,
      "loss": 2.3493,
      "step": 5483
    },
    {
      "epoch": 0.7019070779470113,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0004310170393570917,
      "loss": 0.8746,
      "step": 5484
    },
    {
      "epoch": 0.7020350697555356,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0004306761689480049,
      "loss": 1.1622,
      "step": 5485
    },
    {
      "epoch": 0.7021630615640599,
      "grad_norm": 0.546875,
      "learning_rate": 0.0004303353963864559,
      "loss": 0.8956,
      "step": 5486
    },
    {
      "epoch": 0.7022910533725841,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0004299947217310114,
      "loss": 1.2908,
      "step": 5487
    },
    {
      "epoch": 0.7024190451811084,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00042965414504022215,
      "loss": 1.3897,
      "step": 5488
    },
    {
      "epoch": 0.7025470369896326,
      "grad_norm": 0.609375,
      "learning_rate": 0.0004293136663726217,
      "loss": 1.0237,
      "step": 5489
    },
    {
      "epoch": 0.7026750287981569,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00042897328578672743,
      "loss": 1.3041,
      "step": 5490
    },
    {
      "epoch": 0.7028030206066812,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00042863300334103837,
      "loss": 1.3134,
      "step": 5491
    },
    {
      "epoch": 0.7029310124152054,
      "grad_norm": 0.703125,
      "learning_rate": 0.00042829281909403797,
      "loss": 1.3934,
      "step": 5492
    },
    {
      "epoch": 0.7030590042237297,
      "grad_norm": 0.578125,
      "learning_rate": 0.00042795273310419237,
      "loss": 1.1175,
      "step": 5493
    },
    {
      "epoch": 0.703186996032254,
      "grad_norm": 0.734375,
      "learning_rate": 0.000427612745429951,
      "loss": 1.6743,
      "step": 5494
    },
    {
      "epoch": 0.7033149878407782,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0004272728561297461,
      "loss": 1.2831,
      "step": 5495
    },
    {
      "epoch": 0.7034429796493025,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0004269330652619933,
      "loss": 0.8272,
      "step": 5496
    },
    {
      "epoch": 0.7035709714578267,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00042659337288509104,
      "loss": 1.3547,
      "step": 5497
    },
    {
      "epoch": 0.703698963266351,
      "grad_norm": 0.71875,
      "learning_rate": 0.00042625377905742103,
      "loss": 1.7376,
      "step": 5498
    },
    {
      "epoch": 0.7038269550748752,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00042591428383734843,
      "loss": 1.3953,
      "step": 5499
    },
    {
      "epoch": 0.7039549468833994,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0004255748872832201,
      "loss": 0.6651,
      "step": 5500
    },
    {
      "epoch": 0.7040829386919237,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0004252355894533674,
      "loss": 2.0509,
      "step": 5501
    },
    {
      "epoch": 0.7042109305004479,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00042489639040610395,
      "loss": 1.2592,
      "step": 5502
    },
    {
      "epoch": 0.7043389223089722,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0004245572901997271,
      "loss": 0.9089,
      "step": 5503
    },
    {
      "epoch": 0.7044669141174965,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0004242182888925161,
      "loss": 0.6267,
      "step": 5504
    },
    {
      "epoch": 0.7045949059260207,
      "grad_norm": 0.5625,
      "learning_rate": 0.00042387938654273364,
      "loss": 0.8006,
      "step": 5505
    },
    {
      "epoch": 0.704722897734545,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00042354058320862644,
      "loss": 1.6736,
      "step": 5506
    },
    {
      "epoch": 0.7048508895430693,
      "grad_norm": 0.671875,
      "learning_rate": 0.0004232018789484228,
      "loss": 1.4074,
      "step": 5507
    },
    {
      "epoch": 0.7049788813515935,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00042286327382033505,
      "loss": 1.0363,
      "step": 5508
    },
    {
      "epoch": 0.7051068731601178,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0004225247678825573,
      "loss": 0.8589,
      "step": 5509
    },
    {
      "epoch": 0.705234864968642,
      "grad_norm": 0.75,
      "learning_rate": 0.0004221863611932677,
      "loss": 1.2043,
      "step": 5510
    },
    {
      "epoch": 0.7053628567771663,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00042184805381062664,
      "loss": 1.2715,
      "step": 5511
    },
    {
      "epoch": 0.7054908485856906,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0004215098457927783,
      "loss": 1.4949,
      "step": 5512
    },
    {
      "epoch": 0.7056188403942147,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0004211717371978485,
      "loss": 1.8981,
      "step": 5513
    },
    {
      "epoch": 0.705746832202739,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00042083372808394704,
      "loss": 1.4452,
      "step": 5514
    },
    {
      "epoch": 0.7058748240112632,
      "grad_norm": 0.625,
      "learning_rate": 0.00042049581850916616,
      "loss": 1.2156,
      "step": 5515
    },
    {
      "epoch": 0.7060028158197875,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00042015800853158127,
      "loss": 1.713,
      "step": 5516
    },
    {
      "epoch": 0.7061308076283118,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00041982029820925037,
      "loss": 1.4578,
      "step": 5517
    },
    {
      "epoch": 0.706258799436836,
      "grad_norm": 0.703125,
      "learning_rate": 0.00041948268760021444,
      "loss": 1.2448,
      "step": 5518
    },
    {
      "epoch": 0.7063867912453603,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00041914517676249743,
      "loss": 0.9639,
      "step": 5519
    },
    {
      "epoch": 0.7065147830538846,
      "grad_norm": 0.515625,
      "learning_rate": 0.00041880776575410606,
      "loss": 0.7851,
      "step": 5520
    },
    {
      "epoch": 0.7066427748624088,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0004184704546330302,
      "loss": 1.8938,
      "step": 5521
    },
    {
      "epoch": 0.7067707666709331,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00041813324345724157,
      "loss": 1.0485,
      "step": 5522
    },
    {
      "epoch": 0.7068987584794573,
      "grad_norm": 0.859375,
      "learning_rate": 0.0004177961322846956,
      "loss": 1.401,
      "step": 5523
    },
    {
      "epoch": 0.7070267502879816,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0004174591211733305,
      "loss": 1.2673,
      "step": 5524
    },
    {
      "epoch": 0.7071547420965059,
      "grad_norm": 0.765625,
      "learning_rate": 0.000417122210181067,
      "loss": 1.3863,
      "step": 5525
    },
    {
      "epoch": 0.70728273390503,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00041678539936580906,
      "loss": 1.2057,
      "step": 5526
    },
    {
      "epoch": 0.7074107257135543,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0004164486887854424,
      "loss": 1.603,
      "step": 5527
    },
    {
      "epoch": 0.7075387175220786,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00041611207849783616,
      "loss": 1.5011,
      "step": 5528
    },
    {
      "epoch": 0.7076667093306028,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0004157755685608431,
      "loss": 1.4429,
      "step": 5529
    },
    {
      "epoch": 0.7077947011391271,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0004154391590322978,
      "loss": 1.4545,
      "step": 5530
    },
    {
      "epoch": 0.7079226929476513,
      "grad_norm": 0.6875,
      "learning_rate": 0.00041510284997001704,
      "loss": 1.037,
      "step": 5531
    },
    {
      "epoch": 0.7080506847561756,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00041476664143180134,
      "loss": 1.2166,
      "step": 5532
    },
    {
      "epoch": 0.7081786765646999,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00041443053347543345,
      "loss": 1.4178,
      "step": 5533
    },
    {
      "epoch": 0.7083066683732241,
      "grad_norm": 0.6875,
      "learning_rate": 0.0004140945261586796,
      "loss": 1.562,
      "step": 5534
    },
    {
      "epoch": 0.7084346601817484,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0004137586195392872,
      "loss": 1.0014,
      "step": 5535
    },
    {
      "epoch": 0.7085626519902726,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00041342281367498767,
      "loss": 1.0367,
      "step": 5536
    },
    {
      "epoch": 0.7086906437987969,
      "grad_norm": 0.5625,
      "learning_rate": 0.00041308710862349477,
      "loss": 0.9299,
      "step": 5537
    },
    {
      "epoch": 0.7088186356073212,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00041275150444250463,
      "loss": 1.3789,
      "step": 5538
    },
    {
      "epoch": 0.7089466274158454,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0004124160011896965,
      "loss": 1.6433,
      "step": 5539
    },
    {
      "epoch": 0.7090746192243697,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0004120805989227319,
      "loss": 1.2369,
      "step": 5540
    },
    {
      "epoch": 0.709202611032894,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00041174529769925526,
      "loss": 0.9528,
      "step": 5541
    },
    {
      "epoch": 0.7093306028414181,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0004114100975768935,
      "loss": 1.0805,
      "step": 5542
    },
    {
      "epoch": 0.7094585946499424,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0004110749986132566,
      "loss": 1.4812,
      "step": 5543
    },
    {
      "epoch": 0.7095865864584666,
      "grad_norm": 0.5234375,
      "learning_rate": 0.000410740000865936,
      "loss": 0.7638,
      "step": 5544
    },
    {
      "epoch": 0.7097145782669909,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00041040510439250676,
      "loss": 1.4272,
      "step": 5545
    },
    {
      "epoch": 0.7098425700755152,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00041007030925052645,
      "loss": 0.5393,
      "step": 5546
    },
    {
      "epoch": 0.7099705618840394,
      "grad_norm": 0.609375,
      "learning_rate": 0.00040973561549753493,
      "loss": 1.1129,
      "step": 5547
    },
    {
      "epoch": 0.7100985536925637,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0004094010231910552,
      "loss": 1.0713,
      "step": 5548
    },
    {
      "epoch": 0.7102265455010879,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00040906653238859147,
      "loss": 1.1098,
      "step": 5549
    },
    {
      "epoch": 0.7103545373096122,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00040873214314763207,
      "loss": 1.4736,
      "step": 5550
    },
    {
      "epoch": 0.7104825291181365,
      "grad_norm": 0.671875,
      "learning_rate": 0.00040839785552564654,
      "loss": 1.3915,
      "step": 5551
    },
    {
      "epoch": 0.7106105209266607,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00040806366958008887,
      "loss": 0.9844,
      "step": 5552
    },
    {
      "epoch": 0.710738512735185,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0004077295853683933,
      "loss": 1.3981,
      "step": 5553
    },
    {
      "epoch": 0.7108665045437093,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00040739560294797796,
      "loss": 1.2742,
      "step": 5554
    },
    {
      "epoch": 0.7109944963522334,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00040706172237624304,
      "loss": 1.6626,
      "step": 5555
    },
    {
      "epoch": 0.7111224881607577,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0004067279437105719,
      "loss": 1.3853,
      "step": 5556
    },
    {
      "epoch": 0.7112504799692819,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00040639426700832883,
      "loss": 0.9761,
      "step": 5557
    },
    {
      "epoch": 0.7113784717778062,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00040606069232686215,
      "loss": 0.7802,
      "step": 5558
    },
    {
      "epoch": 0.7115064635863305,
      "grad_norm": 0.625,
      "learning_rate": 0.000405727219723502,
      "loss": 1.2469,
      "step": 5559
    },
    {
      "epoch": 0.7116344553948547,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0004053938492555611,
      "loss": 1.9396,
      "step": 5560
    },
    {
      "epoch": 0.711762447203379,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0004050605809803345,
      "loss": 0.7736,
      "step": 5561
    },
    {
      "epoch": 0.7118904390119032,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0004047274149550998,
      "loss": 1.3022,
      "step": 5562
    },
    {
      "epoch": 0.7120184308204275,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0004043943512371171,
      "loss": 1.2667,
      "step": 5563
    },
    {
      "epoch": 0.7121464226289518,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00040406138988362863,
      "loss": 0.6292,
      "step": 5564
    },
    {
      "epoch": 0.712274414437476,
      "grad_norm": 0.578125,
      "learning_rate": 0.0004037285309518596,
      "loss": 1.0511,
      "step": 5565
    },
    {
      "epoch": 0.7124024062460003,
      "grad_norm": 0.734375,
      "learning_rate": 0.00040339577449901654,
      "loss": 0.9983,
      "step": 5566
    },
    {
      "epoch": 0.7125303980545246,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0004030631205822893,
      "loss": 1.0426,
      "step": 5567
    },
    {
      "epoch": 0.7126583898630487,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0004027305692588501,
      "loss": 1.9369,
      "step": 5568
    },
    {
      "epoch": 0.712786381671573,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00040239812058585304,
      "loss": 1.0793,
      "step": 5569
    },
    {
      "epoch": 0.7129143734800972,
      "grad_norm": 0.703125,
      "learning_rate": 0.00040206577462043524,
      "loss": 1.4634,
      "step": 5570
    },
    {
      "epoch": 0.7130423652886215,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00040173353141971513,
      "loss": 1.2423,
      "step": 5571
    },
    {
      "epoch": 0.7131703570971458,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00040140139104079443,
      "loss": 0.9228,
      "step": 5572
    },
    {
      "epoch": 0.71329834890567,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00040106935354075634,
      "loss": 1.4238,
      "step": 5573
    },
    {
      "epoch": 0.7134263407141943,
      "grad_norm": 0.65625,
      "learning_rate": 0.00040073741897666805,
      "loss": 1.4804,
      "step": 5574
    },
    {
      "epoch": 0.7135543325227185,
      "grad_norm": 0.703125,
      "learning_rate": 0.00040040558740557687,
      "loss": 1.2927,
      "step": 5575
    },
    {
      "epoch": 0.7136823243312428,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0004000738588845138,
      "loss": 1.177,
      "step": 5576
    },
    {
      "epoch": 0.7138103161397671,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00039974223347049166,
      "loss": 1.1808,
      "step": 5577
    },
    {
      "epoch": 0.7139383079482913,
      "grad_norm": 0.5625,
      "learning_rate": 0.0003994107112205058,
      "loss": 1.0247,
      "step": 5578
    },
    {
      "epoch": 0.7140662997568156,
      "grad_norm": 1.21875,
      "learning_rate": 0.0003990792921915339,
      "loss": 1.1284,
      "step": 5579
    },
    {
      "epoch": 0.7141942915653399,
      "grad_norm": 0.953125,
      "learning_rate": 0.00039874797644053517,
      "loss": 2.0415,
      "step": 5580
    },
    {
      "epoch": 0.714322283373864,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0003984167640244518,
      "loss": 1.0648,
      "step": 5581
    },
    {
      "epoch": 0.7144502751823884,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0003980856550002081,
      "loss": 1.5028,
      "step": 5582
    },
    {
      "epoch": 0.7145782669909125,
      "grad_norm": 0.6875,
      "learning_rate": 0.0003977546494247105,
      "loss": 1.5033,
      "step": 5583
    },
    {
      "epoch": 0.7147062587994368,
      "grad_norm": 0.6875,
      "learning_rate": 0.00039742374735484776,
      "loss": 1.2614,
      "step": 5584
    },
    {
      "epoch": 0.7148342506079611,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0003970929488474906,
      "loss": 1.0471,
      "step": 5585
    },
    {
      "epoch": 0.7149622424164853,
      "grad_norm": 0.71875,
      "learning_rate": 0.00039676225395949226,
      "loss": 1.5924,
      "step": 5586
    },
    {
      "epoch": 0.7150902342250096,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00039643166274768826,
      "loss": 0.965,
      "step": 5587
    },
    {
      "epoch": 0.7152182260335338,
      "grad_norm": 0.71875,
      "learning_rate": 0.0003961011752688954,
      "loss": 0.9168,
      "step": 5588
    },
    {
      "epoch": 0.7153462178420581,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0003957707915799137,
      "loss": 1.4505,
      "step": 5589
    },
    {
      "epoch": 0.7154742096505824,
      "grad_norm": 0.66796875,
      "learning_rate": 0.000395440511737525,
      "loss": 1.304,
      "step": 5590
    },
    {
      "epoch": 0.7156022014591066,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0003951103357984931,
      "loss": 1.1661,
      "step": 5591
    },
    {
      "epoch": 0.7157301932676309,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0003947802638195647,
      "loss": 1.6895,
      "step": 5592
    },
    {
      "epoch": 0.7158581850761552,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00039445029585746696,
      "loss": 0.6784,
      "step": 5593
    },
    {
      "epoch": 0.7159861768846794,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0003941204319689108,
      "loss": 0.6789,
      "step": 5594
    },
    {
      "epoch": 0.7161141686932037,
      "grad_norm": 0.703125,
      "learning_rate": 0.0003937906722105887,
      "loss": 1.7009,
      "step": 5595
    },
    {
      "epoch": 0.7162421605017278,
      "grad_norm": 0.625,
      "learning_rate": 0.00039346101663917523,
      "loss": 1.1002,
      "step": 5596
    },
    {
      "epoch": 0.7163701523102521,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0003931314653113269,
      "loss": 1.4465,
      "step": 5597
    },
    {
      "epoch": 0.7164981441187764,
      "grad_norm": 0.59375,
      "learning_rate": 0.00039280201828368266,
      "loss": 0.8636,
      "step": 5598
    },
    {
      "epoch": 0.7166261359273006,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0003924726756128632,
      "loss": 1.5058,
      "step": 5599
    },
    {
      "epoch": 0.7167541277358249,
      "grad_norm": 0.640625,
      "learning_rate": 0.0003921434373554713,
      "loss": 1.2285,
      "step": 5600
    },
    {
      "epoch": 0.7168821195443492,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00039181430356809247,
      "loss": 1.156,
      "step": 5601
    },
    {
      "epoch": 0.7170101113528734,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0003914852743072929,
      "loss": 0.9232,
      "step": 5602
    },
    {
      "epoch": 0.7171381031613977,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0003911563496296219,
      "loss": 1.2116,
      "step": 5603
    },
    {
      "epoch": 0.7172660949699219,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0003908275295916106,
      "loss": 1.0797,
      "step": 5604
    },
    {
      "epoch": 0.7173940867784462,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00039049881424977196,
      "loss": 1.1183,
      "step": 5605
    },
    {
      "epoch": 0.7175220785869705,
      "grad_norm": 0.546875,
      "learning_rate": 0.0003901702036606011,
      "loss": 1.1559,
      "step": 5606
    },
    {
      "epoch": 0.7176500703954947,
      "grad_norm": 0.796875,
      "learning_rate": 0.0003898416978805751,
      "loss": 1.4551,
      "step": 5607
    },
    {
      "epoch": 0.717778062204019,
      "grad_norm": 0.66015625,
      "learning_rate": 0.000389513296966153,
      "loss": 1.5549,
      "step": 5608
    },
    {
      "epoch": 0.7179060540125431,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00038918500097377586,
      "loss": 1.2257,
      "step": 5609
    },
    {
      "epoch": 0.7180340458210674,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00038885680995986695,
      "loss": 1.2189,
      "step": 5610
    },
    {
      "epoch": 0.7181620376295917,
      "grad_norm": 0.59375,
      "learning_rate": 0.0003885287239808305,
      "loss": 1.1249,
      "step": 5611
    },
    {
      "epoch": 0.7182900294381159,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0003882007430930539,
      "loss": 1.0363,
      "step": 5612
    },
    {
      "epoch": 0.7184180212466402,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00038787286735290574,
      "loss": 1.0932,
      "step": 5613
    },
    {
      "epoch": 0.7185460130551645,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00038754509681673745,
      "loss": 0.8398,
      "step": 5614
    },
    {
      "epoch": 0.7186740048636887,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00038721743154088075,
      "loss": 0.6919,
      "step": 5615
    },
    {
      "epoch": 0.718801996672213,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0003868898715816508,
      "loss": 0.7396,
      "step": 5616
    },
    {
      "epoch": 0.7189299884807372,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00038656241699534387,
      "loss": 0.7098,
      "step": 5617
    },
    {
      "epoch": 0.7190579802892615,
      "grad_norm": 0.625,
      "learning_rate": 0.00038623506783823846,
      "loss": 1.3043,
      "step": 5618
    },
    {
      "epoch": 0.7191859720977858,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0003859078241665949,
      "loss": 1.8243,
      "step": 5619
    },
    {
      "epoch": 0.71931396390631,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00038558068603665517,
      "loss": 1.0821,
      "step": 5620
    },
    {
      "epoch": 0.7194419557148343,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00038525365350464347,
      "loss": 1.0427,
      "step": 5621
    },
    {
      "epoch": 0.7195699475233585,
      "grad_norm": 0.640625,
      "learning_rate": 0.00038492672662676544,
      "loss": 1.3117,
      "step": 5622
    },
    {
      "epoch": 0.7196979393318828,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0003845999054592092,
      "loss": 1.3874,
      "step": 5623
    },
    {
      "epoch": 0.719825931140407,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00038427319005814357,
      "loss": 0.9301,
      "step": 5624
    },
    {
      "epoch": 0.7199539229489312,
      "grad_norm": 0.546875,
      "learning_rate": 0.0003839465804797203,
      "loss": 0.961,
      "step": 5625
    },
    {
      "epoch": 0.7200819147574555,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00038362007678007247,
      "loss": 1.2867,
      "step": 5626
    },
    {
      "epoch": 0.7202099065659798,
      "grad_norm": 0.671875,
      "learning_rate": 0.000383293679015315,
      "loss": 1.0528,
      "step": 5627
    },
    {
      "epoch": 0.720337898374504,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0003829673872415448,
      "loss": 1.0089,
      "step": 5628
    },
    {
      "epoch": 0.7204658901830283,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00038264120151484027,
      "loss": 1.2575,
      "step": 5629
    },
    {
      "epoch": 0.7205938819915525,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0003823151218912617,
      "loss": 1.0203,
      "step": 5630
    },
    {
      "epoch": 0.7207218738000768,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00038198914842685127,
      "loss": 1.2928,
      "step": 5631
    },
    {
      "epoch": 0.7208498656086011,
      "grad_norm": 0.640625,
      "learning_rate": 0.0003816632811776332,
      "loss": 0.843,
      "step": 5632
    },
    {
      "epoch": 0.7209778574171253,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0003813375201996121,
      "loss": 1.0477,
      "step": 5633
    },
    {
      "epoch": 0.7211058492256496,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0003810118655487759,
      "loss": 1.2117,
      "step": 5634
    },
    {
      "epoch": 0.7212338410341738,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0003806863172810936,
      "loss": 1.5938,
      "step": 5635
    },
    {
      "epoch": 0.7213618328426981,
      "grad_norm": 0.609375,
      "learning_rate": 0.0003803608754525161,
      "loss": 0.9806,
      "step": 5636
    },
    {
      "epoch": 0.7214898246512224,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00038003554011897535,
      "loss": 1.084,
      "step": 5637
    },
    {
      "epoch": 0.7216178164597465,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0003797103113363859,
      "loss": 0.8848,
      "step": 5638
    },
    {
      "epoch": 0.7217458082682708,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00037938518916064335,
      "loss": 1.9432,
      "step": 5639
    },
    {
      "epoch": 0.7218738000767951,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0003790601736476255,
      "loss": 1.8154,
      "step": 5640
    },
    {
      "epoch": 0.7220017918853193,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0003787352648531913,
      "loss": 1.5037,
      "step": 5641
    },
    {
      "epoch": 0.7221297836938436,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0003784104628331817,
      "loss": 1.5,
      "step": 5642
    },
    {
      "epoch": 0.7222577755023678,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00037808576764341927,
      "loss": 1.1489,
      "step": 5643
    },
    {
      "epoch": 0.7223857673108921,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0003777611793397081,
      "loss": 1.1102,
      "step": 5644
    },
    {
      "epoch": 0.7225137591194164,
      "grad_norm": 0.640625,
      "learning_rate": 0.0003774366979778342,
      "loss": 1.3801,
      "step": 5645
    },
    {
      "epoch": 0.7226417509279406,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0003771123236135644,
      "loss": 1.151,
      "step": 5646
    },
    {
      "epoch": 0.7227697427364649,
      "grad_norm": 0.5234375,
      "learning_rate": 0.000376788056302648,
      "loss": 0.8192,
      "step": 5647
    },
    {
      "epoch": 0.7228977345449891,
      "grad_norm": 0.546875,
      "learning_rate": 0.00037646389610081557,
      "loss": 1.0703,
      "step": 5648
    },
    {
      "epoch": 0.7230257263535134,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0003761398430637796,
      "loss": 0.9556,
      "step": 5649
    },
    {
      "epoch": 0.7231537181620377,
      "grad_norm": 0.6171875,
      "learning_rate": 0.000375815897247233,
      "loss": 1.0318,
      "step": 5650
    },
    {
      "epoch": 0.7232817099705618,
      "grad_norm": 0.52734375,
      "learning_rate": 0.000375492058706852,
      "loss": 1.0351,
      "step": 5651
    },
    {
      "epoch": 0.7234097017790861,
      "grad_norm": 0.703125,
      "learning_rate": 0.00037516832749829313,
      "loss": 1.4582,
      "step": 5652
    },
    {
      "epoch": 0.7235376935876104,
      "grad_norm": 0.96875,
      "learning_rate": 0.0003748447036771949,
      "loss": 2.0068,
      "step": 5653
    },
    {
      "epoch": 0.7236656853961346,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00037452118729917785,
      "loss": 1.6049,
      "step": 5654
    },
    {
      "epoch": 0.7237936772046589,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0003741977784198426,
      "loss": 1.0109,
      "step": 5655
    },
    {
      "epoch": 0.7239216690131831,
      "grad_norm": 0.78125,
      "learning_rate": 0.0003738744770947725,
      "loss": 1.4134,
      "step": 5656
    },
    {
      "epoch": 0.7240496608217074,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00037355128337953237,
      "loss": 0.9463,
      "step": 5657
    },
    {
      "epoch": 0.7241776526302317,
      "grad_norm": 0.65625,
      "learning_rate": 0.00037322819732966853,
      "loss": 1.404,
      "step": 5658
    },
    {
      "epoch": 0.7243056444387559,
      "grad_norm": 0.515625,
      "learning_rate": 0.00037290521900070774,
      "loss": 0.7208,
      "step": 5659
    },
    {
      "epoch": 0.7244336362472802,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00037258234844815964,
      "loss": 1.0128,
      "step": 5660
    },
    {
      "epoch": 0.7245616280558044,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0003722595857275146,
      "loss": 1.3968,
      "step": 5661
    },
    {
      "epoch": 0.7246896198643287,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0003719369308942446,
      "loss": 1.2977,
      "step": 5662
    },
    {
      "epoch": 0.724817611672853,
      "grad_norm": 0.71875,
      "learning_rate": 0.0003716143840038033,
      "loss": 1.3525,
      "step": 5663
    },
    {
      "epoch": 0.7249456034813772,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00037129194511162556,
      "loss": 0.8546,
      "step": 5664
    },
    {
      "epoch": 0.7250735952899015,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0003709696142731277,
      "loss": 1.3476,
      "step": 5665
    },
    {
      "epoch": 0.7252015870984257,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0003706473915437074,
      "loss": 1.1322,
      "step": 5666
    },
    {
      "epoch": 0.7253295789069499,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0003703252769787443,
      "loss": 1.1852,
      "step": 5667
    },
    {
      "epoch": 0.7254575707154742,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0003700032706335984,
      "loss": 1.14,
      "step": 5668
    },
    {
      "epoch": 0.7255855625239984,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0003696813725636119,
      "loss": 1.0576,
      "step": 5669
    },
    {
      "epoch": 0.7257135543325227,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0003693595828241083,
      "loss": 1.3215,
      "step": 5670
    },
    {
      "epoch": 0.725841546141047,
      "grad_norm": 0.734375,
      "learning_rate": 0.00036903790147039286,
      "loss": 0.9585,
      "step": 5671
    },
    {
      "epoch": 0.7259695379495712,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0003687163285577504,
      "loss": 0.9516,
      "step": 5672
    },
    {
      "epoch": 0.7260975297580955,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0003683948641414497,
      "loss": 0.6807,
      "step": 5673
    },
    {
      "epoch": 0.7262255215666198,
      "grad_norm": 0.765625,
      "learning_rate": 0.00036807350827673934,
      "loss": 1.492,
      "step": 5674
    },
    {
      "epoch": 0.726353513375144,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0003677522610188493,
      "loss": 1.5187,
      "step": 5675
    },
    {
      "epoch": 0.7264815051836683,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0003674311224229917,
      "loss": 1.1831,
      "step": 5676
    },
    {
      "epoch": 0.7266094969921925,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00036711009254435854,
      "loss": 1.7742,
      "step": 5677
    },
    {
      "epoch": 0.7267374888007168,
      "grad_norm": 0.671875,
      "learning_rate": 0.0003667891714381246,
      "loss": 1.4276,
      "step": 5678
    },
    {
      "epoch": 0.726865480609241,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0003664683591594452,
      "loss": 1.7183,
      "step": 5679
    },
    {
      "epoch": 0.7269934724177652,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0003661476557634575,
      "loss": 1.1381,
      "step": 5680
    },
    {
      "epoch": 0.7271214642262895,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0003658270613052789,
      "loss": 1.2793,
      "step": 5681
    },
    {
      "epoch": 0.7272494560348137,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0003655065758400091,
      "loss": 1.7224,
      "step": 5682
    },
    {
      "epoch": 0.727377447843338,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0003651861994227288,
      "loss": 0.9903,
      "step": 5683
    },
    {
      "epoch": 0.7275054396518623,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00036486593210849984,
      "loss": 1.4383,
      "step": 5684
    },
    {
      "epoch": 0.7276334314603865,
      "grad_norm": 0.65625,
      "learning_rate": 0.00036454577395236533,
      "loss": 0.9163,
      "step": 5685
    },
    {
      "epoch": 0.7277614232689108,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00036422572500934957,
      "loss": 1.3647,
      "step": 5686
    },
    {
      "epoch": 0.7278894150774351,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0003639057853344583,
      "loss": 1.4449,
      "step": 5687
    },
    {
      "epoch": 0.7280174068859593,
      "grad_norm": 0.640625,
      "learning_rate": 0.0003635859549826783,
      "loss": 1.4701,
      "step": 5688
    },
    {
      "epoch": 0.7281453986944836,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00036326623400897794,
      "loss": 1.3598,
      "step": 5689
    },
    {
      "epoch": 0.7282733905030078,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0003629466224683059,
      "loss": 1.362,
      "step": 5690
    },
    {
      "epoch": 0.7284013823115321,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00036262712041559267,
      "loss": 0.8958,
      "step": 5691
    },
    {
      "epoch": 0.7285293741200564,
      "grad_norm": 0.640625,
      "learning_rate": 0.0003623077279057501,
      "loss": 1.0122,
      "step": 5692
    },
    {
      "epoch": 0.7286573659285805,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0003619884449936708,
      "loss": 1.1048,
      "step": 5693
    },
    {
      "epoch": 0.7287853577371048,
      "grad_norm": 0.5625,
      "learning_rate": 0.00036166927173422926,
      "loss": 0.824,
      "step": 5694
    },
    {
      "epoch": 0.728913349545629,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0003613502081822794,
      "loss": 1.3975,
      "step": 5695
    },
    {
      "epoch": 0.7290413413541533,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0003610312543926585,
      "loss": 1.3874,
      "step": 5696
    },
    {
      "epoch": 0.7291693331626776,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00036071241042018364,
      "loss": 1.6499,
      "step": 5697
    },
    {
      "epoch": 0.7292973249712018,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0003603936763196538,
      "loss": 1.7507,
      "step": 5698
    },
    {
      "epoch": 0.7294253167797261,
      "grad_norm": 0.640625,
      "learning_rate": 0.00036007505214584766,
      "loss": 1.1812,
      "step": 5699
    },
    {
      "epoch": 0.7295533085882504,
      "grad_norm": 0.625,
      "learning_rate": 0.0003597565379535266,
      "loss": 1.1631,
      "step": 5700
    },
    {
      "epoch": 0.7296813003967746,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0003594381337974322,
      "loss": 1.6458,
      "step": 5701
    },
    {
      "epoch": 0.7298092922052989,
      "grad_norm": 0.65625,
      "learning_rate": 0.0003591198397322878,
      "loss": 0.8592,
      "step": 5702
    },
    {
      "epoch": 0.7299372840138231,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00035880165581279666,
      "loss": 1.1907,
      "step": 5703
    },
    {
      "epoch": 0.7300652758223474,
      "grad_norm": 0.5,
      "learning_rate": 0.0003584835820936444,
      "loss": 0.8947,
      "step": 5704
    },
    {
      "epoch": 0.7301932676308717,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00035816561862949694,
      "loss": 1.3307,
      "step": 5705
    },
    {
      "epoch": 0.7303212594393959,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0003578477654750014,
      "loss": 1.1936,
      "step": 5706
    },
    {
      "epoch": 0.7304492512479202,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0003575300226847863,
      "loss": 0.8877,
      "step": 5707
    },
    {
      "epoch": 0.7305772430564443,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0003572123903134606,
      "loss": 1.1707,
      "step": 5708
    },
    {
      "epoch": 0.7307052348649686,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0003568948684156148,
      "loss": 1.3404,
      "step": 5709
    },
    {
      "epoch": 0.7308332266734929,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0003565774570458201,
      "loss": 1.6692,
      "step": 5710
    },
    {
      "epoch": 0.7309612184820171,
      "grad_norm": 0.734375,
      "learning_rate": 0.0003562601562586291,
      "loss": 1.7411,
      "step": 5711
    },
    {
      "epoch": 0.7310892102905414,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00035594296610857437,
      "loss": 1.5402,
      "step": 5712
    },
    {
      "epoch": 0.7312172020990657,
      "grad_norm": 0.671875,
      "learning_rate": 0.0003556258866501706,
      "loss": 1.6075,
      "step": 5713
    },
    {
      "epoch": 0.7313451939075899,
      "grad_norm": 0.75,
      "learning_rate": 0.0003553089179379131,
      "loss": 1.4542,
      "step": 5714
    },
    {
      "epoch": 0.7314731857161142,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00035499206002627794,
      "loss": 1.2959,
      "step": 5715
    },
    {
      "epoch": 0.7316011775246384,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00035467531296972266,
      "loss": 1.4088,
      "step": 5716
    },
    {
      "epoch": 0.7317291693331627,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00035435867682268485,
      "loss": 1.2917,
      "step": 5717
    },
    {
      "epoch": 0.731857161141687,
      "grad_norm": 0.5625,
      "learning_rate": 0.00035404215163958343,
      "loss": 1.1359,
      "step": 5718
    },
    {
      "epoch": 0.7319851529502112,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00035372573747481905,
      "loss": 1.2973,
      "step": 5719
    },
    {
      "epoch": 0.7321131447587355,
      "grad_norm": 0.71875,
      "learning_rate": 0.00035340943438277276,
      "loss": 1.4663,
      "step": 5720
    },
    {
      "epoch": 0.7322411365672596,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0003530932424178055,
      "loss": 1.6634,
      "step": 5721
    },
    {
      "epoch": 0.7323691283757839,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0003527771616342604,
      "loss": 0.9416,
      "step": 5722
    },
    {
      "epoch": 0.7324971201843082,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0003524611920864612,
      "loss": 1.3514,
      "step": 5723
    },
    {
      "epoch": 0.7326251119928324,
      "grad_norm": 0.703125,
      "learning_rate": 0.0003521453338287127,
      "loss": 1.2258,
      "step": 5724
    },
    {
      "epoch": 0.7327531038013567,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0003518295869152994,
      "loss": 1.2931,
      "step": 5725
    },
    {
      "epoch": 0.732881095609881,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00035151395140048814,
      "loss": 1.3164,
      "step": 5726
    },
    {
      "epoch": 0.7330090874184052,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00035119842733852567,
      "loss": 0.9159,
      "step": 5727
    },
    {
      "epoch": 0.7331370792269295,
      "grad_norm": 0.75,
      "learning_rate": 0.00035088301478364026,
      "loss": 1.1459,
      "step": 5728
    },
    {
      "epoch": 0.7332650710354537,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0003505677137900405,
      "loss": 1.4833,
      "step": 5729
    },
    {
      "epoch": 0.733393062843978,
      "grad_norm": 0.6640625,
      "learning_rate": 0.000350252524411916,
      "loss": 1.3555,
      "step": 5730
    },
    {
      "epoch": 0.7335210546525023,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0003499374467034372,
      "loss": 1.4596,
      "step": 5731
    },
    {
      "epoch": 0.7336490464610265,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00034962248071875523,
      "loss": 1.531,
      "step": 5732
    },
    {
      "epoch": 0.7337770382695508,
      "grad_norm": 0.5625,
      "learning_rate": 0.0003493076265120025,
      "loss": 0.9429,
      "step": 5733
    },
    {
      "epoch": 0.733905030078075,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0003489928841372912,
      "loss": 1.1085,
      "step": 5734
    },
    {
      "epoch": 0.7340330218865992,
      "grad_norm": 0.859375,
      "learning_rate": 0.0003486782536487152,
      "loss": 1.3353,
      "step": 5735
    },
    {
      "epoch": 0.7341610136951235,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0003483637351003488,
      "loss": 1.4862,
      "step": 5736
    },
    {
      "epoch": 0.7342890055036477,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00034804932854624713,
      "loss": 1.5332,
      "step": 5737
    },
    {
      "epoch": 0.734416997312172,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00034773503404044646,
      "loss": 1.3282,
      "step": 5738
    },
    {
      "epoch": 0.7345449891206963,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0003474208516369628,
      "loss": 0.9028,
      "step": 5739
    },
    {
      "epoch": 0.7346729809292205,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0003471067813897932,
      "loss": 1.3195,
      "step": 5740
    },
    {
      "epoch": 0.7348009727377448,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0003467928233529167,
      "loss": 1.0851,
      "step": 5741
    },
    {
      "epoch": 0.734928964546269,
      "grad_norm": 0.7734375,
      "learning_rate": 0.000346478977580292,
      "loss": 1.9963,
      "step": 5742
    },
    {
      "epoch": 0.7350569563547933,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00034616524412585793,
      "loss": 1.3965,
      "step": 5743
    },
    {
      "epoch": 0.7351849481633176,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0003458516230435349,
      "loss": 1.4576,
      "step": 5744
    },
    {
      "epoch": 0.7353129399718418,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0003455381143872238,
      "loss": 1.5325,
      "step": 5745
    },
    {
      "epoch": 0.7354409317803661,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00034522471821080624,
      "loss": 1.1844,
      "step": 5746
    },
    {
      "epoch": 0.7355689235888903,
      "grad_norm": 0.9375,
      "learning_rate": 0.00034491143456814475,
      "loss": 1.4208,
      "step": 5747
    },
    {
      "epoch": 0.7356969153974146,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00034459826351308163,
      "loss": 1.2089,
      "step": 5748
    },
    {
      "epoch": 0.7358249072059388,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00034428520509944063,
      "loss": 1.5339,
      "step": 5749
    },
    {
      "epoch": 0.735952899014463,
      "grad_norm": 0.6875,
      "learning_rate": 0.00034397225938102594,
      "loss": 1.158,
      "step": 5750
    },
    {
      "epoch": 0.7360808908229873,
      "grad_norm": 0.640625,
      "learning_rate": 0.0003436594264116224,
      "loss": 1.4649,
      "step": 5751
    },
    {
      "epoch": 0.7362088826315116,
      "grad_norm": 0.609375,
      "learning_rate": 0.00034334670624499534,
      "loss": 1.1591,
      "step": 5752
    },
    {
      "epoch": 0.7363368744400358,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0003430340989348909,
      "loss": 1.155,
      "step": 5753
    },
    {
      "epoch": 0.7364648662485601,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0003427216045350356,
      "loss": 1.2908,
      "step": 5754
    },
    {
      "epoch": 0.7365928580570843,
      "grad_norm": 0.640625,
      "learning_rate": 0.00034240922309913715,
      "loss": 1.1896,
      "step": 5755
    },
    {
      "epoch": 0.7367208498656086,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00034209695468088254,
      "loss": 1.2353,
      "step": 5756
    },
    {
      "epoch": 0.7368488416741329,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0003417847993339407,
      "loss": 1.3299,
      "step": 5757
    },
    {
      "epoch": 0.7369768334826571,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00034147275711196036,
      "loss": 1.3543,
      "step": 5758
    },
    {
      "epoch": 0.7371048252911814,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0003411608280685713,
      "loss": 1.5475,
      "step": 5759
    },
    {
      "epoch": 0.7372328170997057,
      "grad_norm": 0.765625,
      "learning_rate": 0.00034084901225738364,
      "loss": 1.2073,
      "step": 5760
    },
    {
      "epoch": 0.7373608089082299,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0003405373097319875,
      "loss": 0.9131,
      "step": 5761
    },
    {
      "epoch": 0.7374888007167542,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0003402257205459544,
      "loss": 0.8869,
      "step": 5762
    },
    {
      "epoch": 0.7376167925252783,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0003399142447528354,
      "loss": 1.0147,
      "step": 5763
    },
    {
      "epoch": 0.7377447843338026,
      "grad_norm": 0.640625,
      "learning_rate": 0.00033960288240616386,
      "loss": 0.5954,
      "step": 5764
    },
    {
      "epoch": 0.7378727761423269,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00033929163355945157,
      "loss": 0.8415,
      "step": 5765
    },
    {
      "epoch": 0.7380007679508511,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0003389804982661918,
      "loss": 1.5789,
      "step": 5766
    },
    {
      "epoch": 0.7381287597593754,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00033866947657985825,
      "loss": 1.4877,
      "step": 5767
    },
    {
      "epoch": 0.7382567515678996,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0003383585685539051,
      "loss": 1.1663,
      "step": 5768
    },
    {
      "epoch": 0.7383847433764239,
      "grad_norm": 0.640625,
      "learning_rate": 0.00033804777424176733,
      "loss": 1.2081,
      "step": 5769
    },
    {
      "epoch": 0.7385127351849482,
      "grad_norm": 0.703125,
      "learning_rate": 0.00033773709369685925,
      "loss": 1.663,
      "step": 5770
    },
    {
      "epoch": 0.7386407269934724,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00033742652697257657,
      "loss": 1.1859,
      "step": 5771
    },
    {
      "epoch": 0.7387687188019967,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00033711607412229526,
      "loss": 1.5242,
      "step": 5772
    },
    {
      "epoch": 0.738896710610521,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00033680573519937184,
      "loss": 1.3585,
      "step": 5773
    },
    {
      "epoch": 0.7390247024190452,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0003364955102571431,
      "loss": 1.1994,
      "step": 5774
    },
    {
      "epoch": 0.7391526942275695,
      "grad_norm": 0.58203125,
      "learning_rate": 0.000336185399348926,
      "loss": 1.2169,
      "step": 5775
    },
    {
      "epoch": 0.7392806860360936,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0003358754025280182,
      "loss": 1.469,
      "step": 5776
    },
    {
      "epoch": 0.739408677844618,
      "grad_norm": 1.59375,
      "learning_rate": 0.0003355655198476978,
      "loss": 1.4007,
      "step": 5777
    },
    {
      "epoch": 0.7395366696531422,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0003352557513612234,
      "loss": 1.1553,
      "step": 5778
    },
    {
      "epoch": 0.7396646614616664,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00033494609712183323,
      "loss": 0.8752,
      "step": 5779
    },
    {
      "epoch": 0.7397926532701907,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00033463655718274653,
      "loss": 1.7366,
      "step": 5780
    },
    {
      "epoch": 0.7399206450787149,
      "grad_norm": 0.9375,
      "learning_rate": 0.0003343271315971629,
      "loss": 2.1159,
      "step": 5781
    },
    {
      "epoch": 0.7400486368872392,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00033401782041826234,
      "loss": 1.4447,
      "step": 5782
    },
    {
      "epoch": 0.7401766286957635,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0003337086236992044,
      "loss": 1.0371,
      "step": 5783
    },
    {
      "epoch": 0.7403046205042877,
      "grad_norm": 0.84375,
      "learning_rate": 0.0003333995414931299,
      "loss": 1.6133,
      "step": 5784
    },
    {
      "epoch": 0.740432612312812,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0003330905738531597,
      "loss": 0.8074,
      "step": 5785
    },
    {
      "epoch": 0.7405606041213363,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0003327817208323947,
      "loss": 0.8179,
      "step": 5786
    },
    {
      "epoch": 0.7406885959298605,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00033247298248391654,
      "loss": 1.5871,
      "step": 5787
    },
    {
      "epoch": 0.7408165877383848,
      "grad_norm": 0.515625,
      "learning_rate": 0.00033216435886078676,
      "loss": 0.9563,
      "step": 5788
    },
    {
      "epoch": 0.740944579546909,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00033185585001604734,
      "loss": 1.2488,
      "step": 5789
    },
    {
      "epoch": 0.7410725713554333,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0003315474560027206,
      "loss": 1.415,
      "step": 5790
    },
    {
      "epoch": 0.7412005631639575,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0003312391768738094,
      "loss": 0.6866,
      "step": 5791
    },
    {
      "epoch": 0.7413285549724817,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0003309310126822959,
      "loss": 0.9356,
      "step": 5792
    },
    {
      "epoch": 0.741456546781006,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00033062296348114316,
      "loss": 1.3268,
      "step": 5793
    },
    {
      "epoch": 0.7415845385895302,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00033031502932329474,
      "loss": 1.3601,
      "step": 5794
    },
    {
      "epoch": 0.7417125303980545,
      "grad_norm": 0.609375,
      "learning_rate": 0.0003300072102616739,
      "loss": 1.3033,
      "step": 5795
    },
    {
      "epoch": 0.7418405222065788,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00032969950634918456,
      "loss": 0.6575,
      "step": 5796
    },
    {
      "epoch": 0.741968514015103,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00032939191763871047,
      "loss": 1.5261,
      "step": 5797
    },
    {
      "epoch": 0.7420965058236273,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0003290844441831158,
      "loss": 1.8313,
      "step": 5798
    },
    {
      "epoch": 0.7422244976321516,
      "grad_norm": 0.59375,
      "learning_rate": 0.00032877708603524483,
      "loss": 0.889,
      "step": 5799
    },
    {
      "epoch": 0.7423524894406758,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0003284698432479224,
      "loss": 0.9925,
      "step": 5800
    },
    {
      "epoch": 0.7424804812492001,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0003281627158739525,
      "loss": 1.5795,
      "step": 5801
    },
    {
      "epoch": 0.7426084730577243,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0003278557039661202,
      "loss": 1.3945,
      "step": 5802
    },
    {
      "epoch": 0.7427364648662486,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00032754880757719066,
      "loss": 1.2491,
      "step": 5803
    },
    {
      "epoch": 0.7428644566747729,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0003272420267599092,
      "loss": 1.5644,
      "step": 5804
    },
    {
      "epoch": 0.742992448483297,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00032693536156700045,
      "loss": 1.4106,
      "step": 5805
    },
    {
      "epoch": 0.7431204402918213,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00032662881205117,
      "loss": 1.4181,
      "step": 5806
    },
    {
      "epoch": 0.7432484321003455,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0003263223782651036,
      "loss": 1.5786,
      "step": 5807
    },
    {
      "epoch": 0.7433764239088698,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0003260160602614668,
      "loss": 1.6273,
      "step": 5808
    },
    {
      "epoch": 0.7435044157173941,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0003257098580929052,
      "loss": 1.7547,
      "step": 5809
    },
    {
      "epoch": 0.7436324075259183,
      "grad_norm": 0.65625,
      "learning_rate": 0.0003254037718120446,
      "loss": 1.3642,
      "step": 5810
    },
    {
      "epoch": 0.7437603993344426,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00032509780147149117,
      "loss": 1.6065,
      "step": 5811
    },
    {
      "epoch": 0.7438883911429669,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00032479194712383054,
      "loss": 1.0392,
      "step": 5812
    },
    {
      "epoch": 0.7440163829514911,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0003244862088216293,
      "loss": 0.8275,
      "step": 5813
    },
    {
      "epoch": 0.7441443747600154,
      "grad_norm": 0.59375,
      "learning_rate": 0.00032418058661743267,
      "loss": 0.931,
      "step": 5814
    },
    {
      "epoch": 0.7442723665685396,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00032387508056376724,
      "loss": 1.4868,
      "step": 5815
    },
    {
      "epoch": 0.7444003583770639,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00032356969071313925,
      "loss": 1.7398,
      "step": 5816
    },
    {
      "epoch": 0.7445283501855882,
      "grad_norm": 0.65625,
      "learning_rate": 0.0003232644171180349,
      "loss": 1.1891,
      "step": 5817
    },
    {
      "epoch": 0.7446563419941123,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00032295925983092023,
      "loss": 0.9595,
      "step": 5818
    },
    {
      "epoch": 0.7447843338026366,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00032265421890424165,
      "loss": 1.0173,
      "step": 5819
    },
    {
      "epoch": 0.7449123256111608,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00032234929439042527,
      "loss": 1.3434,
      "step": 5820
    },
    {
      "epoch": 0.7450403174196851,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0003220444863418773,
      "loss": 1.6793,
      "step": 5821
    },
    {
      "epoch": 0.7451683092282094,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0003217397948109845,
      "loss": 1.4386,
      "step": 5822
    },
    {
      "epoch": 0.7452963010367336,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00032143521985011214,
      "loss": 1.1843,
      "step": 5823
    },
    {
      "epoch": 0.7454242928452579,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0003211307615116068,
      "loss": 1.5233,
      "step": 5824
    },
    {
      "epoch": 0.7455522846537822,
      "grad_norm": 0.5625,
      "learning_rate": 0.0003208264198477947,
      "loss": 1.0409,
      "step": 5825
    },
    {
      "epoch": 0.7456802764623064,
      "grad_norm": 0.625,
      "learning_rate": 0.000320522194910982,
      "loss": 1.0026,
      "step": 5826
    },
    {
      "epoch": 0.7458082682708307,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0003202180867534541,
      "loss": 1.3006,
      "step": 5827
    },
    {
      "epoch": 0.7459362600793549,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0003199140954274774,
      "loss": 1.2539,
      "step": 5828
    },
    {
      "epoch": 0.7460642518878792,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00031961022098529756,
      "loss": 1.3711,
      "step": 5829
    },
    {
      "epoch": 0.7461922436964035,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00031930646347914037,
      "loss": 1.1335,
      "step": 5830
    },
    {
      "epoch": 0.7463202355049277,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00031900282296121163,
      "loss": 1.3259,
      "step": 5831
    },
    {
      "epoch": 0.746448227313452,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0003186992994836967,
      "loss": 1.5586,
      "step": 5832
    },
    {
      "epoch": 0.7465762191219762,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00031839589309876114,
      "loss": 1.6686,
      "step": 5833
    },
    {
      "epoch": 0.7467042109305004,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0003180926038585502,
      "loss": 0.8806,
      "step": 5834
    },
    {
      "epoch": 0.7468322027390247,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00031778943181518936,
      "loss": 1.1191,
      "step": 5835
    },
    {
      "epoch": 0.7469601945475489,
      "grad_norm": 0.73046875,
      "learning_rate": 0.000317486377020783,
      "loss": 1.8472,
      "step": 5836
    },
    {
      "epoch": 0.7470881863560732,
      "grad_norm": 0.75,
      "learning_rate": 0.0003171834395274165,
      "loss": 1.3216,
      "step": 5837
    },
    {
      "epoch": 0.7472161781645975,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00031688061938715427,
      "loss": 1.2393,
      "step": 5838
    },
    {
      "epoch": 0.7473441699731217,
      "grad_norm": 0.53125,
      "learning_rate": 0.0003165779166520415,
      "loss": 0.8865,
      "step": 5839
    },
    {
      "epoch": 0.747472161781646,
      "grad_norm": 0.671875,
      "learning_rate": 0.0003162753313741015,
      "loss": 1.4656,
      "step": 5840
    },
    {
      "epoch": 0.7476001535901702,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0003159728636053393,
      "loss": 1.0002,
      "step": 5841
    },
    {
      "epoch": 0.7477281453986945,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00031567051339773865,
      "loss": 1.6096,
      "step": 5842
    },
    {
      "epoch": 0.7478561372072188,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0003153682808032634,
      "loss": 0.9653,
      "step": 5843
    },
    {
      "epoch": 0.747984129015743,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0003150661658738575,
      "loss": 1.4388,
      "step": 5844
    },
    {
      "epoch": 0.7481121208242673,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0003147641686614434,
      "loss": 1.1085,
      "step": 5845
    },
    {
      "epoch": 0.7482401126327916,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00031446228921792473,
      "loss": 1.8167,
      "step": 5846
    },
    {
      "epoch": 0.7483681044413157,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0003141605275951844,
      "loss": 1.4167,
      "step": 5847
    },
    {
      "epoch": 0.74849609624984,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0003138588838450853,
      "loss": 1.1688,
      "step": 5848
    },
    {
      "epoch": 0.7486240880583642,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00031355735801946917,
      "loss": 0.8056,
      "step": 5849
    },
    {
      "epoch": 0.7487520798668885,
      "grad_norm": 0.703125,
      "learning_rate": 0.0003132559501701584,
      "loss": 0.8408,
      "step": 5850
    },
    {
      "epoch": 0.7488800716754128,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0003129546603489548,
      "loss": 0.9023,
      "step": 5851
    },
    {
      "epoch": 0.749008063483937,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00031265348860764,
      "loss": 1.3145,
      "step": 5852
    },
    {
      "epoch": 0.7491360552924613,
      "grad_norm": 0.921875,
      "learning_rate": 0.00031235243499797526,
      "loss": 1.8804,
      "step": 5853
    },
    {
      "epoch": 0.7492640471009855,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00031205149957170143,
      "loss": 1.4838,
      "step": 5854
    },
    {
      "epoch": 0.7493920389095098,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0003117506823805393,
      "loss": 1.2222,
      "step": 5855
    },
    {
      "epoch": 0.7495200307180341,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00031144998347618893,
      "loss": 1.2151,
      "step": 5856
    },
    {
      "epoch": 0.7496480225265583,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0003111494029103309,
      "loss": 1.5949,
      "step": 5857
    },
    {
      "epoch": 0.7497760143350826,
      "grad_norm": 0.51171875,
      "learning_rate": 0.000310848940734624,
      "loss": 0.9347,
      "step": 5858
    },
    {
      "epoch": 0.7499040061436069,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00031054859700070794,
      "loss": 0.7799,
      "step": 5859
    },
    {
      "epoch": 0.750031997952131,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0003102483717602017,
      "loss": 1.1463,
      "step": 5860
    },
    {
      "epoch": 0.7501599897606553,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00030994826506470385,
      "loss": 1.6082,
      "step": 5861
    },
    {
      "epoch": 0.7502879815691795,
      "grad_norm": 0.953125,
      "learning_rate": 0.00030964827696579257,
      "loss": 1.8192,
      "step": 5862
    },
    {
      "epoch": 0.7504159733777038,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00030934840751502566,
      "loss": 1.4278,
      "step": 5863
    },
    {
      "epoch": 0.7505439651862281,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00030904865676394057,
      "loss": 1.5621,
      "step": 5864
    },
    {
      "epoch": 0.7506719569947523,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0003087490247640544,
      "loss": 1.3564,
      "step": 5865
    },
    {
      "epoch": 0.7507999488032766,
      "grad_norm": 0.703125,
      "learning_rate": 0.000308449511566864,
      "loss": 1.61,
      "step": 5866
    },
    {
      "epoch": 0.7509279406118008,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00030815011722384487,
      "loss": 0.9814,
      "step": 5867
    },
    {
      "epoch": 0.7510559324203251,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0003078508417864533,
      "loss": 1.3305,
      "step": 5868
    },
    {
      "epoch": 0.7511839242288494,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00030755168530612444,
      "loss": 1.43,
      "step": 5869
    },
    {
      "epoch": 0.7513119160373736,
      "grad_norm": 0.6875,
      "learning_rate": 0.00030725264783427363,
      "loss": 1.5289,
      "step": 5870
    },
    {
      "epoch": 0.7514399078458979,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0003069537294222946,
      "loss": 0.7228,
      "step": 5871
    },
    {
      "epoch": 0.7515678996544222,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00030665493012156176,
      "loss": 1.3002,
      "step": 5872
    },
    {
      "epoch": 0.7516958914629464,
      "grad_norm": 0.65625,
      "learning_rate": 0.00030635624998342836,
      "loss": 1.6737,
      "step": 5873
    },
    {
      "epoch": 0.7518238832714706,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0003060576890592278,
      "loss": 1.0701,
      "step": 5874
    },
    {
      "epoch": 0.7519518750799948,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00030575924740027226,
      "loss": 1.202,
      "step": 5875
    },
    {
      "epoch": 0.7520798668885191,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00030546092505785407,
      "loss": 1.1294,
      "step": 5876
    },
    {
      "epoch": 0.7522078586970434,
      "grad_norm": 0.703125,
      "learning_rate": 0.0003051627220832446,
      "loss": 1.4866,
      "step": 5877
    },
    {
      "epoch": 0.7523358505055676,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00030486463852769487,
      "loss": 1.3078,
      "step": 5878
    },
    {
      "epoch": 0.7524638423140919,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0003045666744424358,
      "loss": 1.1349,
      "step": 5879
    },
    {
      "epoch": 0.7525918341226161,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00030426882987867653,
      "loss": 0.8171,
      "step": 5880
    },
    {
      "epoch": 0.7527198259311404,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00030397110488760706,
      "loss": 1.5926,
      "step": 5881
    },
    {
      "epoch": 0.7528478177396647,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00030367349952039604,
      "loss": 1.418,
      "step": 5882
    },
    {
      "epoch": 0.7529758095481889,
      "grad_norm": 0.484375,
      "learning_rate": 0.0003033760138281918,
      "loss": 0.7875,
      "step": 5883
    },
    {
      "epoch": 0.7531038013567132,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0003030786478621226,
      "loss": 1.1093,
      "step": 5884
    },
    {
      "epoch": 0.7532317931652375,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00030278140167329436,
      "loss": 1.2292,
      "step": 5885
    },
    {
      "epoch": 0.7533597849737617,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00030248427531279475,
      "loss": 1.4123,
      "step": 5886
    },
    {
      "epoch": 0.753487776782286,
      "grad_norm": 0.65625,
      "learning_rate": 0.0003021872688316896,
      "loss": 1.6505,
      "step": 5887
    },
    {
      "epoch": 0.7536157685908101,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0003018903822810244,
      "loss": 1.4422,
      "step": 5888
    },
    {
      "epoch": 0.7537437603993344,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00030159361571182333,
      "loss": 1.389,
      "step": 5889
    },
    {
      "epoch": 0.7538717522078587,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00030129696917509087,
      "loss": 1.575,
      "step": 5890
    },
    {
      "epoch": 0.7539997440163829,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0003010004427218106,
      "loss": 0.7569,
      "step": 5891
    },
    {
      "epoch": 0.7541277358249072,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00030070403640294565,
      "loss": 1.5979,
      "step": 5892
    },
    {
      "epoch": 0.7542557276334314,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00030040775026943755,
      "loss": 1.8735,
      "step": 5893
    },
    {
      "epoch": 0.7543837194419557,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0003001115843722083,
      "loss": 0.7654,
      "step": 5894
    },
    {
      "epoch": 0.75451171125048,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00029981553876215884,
      "loss": 1.4183,
      "step": 5895
    },
    {
      "epoch": 0.7546397030590042,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002995196134901694,
      "loss": 1.1844,
      "step": 5896
    },
    {
      "epoch": 0.7547676948675285,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002992238086070994,
      "loss": 1.5381,
      "step": 5897
    },
    {
      "epoch": 0.7548956866760528,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002989281241637879,
      "loss": 0.838,
      "step": 5898
    },
    {
      "epoch": 0.755023678484577,
      "grad_norm": 0.6328125,
      "learning_rate": 0.000298632560211053,
      "loss": 1.2481,
      "step": 5899
    },
    {
      "epoch": 0.7551516702931013,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002983371167996921,
      "loss": 0.9281,
      "step": 5900
    },
    {
      "epoch": 0.7552796621016254,
      "grad_norm": 0.640625,
      "learning_rate": 0.00029804179398048236,
      "loss": 1.2941,
      "step": 5901
    },
    {
      "epoch": 0.7554076539101497,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00029774659180417906,
      "loss": 1.0686,
      "step": 5902
    },
    {
      "epoch": 0.755535645718674,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00029745151032151785,
      "loss": 1.3726,
      "step": 5903
    },
    {
      "epoch": 0.7556636375271982,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002971565495832133,
      "loss": 1.0257,
      "step": 5904
    },
    {
      "epoch": 0.7557916293357225,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0002968617096399592,
      "loss": 1.3615,
      "step": 5905
    },
    {
      "epoch": 0.7559196211442468,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00029656699054242874,
      "loss": 1.4151,
      "step": 5906
    },
    {
      "epoch": 0.756047612952771,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00029627239234127377,
      "loss": 0.9974,
      "step": 5907
    },
    {
      "epoch": 0.7561756047612953,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002959779150871256,
      "loss": 0.8485,
      "step": 5908
    },
    {
      "epoch": 0.7563035965698195,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002956835588305957,
      "loss": 1.5071,
      "step": 5909
    },
    {
      "epoch": 0.7564315883783438,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002953893236222739,
      "loss": 1.3791,
      "step": 5910
    },
    {
      "epoch": 0.7565595801868681,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0002950952095127286,
      "loss": 1.7148,
      "step": 5911
    },
    {
      "epoch": 0.7566875719953923,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002948012165525085,
      "loss": 1.2198,
      "step": 5912
    },
    {
      "epoch": 0.7568155638039166,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002945073447921409,
      "loss": 1.1494,
      "step": 5913
    },
    {
      "epoch": 0.7569435556124408,
      "grad_norm": 0.6875,
      "learning_rate": 0.00029421359428213266,
      "loss": 1.0119,
      "step": 5914
    },
    {
      "epoch": 0.757071547420965,
      "grad_norm": 0.734375,
      "learning_rate": 0.0002939199650729698,
      "loss": 1.6029,
      "step": 5915
    },
    {
      "epoch": 0.7571995392294893,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00029362645721511647,
      "loss": 1.5707,
      "step": 5916
    },
    {
      "epoch": 0.7573275310380135,
      "grad_norm": 0.6875,
      "learning_rate": 0.00029333307075901736,
      "loss": 1.3455,
      "step": 5917
    },
    {
      "epoch": 0.7574555228465378,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002930398057550955,
      "loss": 1.0185,
      "step": 5918
    },
    {
      "epoch": 0.7575835146550621,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002927466622537532,
      "loss": 1.1452,
      "step": 5919
    },
    {
      "epoch": 0.7577115064635863,
      "grad_norm": 0.85546875,
      "learning_rate": 0.000292453640305372,
      "loss": 1.3773,
      "step": 5920
    },
    {
      "epoch": 0.7578394982721106,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00029216073996031255,
      "loss": 1.7001,
      "step": 5921
    },
    {
      "epoch": 0.7579674900806348,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002918679612689145,
      "loss": 1.3534,
      "step": 5922
    },
    {
      "epoch": 0.7580954818891591,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00029157530428149684,
      "loss": 1.3839,
      "step": 5923
    },
    {
      "epoch": 0.7582234736976834,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002912827690483568,
      "loss": 1.2468,
      "step": 5924
    },
    {
      "epoch": 0.7583514655062076,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002909903556197717,
      "loss": 1.5415,
      "step": 5925
    },
    {
      "epoch": 0.7584794573147319,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002906980640459975,
      "loss": 1.5097,
      "step": 5926
    },
    {
      "epoch": 0.7586074491232561,
      "grad_norm": 0.8125,
      "learning_rate": 0.0002904058943772693,
      "loss": 0.8155,
      "step": 5927
    },
    {
      "epoch": 0.7587354409317804,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00029011384666380145,
      "loss": 1.2594,
      "step": 5928
    },
    {
      "epoch": 0.7588634327403047,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002898219209557864,
      "loss": 1.0556,
      "step": 5929
    },
    {
      "epoch": 0.7589914245488288,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0002895301173033964,
      "loss": 1.6612,
      "step": 5930
    },
    {
      "epoch": 0.7591194163573531,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00028923843575678334,
      "loss": 0.7407,
      "step": 5931
    },
    {
      "epoch": 0.7592474081658774,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002889468763660772,
      "loss": 0.8497,
      "step": 5932
    },
    {
      "epoch": 0.7593753999744016,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002886554391813868,
      "loss": 1.3403,
      "step": 5933
    },
    {
      "epoch": 0.7595033917829259,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002883641242528006,
      "loss": 0.7408,
      "step": 5934
    },
    {
      "epoch": 0.7596313835914501,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00028807293163038574,
      "loss": 0.7503,
      "step": 5935
    },
    {
      "epoch": 0.7597593753999744,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002877818613641884,
      "loss": 1.1935,
      "step": 5936
    },
    {
      "epoch": 0.7598873672084987,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00028749091350423406,
      "loss": 1.4146,
      "step": 5937
    },
    {
      "epoch": 0.7600153590170229,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002872000881005262,
      "loss": 1.228,
      "step": 5938
    },
    {
      "epoch": 0.7601433508255472,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00028690938520304824,
      "loss": 1.4191,
      "step": 5939
    },
    {
      "epoch": 0.7602713426340714,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002866188048617622,
      "loss": 1.7078,
      "step": 5940
    },
    {
      "epoch": 0.7603993344425957,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00028632834712660884,
      "loss": 1.0708,
      "step": 5941
    },
    {
      "epoch": 0.76052732625112,
      "grad_norm": 0.578125,
      "learning_rate": 0.00028603801204750824,
      "loss": 0.8285,
      "step": 5942
    },
    {
      "epoch": 0.7606553180596441,
      "grad_norm": 0.625,
      "learning_rate": 0.00028574779967435916,
      "loss": 0.6676,
      "step": 5943
    },
    {
      "epoch": 0.7607833098681684,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00028545771005703924,
      "loss": 1.6365,
      "step": 5944
    },
    {
      "epoch": 0.7609113016766927,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002851677432454052,
      "loss": 0.7824,
      "step": 5945
    },
    {
      "epoch": 0.7610392934852169,
      "grad_norm": 0.734375,
      "learning_rate": 0.0002848778992892928,
      "loss": 1.5546,
      "step": 5946
    },
    {
      "epoch": 0.7611672852937412,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002845881782385156,
      "loss": 1.9672,
      "step": 5947
    },
    {
      "epoch": 0.7612952771022654,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00028429858014286744,
      "loss": 1.9262,
      "step": 5948
    },
    {
      "epoch": 0.7614232689107897,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00028400910505212043,
      "loss": 1.183,
      "step": 5949
    },
    {
      "epoch": 0.761551260719314,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002837197530160257,
      "loss": 1.7212,
      "step": 5950
    },
    {
      "epoch": 0.7616792525278382,
      "grad_norm": 0.765625,
      "learning_rate": 0.00028343052408431267,
      "loss": 1.3336,
      "step": 5951
    },
    {
      "epoch": 0.7618072443363625,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002831414183066903,
      "loss": 1.1047,
      "step": 5952
    },
    {
      "epoch": 0.7619352361448867,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002828524357328456,
      "loss": 1.3185,
      "step": 5953
    },
    {
      "epoch": 0.762063227953411,
      "grad_norm": 0.5625,
      "learning_rate": 0.000282563576412446,
      "loss": 0.9853,
      "step": 5954
    },
    {
      "epoch": 0.7621912197619353,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00028227484039513575,
      "loss": 1.2432,
      "step": 5955
    },
    {
      "epoch": 0.7623192115704595,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00028198622773053904,
      "loss": 1.1078,
      "step": 5956
    },
    {
      "epoch": 0.7624472033789838,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00028169773846825884,
      "loss": 1.2796,
      "step": 5957
    },
    {
      "epoch": 0.762575195187508,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002814093726578765,
      "loss": 0.8792,
      "step": 5958
    },
    {
      "epoch": 0.7627031869960322,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002811211303489527,
      "loss": 1.1183,
      "step": 5959
    },
    {
      "epoch": 0.7628311788045565,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002808330115910259,
      "loss": 1.2995,
      "step": 5960
    },
    {
      "epoch": 0.7629591706130807,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002805450164336145,
      "loss": 1.5348,
      "step": 5961
    },
    {
      "epoch": 0.763087162421605,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00028025714492621503,
      "loss": 1.2319,
      "step": 5962
    },
    {
      "epoch": 0.7632151542301293,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00027996939711830274,
      "loss": 1.1613,
      "step": 5963
    },
    {
      "epoch": 0.7633431460386535,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00027968177305933193,
      "loss": 1.5064,
      "step": 5964
    },
    {
      "epoch": 0.7634711378471778,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002793942727987354,
      "loss": 1.5781,
      "step": 5965
    },
    {
      "epoch": 0.763599129655702,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002791068963859248,
      "loss": 1.2689,
      "step": 5966
    },
    {
      "epoch": 0.7637271214642263,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00027881964387029034,
      "loss": 1.3662,
      "step": 5967
    },
    {
      "epoch": 0.7638551132727506,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002785325153012014,
      "loss": 1.601,
      "step": 5968
    },
    {
      "epoch": 0.7639831050812748,
      "grad_norm": 0.75,
      "learning_rate": 0.00027824551072800496,
      "loss": 1.6389,
      "step": 5969
    },
    {
      "epoch": 0.7641110968897991,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00027795863020002777,
      "loss": 1.415,
      "step": 5970
    },
    {
      "epoch": 0.7642390886983234,
      "grad_norm": 1.015625,
      "learning_rate": 0.000277671873766575,
      "loss": 1.1059,
      "step": 5971
    },
    {
      "epoch": 0.7643670805068475,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0002773852414769306,
      "loss": 1.081,
      "step": 5972
    },
    {
      "epoch": 0.7644950723153718,
      "grad_norm": 0.8125,
      "learning_rate": 0.0002770987333803564,
      "loss": 1.4324,
      "step": 5973
    },
    {
      "epoch": 0.764623064123896,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002768123495260938,
      "loss": 1.074,
      "step": 5974
    },
    {
      "epoch": 0.7647510559324203,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00027652608996336236,
      "loss": 1.1266,
      "step": 5975
    },
    {
      "epoch": 0.7648790477409446,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00027623995474136063,
      "loss": 0.9929,
      "step": 5976
    },
    {
      "epoch": 0.7650070395494688,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00027595394390926553,
      "loss": 1.0579,
      "step": 5977
    },
    {
      "epoch": 0.7651350313579931,
      "grad_norm": 0.828125,
      "learning_rate": 0.0002756680575162327,
      "loss": 1.402,
      "step": 5978
    },
    {
      "epoch": 0.7652630231665174,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00027538229561139616,
      "loss": 0.7966,
      "step": 5979
    },
    {
      "epoch": 0.7653910149750416,
      "grad_norm": 0.62890625,
      "learning_rate": 0.000275096658243869,
      "loss": 1.4967,
      "step": 5980
    },
    {
      "epoch": 0.7655190067835659,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002748111454627428,
      "loss": 0.9519,
      "step": 5981
    },
    {
      "epoch": 0.7656469985920901,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002745257573170868,
      "loss": 0.8974,
      "step": 5982
    },
    {
      "epoch": 0.7657749904006144,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002742404938559502,
      "loss": 1.3709,
      "step": 5983
    },
    {
      "epoch": 0.7659029822091387,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0002739553551283599,
      "loss": 1.4114,
      "step": 5984
    },
    {
      "epoch": 0.7660309740176628,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002736703411833218,
      "loss": 0.9999,
      "step": 5985
    },
    {
      "epoch": 0.7661589658261871,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00027338545206982,
      "loss": 1.0199,
      "step": 5986
    },
    {
      "epoch": 0.7662869576347113,
      "grad_norm": 0.703125,
      "learning_rate": 0.00027310068783681753,
      "loss": 1.8409,
      "step": 5987
    },
    {
      "epoch": 0.7664149494432356,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0002728160485332556,
      "loss": 1.0862,
      "step": 5988
    },
    {
      "epoch": 0.7665429412517599,
      "grad_norm": 0.7734375,
      "learning_rate": 0.000272531534208054,
      "loss": 1.7566,
      "step": 5989
    },
    {
      "epoch": 0.7666709330602841,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002722471449101117,
      "loss": 0.9934,
      "step": 5990
    },
    {
      "epoch": 0.7667989248688084,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002719628806883048,
      "loss": 0.8887,
      "step": 5991
    },
    {
      "epoch": 0.7669269166773327,
      "grad_norm": 0.56640625,
      "learning_rate": 0.000271678741591489,
      "loss": 1.1427,
      "step": 5992
    },
    {
      "epoch": 0.7670549084858569,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00027139472766849837,
      "loss": 1.0461,
      "step": 5993
    },
    {
      "epoch": 0.7671829002943812,
      "grad_norm": 0.703125,
      "learning_rate": 0.00027111083896814547,
      "loss": 1.1651,
      "step": 5994
    },
    {
      "epoch": 0.7673108921029054,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002708270755392207,
      "loss": 0.9382,
      "step": 5995
    },
    {
      "epoch": 0.7674388839114297,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002705434374304937,
      "loss": 1.1578,
      "step": 5996
    },
    {
      "epoch": 0.767566875719954,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00027025992469071215,
      "loss": 1.9144,
      "step": 5997
    },
    {
      "epoch": 0.7676948675284782,
      "grad_norm": 0.609375,
      "learning_rate": 0.00026997653736860215,
      "loss": 1.2737,
      "step": 5998
    },
    {
      "epoch": 0.7678228593370025,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00026969327551286924,
      "loss": 1.3793,
      "step": 5999
    },
    {
      "epoch": 0.7679508511455266,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002694101391721957,
      "loss": 1.0844,
      "step": 6000
    },
    {
      "epoch": 0.7680788429540509,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00026912712839524336,
      "loss": 1.0947,
      "step": 6001
    },
    {
      "epoch": 0.7682068347625752,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00026884424323065227,
      "loss": 0.8226,
      "step": 6002
    },
    {
      "epoch": 0.7683348265710994,
      "grad_norm": 0.6015625,
      "learning_rate": 0.000268561483727041,
      "loss": 0.9825,
      "step": 6003
    },
    {
      "epoch": 0.7684628183796237,
      "grad_norm": 0.90625,
      "learning_rate": 0.000268278849933006,
      "loss": 1.6239,
      "step": 6004
    },
    {
      "epoch": 0.768590810188148,
      "grad_norm": 0.609375,
      "learning_rate": 0.00026799634189712265,
      "loss": 1.0631,
      "step": 6005
    },
    {
      "epoch": 0.7687188019966722,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00026771395966794444,
      "loss": 1.3838,
      "step": 6006
    },
    {
      "epoch": 0.7688467938051965,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00026743170329400343,
      "loss": 1.4818,
      "step": 6007
    },
    {
      "epoch": 0.7689747856137207,
      "grad_norm": 0.703125,
      "learning_rate": 0.00026714957282381003,
      "loss": 1.3851,
      "step": 6008
    },
    {
      "epoch": 0.769102777422245,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002668675683058528,
      "loss": 1.2861,
      "step": 6009
    },
    {
      "epoch": 0.7692307692307693,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00026658568978859886,
      "loss": 0.8006,
      "step": 6010
    },
    {
      "epoch": 0.7693587610392935,
      "grad_norm": 0.734375,
      "learning_rate": 0.00026630393732049353,
      "loss": 1.4396,
      "step": 6011
    },
    {
      "epoch": 0.7694867528478178,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00026602231094996097,
      "loss": 1.4374,
      "step": 6012
    },
    {
      "epoch": 0.7696147446563419,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002657408107254027,
      "loss": 1.33,
      "step": 6013
    },
    {
      "epoch": 0.7697427364648662,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002654594366951991,
      "loss": 0.9101,
      "step": 6014
    },
    {
      "epoch": 0.7698707282733905,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00026517818890770917,
      "loss": 1.1259,
      "step": 6015
    },
    {
      "epoch": 0.7699987200819147,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002648970674112701,
      "loss": 1.1279,
      "step": 6016
    },
    {
      "epoch": 0.770126711890439,
      "grad_norm": 0.8125,
      "learning_rate": 0.00026461607225419635,
      "loss": 2.1696,
      "step": 6017
    },
    {
      "epoch": 0.7702547036989633,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00026433520348478223,
      "loss": 1.5144,
      "step": 6018
    },
    {
      "epoch": 0.7703826955074875,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0002640544611512993,
      "loss": 2.0164,
      "step": 6019
    },
    {
      "epoch": 0.7705106873160118,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00026377384530199775,
      "loss": 0.786,
      "step": 6020
    },
    {
      "epoch": 0.770638679124536,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00026349335598510604,
      "loss": 0.7691,
      "step": 6021
    },
    {
      "epoch": 0.7707666709330603,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00026321299324883065,
      "loss": 1.4258,
      "step": 6022
    },
    {
      "epoch": 0.7708946627415846,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002629327571413567,
      "loss": 1.2273,
      "step": 6023
    },
    {
      "epoch": 0.7710226545501088,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0002626526477108471,
      "loss": 1.177,
      "step": 6024
    },
    {
      "epoch": 0.7711506463586331,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002623726650054434,
      "loss": 1.2603,
      "step": 6025
    },
    {
      "epoch": 0.7712786381671572,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00026209280907326494,
      "loss": 1.0818,
      "step": 6026
    },
    {
      "epoch": 0.7714066299756815,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00026181307996240964,
      "loss": 1.1065,
      "step": 6027
    },
    {
      "epoch": 0.7715346217842058,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002615334777209534,
      "loss": 1.291,
      "step": 6028
    },
    {
      "epoch": 0.77166261359273,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002612540023969504,
      "loss": 1.0551,
      "step": 6029
    },
    {
      "epoch": 0.7717906054012543,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00026097465403843314,
      "loss": 1.0369,
      "step": 6030
    },
    {
      "epoch": 0.7719185972097786,
      "grad_norm": 0.55078125,
      "learning_rate": 0.000260695432693412,
      "loss": 0.8425,
      "step": 6031
    },
    {
      "epoch": 0.7720465890183028,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002604163384098759,
      "loss": 1.0374,
      "step": 6032
    },
    {
      "epoch": 0.7721745808268271,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00026013737123579153,
      "loss": 0.9619,
      "step": 6033
    },
    {
      "epoch": 0.7723025726353513,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002598585312191043,
      "loss": 1.4319,
      "step": 6034
    },
    {
      "epoch": 0.7724305644438756,
      "grad_norm": 0.734375,
      "learning_rate": 0.00025957981840773683,
      "loss": 1.2935,
      "step": 6035
    },
    {
      "epoch": 0.7725585562523999,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002593012328495906,
      "loss": 1.5247,
      "step": 6036
    },
    {
      "epoch": 0.7726865480609241,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002590227745925453,
      "loss": 1.0064,
      "step": 6037
    },
    {
      "epoch": 0.7728145398694484,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002587444436844586,
      "loss": 1.6959,
      "step": 6038
    },
    {
      "epoch": 0.7729425316779726,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00025846624017316565,
      "loss": 1.4747,
      "step": 6039
    },
    {
      "epoch": 0.7730705234864969,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0002581881641064806,
      "loss": 1.6392,
      "step": 6040
    },
    {
      "epoch": 0.7731985152950211,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002579102155321952,
      "loss": 1.0542,
      "step": 6041
    },
    {
      "epoch": 0.7733265071035453,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00025763239449807954,
      "loss": 0.819,
      "step": 6042
    },
    {
      "epoch": 0.7734544989120696,
      "grad_norm": 0.59375,
      "learning_rate": 0.00025735470105188164,
      "loss": 1.1657,
      "step": 6043
    },
    {
      "epoch": 0.7735824907205939,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002570771352413275,
      "loss": 0.8697,
      "step": 6044
    },
    {
      "epoch": 0.7737104825291181,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002567996971141214,
      "loss": 1.5477,
      "step": 6045
    },
    {
      "epoch": 0.7738384743376424,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002565223867179457,
      "loss": 1.0697,
      "step": 6046
    },
    {
      "epoch": 0.7739664661461666,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002562452041004608,
      "loss": 1.0237,
      "step": 6047
    },
    {
      "epoch": 0.7740944579546909,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002559681493093046,
      "loss": 1.1713,
      "step": 6048
    },
    {
      "epoch": 0.7742224497632152,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002556912223920936,
      "loss": 1.5754,
      "step": 6049
    },
    {
      "epoch": 0.7743504415717394,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0002554144233964224,
      "loss": 2.339,
      "step": 6050
    },
    {
      "epoch": 0.7744784333802637,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00025513775236986314,
      "loss": 1.1741,
      "step": 6051
    },
    {
      "epoch": 0.774606425188788,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002548612093599664,
      "loss": 1.0942,
      "step": 6052
    },
    {
      "epoch": 0.7747344169973122,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002545847944142604,
      "loss": 1.1793,
      "step": 6053
    },
    {
      "epoch": 0.7748624088058365,
      "grad_norm": 0.59375,
      "learning_rate": 0.00025430850758025183,
      "loss": 1.0242,
      "step": 6054
    },
    {
      "epoch": 0.7749904006143606,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002540323489054248,
      "loss": 1.6064,
      "step": 6055
    },
    {
      "epoch": 0.7751183924228849,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002537563184372421,
      "loss": 0.983,
      "step": 6056
    },
    {
      "epoch": 0.7752463842314092,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00025348041622314335,
      "loss": 1.5031,
      "step": 6057
    },
    {
      "epoch": 0.7753743760399334,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002532046423105472,
      "loss": 1.8124,
      "step": 6058
    },
    {
      "epoch": 0.7755023678484577,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002529289967468499,
      "loss": 1.6467,
      "step": 6059
    },
    {
      "epoch": 0.7756303596569819,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00025265347957942574,
      "loss": 1.4106,
      "step": 6060
    },
    {
      "epoch": 0.7757583514655062,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00025237809085562645,
      "loss": 1.2299,
      "step": 6061
    },
    {
      "epoch": 0.7758863432740305,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002521028306227822,
      "loss": 1.4466,
      "step": 6062
    },
    {
      "epoch": 0.7760143350825547,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002518276989282009,
      "loss": 1.2504,
      "step": 6063
    },
    {
      "epoch": 0.776142326891079,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002515526958191685,
      "loss": 0.9863,
      "step": 6064
    },
    {
      "epoch": 0.7762703186996033,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002512778213429486,
      "loss": 1.4729,
      "step": 6065
    },
    {
      "epoch": 0.7763983105081275,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00025100307554678304,
      "loss": 1.2179,
      "step": 6066
    },
    {
      "epoch": 0.7765263023166518,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00025072845847789126,
      "loss": 1.6835,
      "step": 6067
    },
    {
      "epoch": 0.776654294125176,
      "grad_norm": 0.625,
      "learning_rate": 0.0002504539701834706,
      "loss": 1.0867,
      "step": 6068
    },
    {
      "epoch": 0.7767822859337002,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002501796107106966,
      "loss": 1.3428,
      "step": 6069
    },
    {
      "epoch": 0.7769102777422245,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00024990538010672206,
      "loss": 0.9998,
      "step": 6070
    },
    {
      "epoch": 0.7770382695507487,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002496312784186778,
      "loss": 0.9903,
      "step": 6071
    },
    {
      "epoch": 0.777166261359273,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0002493573056936731,
      "loss": 1.2281,
      "step": 6072
    },
    {
      "epoch": 0.7772942531677972,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002490834619787943,
      "loss": 1.2963,
      "step": 6073
    },
    {
      "epoch": 0.7774222449763215,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00024880974732110627,
      "loss": 1.725,
      "step": 6074
    },
    {
      "epoch": 0.7775502367848458,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002485361617676506,
      "loss": 1.2637,
      "step": 6075
    },
    {
      "epoch": 0.77767822859337,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00024826270536544803,
      "loss": 1.3388,
      "step": 6076
    },
    {
      "epoch": 0.7778062204018943,
      "grad_norm": 0.5625,
      "learning_rate": 0.00024798937816149626,
      "loss": 1.0382,
      "step": 6077
    },
    {
      "epoch": 0.7779342122104186,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002477161802027714,
      "loss": 1.4794,
      "step": 6078
    },
    {
      "epoch": 0.7780622040189428,
      "grad_norm": 0.671875,
      "learning_rate": 0.00024744311153622635,
      "loss": 1.851,
      "step": 6079
    },
    {
      "epoch": 0.7781901958274671,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002471701722087926,
      "loss": 2.0376,
      "step": 6080
    },
    {
      "epoch": 0.7783181876359913,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002468973622673791,
      "loss": 1.0485,
      "step": 6081
    },
    {
      "epoch": 0.7784461794445156,
      "grad_norm": 0.671875,
      "learning_rate": 0.00024662468175887296,
      "loss": 1.2498,
      "step": 6082
    },
    {
      "epoch": 0.7785741712530398,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002463521307301388,
      "loss": 0.8574,
      "step": 6083
    },
    {
      "epoch": 0.778702163061564,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002460797092280186,
      "loss": 1.1963,
      "step": 6084
    },
    {
      "epoch": 0.7788301548700883,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002458074172993324,
      "loss": 1.1001,
      "step": 6085
    },
    {
      "epoch": 0.7789581466786125,
      "grad_norm": 0.6875,
      "learning_rate": 0.00024553525499087814,
      "loss": 1.6738,
      "step": 6086
    },
    {
      "epoch": 0.7790861384871368,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0002452632223494313,
      "loss": 1.3709,
      "step": 6087
    },
    {
      "epoch": 0.7792141302956611,
      "grad_norm": 0.75,
      "learning_rate": 0.00024499131942174525,
      "loss": 1.2421,
      "step": 6088
    },
    {
      "epoch": 0.7793421221041853,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002447195462545506,
      "loss": 1.6077,
      "step": 6089
    },
    {
      "epoch": 0.7794701139127096,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00024444790289455633,
      "loss": 1.0279,
      "step": 6090
    },
    {
      "epoch": 0.7795981057212339,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00024417638938844876,
      "loss": 1.0816,
      "step": 6091
    },
    {
      "epoch": 0.7797260975297581,
      "grad_norm": 0.609375,
      "learning_rate": 0.00024390500578289154,
      "loss": 1.0834,
      "step": 6092
    },
    {
      "epoch": 0.7798540893382824,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00024363375212452655,
      "loss": 1.3494,
      "step": 6093
    },
    {
      "epoch": 0.7799820811468066,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00024336262845997303,
      "loss": 0.7065,
      "step": 6094
    },
    {
      "epoch": 0.7801100729553309,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00024309163483582808,
      "loss": 0.8325,
      "step": 6095
    },
    {
      "epoch": 0.7802380647638552,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00024282077129866665,
      "loss": 1.3516,
      "step": 6096
    },
    {
      "epoch": 0.7803660565723793,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002425500378950405,
      "loss": 1.2968,
      "step": 6097
    },
    {
      "epoch": 0.7804940483809036,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002422794346714794,
      "loss": 1.1925,
      "step": 6098
    },
    {
      "epoch": 0.7806220401894278,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002420089616744916,
      "loss": 1.0823,
      "step": 6099
    },
    {
      "epoch": 0.7807500319979521,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00024173861895056227,
      "loss": 1.5705,
      "step": 6100
    },
    {
      "epoch": 0.7808780238064764,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00024146840654615354,
      "loss": 1.1452,
      "step": 6101
    },
    {
      "epoch": 0.7810060156150006,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002411983245077062,
      "loss": 1.369,
      "step": 6102
    },
    {
      "epoch": 0.7811340074235249,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00024092837288163804,
      "loss": 0.8354,
      "step": 6103
    },
    {
      "epoch": 0.7812619992320492,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00024065855171434493,
      "loss": 1.6902,
      "step": 6104
    },
    {
      "epoch": 0.7813899910405734,
      "grad_norm": 0.546875,
      "learning_rate": 0.00024038886105220004,
      "loss": 0.7931,
      "step": 6105
    },
    {
      "epoch": 0.7815179828490977,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00024011930094155353,
      "loss": 1.6558,
      "step": 6106
    },
    {
      "epoch": 0.7816459746576219,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00023984987142873415,
      "loss": 1.0707,
      "step": 6107
    },
    {
      "epoch": 0.7817739664661462,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002395805725600476,
      "loss": 1.4335,
      "step": 6108
    },
    {
      "epoch": 0.7819019582746705,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00023931140438177722,
      "loss": 1.21,
      "step": 6109
    },
    {
      "epoch": 0.7820299500831946,
      "grad_norm": 0.81640625,
      "learning_rate": 0.000239042366940184,
      "loss": 1.3239,
      "step": 6110
    },
    {
      "epoch": 0.7821579418917189,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00023877346028150636,
      "loss": 1.6738,
      "step": 6111
    },
    {
      "epoch": 0.7822859337002431,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002385046844519604,
      "loss": 1.2386,
      "step": 6112
    },
    {
      "epoch": 0.7824139255087674,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00023823603949773943,
      "loss": 0.9186,
      "step": 6113
    },
    {
      "epoch": 0.7825419173172917,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00023796752546501477,
      "loss": 0.7933,
      "step": 6114
    },
    {
      "epoch": 0.7826699091258159,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00023769914239993428,
      "loss": 1.1618,
      "step": 6115
    },
    {
      "epoch": 0.7827979009343402,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00023743089034862443,
      "loss": 1.4325,
      "step": 6116
    },
    {
      "epoch": 0.7829258927428645,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00023716276935718862,
      "loss": 1.4488,
      "step": 6117
    },
    {
      "epoch": 0.7830538845513887,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0002368947794717079,
      "loss": 1.5559,
      "step": 6118
    },
    {
      "epoch": 0.783181876359913,
      "grad_norm": 0.734375,
      "learning_rate": 0.00023662692073824032,
      "loss": 1.7343,
      "step": 6119
    },
    {
      "epoch": 0.7833098681684372,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002363591932028216,
      "loss": 1.2624,
      "step": 6120
    },
    {
      "epoch": 0.7834378599769615,
      "grad_norm": 0.765625,
      "learning_rate": 0.00023609159691146574,
      "loss": 1.1486,
      "step": 6121
    },
    {
      "epoch": 0.7835658517854858,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002358241319101635,
      "loss": 1.0536,
      "step": 6122
    },
    {
      "epoch": 0.78369384359401,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0002355567982448825,
      "loss": 1.7611,
      "step": 6123
    },
    {
      "epoch": 0.7838218354025343,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00023528959596156852,
      "loss": 1.7737,
      "step": 6124
    },
    {
      "epoch": 0.7839498272110585,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00023502252510614486,
      "loss": 0.9668,
      "step": 6125
    },
    {
      "epoch": 0.7840778190195827,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002347555857245117,
      "loss": 1.5643,
      "step": 6126
    },
    {
      "epoch": 0.784205810828107,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002344887778625474,
      "loss": 1.1915,
      "step": 6127
    },
    {
      "epoch": 0.7843338026366312,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00023422210156610657,
      "loss": 1.3524,
      "step": 6128
    },
    {
      "epoch": 0.7844617944451555,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0002339555568810221,
      "loss": 1.1542,
      "step": 6129
    },
    {
      "epoch": 0.7845897862536798,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00023368914385310413,
      "loss": 1.1752,
      "step": 6130
    },
    {
      "epoch": 0.784717778062204,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00023342286252813992,
      "loss": 1.2485,
      "step": 6131
    },
    {
      "epoch": 0.7848457698707283,
      "grad_norm": 0.546875,
      "learning_rate": 0.00023315671295189434,
      "loss": 1.07,
      "step": 6132
    },
    {
      "epoch": 0.7849737616792525,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00023289069517010952,
      "loss": 1.169,
      "step": 6133
    },
    {
      "epoch": 0.7851017534877768,
      "grad_norm": 0.75,
      "learning_rate": 0.00023262480922850483,
      "loss": 1.2255,
      "step": 6134
    },
    {
      "epoch": 0.7852297452963011,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00023235905517277722,
      "loss": 0.8516,
      "step": 6135
    },
    {
      "epoch": 0.7853577371048253,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00023209343304860108,
      "loss": 1.0404,
      "step": 6136
    },
    {
      "epoch": 0.7854857289133496,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00023182794290162723,
      "loss": 2.2663,
      "step": 6137
    },
    {
      "epoch": 0.7856137207218739,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002315625847774848,
      "loss": 1.4601,
      "step": 6138
    },
    {
      "epoch": 0.785741712530398,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0002312973587217798,
      "loss": 1.6266,
      "step": 6139
    },
    {
      "epoch": 0.7858697043389223,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002310322647800962,
      "loss": 1.3143,
      "step": 6140
    },
    {
      "epoch": 0.7859976961474465,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00023076730299799377,
      "loss": 1.3663,
      "step": 6141
    },
    {
      "epoch": 0.7861256879559708,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00023050247342101094,
      "loss": 0.7454,
      "step": 6142
    },
    {
      "epoch": 0.7862536797644951,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002302377760946628,
      "loss": 1.3628,
      "step": 6143
    },
    {
      "epoch": 0.7863816715730193,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00022997321106444247,
      "loss": 1.6593,
      "step": 6144
    },
    {
      "epoch": 0.7865096633815436,
      "grad_norm": 0.671875,
      "learning_rate": 0.00022970877837581917,
      "loss": 1.2821,
      "step": 6145
    },
    {
      "epoch": 0.7866376551900678,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00022944447807423995,
      "loss": 1.0893,
      "step": 6146
    },
    {
      "epoch": 0.7867656469985921,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0002291803102051294,
      "loss": 1.6324,
      "step": 6147
    },
    {
      "epoch": 0.7868936388071164,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00022891627481388878,
      "loss": 1.1119,
      "step": 6148
    },
    {
      "epoch": 0.7870216306156406,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00022865237194589737,
      "loss": 2.2316,
      "step": 6149
    },
    {
      "epoch": 0.7871496224241649,
      "grad_norm": 0.5625,
      "learning_rate": 0.00022838860164651044,
      "loss": 0.9043,
      "step": 6150
    },
    {
      "epoch": 0.7872776142326892,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00022812496396106154,
      "loss": 1.317,
      "step": 6151
    },
    {
      "epoch": 0.7874056060412133,
      "grad_norm": 0.6015625,
      "learning_rate": 0.000227861458934861,
      "loss": 1.0483,
      "step": 6152
    },
    {
      "epoch": 0.7875335978497376,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002275980866131966,
      "loss": 0.7755,
      "step": 6153
    },
    {
      "epoch": 0.7876615896582618,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002273348470413329,
      "loss": 1.1493,
      "step": 6154
    },
    {
      "epoch": 0.7877895814667861,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00022707174026451217,
      "loss": 1.2476,
      "step": 6155
    },
    {
      "epoch": 0.7879175732753104,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00022680876632795343,
      "loss": 1.2417,
      "step": 6156
    },
    {
      "epoch": 0.7880455650838346,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00022654592527685303,
      "loss": 1.2617,
      "step": 6157
    },
    {
      "epoch": 0.7881735568923589,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00022628321715638478,
      "loss": 0.8614,
      "step": 6158
    },
    {
      "epoch": 0.7883015487008831,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002260206420116987,
      "loss": 1.1341,
      "step": 6159
    },
    {
      "epoch": 0.7884295405094074,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002257581998879229,
      "loss": 1.0782,
      "step": 6160
    },
    {
      "epoch": 0.7885575323179317,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0002254958908301624,
      "loss": 1.5362,
      "step": 6161
    },
    {
      "epoch": 0.7886855241264559,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00022523371488349954,
      "loss": 1.0229,
      "step": 6162
    },
    {
      "epoch": 0.7888135159349802,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00022497167209299275,
      "loss": 1.1013,
      "step": 6163
    },
    {
      "epoch": 0.7889415077435045,
      "grad_norm": 0.6875,
      "learning_rate": 0.00022470976250367882,
      "loss": 1.3694,
      "step": 6164
    },
    {
      "epoch": 0.7890694995520287,
      "grad_norm": 0.671875,
      "learning_rate": 0.00022444798616057116,
      "loss": 1.712,
      "step": 6165
    },
    {
      "epoch": 0.789197491360553,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0002241863431086598,
      "loss": 1.6403,
      "step": 6166
    },
    {
      "epoch": 0.7893254831690771,
      "grad_norm": 0.734375,
      "learning_rate": 0.00022392483339291348,
      "loss": 1.2938,
      "step": 6167
    },
    {
      "epoch": 0.7894534749776014,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00022366345705827572,
      "loss": 1.2758,
      "step": 6168
    },
    {
      "epoch": 0.7895814667861257,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00022340221414966888,
      "loss": 1.361,
      "step": 6169
    },
    {
      "epoch": 0.7897094585946499,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00022314110471199168,
      "loss": 1.152,
      "step": 6170
    },
    {
      "epoch": 0.7898374504031742,
      "grad_norm": 0.625,
      "learning_rate": 0.0002228801287901202,
      "loss": 1.3651,
      "step": 6171
    },
    {
      "epoch": 0.7899654422116984,
      "grad_norm": 0.71875,
      "learning_rate": 0.00022261928642890684,
      "loss": 1.7125,
      "step": 6172
    },
    {
      "epoch": 0.7900934340202227,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00022235857767318202,
      "loss": 1.3596,
      "step": 6173
    },
    {
      "epoch": 0.790221425828747,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002220980025677526,
      "loss": 0.8571,
      "step": 6174
    },
    {
      "epoch": 0.7903494176372712,
      "grad_norm": 0.859375,
      "learning_rate": 0.00022183756115740273,
      "loss": 2.0979,
      "step": 6175
    },
    {
      "epoch": 0.7904774094457955,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002215772534868934,
      "loss": 1.4947,
      "step": 6176
    },
    {
      "epoch": 0.7906054012543198,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00022131707960096293,
      "loss": 1.573,
      "step": 6177
    },
    {
      "epoch": 0.790733393062844,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002210570395443261,
      "loss": 1.5354,
      "step": 6178
    },
    {
      "epoch": 0.7908613848713683,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00022079713336167528,
      "loss": 1.2702,
      "step": 6179
    },
    {
      "epoch": 0.7909893766798924,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00022053736109767976,
      "loss": 1.5319,
      "step": 6180
    },
    {
      "epoch": 0.7911173684884167,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002202777227969851,
      "loss": 1.2561,
      "step": 6181
    },
    {
      "epoch": 0.791245360296941,
      "grad_norm": 0.59375,
      "learning_rate": 0.00022001821850421444,
      "loss": 0.7871,
      "step": 6182
    },
    {
      "epoch": 0.7913733521054652,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002197588482639681,
      "loss": 0.8407,
      "step": 6183
    },
    {
      "epoch": 0.7915013439139895,
      "grad_norm": 0.609375,
      "learning_rate": 0.00021949961212082303,
      "loss": 1.2543,
      "step": 6184
    },
    {
      "epoch": 0.7916293357225137,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002192405101193329,
      "loss": 0.9715,
      "step": 6185
    },
    {
      "epoch": 0.791757327531038,
      "grad_norm": 0.59375,
      "learning_rate": 0.00021898154230402868,
      "loss": 1.1489,
      "step": 6186
    },
    {
      "epoch": 0.7918853193395623,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00021872270871941825,
      "loss": 1.1173,
      "step": 6187
    },
    {
      "epoch": 0.7920133111480865,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002184640094099861,
      "loss": 1.9079,
      "step": 6188
    },
    {
      "epoch": 0.7921413029566108,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00021820544442019475,
      "loss": 1.3149,
      "step": 6189
    },
    {
      "epoch": 0.7922692947651351,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002179470137944819,
      "loss": 1.0584,
      "step": 6190
    },
    {
      "epoch": 0.7923972865736593,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00021768871757726327,
      "loss": 1.063,
      "step": 6191
    },
    {
      "epoch": 0.7925252783821836,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00021743055581293147,
      "loss": 1.6331,
      "step": 6192
    },
    {
      "epoch": 0.7926532701907077,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00021717252854585589,
      "loss": 1.3583,
      "step": 6193
    },
    {
      "epoch": 0.792781261999232,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00021691463582038217,
      "loss": 1.4645,
      "step": 6194
    },
    {
      "epoch": 0.7929092538077563,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00021665687768083363,
      "loss": 1.1755,
      "step": 6195
    },
    {
      "epoch": 0.7930372456162805,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002163992541715102,
      "loss": 1.961,
      "step": 6196
    },
    {
      "epoch": 0.7931652374248048,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00021614176533668873,
      "loss": 0.839,
      "step": 6197
    },
    {
      "epoch": 0.7932932292333291,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00021588441122062275,
      "loss": 1.1729,
      "step": 6198
    },
    {
      "epoch": 0.7934212210418533,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00021562719186754288,
      "loss": 1.2731,
      "step": 6199
    },
    {
      "epoch": 0.7935492128503776,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002153701073216563,
      "loss": 1.1369,
      "step": 6200
    },
    {
      "epoch": 0.7936772046589018,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002151131576271471,
      "loss": 0.8436,
      "step": 6201
    },
    {
      "epoch": 0.7938051964674261,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002148563428281768,
      "loss": 1.3356,
      "step": 6202
    },
    {
      "epoch": 0.7939331882759504,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00021459966296888234,
      "loss": 0.9569,
      "step": 6203
    },
    {
      "epoch": 0.7940611800844746,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002143431180933788,
      "loss": 1.0976,
      "step": 6204
    },
    {
      "epoch": 0.7941891718929989,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002140867082457575,
      "loss": 1.3126,
      "step": 6205
    },
    {
      "epoch": 0.794317163701523,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002138304334700869,
      "loss": 0.7462,
      "step": 6206
    },
    {
      "epoch": 0.7944451555100474,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00021357429381041148,
      "loss": 1.1364,
      "step": 6207
    },
    {
      "epoch": 0.7945731473185716,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00021331828931075324,
      "loss": 1.0402,
      "step": 6208
    },
    {
      "epoch": 0.7947011391270958,
      "grad_norm": 0.53125,
      "learning_rate": 0.00021306242001511066,
      "loss": 0.9949,
      "step": 6209
    },
    {
      "epoch": 0.7948291309356201,
      "grad_norm": 0.734375,
      "learning_rate": 0.0002128066859674591,
      "loss": 1.4255,
      "step": 6210
    },
    {
      "epoch": 0.7949571227441444,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00021255108721175066,
      "loss": 0.9923,
      "step": 6211
    },
    {
      "epoch": 0.7950851145526686,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00021229562379191392,
      "loss": 0.9512,
      "step": 6212
    },
    {
      "epoch": 0.7952131063611929,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00021204029575185458,
      "loss": 1.2956,
      "step": 6213
    },
    {
      "epoch": 0.7953410981697171,
      "grad_norm": 0.59375,
      "learning_rate": 0.00021178510313545485,
      "loss": 0.8675,
      "step": 6214
    },
    {
      "epoch": 0.7954690899782414,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00021153004598657412,
      "loss": 1.4963,
      "step": 6215
    },
    {
      "epoch": 0.7955970817867657,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00021127512434904728,
      "loss": 0.7326,
      "step": 6216
    },
    {
      "epoch": 0.7957250735952899,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0002110203382666873,
      "loss": 0.7626,
      "step": 6217
    },
    {
      "epoch": 0.7958530654038142,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002107656877832832,
      "loss": 1.0359,
      "step": 6218
    },
    {
      "epoch": 0.7959810572123384,
      "grad_norm": 0.921875,
      "learning_rate": 0.0002105111729426007,
      "loss": 1.3033,
      "step": 6219
    },
    {
      "epoch": 0.7961090490208627,
      "grad_norm": 0.546875,
      "learning_rate": 0.00021025679378838246,
      "loss": 1.1477,
      "step": 6220
    },
    {
      "epoch": 0.796237040829387,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00021000255036434767,
      "loss": 1.701,
      "step": 6221
    },
    {
      "epoch": 0.7963650326379111,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00020974844271419202,
      "loss": 1.0098,
      "step": 6222
    },
    {
      "epoch": 0.7964930244464354,
      "grad_norm": 0.578125,
      "learning_rate": 0.00020949447088158813,
      "loss": 1.1743,
      "step": 6223
    },
    {
      "epoch": 0.7966210162549597,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002092406349101855,
      "loss": 0.9741,
      "step": 6224
    },
    {
      "epoch": 0.7967490080634839,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00020898693484360932,
      "loss": 1.5345,
      "step": 6225
    },
    {
      "epoch": 0.7968769998720082,
      "grad_norm": 0.734375,
      "learning_rate": 0.0002087333707254625,
      "loss": 1.1976,
      "step": 6226
    },
    {
      "epoch": 0.7970049916805324,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002084799425993239,
      "loss": 1.3495,
      "step": 6227
    },
    {
      "epoch": 0.7971329834890567,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00020822665050874967,
      "loss": 0.9547,
      "step": 6228
    },
    {
      "epoch": 0.797260975297581,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00020797349449727166,
      "loss": 1.5635,
      "step": 6229
    },
    {
      "epoch": 0.7973889671061052,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00020772047460839905,
      "loss": 1.361,
      "step": 6230
    },
    {
      "epoch": 0.7975169589146295,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002074675908856174,
      "loss": 1.4314,
      "step": 6231
    },
    {
      "epoch": 0.7976449507231537,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00020721484337238893,
      "loss": 1.4621,
      "step": 6232
    },
    {
      "epoch": 0.797772942531678,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00020696223211215238,
      "loss": 1.4921,
      "step": 6233
    },
    {
      "epoch": 0.7979009343402023,
      "grad_norm": 0.5625,
      "learning_rate": 0.00020670975714832309,
      "loss": 0.6698,
      "step": 6234
    },
    {
      "epoch": 0.7980289261487264,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002064574185242929,
      "loss": 1.2611,
      "step": 6235
    },
    {
      "epoch": 0.7981569179572507,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00020620521628343048,
      "loss": 0.9187,
      "step": 6236
    },
    {
      "epoch": 0.798284909765775,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00020595315046908092,
      "loss": 0.9244,
      "step": 6237
    },
    {
      "epoch": 0.7984129015742992,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00020570122112456546,
      "loss": 1.201,
      "step": 6238
    },
    {
      "epoch": 0.7985408933828235,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00020544942829318246,
      "loss": 1.0484,
      "step": 6239
    },
    {
      "epoch": 0.7986688851913477,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00020519777201820667,
      "loss": 1.8598,
      "step": 6240
    },
    {
      "epoch": 0.798796876999872,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0002049462523428891,
      "loss": 1.3357,
      "step": 6241
    },
    {
      "epoch": 0.7989248688083963,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00020469486931045766,
      "loss": 1.7176,
      "step": 6242
    },
    {
      "epoch": 0.7990528606169205,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00020444362296411666,
      "loss": 1.4896,
      "step": 6243
    },
    {
      "epoch": 0.7991808524254448,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002041925133470467,
      "loss": 1.2462,
      "step": 6244
    },
    {
      "epoch": 0.799308844233969,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00020394154050240509,
      "loss": 1.3411,
      "step": 6245
    },
    {
      "epoch": 0.7994368360424933,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00020369070447332583,
      "loss": 1.0707,
      "step": 6246
    },
    {
      "epoch": 0.7995648278510176,
      "grad_norm": 0.65625,
      "learning_rate": 0.00020344000530291874,
      "loss": 1.0125,
      "step": 6247
    },
    {
      "epoch": 0.7996928196595418,
      "grad_norm": 0.75,
      "learning_rate": 0.00020318944303427067,
      "loss": 1.7938,
      "step": 6248
    },
    {
      "epoch": 0.799820811468066,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00020293901771044486,
      "loss": 1.9149,
      "step": 6249
    },
    {
      "epoch": 0.7999488032765903,
      "grad_norm": 0.765625,
      "learning_rate": 0.00020268872937448114,
      "loss": 1.3927,
      "step": 6250
    },
    {
      "epoch": 0.8000767950851145,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002024385780693956,
      "loss": 1.2099,
      "step": 6251
    },
    {
      "epoch": 0.8002047868936388,
      "grad_norm": 0.796875,
      "learning_rate": 0.00020218856383818042,
      "loss": 0.8402,
      "step": 6252
    },
    {
      "epoch": 0.800332778702163,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00020193868672380478,
      "loss": 1.235,
      "step": 6253
    },
    {
      "epoch": 0.8004607705106873,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00020168894676921424,
      "loss": 0.7111,
      "step": 6254
    },
    {
      "epoch": 0.8005887623192116,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00020143934401733055,
      "loss": 1.1244,
      "step": 6255
    },
    {
      "epoch": 0.8007167541277358,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00020118987851105207,
      "loss": 1.8556,
      "step": 6256
    },
    {
      "epoch": 0.8008447459362601,
      "grad_norm": 0.75,
      "learning_rate": 0.00020094055029325352,
      "loss": 1.4815,
      "step": 6257
    },
    {
      "epoch": 0.8009727377447843,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00020069135940678585,
      "loss": 1.5262,
      "step": 6258
    },
    {
      "epoch": 0.8011007295533086,
      "grad_norm": 0.73828125,
      "learning_rate": 0.000200442305894477,
      "loss": 1.5229,
      "step": 6259
    },
    {
      "epoch": 0.8012287213618329,
      "grad_norm": 1.3125,
      "learning_rate": 0.00020019338979913027,
      "loss": 1.1487,
      "step": 6260
    },
    {
      "epoch": 0.8013567131703571,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019994461116352603,
      "loss": 1.1138,
      "step": 6261
    },
    {
      "epoch": 0.8014847049788814,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019969597003042117,
      "loss": 1.1266,
      "step": 6262
    },
    {
      "epoch": 0.8016126967874057,
      "grad_norm": 0.625,
      "learning_rate": 0.00019944746644254862,
      "loss": 1.4253,
      "step": 6263
    },
    {
      "epoch": 0.8017406885959298,
      "grad_norm": 0.78515625,
      "learning_rate": 0.000199199100442618,
      "loss": 1.4457,
      "step": 6264
    },
    {
      "epoch": 0.8018686804044541,
      "grad_norm": 0.671875,
      "learning_rate": 0.00019895087207331418,
      "loss": 1.1049,
      "step": 6265
    },
    {
      "epoch": 0.8019966722129783,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00019870278137730014,
      "loss": 1.9353,
      "step": 6266
    },
    {
      "epoch": 0.8021246640215026,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00019845482839721395,
      "loss": 1.601,
      "step": 6267
    },
    {
      "epoch": 0.8022526558300269,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001982070131756707,
      "loss": 1.221,
      "step": 6268
    },
    {
      "epoch": 0.8023806476385511,
      "grad_norm": 0.578125,
      "learning_rate": 0.00019795933575526072,
      "loss": 1.02,
      "step": 6269
    },
    {
      "epoch": 0.8025086394470754,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019771179617855195,
      "loss": 0.8447,
      "step": 6270
    },
    {
      "epoch": 0.8026366312555996,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019746439448808784,
      "loss": 1.0898,
      "step": 6271
    },
    {
      "epoch": 0.8027646230641239,
      "grad_norm": 0.703125,
      "learning_rate": 0.00019721713072638858,
      "loss": 1.1416,
      "step": 6272
    },
    {
      "epoch": 0.8028926148726482,
      "grad_norm": 0.765625,
      "learning_rate": 0.00019697000493595053,
      "loss": 1.4194,
      "step": 6273
    },
    {
      "epoch": 0.8030206066811724,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001967230171592459,
      "loss": 1.3946,
      "step": 6274
    },
    {
      "epoch": 0.8031485984896967,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00019647616743872375,
      "loss": 1.312,
      "step": 6275
    },
    {
      "epoch": 0.803276590298221,
      "grad_norm": 0.609375,
      "learning_rate": 0.00019622945581680918,
      "loss": 1.0904,
      "step": 6276
    },
    {
      "epoch": 0.8034045821067451,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00019598288233590368,
      "loss": 1.0389,
      "step": 6277
    },
    {
      "epoch": 0.8035325739152694,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00019573644703838467,
      "loss": 1.8005,
      "step": 6278
    },
    {
      "epoch": 0.8036605657237936,
      "grad_norm": 0.6875,
      "learning_rate": 0.00019549014996660642,
      "loss": 1.5391,
      "step": 6279
    },
    {
      "epoch": 0.8037885575323179,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001952439911628987,
      "loss": 1.8258,
      "step": 6280
    },
    {
      "epoch": 0.8039165493408422,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00019499797066956814,
      "loss": 1.3678,
      "step": 6281
    },
    {
      "epoch": 0.8040445411493664,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00019475208852889759,
      "loss": 1.1926,
      "step": 6282
    },
    {
      "epoch": 0.8041725329578907,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00019450634478314523,
      "loss": 1.1444,
      "step": 6283
    },
    {
      "epoch": 0.804300524766415,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001942607394745466,
      "loss": 0.9583,
      "step": 6284
    },
    {
      "epoch": 0.8044285165749392,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001940152726453127,
      "loss": 1.5784,
      "step": 6285
    },
    {
      "epoch": 0.8045565083834635,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001937699443376314,
      "loss": 1.0851,
      "step": 6286
    },
    {
      "epoch": 0.8046845001919877,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019352475459366548,
      "loss": 1.4298,
      "step": 6287
    },
    {
      "epoch": 0.804812492000512,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00019327970345555566,
      "loss": 1.3014,
      "step": 6288
    },
    {
      "epoch": 0.8049404838090363,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00019303479096541764,
      "loss": 1.2683,
      "step": 6289
    },
    {
      "epoch": 0.8050684756175605,
      "grad_norm": 0.5,
      "learning_rate": 0.00019279001716534383,
      "loss": 0.6699,
      "step": 6290
    },
    {
      "epoch": 0.8051964674260847,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00019254538209740213,
      "loss": 0.7029,
      "step": 6291
    },
    {
      "epoch": 0.8053244592346089,
      "grad_norm": 0.671875,
      "learning_rate": 0.00019230088580363724,
      "loss": 1.1885,
      "step": 6292
    },
    {
      "epoch": 0.8054524510431332,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001920565283260699,
      "loss": 1.7332,
      "step": 6293
    },
    {
      "epoch": 0.8055804428516575,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019181230970669684,
      "loss": 1.7608,
      "step": 6294
    },
    {
      "epoch": 0.8057084346601817,
      "grad_norm": 0.71875,
      "learning_rate": 0.00019156822998749124,
      "loss": 1.644,
      "step": 6295
    },
    {
      "epoch": 0.805836426468706,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00019132428921040168,
      "loss": 1.7265,
      "step": 6296
    },
    {
      "epoch": 0.8059644182772303,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00019108048741735362,
      "loss": 1.5577,
      "step": 6297
    },
    {
      "epoch": 0.8060924100857545,
      "grad_norm": 0.78125,
      "learning_rate": 0.00019083682465024832,
      "loss": 1.5605,
      "step": 6298
    },
    {
      "epoch": 0.8062204018942788,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00019059330095096318,
      "loss": 1.7285,
      "step": 6299
    },
    {
      "epoch": 0.806348393702803,
      "grad_norm": 0.640625,
      "learning_rate": 0.00019034991636135158,
      "loss": 1.5219,
      "step": 6300
    },
    {
      "epoch": 0.8064763855113273,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00019010667092324342,
      "loss": 1.1018,
      "step": 6301
    },
    {
      "epoch": 0.8066043773198516,
      "grad_norm": 0.6328125,
      "learning_rate": 0.000189863564678444,
      "loss": 1.2427,
      "step": 6302
    },
    {
      "epoch": 0.8067323691283758,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001896205976687354,
      "loss": 1.3295,
      "step": 6303
    },
    {
      "epoch": 0.8068603609369001,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001893777699358755,
      "loss": 1.2346,
      "step": 6304
    },
    {
      "epoch": 0.8069883527454242,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00018913508152159774,
      "loss": 1.3036,
      "step": 6305
    },
    {
      "epoch": 0.8071163445539485,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00018889253246761229,
      "loss": 1.2253,
      "step": 6306
    },
    {
      "epoch": 0.8072443363624728,
      "grad_norm": 0.59375,
      "learning_rate": 0.00018865012281560524,
      "loss": 1.0526,
      "step": 6307
    },
    {
      "epoch": 0.807372328170997,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001884078526072387,
      "loss": 1.4635,
      "step": 6308
    },
    {
      "epoch": 0.8075003199795213,
      "grad_norm": 0.65625,
      "learning_rate": 0.00018816572188415026,
      "loss": 1.3347,
      "step": 6309
    },
    {
      "epoch": 0.8076283117880456,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001879237306879542,
      "loss": 1.2139,
      "step": 6310
    },
    {
      "epoch": 0.8077563035965698,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00018768187906024103,
      "loss": 1.2885,
      "step": 6311
    },
    {
      "epoch": 0.8078842954050941,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00018744016704257683,
      "loss": 1.2732,
      "step": 6312
    },
    {
      "epoch": 0.8080122872136183,
      "grad_norm": 0.75,
      "learning_rate": 0.00018719859467650313,
      "loss": 1.674,
      "step": 6313
    },
    {
      "epoch": 0.8081402790221426,
      "grad_norm": 0.609375,
      "learning_rate": 0.00018695716200353861,
      "loss": 1.1499,
      "step": 6314
    },
    {
      "epoch": 0.8082682708306669,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001867158690651771,
      "loss": 1.6537,
      "step": 6315
    },
    {
      "epoch": 0.8083962626391911,
      "grad_norm": 0.62890625,
      "learning_rate": 0.000186474715902889,
      "loss": 1.3217,
      "step": 6316
    },
    {
      "epoch": 0.8085242544477154,
      "grad_norm": 0.578125,
      "learning_rate": 0.00018623370255812033,
      "loss": 1.1562,
      "step": 6317
    },
    {
      "epoch": 0.8086522462562395,
      "grad_norm": 0.765625,
      "learning_rate": 0.00018599282907229286,
      "loss": 1.5663,
      "step": 6318
    },
    {
      "epoch": 0.8087802380647638,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001857520954868047,
      "loss": 1.3423,
      "step": 6319
    },
    {
      "epoch": 0.8089082298732881,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00018551150184303002,
      "loss": 1.202,
      "step": 6320
    },
    {
      "epoch": 0.8090362216818123,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00018527104818231855,
      "loss": 1.0599,
      "step": 6321
    },
    {
      "epoch": 0.8091642134903366,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00018503073454599617,
      "loss": 0.8826,
      "step": 6322
    },
    {
      "epoch": 0.8092922052988609,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00018479056097536485,
      "loss": 1.3705,
      "step": 6323
    },
    {
      "epoch": 0.8094201971073851,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00018455052751170209,
      "loss": 0.9006,
      "step": 6324
    },
    {
      "epoch": 0.8095481889159094,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001843106341962617,
      "loss": 1.1084,
      "step": 6325
    },
    {
      "epoch": 0.8096761807244336,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00018407088107027348,
      "loss": 1.2032,
      "step": 6326
    },
    {
      "epoch": 0.8098041725329579,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00018383126817494234,
      "loss": 1.4822,
      "step": 6327
    },
    {
      "epoch": 0.8099321643414822,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001835917955514499,
      "loss": 0.9276,
      "step": 6328
    },
    {
      "epoch": 0.8100601561500064,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00018335246324095345,
      "loss": 1.2978,
      "step": 6329
    },
    {
      "epoch": 0.8101881479585307,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00018311327128458631,
      "loss": 1.5541,
      "step": 6330
    },
    {
      "epoch": 0.8103161397670549,
      "grad_norm": 0.5625,
      "learning_rate": 0.00018287421972345696,
      "loss": 1.0109,
      "step": 6331
    },
    {
      "epoch": 0.8104441315755792,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00018263530859865073,
      "loss": 1.334,
      "step": 6332
    },
    {
      "epoch": 0.8105721233841034,
      "grad_norm": 0.5,
      "learning_rate": 0.000182396537951228,
      "loss": 0.8277,
      "step": 6333
    },
    {
      "epoch": 0.8107001151926276,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00018215790782222585,
      "loss": 2.1084,
      "step": 6334
    },
    {
      "epoch": 0.8108281070011519,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001819194182526568,
      "loss": 1.0032,
      "step": 6335
    },
    {
      "epoch": 0.8109560988096762,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001816810692835087,
      "loss": 1.3164,
      "step": 6336
    },
    {
      "epoch": 0.8110840906182004,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001814428609557458,
      "loss": 1.1235,
      "step": 6337
    },
    {
      "epoch": 0.8112120824267247,
      "grad_norm": 0.578125,
      "learning_rate": 0.00018120479331030815,
      "loss": 0.7769,
      "step": 6338
    },
    {
      "epoch": 0.8113400742352489,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00018096686638811177,
      "loss": 0.7047,
      "step": 6339
    },
    {
      "epoch": 0.8114680660437732,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00018072908023004775,
      "loss": 1.4465,
      "step": 6340
    },
    {
      "epoch": 0.8115960578522975,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00018049143487698372,
      "loss": 1.0473,
      "step": 6341
    },
    {
      "epoch": 0.8117240496608217,
      "grad_norm": 0.62109375,
      "learning_rate": 0.000180253930369763,
      "loss": 1.1467,
      "step": 6342
    },
    {
      "epoch": 0.811852041469346,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00018001656674920453,
      "loss": 1.1201,
      "step": 6343
    },
    {
      "epoch": 0.8119800332778702,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00017977934405610308,
      "loss": 1.0409,
      "step": 6344
    },
    {
      "epoch": 0.8121080250863945,
      "grad_norm": 0.609375,
      "learning_rate": 0.00017954226233122929,
      "loss": 1.0965,
      "step": 6345
    },
    {
      "epoch": 0.8122360168949188,
      "grad_norm": 1.890625,
      "learning_rate": 0.0001793053216153294,
      "loss": 1.2649,
      "step": 6346
    },
    {
      "epoch": 0.8123640087034429,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001790685219491256,
      "loss": 1.3704,
      "step": 6347
    },
    {
      "epoch": 0.8124920005119672,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001788318633733159,
      "loss": 1.3152,
      "step": 6348
    },
    {
      "epoch": 0.8126199923204915,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00017859534592857353,
      "loss": 0.7551,
      "step": 6349
    },
    {
      "epoch": 0.8127479841290157,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00017835896965554798,
      "loss": 1.388,
      "step": 6350
    },
    {
      "epoch": 0.81287597593754,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001781227345948645,
      "loss": 1.4418,
      "step": 6351
    },
    {
      "epoch": 0.8130039677460642,
      "grad_norm": 0.78125,
      "learning_rate": 0.00017788664078712414,
      "loss": 1.8132,
      "step": 6352
    },
    {
      "epoch": 0.8131319595545885,
      "grad_norm": 0.640625,
      "learning_rate": 0.00017765068827290276,
      "loss": 0.7896,
      "step": 6353
    },
    {
      "epoch": 0.8132599513631128,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00017741487709275317,
      "loss": 0.9079,
      "step": 6354
    },
    {
      "epoch": 0.813387943171637,
      "grad_norm": 0.71875,
      "learning_rate": 0.00017717920728720283,
      "loss": 1.3667,
      "step": 6355
    },
    {
      "epoch": 0.8135159349801613,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001769436788967561,
      "loss": 1.5344,
      "step": 6356
    },
    {
      "epoch": 0.8136439267886856,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001767082919618922,
      "loss": 1.2828,
      "step": 6357
    },
    {
      "epoch": 0.8137719185972098,
      "grad_norm": 0.640625,
      "learning_rate": 0.00017647304652306596,
      "loss": 1.4403,
      "step": 6358
    },
    {
      "epoch": 0.8138999104057341,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00017623794262070804,
      "loss": 1.3553,
      "step": 6359
    },
    {
      "epoch": 0.8140279022142582,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001760029802952251,
      "loss": 0.7889,
      "step": 6360
    },
    {
      "epoch": 0.8141558940227825,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00017576815958699932,
      "loss": 1.7405,
      "step": 6361
    },
    {
      "epoch": 0.8142838858313068,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00017553348053638807,
      "loss": 0.9437,
      "step": 6362
    },
    {
      "epoch": 0.814411877639831,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00017529894318372497,
      "loss": 1.1158,
      "step": 6363
    },
    {
      "epoch": 0.8145398694483553,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001750645475693191,
      "loss": 1.3908,
      "step": 6364
    },
    {
      "epoch": 0.8146678612568795,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00017483029373345504,
      "loss": 1.1871,
      "step": 6365
    },
    {
      "epoch": 0.8147958530654038,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00017459618171639324,
      "loss": 1.4117,
      "step": 6366
    },
    {
      "epoch": 0.8149238448739281,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00017436221155836952,
      "loss": 1.5437,
      "step": 6367
    },
    {
      "epoch": 0.8150518366824523,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00017412838329959558,
      "loss": 1.2214,
      "step": 6368
    },
    {
      "epoch": 0.8151798284909766,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00017389469698025851,
      "loss": 1.3425,
      "step": 6369
    },
    {
      "epoch": 0.8153078202995009,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0001736611526405214,
      "loss": 0.9628,
      "step": 6370
    },
    {
      "epoch": 0.8154358121080251,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00017342775032052205,
      "loss": 1.3132,
      "step": 6371
    },
    {
      "epoch": 0.8155638039165494,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0001731944900603748,
      "loss": 0.8973,
      "step": 6372
    },
    {
      "epoch": 0.8156917957250736,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00017296137190016915,
      "loss": 1.5198,
      "step": 6373
    },
    {
      "epoch": 0.8158197875335979,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00017272839587997046,
      "loss": 0.9456,
      "step": 6374
    },
    {
      "epoch": 0.8159477793421221,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00017249556203981898,
      "loss": 1.0797,
      "step": 6375
    },
    {
      "epoch": 0.8160757711506463,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001722628704197312,
      "loss": 1.2326,
      "step": 6376
    },
    {
      "epoch": 0.8162037629591706,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00017203032105969895,
      "loss": 1.5767,
      "step": 6377
    },
    {
      "epoch": 0.8163317547676948,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00017179791399968947,
      "loss": 0.6635,
      "step": 6378
    },
    {
      "epoch": 0.8164597465762191,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00017156564927964634,
      "loss": 1.0486,
      "step": 6379
    },
    {
      "epoch": 0.8165877383847434,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00017133352693948724,
      "loss": 1.2695,
      "step": 6380
    },
    {
      "epoch": 0.8167157301932676,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00017110154701910642,
      "loss": 1.2817,
      "step": 6381
    },
    {
      "epoch": 0.8168437220017919,
      "grad_norm": 0.59375,
      "learning_rate": 0.00017086970955837344,
      "loss": 1.0703,
      "step": 6382
    },
    {
      "epoch": 0.8169717138103162,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00017063801459713356,
      "loss": 1.6211,
      "step": 6383
    },
    {
      "epoch": 0.8170997056188404,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00017040646217520683,
      "loss": 1.2753,
      "step": 6384
    },
    {
      "epoch": 0.8172276974273647,
      "grad_norm": 0.84375,
      "learning_rate": 0.00017017505233238962,
      "loss": 1.6021,
      "step": 6385
    },
    {
      "epoch": 0.8173556892358889,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00016994378510845333,
      "loss": 1.1946,
      "step": 6386
    },
    {
      "epoch": 0.8174836810444132,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00016971266054314493,
      "loss": 1.462,
      "step": 6387
    },
    {
      "epoch": 0.8176116728529375,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00016948167867618713,
      "loss": 0.7741,
      "step": 6388
    },
    {
      "epoch": 0.8177396646614616,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00016925083954727772,
      "loss": 1.4196,
      "step": 6389
    },
    {
      "epoch": 0.8178676564699859,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00016902014319609027,
      "loss": 1.2514,
      "step": 6390
    },
    {
      "epoch": 0.8179956482785101,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00016878958966227365,
      "loss": 1.0158,
      "step": 6391
    },
    {
      "epoch": 0.8181236400870344,
      "grad_norm": 0.90625,
      "learning_rate": 0.00016855917898545258,
      "loss": 1.8463,
      "step": 6392
    },
    {
      "epoch": 0.8182516318955587,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0001683289112052262,
      "loss": 0.9649,
      "step": 6393
    },
    {
      "epoch": 0.8183796237040829,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00016809878636117004,
      "loss": 1.2049,
      "step": 6394
    },
    {
      "epoch": 0.8185076155126072,
      "grad_norm": 0.59375,
      "learning_rate": 0.00016786880449283492,
      "loss": 1.022,
      "step": 6395
    },
    {
      "epoch": 0.8186356073211315,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00016763896563974713,
      "loss": 1.3466,
      "step": 6396
    },
    {
      "epoch": 0.8187635991296557,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001674092698414077,
      "loss": 0.9601,
      "step": 6397
    },
    {
      "epoch": 0.81889159093818,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00016717971713729374,
      "loss": 0.9671,
      "step": 6398
    },
    {
      "epoch": 0.8190195827467042,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001669503075668578,
      "loss": 1.6433,
      "step": 6399
    },
    {
      "epoch": 0.8191475745552285,
      "grad_norm": 0.609375,
      "learning_rate": 0.00016672104116952747,
      "loss": 1.2212,
      "step": 6400
    },
    {
      "epoch": 0.8192755663637528,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00016649191798470607,
      "loss": 1.2913,
      "step": 6401
    },
    {
      "epoch": 0.819403558172277,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00016626293805177195,
      "loss": 1.5849,
      "step": 6402
    },
    {
      "epoch": 0.8195315499808012,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00016603410141007914,
      "loss": 1.4943,
      "step": 6403
    },
    {
      "epoch": 0.8196595417893254,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00016580540809895682,
      "loss": 1.0715,
      "step": 6404
    },
    {
      "epoch": 0.8197875335978497,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00016557685815771006,
      "loss": 1.0771,
      "step": 6405
    },
    {
      "epoch": 0.819915525406374,
      "grad_norm": 0.578125,
      "learning_rate": 0.00016534845162561817,
      "loss": 1.0131,
      "step": 6406
    },
    {
      "epoch": 0.8200435172148982,
      "grad_norm": 0.5859375,
      "learning_rate": 0.000165120188541937,
      "loss": 1.1404,
      "step": 6407
    },
    {
      "epoch": 0.8201715090234225,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00016489206894589715,
      "loss": 1.1012,
      "step": 6408
    },
    {
      "epoch": 0.8202995008319468,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001646640928767047,
      "loss": 1.4369,
      "step": 6409
    },
    {
      "epoch": 0.820427492640471,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00016443626037354088,
      "loss": 1.2064,
      "step": 6410
    },
    {
      "epoch": 0.8205554844489953,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00016420857147556267,
      "loss": 1.4007,
      "step": 6411
    },
    {
      "epoch": 0.8206834762575195,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00016398102622190193,
      "loss": 1.169,
      "step": 6412
    },
    {
      "epoch": 0.8208114680660438,
      "grad_norm": 0.67578125,
      "learning_rate": 0.000163753624651666,
      "loss": 1.7751,
      "step": 6413
    },
    {
      "epoch": 0.8209394598745681,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00016352636680393786,
      "loss": 0.9779,
      "step": 6414
    },
    {
      "epoch": 0.8210674516830923,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001632992527177749,
      "loss": 1.4454,
      "step": 6415
    },
    {
      "epoch": 0.8211954434916166,
      "grad_norm": 0.828125,
      "learning_rate": 0.00016307228243221062,
      "loss": 1.2393,
      "step": 6416
    },
    {
      "epoch": 0.8213234353001407,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00016284545598625356,
      "loss": 2.0656,
      "step": 6417
    },
    {
      "epoch": 0.821451427108665,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00016261877341888754,
      "loss": 1.1972,
      "step": 6418
    },
    {
      "epoch": 0.8215794189171893,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00016239223476907183,
      "loss": 1.5271,
      "step": 6419
    },
    {
      "epoch": 0.8217074107257135,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001621658400757403,
      "loss": 1.345,
      "step": 6420
    },
    {
      "epoch": 0.8218354025342378,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00016193958937780285,
      "loss": 1.4861,
      "step": 6421
    },
    {
      "epoch": 0.8219633943427621,
      "grad_norm": 0.75,
      "learning_rate": 0.00016171348271414433,
      "loss": 1.9997,
      "step": 6422
    },
    {
      "epoch": 0.8220913861512863,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00016148752012362465,
      "loss": 1.0531,
      "step": 6423
    },
    {
      "epoch": 0.8222193779598106,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00016126170164507947,
      "loss": 1.3409,
      "step": 6424
    },
    {
      "epoch": 0.8223473697683348,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00016103602731731915,
      "loss": 1.6066,
      "step": 6425
    },
    {
      "epoch": 0.8224753615768591,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00016081049717912944,
      "loss": 1.0628,
      "step": 6426
    },
    {
      "epoch": 0.8226033533853834,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00016058511126927177,
      "loss": 1.2483,
      "step": 6427
    },
    {
      "epoch": 0.8227313451939076,
      "grad_norm": 0.65625,
      "learning_rate": 0.00016035986962648175,
      "loss": 1.3517,
      "step": 6428
    },
    {
      "epoch": 0.8228593370024319,
      "grad_norm": 1.703125,
      "learning_rate": 0.00016013477228947104,
      "loss": 1.4293,
      "step": 6429
    },
    {
      "epoch": 0.8229873288109562,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00015990981929692638,
      "loss": 1.8429,
      "step": 6430
    },
    {
      "epoch": 0.8231153206194803,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001596850106875095,
      "loss": 1.739,
      "step": 6431
    },
    {
      "epoch": 0.8232433124280046,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015946034649985753,
      "loss": 1.0612,
      "step": 6432
    },
    {
      "epoch": 0.8233713042365288,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015923582677258253,
      "loss": 1.122,
      "step": 6433
    },
    {
      "epoch": 0.8234992960450531,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001590114515442719,
      "loss": 1.3149,
      "step": 6434
    },
    {
      "epoch": 0.8236272878535774,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015878722085348805,
      "loss": 1.3431,
      "step": 6435
    },
    {
      "epoch": 0.8237552796621016,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015856313473876915,
      "loss": 1.4255,
      "step": 6436
    },
    {
      "epoch": 0.8238832714706259,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00015833919323862735,
      "loss": 2.2158,
      "step": 6437
    },
    {
      "epoch": 0.8240112632791501,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00015811539639155092,
      "loss": 0.6549,
      "step": 6438
    },
    {
      "epoch": 0.8241392550876744,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00015789174423600305,
      "loss": 0.5742,
      "step": 6439
    },
    {
      "epoch": 0.8242672468961987,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00015766823681042185,
      "loss": 0.8712,
      "step": 6440
    },
    {
      "epoch": 0.8243952387047229,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00015744487415322107,
      "loss": 0.952,
      "step": 6441
    },
    {
      "epoch": 0.8245232305132472,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015722165630278873,
      "loss": 1.8648,
      "step": 6442
    },
    {
      "epoch": 0.8246512223217715,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001569985832974885,
      "loss": 1.5168,
      "step": 6443
    },
    {
      "epoch": 0.8247792141302956,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001567756551756594,
      "loss": 1.1286,
      "step": 6444
    },
    {
      "epoch": 0.8249072059388199,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015655287197561497,
      "loss": 2.0485,
      "step": 6445
    },
    {
      "epoch": 0.8250351977473441,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015633023373564414,
      "loss": 1.6017,
      "step": 6446
    },
    {
      "epoch": 0.8251631895558684,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001561077404940111,
      "loss": 1.0998,
      "step": 6447
    },
    {
      "epoch": 0.8252911813643927,
      "grad_norm": 0.84375,
      "learning_rate": 0.00015588539228895482,
      "loss": 1.6422,
      "step": 6448
    },
    {
      "epoch": 0.8254191731729169,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015566318915868937,
      "loss": 1.9149,
      "step": 6449
    },
    {
      "epoch": 0.8255471649814412,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00015544113114140425,
      "loss": 1.1797,
      "step": 6450
    },
    {
      "epoch": 0.8256751567899654,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001552192182752633,
      "loss": 1.2452,
      "step": 6451
    },
    {
      "epoch": 0.8258031485984897,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001549974505984061,
      "loss": 0.7402,
      "step": 6452
    },
    {
      "epoch": 0.825931140407014,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00015477582814894686,
      "loss": 1.2192,
      "step": 6453
    },
    {
      "epoch": 0.8260591322155382,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001545543509649754,
      "loss": 1.261,
      "step": 6454
    },
    {
      "epoch": 0.8261871240240625,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00015433301908455543,
      "loss": 0.5256,
      "step": 6455
    },
    {
      "epoch": 0.8263151158325868,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015411183254572714,
      "loss": 1.3297,
      "step": 6456
    },
    {
      "epoch": 0.826443107641111,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015389079138650465,
      "loss": 1.7213,
      "step": 6457
    },
    {
      "epoch": 0.8265710994496352,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00015366989564487787,
      "loss": 1.3736,
      "step": 6458
    },
    {
      "epoch": 0.8266990912581594,
      "grad_norm": 2.46875,
      "learning_rate": 0.00015344914535881082,
      "loss": 1.5278,
      "step": 6459
    },
    {
      "epoch": 0.8268270830666837,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00015322854056624313,
      "loss": 1.0237,
      "step": 6460
    },
    {
      "epoch": 0.826955074875208,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00015300808130508947,
      "loss": 1.8613,
      "step": 6461
    },
    {
      "epoch": 0.8270830666837322,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015278776761323932,
      "loss": 1.1849,
      "step": 6462
    },
    {
      "epoch": 0.8272110584922565,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015256759952855738,
      "loss": 1.4194,
      "step": 6463
    },
    {
      "epoch": 0.8273390503007807,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00015234757708888248,
      "loss": 2.0083,
      "step": 6464
    },
    {
      "epoch": 0.827467042109305,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00015212770033202949,
      "loss": 1.2209,
      "step": 6465
    },
    {
      "epoch": 0.8275950339178293,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00015190796929578777,
      "loss": 1.0074,
      "step": 6466
    },
    {
      "epoch": 0.8277230257263535,
      "grad_norm": 0.546875,
      "learning_rate": 0.00015168838401792152,
      "loss": 1.1444,
      "step": 6467
    },
    {
      "epoch": 0.8278510175348778,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001514689445361701,
      "loss": 2.1477,
      "step": 6468
    },
    {
      "epoch": 0.8279790093434021,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015124965088824792,
      "loss": 1.3546,
      "step": 6469
    },
    {
      "epoch": 0.8281070011519263,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00015103050311184385,
      "loss": 1.1014,
      "step": 6470
    },
    {
      "epoch": 0.8282349929604506,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001508115012446223,
      "loss": 0.7142,
      "step": 6471
    },
    {
      "epoch": 0.8283629847689747,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015059264532422224,
      "loss": 1.6988,
      "step": 6472
    },
    {
      "epoch": 0.828490976577499,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00015037393538825728,
      "loss": 1.3576,
      "step": 6473
    },
    {
      "epoch": 0.8286189683860233,
      "grad_norm": 0.546875,
      "learning_rate": 0.00015015537147431657,
      "loss": 0.9949,
      "step": 6474
    },
    {
      "epoch": 0.8287469601945475,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014993695361996374,
      "loss": 1.876,
      "step": 6475
    },
    {
      "epoch": 0.8288749520030718,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00014971868186273784,
      "loss": 1.2106,
      "step": 6476
    },
    {
      "epoch": 0.829002943811596,
      "grad_norm": 0.53125,
      "learning_rate": 0.00014950055624015146,
      "loss": 0.7975,
      "step": 6477
    },
    {
      "epoch": 0.8291309356201203,
      "grad_norm": 0.5625,
      "learning_rate": 0.00014928257678969393,
      "loss": 0.9946,
      "step": 6478
    },
    {
      "epoch": 0.8292589274286446,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00014906474354882827,
      "loss": 0.7791,
      "step": 6479
    },
    {
      "epoch": 0.8293869192371688,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00014884705655499297,
      "loss": 0.8972,
      "step": 6480
    },
    {
      "epoch": 0.8295149110456931,
      "grad_norm": 0.59375,
      "learning_rate": 0.00014862951584560037,
      "loss": 0.9461,
      "step": 6481
    },
    {
      "epoch": 0.8296429028542174,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014841212145803883,
      "loss": 1.4915,
      "step": 6482
    },
    {
      "epoch": 0.8297708946627416,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00014819487342967097,
      "loss": 0.6069,
      "step": 6483
    },
    {
      "epoch": 0.8298988864712659,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014797777179783456,
      "loss": 1.5834,
      "step": 6484
    },
    {
      "epoch": 0.83002687827979,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001477608165998421,
      "loss": 1.7548,
      "step": 6485
    },
    {
      "epoch": 0.8301548700883143,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001475440078729804,
      "loss": 0.9615,
      "step": 6486
    },
    {
      "epoch": 0.8302828618968386,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001473273456545119,
      "loss": 0.8091,
      "step": 6487
    },
    {
      "epoch": 0.8304108537053628,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00014711082998167336,
      "loss": 0.9273,
      "step": 6488
    },
    {
      "epoch": 0.8305388455138871,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001468944608916767,
      "loss": 1.007,
      "step": 6489
    },
    {
      "epoch": 0.8306668373224113,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014667823842170834,
      "loss": 1.6675,
      "step": 6490
    },
    {
      "epoch": 0.8307948291309356,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014646216260892964,
      "loss": 1.2441,
      "step": 6491
    },
    {
      "epoch": 0.8309228209394599,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00014624623349047673,
      "loss": 0.9493,
      "step": 6492
    },
    {
      "epoch": 0.8310508127479841,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00014603045110346059,
      "loss": 1.2221,
      "step": 6493
    },
    {
      "epoch": 0.8311788045565084,
      "grad_norm": 0.63671875,
      "learning_rate": 0.000145814815484967,
      "loss": 1.1692,
      "step": 6494
    },
    {
      "epoch": 0.8313067963650327,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001455993266720562,
      "loss": 1.2411,
      "step": 6495
    },
    {
      "epoch": 0.8314347881735569,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001453839847017635,
      "loss": 1.3437,
      "step": 6496
    },
    {
      "epoch": 0.8315627799820812,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00014516878961109902,
      "loss": 1.4947,
      "step": 6497
    },
    {
      "epoch": 0.8316907717906054,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014495374143704766,
      "loss": 0.9976,
      "step": 6498
    },
    {
      "epoch": 0.8318187635991297,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014473884021656857,
      "loss": 1.7582,
      "step": 6499
    },
    {
      "epoch": 0.831946755407654,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014452408598659595,
      "loss": 1.5204,
      "step": 6500
    },
    {
      "epoch": 0.8320747472161781,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001443094787840392,
      "loss": 1.1701,
      "step": 6501
    },
    {
      "epoch": 0.8322027390247024,
      "grad_norm": 0.63671875,
      "learning_rate": 0.000144095018645782,
      "loss": 1.3076,
      "step": 6502
    },
    {
      "epoch": 0.8323307308332267,
      "grad_norm": 0.5625,
      "learning_rate": 0.00014388070560868282,
      "loss": 0.9604,
      "step": 6503
    },
    {
      "epoch": 0.8324587226417509,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014366653970957456,
      "loss": 1.2016,
      "step": 6504
    },
    {
      "epoch": 0.8325867144502752,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00014345252098526518,
      "loss": 1.0339,
      "step": 6505
    },
    {
      "epoch": 0.8327147062587994,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001432386494725374,
      "loss": 1.6986,
      "step": 6506
    },
    {
      "epoch": 0.8328426980673237,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001430249252081487,
      "loss": 1.1759,
      "step": 6507
    },
    {
      "epoch": 0.832970689875848,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014281134822883058,
      "loss": 1.1509,
      "step": 6508
    },
    {
      "epoch": 0.8330986816843722,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001425979185712899,
      "loss": 0.8378,
      "step": 6509
    },
    {
      "epoch": 0.8332266734928965,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001423846362722081,
      "loss": 1.46,
      "step": 6510
    },
    {
      "epoch": 0.8333546653014207,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014217150136824119,
      "loss": 1.1725,
      "step": 6511
    },
    {
      "epoch": 0.833482657109945,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014195851389601976,
      "loss": 1.4198,
      "step": 6512
    },
    {
      "epoch": 0.8336106489184693,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001417456738921492,
      "loss": 1.0739,
      "step": 6513
    },
    {
      "epoch": 0.8337386407269934,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00014153298139320969,
      "loss": 1.4247,
      "step": 6514
    },
    {
      "epoch": 0.8338666325355177,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00014132043643575565,
      "loss": 1.112,
      "step": 6515
    },
    {
      "epoch": 0.833994624344042,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014110803905631664,
      "loss": 1.1553,
      "step": 6516
    },
    {
      "epoch": 0.8341226161525662,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00014089578929139636,
      "loss": 0.9801,
      "step": 6517
    },
    {
      "epoch": 0.8342506079610905,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014068368717747338,
      "loss": 1.8055,
      "step": 6518
    },
    {
      "epoch": 0.8343785997696147,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014047173275100088,
      "loss": 1.7322,
      "step": 6519
    },
    {
      "epoch": 0.834506591578139,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001402599260484071,
      "loss": 1.3155,
      "step": 6520
    },
    {
      "epoch": 0.8346345833866633,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014004826710609377,
      "loss": 1.4031,
      "step": 6521
    },
    {
      "epoch": 0.8347625751951875,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013983675596043833,
      "loss": 1.4333,
      "step": 6522
    },
    {
      "epoch": 0.8348905670037118,
      "grad_norm": 0.53125,
      "learning_rate": 0.00013962539264779207,
      "loss": 0.8931,
      "step": 6523
    },
    {
      "epoch": 0.835018558812236,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001394141772044817,
      "loss": 1.7468,
      "step": 6524
    },
    {
      "epoch": 0.8351465506207603,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013920310966680793,
      "loss": 1.2484,
      "step": 6525
    },
    {
      "epoch": 0.8352745424292846,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001389921900710458,
      "loss": 1.1519,
      "step": 6526
    },
    {
      "epoch": 0.8354025342378087,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00013878141845344538,
      "loss": 0.8406,
      "step": 6527
    },
    {
      "epoch": 0.835530526046333,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001385707948502314,
      "loss": 1.0931,
      "step": 6528
    },
    {
      "epoch": 0.8356585178548573,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013836031929760295,
      "loss": 1.6377,
      "step": 6529
    },
    {
      "epoch": 0.8357865096633815,
      "grad_norm": 0.546875,
      "learning_rate": 0.00013814999183173326,
      "loss": 0.8193,
      "step": 6530
    },
    {
      "epoch": 0.8359145014719058,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00013793981248877085,
      "loss": 1.1256,
      "step": 6531
    },
    {
      "epoch": 0.83604249328043,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00013772978130483838,
      "loss": 1.6735,
      "step": 6532
    },
    {
      "epoch": 0.8361704850889543,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00013751989831603306,
      "loss": 0.8772,
      "step": 6533
    },
    {
      "epoch": 0.8362984768974786,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013731016355842684,
      "loss": 1.5017,
      "step": 6534
    },
    {
      "epoch": 0.8364264687060028,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013710057706806588,
      "loss": 1.5434,
      "step": 6535
    },
    {
      "epoch": 0.8365544605145271,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00013689113888097116,
      "loss": 1.2617,
      "step": 6536
    },
    {
      "epoch": 0.8366824523230513,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013668184903313786,
      "loss": 1.5469,
      "step": 6537
    },
    {
      "epoch": 0.8368104441315756,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001364727075605362,
      "loss": 1.4573,
      "step": 6538
    },
    {
      "epoch": 0.8369384359400999,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00013626371449911002,
      "loss": 1.0843,
      "step": 6539
    },
    {
      "epoch": 0.837066427748624,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00013605486988477845,
      "loss": 1.2193,
      "step": 6540
    },
    {
      "epoch": 0.8371944195571484,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013584617375343478,
      "loss": 1.7929,
      "step": 6541
    },
    {
      "epoch": 0.8373224113656726,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00013563762614094698,
      "loss": 1.819,
      "step": 6542
    },
    {
      "epoch": 0.8374504031741968,
      "grad_norm": 0.890625,
      "learning_rate": 0.00013542922708315708,
      "loss": 1.8808,
      "step": 6543
    },
    {
      "epoch": 0.8375783949827211,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013522097661588197,
      "loss": 1.7176,
      "step": 6544
    },
    {
      "epoch": 0.8377063867912453,
      "grad_norm": 0.640625,
      "learning_rate": 0.00013501287477491253,
      "loss": 1.2785,
      "step": 6545
    },
    {
      "epoch": 0.8378343785997696,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013480492159601508,
      "loss": 1.4505,
      "step": 6546
    },
    {
      "epoch": 0.8379623704082939,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00013459711711492962,
      "loss": 0.8364,
      "step": 6547
    },
    {
      "epoch": 0.8380903622168181,
      "grad_norm": 0.640625,
      "learning_rate": 0.00013438946136737028,
      "loss": 1.2178,
      "step": 6548
    },
    {
      "epoch": 0.8382183540253424,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00013418195438902637,
      "loss": 1.4079,
      "step": 6549
    },
    {
      "epoch": 0.8383463458338666,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001339745962155613,
      "loss": 1.341,
      "step": 6550
    },
    {
      "epoch": 0.8384743376423909,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00013376738688261315,
      "loss": 1.1028,
      "step": 6551
    },
    {
      "epoch": 0.8386023294509152,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013356032642579375,
      "loss": 1.1919,
      "step": 6552
    },
    {
      "epoch": 0.8387303212594394,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013335341488068998,
      "loss": 1.6525,
      "step": 6553
    },
    {
      "epoch": 0.8388583130679637,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013314665228286303,
      "loss": 1.9218,
      "step": 6554
    },
    {
      "epoch": 0.838986304876488,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013294003866784842,
      "loss": 1.4924,
      "step": 6555
    },
    {
      "epoch": 0.8391142966850121,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013273357407115593,
      "loss": 1.8948,
      "step": 6556
    },
    {
      "epoch": 0.8392422884935364,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00013252725852826996,
      "loss": 0.987,
      "step": 6557
    },
    {
      "epoch": 0.8393702803020606,
      "grad_norm": 0.640625,
      "learning_rate": 0.00013232109207464914,
      "loss": 1.0167,
      "step": 6558
    },
    {
      "epoch": 0.8394982721105849,
      "grad_norm": 0.59375,
      "learning_rate": 0.00013211507474572649,
      "loss": 0.9898,
      "step": 6559
    },
    {
      "epoch": 0.8396262639191092,
      "grad_norm": 0.609375,
      "learning_rate": 0.00013190920657690975,
      "loss": 1.2531,
      "step": 6560
    },
    {
      "epoch": 0.8397542557276334,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001317034876035802,
      "loss": 0.9133,
      "step": 6561
    },
    {
      "epoch": 0.8398822475361577,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001314979178610942,
      "loss": 1.8721,
      "step": 6562
    },
    {
      "epoch": 0.8400102393446819,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001312924973847822,
      "loss": 1.2701,
      "step": 6563
    },
    {
      "epoch": 0.8401382311532062,
      "grad_norm": 0.625,
      "learning_rate": 0.00013108722620994918,
      "loss": 1.514,
      "step": 6564
    },
    {
      "epoch": 0.8402662229617305,
      "grad_norm": 0.546875,
      "learning_rate": 0.00013088210437187431,
      "loss": 0.8356,
      "step": 6565
    },
    {
      "epoch": 0.8403942147702547,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00013067713190581098,
      "loss": 1.4904,
      "step": 6566
    },
    {
      "epoch": 0.840522206578779,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013047230884698701,
      "loss": 1.2747,
      "step": 6567
    },
    {
      "epoch": 0.8406501983873033,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013026763523060436,
      "loss": 1.2733,
      "step": 6568
    },
    {
      "epoch": 0.8407781901958274,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00013006311109184032,
      "loss": 1.0725,
      "step": 6569
    },
    {
      "epoch": 0.8409061820043517,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001298587364658449,
      "loss": 0.9015,
      "step": 6570
    },
    {
      "epoch": 0.8410341738128759,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00012965451138774342,
      "loss": 1.1163,
      "step": 6571
    },
    {
      "epoch": 0.8411621656214002,
      "grad_norm": 0.609375,
      "learning_rate": 0.00012945043589263538,
      "loss": 1.1832,
      "step": 6572
    },
    {
      "epoch": 0.8412901574299245,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00012924651001559463,
      "loss": 0.7039,
      "step": 6573
    },
    {
      "epoch": 0.8414181492384487,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012904273379166854,
      "loss": 1.4701,
      "step": 6574
    },
    {
      "epoch": 0.841546141046973,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00012883910725587988,
      "loss": 1.4424,
      "step": 6575
    },
    {
      "epoch": 0.8416741328554973,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001286356304432249,
      "loss": 1.3802,
      "step": 6576
    },
    {
      "epoch": 0.8418021246640215,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001284323033886745,
      "loss": 1.3473,
      "step": 6577
    },
    {
      "epoch": 0.8419301164725458,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012822912612717386,
      "loss": 1.8812,
      "step": 6578
    },
    {
      "epoch": 0.84205810828107,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012802609869364201,
      "loss": 1.4463,
      "step": 6579
    },
    {
      "epoch": 0.8421861000895943,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00012782322112297274,
      "loss": 1.0496,
      "step": 6580
    },
    {
      "epoch": 0.8423140918981186,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0001276204934500338,
      "loss": 0.8084,
      "step": 6581
    },
    {
      "epoch": 0.8424420837066428,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012741791570966744,
      "loss": 1.505,
      "step": 6582
    },
    {
      "epoch": 0.842570075515167,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012721548793668937,
      "loss": 1.6454,
      "step": 6583
    },
    {
      "epoch": 0.8426980673236912,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001270132101658905,
      "loss": 0.766,
      "step": 6584
    },
    {
      "epoch": 0.8428260591322155,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012681108243203543,
      "loss": 1.6209,
      "step": 6585
    },
    {
      "epoch": 0.8429540509407398,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012660910476986322,
      "loss": 1.3398,
      "step": 6586
    },
    {
      "epoch": 0.843082042749264,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012640727721408696,
      "loss": 1.5558,
      "step": 6587
    },
    {
      "epoch": 0.8432100345577883,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00012620559979939383,
      "loss": 0.7033,
      "step": 6588
    },
    {
      "epoch": 0.8433380263663126,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012600407256044543,
      "loss": 1.4392,
      "step": 6589
    },
    {
      "epoch": 0.8434660181748368,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012580269553187762,
      "loss": 1.3851,
      "step": 6590
    },
    {
      "epoch": 0.8435940099833611,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012560146874830014,
      "loss": 1.2076,
      "step": 6591
    },
    {
      "epoch": 0.8437220017918853,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012540039224429722,
      "loss": 1.2536,
      "step": 6592
    },
    {
      "epoch": 0.8438499936004096,
      "grad_norm": 0.72265625,
      "learning_rate": 0.000125199466054427,
      "loss": 1.4267,
      "step": 6593
    },
    {
      "epoch": 0.8439779854089339,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012499869021322197,
      "loss": 1.8403,
      "step": 6594
    },
    {
      "epoch": 0.8441059772174581,
      "grad_norm": 0.859375,
      "learning_rate": 0.00012479806475518894,
      "loss": 1.6941,
      "step": 6595
    },
    {
      "epoch": 0.8442339690259824,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001245975897148083,
      "loss": 0.9271,
      "step": 6596
    },
    {
      "epoch": 0.8443619608345065,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00012439726512653492,
      "loss": 0.946,
      "step": 6597
    },
    {
      "epoch": 0.8444899526430308,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012419709102479804,
      "loss": 1.0246,
      "step": 6598
    },
    {
      "epoch": 0.8446179444515551,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012399706744400075,
      "loss": 1.5801,
      "step": 6599
    },
    {
      "epoch": 0.8447459362600793,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012379719441852033,
      "loss": 1.1873,
      "step": 6600
    },
    {
      "epoch": 0.8448739280686036,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012359747198270832,
      "loss": 1.2704,
      "step": 6601
    },
    {
      "epoch": 0.8450019198771279,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012339790017089003,
      "loss": 1.8863,
      "step": 6602
    },
    {
      "epoch": 0.8451299116856521,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012319847901736536,
      "loss": 1.8426,
      "step": 6603
    },
    {
      "epoch": 0.8452579034941764,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001229992085564081,
      "loss": 1.5247,
      "step": 6604
    },
    {
      "epoch": 0.8453858953027006,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001228000888222659,
      "loss": 1.286,
      "step": 6605
    },
    {
      "epoch": 0.8455138871112249,
      "grad_norm": 0.609375,
      "learning_rate": 0.00012260111984916067,
      "loss": 1.0678,
      "step": 6606
    },
    {
      "epoch": 0.8456418789197492,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00012240230167128864,
      "loss": 0.9746,
      "step": 6607
    },
    {
      "epoch": 0.8457698707282734,
      "grad_norm": 0.578125,
      "learning_rate": 0.00012220363432281978,
      "loss": 0.8899,
      "step": 6608
    },
    {
      "epoch": 0.8458978625367977,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0001220051178378987,
      "loss": 0.8067,
      "step": 6609
    },
    {
      "epoch": 0.8460258543453218,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00012180675225064308,
      "loss": 1.0965,
      "step": 6610
    },
    {
      "epoch": 0.8461538461538461,
      "grad_norm": 0.578125,
      "learning_rate": 0.00012160853759514568,
      "loss": 0.9568,
      "step": 6611
    },
    {
      "epoch": 0.8462818379623704,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00012141047390547266,
      "loss": 1.4542,
      "step": 6612
    },
    {
      "epoch": 0.8464098297708946,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012121256121566471,
      "loss": 1.5142,
      "step": 6613
    },
    {
      "epoch": 0.8465378215794189,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012101479955973616,
      "loss": 1.0691,
      "step": 6614
    },
    {
      "epoch": 0.8466658133879432,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001208171889716756,
      "loss": 1.2965,
      "step": 6615
    },
    {
      "epoch": 0.8467938051964674,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012061972948544576,
      "loss": 0.9922,
      "step": 6616
    },
    {
      "epoch": 0.8469217970049917,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00012042242113498303,
      "loss": 1.1313,
      "step": 6617
    },
    {
      "epoch": 0.8470497888135159,
      "grad_norm": 0.625,
      "learning_rate": 0.00012022526395419842,
      "loss": 1.345,
      "step": 6618
    },
    {
      "epoch": 0.8471777806220402,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012002825797697603,
      "loss": 1.2863,
      "step": 6619
    },
    {
      "epoch": 0.8473057724305645,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011983140323717501,
      "loss": 1.7877,
      "step": 6620
    },
    {
      "epoch": 0.8474337642390887,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011963469976862773,
      "loss": 1.242,
      "step": 6621
    },
    {
      "epoch": 0.847561756047613,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00011943814760514105,
      "loss": 1.0596,
      "step": 6622
    },
    {
      "epoch": 0.8476897478561372,
      "grad_norm": 0.625,
      "learning_rate": 0.0001192417467804956,
      "loss": 1.6958,
      "step": 6623
    },
    {
      "epoch": 0.8478177396646615,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00011904549732844605,
      "loss": 1.1549,
      "step": 6624
    },
    {
      "epoch": 0.8479457314731857,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011884939928272109,
      "loss": 1.4674,
      "step": 6625
    },
    {
      "epoch": 0.8480737232817099,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00011865345267702354,
      "loss": 1.0368,
      "step": 6626
    },
    {
      "epoch": 0.8482017150902342,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00011845765754502957,
      "loss": 1.2033,
      "step": 6627
    },
    {
      "epoch": 0.8483297068987585,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00011826201392038993,
      "loss": 1.028,
      "step": 6628
    },
    {
      "epoch": 0.8484576987072827,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00011806652183672917,
      "loss": 1.0857,
      "step": 6629
    },
    {
      "epoch": 0.848585690515807,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011787118132764585,
      "loss": 0.9778,
      "step": 6630
    },
    {
      "epoch": 0.8487136823243312,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011767599242671245,
      "loss": 1.5107,
      "step": 6631
    },
    {
      "epoch": 0.8488416741328555,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00011748095516747515,
      "loss": 1.2956,
      "step": 6632
    },
    {
      "epoch": 0.8489696659413798,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00011728606958345422,
      "loss": 0.9219,
      "step": 6633
    },
    {
      "epoch": 0.849097657749904,
      "grad_norm": 0.75,
      "learning_rate": 0.00011709133570814423,
      "loss": 0.8827,
      "step": 6634
    },
    {
      "epoch": 0.8492256495584283,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011689675357501306,
      "loss": 1.2553,
      "step": 6635
    },
    {
      "epoch": 0.8493536413669525,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011670232321750296,
      "loss": 1.1499,
      "step": 6636
    },
    {
      "epoch": 0.8494816331754768,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00011650804466902987,
      "loss": 1.0334,
      "step": 6637
    },
    {
      "epoch": 0.849609624984001,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011631391796298385,
      "loss": 1.2121,
      "step": 6638
    },
    {
      "epoch": 0.8497376167925252,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011611994313272856,
      "loss": 1.2727,
      "step": 6639
    },
    {
      "epoch": 0.8498656086010495,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011592612021160199,
      "loss": 1.8812,
      "step": 6640
    },
    {
      "epoch": 0.8499936004095738,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00011573244923291525,
      "loss": 0.7377,
      "step": 6641
    },
    {
      "epoch": 0.850121592218098,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00011553893022995432,
      "loss": 1.5403,
      "step": 6642
    },
    {
      "epoch": 0.8502495840266223,
      "grad_norm": 0.59375,
      "learning_rate": 0.00011534556323597823,
      "loss": 0.9521,
      "step": 6643
    },
    {
      "epoch": 0.8503775758351465,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011515234828422072,
      "loss": 1.6086,
      "step": 6644
    },
    {
      "epoch": 0.8505055676436708,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011495928540788814,
      "loss": 1.6568,
      "step": 6645
    },
    {
      "epoch": 0.8506335594521951,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011476637464016215,
      "loss": 2.1636,
      "step": 6646
    },
    {
      "epoch": 0.8507615512607193,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011457361601419736,
      "loss": 1.4908,
      "step": 6647
    },
    {
      "epoch": 0.8508895430692436,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011438100956312247,
      "loss": 1.1851,
      "step": 6648
    },
    {
      "epoch": 0.8510175348777679,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00011418855532004025,
      "loss": 1.4554,
      "step": 6649
    },
    {
      "epoch": 0.8511455266862921,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011399625331802676,
      "loss": 1.5176,
      "step": 6650
    },
    {
      "epoch": 0.8512735184948164,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011380410359013226,
      "loss": 1.6087,
      "step": 6651
    },
    {
      "epoch": 0.8514015103033405,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00011361210616938089,
      "loss": 1.2715,
      "step": 6652
    },
    {
      "epoch": 0.8515295021118648,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011342026108877069,
      "loss": 1.2774,
      "step": 6653
    },
    {
      "epoch": 0.8516574939203891,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00011322856838127294,
      "loss": 0.9748,
      "step": 6654
    },
    {
      "epoch": 0.8517854857289133,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0001130370280798334,
      "loss": 0.6351,
      "step": 6655
    },
    {
      "epoch": 0.8519134775374376,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011284564021737131,
      "loss": 1.1854,
      "step": 6656
    },
    {
      "epoch": 0.8520414693459618,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011265440482677991,
      "loss": 1.0922,
      "step": 6657
    },
    {
      "epoch": 0.8521694611544861,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001124633219409259,
      "loss": 1.1796,
      "step": 6658
    },
    {
      "epoch": 0.8522974529630104,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011227239159265012,
      "loss": 1.0794,
      "step": 6659
    },
    {
      "epoch": 0.8524254447715346,
      "grad_norm": 0.56640625,
      "learning_rate": 0.000112081613814767,
      "loss": 0.9938,
      "step": 6660
    },
    {
      "epoch": 0.8525534365800589,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00011189098864006486,
      "loss": 0.9382,
      "step": 6661
    },
    {
      "epoch": 0.8526814283885832,
      "grad_norm": 0.75,
      "learning_rate": 0.00011170051610130582,
      "loss": 1.672,
      "step": 6662
    },
    {
      "epoch": 0.8528094201971074,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00011151019623122538,
      "loss": 0.8439,
      "step": 6663
    },
    {
      "epoch": 0.8529374120056317,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011132002906253314,
      "loss": 1.1408,
      "step": 6664
    },
    {
      "epoch": 0.8530654038141559,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011113001462791261,
      "loss": 1.8365,
      "step": 6665
    },
    {
      "epoch": 0.8531933956226802,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011094015296002102,
      "loss": 1.2838,
      "step": 6666
    },
    {
      "epoch": 0.8533213874312044,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011075044409148849,
      "loss": 1.5028,
      "step": 6667
    },
    {
      "epoch": 0.8534493792397286,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00011056088805492015,
      "loss": 0.9667,
      "step": 6668
    },
    {
      "epoch": 0.8535773710482529,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011037148488289416,
      "loss": 1.3947,
      "step": 6669
    },
    {
      "epoch": 0.8537053628567771,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001101822346079625,
      "loss": 1.1058,
      "step": 6670
    },
    {
      "epoch": 0.8538333546653014,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010999313726265104,
      "loss": 0.9749,
      "step": 6671
    },
    {
      "epoch": 0.8539613464738257,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00010980419287945898,
      "loss": 0.743,
      "step": 6672
    },
    {
      "epoch": 0.8540893382823499,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010961540149085957,
      "loss": 1.5729,
      "step": 6673
    },
    {
      "epoch": 0.8542173300908742,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00010942676312929967,
      "loss": 0.8338,
      "step": 6674
    },
    {
      "epoch": 0.8543453218993985,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00010923827782720009,
      "loss": 1.2516,
      "step": 6675
    },
    {
      "epoch": 0.8544733137079227,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010904994561695458,
      "loss": 1.7335,
      "step": 6676
    },
    {
      "epoch": 0.854601305516447,
      "grad_norm": 0.59375,
      "learning_rate": 0.00010886176653093149,
      "loss": 0.8554,
      "step": 6677
    },
    {
      "epoch": 0.8547292973249712,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010867374060147217,
      "loss": 1.6133,
      "step": 6678
    },
    {
      "epoch": 0.8548572891334955,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001084858678608922,
      "loss": 1.4035,
      "step": 6679
    },
    {
      "epoch": 0.8549852809420198,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00010829814834148044,
      "loss": 1.0416,
      "step": 6680
    },
    {
      "epoch": 0.8551132727505439,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010811058207549951,
      "loss": 1.4787,
      "step": 6681
    },
    {
      "epoch": 0.8552412645590682,
      "grad_norm": 0.5625,
      "learning_rate": 0.00010792316909518573,
      "loss": 1.1867,
      "step": 6682
    },
    {
      "epoch": 0.8553692563675924,
      "grad_norm": 0.875,
      "learning_rate": 0.00010773590943274902,
      "loss": 1.2604,
      "step": 6683
    },
    {
      "epoch": 0.8554972481761167,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001075488031203733,
      "loss": 1.5274,
      "step": 6684
    },
    {
      "epoch": 0.855625239984641,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00010736185019021538,
      "loss": 1.3379,
      "step": 6685
    },
    {
      "epoch": 0.8557532317931652,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010717505067440637,
      "loss": 1.4411,
      "step": 6686
    },
    {
      "epoch": 0.8558812236016895,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00010698840460505066,
      "loss": 1.2169,
      "step": 6687
    },
    {
      "epoch": 0.8560092154102138,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010680191201422684,
      "loss": 1.2435,
      "step": 6688
    },
    {
      "epoch": 0.856137207218738,
      "grad_norm": 0.875,
      "learning_rate": 0.00010661557293398604,
      "loss": 1.5361,
      "step": 6689
    },
    {
      "epoch": 0.8562651990272623,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010642938739635388,
      "loss": 1.7479,
      "step": 6690
    },
    {
      "epoch": 0.8563931908357865,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010624335543332964,
      "loss": 1.2252,
      "step": 6691
    },
    {
      "epoch": 0.8565211826443108,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010605747707688573,
      "loss": 1.7671,
      "step": 6692
    },
    {
      "epoch": 0.8566491744528351,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010587175235896863,
      "loss": 1.6284,
      "step": 6693
    },
    {
      "epoch": 0.8567771662613592,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00010568618131149765,
      "loss": 1.057,
      "step": 6694
    },
    {
      "epoch": 0.8569051580698835,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010550076396636644,
      "loss": 1.2736,
      "step": 6695
    },
    {
      "epoch": 0.8570331498784077,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00010531550035544202,
      "loss": 1.0757,
      "step": 6696
    },
    {
      "epoch": 0.857161141686932,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010513039051056505,
      "loss": 1.1896,
      "step": 6697
    },
    {
      "epoch": 0.8572891334954563,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00010494543446354932,
      "loss": 1.0847,
      "step": 6698
    },
    {
      "epoch": 0.8574171253039805,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001047606322461827,
      "loss": 1.3578,
      "step": 6699
    },
    {
      "epoch": 0.8575451171125048,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001045759838902266,
      "loss": 1.4909,
      "step": 6700
    },
    {
      "epoch": 0.8576731089210291,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00010439148942741561,
      "loss": 1.1146,
      "step": 6701
    },
    {
      "epoch": 0.8578011007295533,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00010420714888945815,
      "loss": 0.8912,
      "step": 6702
    },
    {
      "epoch": 0.8579290925380776,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00010402296230803621,
      "loss": 1.0122,
      "step": 6703
    },
    {
      "epoch": 0.8580570843466018,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010383892971480524,
      "loss": 1.2944,
      "step": 6704
    },
    {
      "epoch": 0.8581850761551261,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010365505114139418,
      "loss": 1.5418,
      "step": 6705
    },
    {
      "epoch": 0.8583130679636504,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010347132661940573,
      "loss": 1.1468,
      "step": 6706
    },
    {
      "epoch": 0.8584410597721746,
      "grad_norm": 0.609375,
      "learning_rate": 0.00010328775618041564,
      "loss": 0.8764,
      "step": 6707
    },
    {
      "epoch": 0.8585690515806988,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00010310433985597356,
      "loss": 1.2299,
      "step": 6708
    },
    {
      "epoch": 0.858697043389223,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010292107767760262,
      "loss": 1.2679,
      "step": 6709
    },
    {
      "epoch": 0.8588250351977473,
      "grad_norm": 0.625,
      "learning_rate": 0.0001027379696767995,
      "loss": 1.8233,
      "step": 6710
    },
    {
      "epoch": 0.8589530270062716,
      "grad_norm": 0.59375,
      "learning_rate": 0.00010255501588503391,
      "loss": 1.0516,
      "step": 6711
    },
    {
      "epoch": 0.8590810188147958,
      "grad_norm": 0.578125,
      "learning_rate": 0.00010237221633374983,
      "loss": 0.9168,
      "step": 6712
    },
    {
      "epoch": 0.8592090106233201,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010218957105436377,
      "loss": 1.0642,
      "step": 6713
    },
    {
      "epoch": 0.8593370024318444,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010200708007826698,
      "loss": 1.4872,
      "step": 6714
    },
    {
      "epoch": 0.8594649942403686,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010182474343682346,
      "loss": 1.0391,
      "step": 6715
    },
    {
      "epoch": 0.8595929860488929,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00010164256116137017,
      "loss": 1.0359,
      "step": 6716
    },
    {
      "epoch": 0.8597209778574171,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010146053328321847,
      "loss": 1.7371,
      "step": 6717
    },
    {
      "epoch": 0.8598489696659414,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00010127865983365259,
      "loss": 0.8946,
      "step": 6718
    },
    {
      "epoch": 0.8599769614744657,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00010109694084393084,
      "loss": 1.1656,
      "step": 6719
    },
    {
      "epoch": 0.8601049532829899,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010091537634528403,
      "loss": 1.6508,
      "step": 6720
    },
    {
      "epoch": 0.8602329450915142,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001007339663689173,
      "loss": 1.1272,
      "step": 6721
    },
    {
      "epoch": 0.8603609369000383,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010055271094600871,
      "loss": 1.1252,
      "step": 6722
    },
    {
      "epoch": 0.8604889287085626,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010037161010771013,
      "loss": 1.8741,
      "step": 6723
    },
    {
      "epoch": 0.8606169205170869,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010019066388514663,
      "loss": 1.241,
      "step": 6724
    },
    {
      "epoch": 0.8607449123256111,
      "grad_norm": 0.625,
      "learning_rate": 0.00010000987230941671,
      "loss": 1.1193,
      "step": 6725
    },
    {
      "epoch": 0.8608729041341354,
      "grad_norm": 0.671875,
      "learning_rate": 9.982923541159238e-05,
      "loss": 1.2572,
      "step": 6726
    },
    {
      "epoch": 0.8610008959426597,
      "grad_norm": 0.50390625,
      "learning_rate": 9.964875322271893e-05,
      "loss": 0.9744,
      "step": 6727
    },
    {
      "epoch": 0.8611288877511839,
      "grad_norm": 0.6875,
      "learning_rate": 9.946842577381543e-05,
      "loss": 1.5349,
      "step": 6728
    },
    {
      "epoch": 0.8612568795597082,
      "grad_norm": 0.578125,
      "learning_rate": 9.928825309587374e-05,
      "loss": 0.9812,
      "step": 6729
    },
    {
      "epoch": 0.8613848713682324,
      "grad_norm": 0.69921875,
      "learning_rate": 9.910823521985957e-05,
      "loss": 1.1783,
      "step": 6730
    },
    {
      "epoch": 0.8615128631767567,
      "grad_norm": 0.61328125,
      "learning_rate": 9.89283721767118e-05,
      "loss": 1.3829,
      "step": 6731
    },
    {
      "epoch": 0.861640854985281,
      "grad_norm": 1.1875,
      "learning_rate": 9.874866399734305e-05,
      "loss": 1.3069,
      "step": 6732
    },
    {
      "epoch": 0.8617688467938052,
      "grad_norm": 0.6796875,
      "learning_rate": 9.856911071263918e-05,
      "loss": 1.6364,
      "step": 6733
    },
    {
      "epoch": 0.8618968386023295,
      "grad_norm": 0.7578125,
      "learning_rate": 9.838971235345894e-05,
      "loss": 1.7389,
      "step": 6734
    },
    {
      "epoch": 0.8620248304108538,
      "grad_norm": 0.59375,
      "learning_rate": 9.821046895063479e-05,
      "loss": 1.1525,
      "step": 6735
    },
    {
      "epoch": 0.862152822219378,
      "grad_norm": 0.7421875,
      "learning_rate": 9.803138053497319e-05,
      "loss": 1.1265,
      "step": 6736
    },
    {
      "epoch": 0.8622808140279022,
      "grad_norm": 0.80859375,
      "learning_rate": 9.785244713725316e-05,
      "loss": 1.54,
      "step": 6737
    },
    {
      "epoch": 0.8624088058364264,
      "grad_norm": 0.74609375,
      "learning_rate": 9.767366878822703e-05,
      "loss": 1.3344,
      "step": 6738
    },
    {
      "epoch": 0.8625367976449507,
      "grad_norm": 0.56640625,
      "learning_rate": 9.749504551862087e-05,
      "loss": 0.8843,
      "step": 6739
    },
    {
      "epoch": 0.862664789453475,
      "grad_norm": 0.625,
      "learning_rate": 9.731657735913424e-05,
      "loss": 1.0735,
      "step": 6740
    },
    {
      "epoch": 0.8627927812619992,
      "grad_norm": 0.578125,
      "learning_rate": 9.71382643404397e-05,
      "loss": 1.0285,
      "step": 6741
    },
    {
      "epoch": 0.8629207730705235,
      "grad_norm": 0.67578125,
      "learning_rate": 9.696010649318288e-05,
      "loss": 1.3336,
      "step": 6742
    },
    {
      "epoch": 0.8630487648790477,
      "grad_norm": 0.75390625,
      "learning_rate": 9.678210384798347e-05,
      "loss": 1.1844,
      "step": 6743
    },
    {
      "epoch": 0.863176756687572,
      "grad_norm": 0.62109375,
      "learning_rate": 9.660425643543403e-05,
      "loss": 0.8923,
      "step": 6744
    },
    {
      "epoch": 0.8633047484960963,
      "grad_norm": 0.7265625,
      "learning_rate": 9.642656428610042e-05,
      "loss": 1.4477,
      "step": 6745
    },
    {
      "epoch": 0.8634327403046205,
      "grad_norm": 0.53515625,
      "learning_rate": 9.6249027430522e-05,
      "loss": 0.7757,
      "step": 6746
    },
    {
      "epoch": 0.8635607321131448,
      "grad_norm": 0.62890625,
      "learning_rate": 9.607164589921136e-05,
      "loss": 1.2902,
      "step": 6747
    },
    {
      "epoch": 0.8636887239216691,
      "grad_norm": 1.265625,
      "learning_rate": 9.589441972265445e-05,
      "loss": 1.1742,
      "step": 6748
    },
    {
      "epoch": 0.8638167157301933,
      "grad_norm": 0.625,
      "learning_rate": 9.571734893131034e-05,
      "loss": 0.848,
      "step": 6749
    },
    {
      "epoch": 0.8639447075387175,
      "grad_norm": 0.6484375,
      "learning_rate": 9.554043355561182e-05,
      "loss": 1.0286,
      "step": 6750
    },
    {
      "epoch": 0.8640726993472417,
      "grad_norm": 0.64453125,
      "learning_rate": 9.536367362596421e-05,
      "loss": 1.1007,
      "step": 6751
    },
    {
      "epoch": 0.864200691155766,
      "grad_norm": 0.671875,
      "learning_rate": 9.518706917274688e-05,
      "loss": 1.2003,
      "step": 6752
    },
    {
      "epoch": 0.8643286829642903,
      "grad_norm": 0.6640625,
      "learning_rate": 9.501062022631201e-05,
      "loss": 1.4875,
      "step": 6753
    },
    {
      "epoch": 0.8644566747728145,
      "grad_norm": 0.65234375,
      "learning_rate": 9.483432681698522e-05,
      "loss": 1.4858,
      "step": 6754
    },
    {
      "epoch": 0.8645846665813388,
      "grad_norm": 0.69140625,
      "learning_rate": 9.465818897506573e-05,
      "loss": 1.1671,
      "step": 6755
    },
    {
      "epoch": 0.864712658389863,
      "grad_norm": 0.5,
      "learning_rate": 9.448220673082519e-05,
      "loss": 0.7775,
      "step": 6756
    },
    {
      "epoch": 0.8648406501983873,
      "grad_norm": 0.6640625,
      "learning_rate": 9.430638011450931e-05,
      "loss": 1.441,
      "step": 6757
    },
    {
      "epoch": 0.8649686420069116,
      "grad_norm": 0.71484375,
      "learning_rate": 9.413070915633636e-05,
      "loss": 1.618,
      "step": 6758
    },
    {
      "epoch": 0.8650966338154358,
      "grad_norm": 0.5859375,
      "learning_rate": 9.395519388649909e-05,
      "loss": 1.3146,
      "step": 6759
    },
    {
      "epoch": 0.8652246256239601,
      "grad_norm": 0.76171875,
      "learning_rate": 9.37798343351618e-05,
      "loss": 1.2451,
      "step": 6760
    },
    {
      "epoch": 0.8653526174324844,
      "grad_norm": 0.734375,
      "learning_rate": 9.360463053246315e-05,
      "loss": 1.412,
      "step": 6761
    },
    {
      "epoch": 0.8654806092410086,
      "grad_norm": 0.6796875,
      "learning_rate": 9.342958250851486e-05,
      "loss": 1.4375,
      "step": 6762
    },
    {
      "epoch": 0.8656086010495329,
      "grad_norm": 0.6484375,
      "learning_rate": 9.325469029340195e-05,
      "loss": 1.3324,
      "step": 6763
    },
    {
      "epoch": 0.865736592858057,
      "grad_norm": 0.58984375,
      "learning_rate": 9.307995391718193e-05,
      "loss": 0.9888,
      "step": 6764
    },
    {
      "epoch": 0.8658645846665813,
      "grad_norm": 0.62109375,
      "learning_rate": 9.290537340988647e-05,
      "loss": 1.3549,
      "step": 6765
    },
    {
      "epoch": 0.8659925764751056,
      "grad_norm": 0.7109375,
      "learning_rate": 9.273094880151988e-05,
      "loss": 1.5572,
      "step": 6766
    },
    {
      "epoch": 0.8661205682836298,
      "grad_norm": 0.6796875,
      "learning_rate": 9.255668012206008e-05,
      "loss": 1.4883,
      "step": 6767
    },
    {
      "epoch": 0.8662485600921541,
      "grad_norm": 0.6171875,
      "learning_rate": 9.238256740145778e-05,
      "loss": 1.348,
      "step": 6768
    },
    {
      "epoch": 0.8663765519006783,
      "grad_norm": 0.62890625,
      "learning_rate": 9.220861066963715e-05,
      "loss": 1.3124,
      "step": 6769
    },
    {
      "epoch": 0.8665045437092026,
      "grad_norm": 0.578125,
      "learning_rate": 9.203480995649549e-05,
      "loss": 1.0503,
      "step": 6770
    },
    {
      "epoch": 0.8666325355177269,
      "grad_norm": 0.59375,
      "learning_rate": 9.186116529190325e-05,
      "loss": 1.0971,
      "step": 6771
    },
    {
      "epoch": 0.8667605273262511,
      "grad_norm": 0.6640625,
      "learning_rate": 9.168767670570432e-05,
      "loss": 1.2438,
      "step": 6772
    },
    {
      "epoch": 0.8668885191347754,
      "grad_norm": 0.6484375,
      "learning_rate": 9.151434422771498e-05,
      "loss": 1.0615,
      "step": 6773
    },
    {
      "epoch": 0.8670165109432997,
      "grad_norm": 0.70703125,
      "learning_rate": 9.134116788772562e-05,
      "loss": 1.907,
      "step": 6774
    },
    {
      "epoch": 0.8671445027518239,
      "grad_norm": 0.66796875,
      "learning_rate": 9.116814771549941e-05,
      "loss": 1.3275,
      "step": 6775
    },
    {
      "epoch": 0.8672724945603482,
      "grad_norm": 0.578125,
      "learning_rate": 9.099528374077249e-05,
      "loss": 0.9385,
      "step": 6776
    },
    {
      "epoch": 0.8674004863688723,
      "grad_norm": 0.62890625,
      "learning_rate": 9.08225759932546e-05,
      "loss": 1.3804,
      "step": 6777
    },
    {
      "epoch": 0.8675284781773966,
      "grad_norm": 0.7421875,
      "learning_rate": 9.065002450262817e-05,
      "loss": 1.279,
      "step": 6778
    },
    {
      "epoch": 0.8676564699859209,
      "grad_norm": 0.67578125,
      "learning_rate": 9.047762929854885e-05,
      "loss": 1.5356,
      "step": 6779
    },
    {
      "epoch": 0.8677844617944451,
      "grad_norm": 0.70703125,
      "learning_rate": 9.030539041064568e-05,
      "loss": 1.0181,
      "step": 6780
    },
    {
      "epoch": 0.8679124536029694,
      "grad_norm": 0.92578125,
      "learning_rate": 9.013330786852092e-05,
      "loss": 1.6626,
      "step": 6781
    },
    {
      "epoch": 0.8680404454114936,
      "grad_norm": 0.5703125,
      "learning_rate": 8.99613817017495e-05,
      "loss": 1.2216,
      "step": 6782
    },
    {
      "epoch": 0.8681684372200179,
      "grad_norm": 0.83203125,
      "learning_rate": 8.978961193987989e-05,
      "loss": 0.8813,
      "step": 6783
    },
    {
      "epoch": 0.8682964290285422,
      "grad_norm": 0.62109375,
      "learning_rate": 8.961799861243324e-05,
      "loss": 0.9819,
      "step": 6784
    },
    {
      "epoch": 0.8684244208370664,
      "grad_norm": 0.625,
      "learning_rate": 8.944654174890432e-05,
      "loss": 1.1613,
      "step": 6785
    },
    {
      "epoch": 0.8685524126455907,
      "grad_norm": 0.609375,
      "learning_rate": 8.927524137876097e-05,
      "loss": 1.0209,
      "step": 6786
    },
    {
      "epoch": 0.868680404454115,
      "grad_norm": 0.61328125,
      "learning_rate": 8.910409753144344e-05,
      "loss": 1.5691,
      "step": 6787
    },
    {
      "epoch": 0.8688083962626392,
      "grad_norm": 0.828125,
      "learning_rate": 8.893311023636574e-05,
      "loss": 0.8361,
      "step": 6788
    },
    {
      "epoch": 0.8689363880711635,
      "grad_norm": 0.64453125,
      "learning_rate": 8.876227952291493e-05,
      "loss": 1.1491,
      "step": 6789
    },
    {
      "epoch": 0.8690643798796877,
      "grad_norm": 0.6953125,
      "learning_rate": 8.859160542045108e-05,
      "loss": 1.8159,
      "step": 6790
    },
    {
      "epoch": 0.869192371688212,
      "grad_norm": 0.640625,
      "learning_rate": 8.842108795830717e-05,
      "loss": 1.3219,
      "step": 6791
    },
    {
      "epoch": 0.8693203634967362,
      "grad_norm": 0.578125,
      "learning_rate": 8.825072716578942e-05,
      "loss": 0.8719,
      "step": 6792
    },
    {
      "epoch": 0.8694483553052604,
      "grad_norm": 0.53125,
      "learning_rate": 8.808052307217707e-05,
      "loss": 0.982,
      "step": 6793
    },
    {
      "epoch": 0.8695763471137847,
      "grad_norm": 0.75,
      "learning_rate": 8.791047570672273e-05,
      "loss": 1.556,
      "step": 6794
    },
    {
      "epoch": 0.8697043389223089,
      "grad_norm": 0.640625,
      "learning_rate": 8.774058509865146e-05,
      "loss": 1.4801,
      "step": 6795
    },
    {
      "epoch": 0.8698323307308332,
      "grad_norm": 0.59765625,
      "learning_rate": 8.75708512771617e-05,
      "loss": 1.0036,
      "step": 6796
    },
    {
      "epoch": 0.8699603225393575,
      "grad_norm": 0.69921875,
      "learning_rate": 8.740127427142519e-05,
      "loss": 2.2034,
      "step": 6797
    },
    {
      "epoch": 0.8700883143478817,
      "grad_norm": 0.73828125,
      "learning_rate": 8.723185411058631e-05,
      "loss": 1.6616,
      "step": 6798
    },
    {
      "epoch": 0.870216306156406,
      "grad_norm": 0.69140625,
      "learning_rate": 8.7062590823763e-05,
      "loss": 1.2769,
      "step": 6799
    },
    {
      "epoch": 0.8703442979649303,
      "grad_norm": 0.69140625,
      "learning_rate": 8.689348444004541e-05,
      "loss": 1.3433,
      "step": 6800
    },
    {
      "epoch": 0.8704722897734545,
      "grad_norm": 0.6015625,
      "learning_rate": 8.672453498849753e-05,
      "loss": 1.3435,
      "step": 6801
    },
    {
      "epoch": 0.8706002815819788,
      "grad_norm": 0.58203125,
      "learning_rate": 8.65557424981559e-05,
      "loss": 1.109,
      "step": 6802
    },
    {
      "epoch": 0.870728273390503,
      "grad_norm": 0.63671875,
      "learning_rate": 8.638710699803043e-05,
      "loss": 0.8641,
      "step": 6803
    },
    {
      "epoch": 0.8708562651990273,
      "grad_norm": 0.73046875,
      "learning_rate": 8.621862851710371e-05,
      "loss": 1.6044,
      "step": 6804
    },
    {
      "epoch": 0.8709842570075516,
      "grad_norm": 0.6796875,
      "learning_rate": 8.605030708433148e-05,
      "loss": 1.6729,
      "step": 6805
    },
    {
      "epoch": 0.8711122488160757,
      "grad_norm": 0.7578125,
      "learning_rate": 8.58821427286427e-05,
      "loss": 1.4672,
      "step": 6806
    },
    {
      "epoch": 0.8712402406246,
      "grad_norm": 0.54296875,
      "learning_rate": 8.571413547893902e-05,
      "loss": 0.8732,
      "step": 6807
    },
    {
      "epoch": 0.8713682324331243,
      "grad_norm": 0.6640625,
      "learning_rate": 8.554628536409536e-05,
      "loss": 1.8729,
      "step": 6808
    },
    {
      "epoch": 0.8714962242416485,
      "grad_norm": 0.70703125,
      "learning_rate": 8.53785924129592e-05,
      "loss": 1.4049,
      "step": 6809
    },
    {
      "epoch": 0.8716242160501728,
      "grad_norm": 0.703125,
      "learning_rate": 8.521105665435147e-05,
      "loss": 1.4966,
      "step": 6810
    },
    {
      "epoch": 0.871752207858697,
      "grad_norm": 0.625,
      "learning_rate": 8.504367811706581e-05,
      "loss": 1.4253,
      "step": 6811
    },
    {
      "epoch": 0.8718801996672213,
      "grad_norm": 0.62890625,
      "learning_rate": 8.48764568298691e-05,
      "loss": 1.0728,
      "step": 6812
    },
    {
      "epoch": 0.8720081914757456,
      "grad_norm": 0.65625,
      "learning_rate": 8.470939282150092e-05,
      "loss": 1.1335,
      "step": 6813
    },
    {
      "epoch": 0.8721361832842698,
      "grad_norm": 0.8125,
      "learning_rate": 8.454248612067394e-05,
      "loss": 1.3313,
      "step": 6814
    },
    {
      "epoch": 0.8722641750927941,
      "grad_norm": 0.671875,
      "learning_rate": 8.43757367560738e-05,
      "loss": 1.3326,
      "step": 6815
    },
    {
      "epoch": 0.8723921669013183,
      "grad_norm": 0.8515625,
      "learning_rate": 8.42091447563591e-05,
      "loss": 1.263,
      "step": 6816
    },
    {
      "epoch": 0.8725201587098426,
      "grad_norm": 0.7265625,
      "learning_rate": 8.404271015016151e-05,
      "loss": 1.4037,
      "step": 6817
    },
    {
      "epoch": 0.8726481505183669,
      "grad_norm": 0.65625,
      "learning_rate": 8.387643296608515e-05,
      "loss": 1.4081,
      "step": 6818
    },
    {
      "epoch": 0.872776142326891,
      "grad_norm": 0.54296875,
      "learning_rate": 8.371031323270773e-05,
      "loss": 0.7403,
      "step": 6819
    },
    {
      "epoch": 0.8729041341354153,
      "grad_norm": 0.60546875,
      "learning_rate": 8.354435097857949e-05,
      "loss": 1.2833,
      "step": 6820
    },
    {
      "epoch": 0.8730321259439396,
      "grad_norm": 0.625,
      "learning_rate": 8.337854623222407e-05,
      "loss": 1.2002,
      "step": 6821
    },
    {
      "epoch": 0.8731601177524638,
      "grad_norm": 0.796875,
      "learning_rate": 8.321289902213724e-05,
      "loss": 1.2939,
      "step": 6822
    },
    {
      "epoch": 0.8732881095609881,
      "grad_norm": 0.6171875,
      "learning_rate": 8.304740937678835e-05,
      "loss": 1.1573,
      "step": 6823
    },
    {
      "epoch": 0.8734161013695123,
      "grad_norm": 0.78125,
      "learning_rate": 8.28820773246195e-05,
      "loss": 1.5487,
      "step": 6824
    },
    {
      "epoch": 0.8735440931780366,
      "grad_norm": 0.53515625,
      "learning_rate": 8.271690289404565e-05,
      "loss": 0.9573,
      "step": 6825
    },
    {
      "epoch": 0.8736720849865609,
      "grad_norm": 0.671875,
      "learning_rate": 8.255188611345476e-05,
      "loss": 1.2494,
      "step": 6826
    },
    {
      "epoch": 0.8738000767950851,
      "grad_norm": 1.203125,
      "learning_rate": 8.23870270112076e-05,
      "loss": 1.1847,
      "step": 6827
    },
    {
      "epoch": 0.8739280686036094,
      "grad_norm": 0.67578125,
      "learning_rate": 8.222232561563791e-05,
      "loss": 1.2507,
      "step": 6828
    },
    {
      "epoch": 0.8740560604121336,
      "grad_norm": 0.640625,
      "learning_rate": 8.20577819550522e-05,
      "loss": 1.2206,
      "step": 6829
    },
    {
      "epoch": 0.8741840522206579,
      "grad_norm": 0.68359375,
      "learning_rate": 8.189339605773039e-05,
      "loss": 1.4335,
      "step": 6830
    },
    {
      "epoch": 0.8743120440291822,
      "grad_norm": 0.6484375,
      "learning_rate": 8.17291679519242e-05,
      "loss": 1.3523,
      "step": 6831
    },
    {
      "epoch": 0.8744400358377064,
      "grad_norm": 0.56640625,
      "learning_rate": 8.156509766585928e-05,
      "loss": 1.6037,
      "step": 6832
    },
    {
      "epoch": 0.8745680276462307,
      "grad_norm": 0.91796875,
      "learning_rate": 8.140118522773376e-05,
      "loss": 1.5793,
      "step": 6833
    },
    {
      "epoch": 0.874696019454755,
      "grad_norm": 0.6640625,
      "learning_rate": 8.123743066571854e-05,
      "loss": 1.5232,
      "step": 6834
    },
    {
      "epoch": 0.8748240112632791,
      "grad_norm": 0.734375,
      "learning_rate": 8.107383400795753e-05,
      "loss": 1.518,
      "step": 6835
    },
    {
      "epoch": 0.8749520030718034,
      "grad_norm": 0.6953125,
      "learning_rate": 8.09103952825676e-05,
      "loss": 1.8769,
      "step": 6836
    },
    {
      "epoch": 0.8750799948803276,
      "grad_norm": 0.640625,
      "learning_rate": 8.074711451763816e-05,
      "loss": 1.0603,
      "step": 6837
    },
    {
      "epoch": 0.8752079866888519,
      "grad_norm": 0.61328125,
      "learning_rate": 8.058399174123176e-05,
      "loss": 1.1959,
      "step": 6838
    },
    {
      "epoch": 0.8753359784973762,
      "grad_norm": 0.69140625,
      "learning_rate": 8.042102698138399e-05,
      "loss": 1.1805,
      "step": 6839
    },
    {
      "epoch": 0.8754639703059004,
      "grad_norm": 0.625,
      "learning_rate": 8.02582202661023e-05,
      "loss": 1.1124,
      "step": 6840
    },
    {
      "epoch": 0.8755919621144247,
      "grad_norm": 0.70703125,
      "learning_rate": 8.009557162336823e-05,
      "loss": 1.2078,
      "step": 6841
    },
    {
      "epoch": 0.8757199539229489,
      "grad_norm": 0.5390625,
      "learning_rate": 7.993308108113528e-05,
      "loss": 0.9173,
      "step": 6842
    },
    {
      "epoch": 0.8758479457314732,
      "grad_norm": 0.69921875,
      "learning_rate": 7.977074866733036e-05,
      "loss": 1.8103,
      "step": 6843
    },
    {
      "epoch": 0.8759759375399975,
      "grad_norm": 0.6796875,
      "learning_rate": 7.960857440985269e-05,
      "loss": 1.6248,
      "step": 6844
    },
    {
      "epoch": 0.8761039293485217,
      "grad_norm": 0.640625,
      "learning_rate": 7.944655833657466e-05,
      "loss": 1.1769,
      "step": 6845
    },
    {
      "epoch": 0.876231921157046,
      "grad_norm": 0.6953125,
      "learning_rate": 7.928470047534119e-05,
      "loss": 1.5234,
      "step": 6846
    },
    {
      "epoch": 0.8763599129655703,
      "grad_norm": 0.703125,
      "learning_rate": 7.912300085397051e-05,
      "loss": 1.2149,
      "step": 6847
    },
    {
      "epoch": 0.8764879047740944,
      "grad_norm": 0.59765625,
      "learning_rate": 7.896145950025302e-05,
      "loss": 1.0309,
      "step": 6848
    },
    {
      "epoch": 0.8766158965826187,
      "grad_norm": 0.74609375,
      "learning_rate": 7.880007644195242e-05,
      "loss": 1.346,
      "step": 6849
    },
    {
      "epoch": 0.8767438883911429,
      "grad_norm": 0.6640625,
      "learning_rate": 7.863885170680484e-05,
      "loss": 1.5738,
      "step": 6850
    },
    {
      "epoch": 0.8768718801996672,
      "grad_norm": 0.625,
      "learning_rate": 7.847778532251949e-05,
      "loss": 1.5638,
      "step": 6851
    },
    {
      "epoch": 0.8769998720081915,
      "grad_norm": 0.66015625,
      "learning_rate": 7.831687731677839e-05,
      "loss": 1.3526,
      "step": 6852
    },
    {
      "epoch": 0.8771278638167157,
      "grad_norm": 0.5234375,
      "learning_rate": 7.81561277172359e-05,
      "loss": 0.7733,
      "step": 6853
    },
    {
      "epoch": 0.87725585562524,
      "grad_norm": 0.515625,
      "learning_rate": 7.799553655151947e-05,
      "loss": 0.846,
      "step": 6854
    },
    {
      "epoch": 0.8773838474337642,
      "grad_norm": 0.65625,
      "learning_rate": 7.783510384722936e-05,
      "loss": 1.1026,
      "step": 6855
    },
    {
      "epoch": 0.8775118392422885,
      "grad_norm": 0.6015625,
      "learning_rate": 7.767482963193883e-05,
      "loss": 1.0971,
      "step": 6856
    },
    {
      "epoch": 0.8776398310508128,
      "grad_norm": 0.5703125,
      "learning_rate": 7.751471393319287e-05,
      "loss": 0.8213,
      "step": 6857
    },
    {
      "epoch": 0.877767822859337,
      "grad_norm": 0.625,
      "learning_rate": 7.735475677851067e-05,
      "loss": 0.9826,
      "step": 6858
    },
    {
      "epoch": 0.8778958146678613,
      "grad_norm": 0.640625,
      "learning_rate": 7.719495819538325e-05,
      "loss": 1.3639,
      "step": 6859
    },
    {
      "epoch": 0.8780238064763856,
      "grad_norm": 0.59765625,
      "learning_rate": 7.703531821127441e-05,
      "loss": 0.8795,
      "step": 6860
    },
    {
      "epoch": 0.8781517982849097,
      "grad_norm": 0.66796875,
      "learning_rate": 7.687583685362131e-05,
      "loss": 1.5645,
      "step": 6861
    },
    {
      "epoch": 0.878279790093434,
      "grad_norm": 0.62109375,
      "learning_rate": 7.671651414983294e-05,
      "loss": 1.4325,
      "step": 6862
    },
    {
      "epoch": 0.8784077819019582,
      "grad_norm": 0.5625,
      "learning_rate": 7.655735012729159e-05,
      "loss": 0.7625,
      "step": 6863
    },
    {
      "epoch": 0.8785357737104825,
      "grad_norm": 0.81640625,
      "learning_rate": 7.639834481335228e-05,
      "loss": 1.4458,
      "step": 6864
    },
    {
      "epoch": 0.8786637655190068,
      "grad_norm": 0.5625,
      "learning_rate": 7.623949823534282e-05,
      "loss": 0.9235,
      "step": 6865
    },
    {
      "epoch": 0.878791757327531,
      "grad_norm": 0.6015625,
      "learning_rate": 7.608081042056325e-05,
      "loss": 1.3002,
      "step": 6866
    },
    {
      "epoch": 0.8789197491360553,
      "grad_norm": 0.58984375,
      "learning_rate": 7.592228139628676e-05,
      "loss": 1.2002,
      "step": 6867
    },
    {
      "epoch": 0.8790477409445795,
      "grad_norm": 0.59375,
      "learning_rate": 7.576391118975912e-05,
      "loss": 1.2567,
      "step": 6868
    },
    {
      "epoch": 0.8791757327531038,
      "grad_norm": 0.65625,
      "learning_rate": 7.560569982819876e-05,
      "loss": 1.4303,
      "step": 6869
    },
    {
      "epoch": 0.8793037245616281,
      "grad_norm": 0.72265625,
      "learning_rate": 7.544764733879705e-05,
      "loss": 1.3712,
      "step": 6870
    },
    {
      "epoch": 0.8794317163701523,
      "grad_norm": 0.6875,
      "learning_rate": 7.528975374871782e-05,
      "loss": 1.4614,
      "step": 6871
    },
    {
      "epoch": 0.8795597081786766,
      "grad_norm": 0.671875,
      "learning_rate": 7.51320190850976e-05,
      "loss": 1.177,
      "step": 6872
    },
    {
      "epoch": 0.8796876999872009,
      "grad_norm": 0.65625,
      "learning_rate": 7.497444337504556e-05,
      "loss": 1.388,
      "step": 6873
    },
    {
      "epoch": 0.879815691795725,
      "grad_norm": 0.640625,
      "learning_rate": 7.481702664564405e-05,
      "loss": 1.1488,
      "step": 6874
    },
    {
      "epoch": 0.8799436836042493,
      "grad_norm": 0.578125,
      "learning_rate": 7.46597689239471e-05,
      "loss": 0.7559,
      "step": 6875
    },
    {
      "epoch": 0.8800716754127735,
      "grad_norm": 0.66015625,
      "learning_rate": 7.450267023698231e-05,
      "loss": 1.4581,
      "step": 6876
    },
    {
      "epoch": 0.8801996672212978,
      "grad_norm": 0.65625,
      "learning_rate": 7.434573061174966e-05,
      "loss": 1.4389,
      "step": 6877
    },
    {
      "epoch": 0.8803276590298221,
      "grad_norm": 1.0390625,
      "learning_rate": 7.418895007522187e-05,
      "loss": 1.9565,
      "step": 6878
    },
    {
      "epoch": 0.8804556508383463,
      "grad_norm": 0.49609375,
      "learning_rate": 7.403232865434384e-05,
      "loss": 0.7029,
      "step": 6879
    },
    {
      "epoch": 0.8805836426468706,
      "grad_norm": 0.83984375,
      "learning_rate": 7.387586637603361e-05,
      "loss": 1.7824,
      "step": 6880
    },
    {
      "epoch": 0.8807116344553949,
      "grad_norm": 0.79296875,
      "learning_rate": 7.37195632671821e-05,
      "loss": 1.734,
      "step": 6881
    },
    {
      "epoch": 0.8808396262639191,
      "grad_norm": 0.73046875,
      "learning_rate": 7.356341935465227e-05,
      "loss": 1.5655,
      "step": 6882
    },
    {
      "epoch": 0.8809676180724434,
      "grad_norm": 0.8671875,
      "learning_rate": 7.340743466528022e-05,
      "loss": 1.3738,
      "step": 6883
    },
    {
      "epoch": 0.8810956098809676,
      "grad_norm": 0.6171875,
      "learning_rate": 7.325160922587403e-05,
      "loss": 1.0439,
      "step": 6884
    },
    {
      "epoch": 0.8812236016894919,
      "grad_norm": 0.609375,
      "learning_rate": 7.30959430632151e-05,
      "loss": 1.0681,
      "step": 6885
    },
    {
      "epoch": 0.8813515934980162,
      "grad_norm": 0.51171875,
      "learning_rate": 7.294043620405721e-05,
      "loss": 0.7244,
      "step": 6886
    },
    {
      "epoch": 0.8814795853065404,
      "grad_norm": 1.421875,
      "learning_rate": 7.278508867512679e-05,
      "loss": 1.4743,
      "step": 6887
    },
    {
      "epoch": 0.8816075771150647,
      "grad_norm": 0.7578125,
      "learning_rate": 7.262990050312258e-05,
      "loss": 1.4483,
      "step": 6888
    },
    {
      "epoch": 0.8817355689235888,
      "grad_norm": 0.78515625,
      "learning_rate": 7.247487171471622e-05,
      "loss": 2.3103,
      "step": 6889
    },
    {
      "epoch": 0.8818635607321131,
      "grad_norm": 0.84765625,
      "learning_rate": 7.232000233655212e-05,
      "loss": 1.3367,
      "step": 6890
    },
    {
      "epoch": 0.8819915525406374,
      "grad_norm": 0.90625,
      "learning_rate": 7.216529239524705e-05,
      "loss": 1.6603,
      "step": 6891
    },
    {
      "epoch": 0.8821195443491616,
      "grad_norm": 0.76171875,
      "learning_rate": 7.20107419173902e-05,
      "loss": 1.3032,
      "step": 6892
    },
    {
      "epoch": 0.8822475361576859,
      "grad_norm": 0.6484375,
      "learning_rate": 7.18563509295439e-05,
      "loss": 1.3874,
      "step": 6893
    },
    {
      "epoch": 0.8823755279662102,
      "grad_norm": 0.66796875,
      "learning_rate": 7.170211945824256e-05,
      "loss": 1.2631,
      "step": 6894
    },
    {
      "epoch": 0.8825035197747344,
      "grad_norm": 0.72265625,
      "learning_rate": 7.154804752999344e-05,
      "loss": 1.359,
      "step": 6895
    },
    {
      "epoch": 0.8826315115832587,
      "grad_norm": 0.55859375,
      "learning_rate": 7.139413517127635e-05,
      "loss": 1.2823,
      "step": 6896
    },
    {
      "epoch": 0.8827595033917829,
      "grad_norm": 0.6875,
      "learning_rate": 7.124038240854347e-05,
      "loss": 1.3112,
      "step": 6897
    },
    {
      "epoch": 0.8828874952003072,
      "grad_norm": 0.671875,
      "learning_rate": 7.108678926821966e-05,
      "loss": 1.2209,
      "step": 6898
    },
    {
      "epoch": 0.8830154870088315,
      "grad_norm": 0.66796875,
      "learning_rate": 7.093335577670257e-05,
      "loss": 1.3642,
      "step": 6899
    },
    {
      "epoch": 0.8831434788173557,
      "grad_norm": 0.93359375,
      "learning_rate": 7.078008196036223e-05,
      "loss": 1.6066,
      "step": 6900
    },
    {
      "epoch": 0.88327147062588,
      "grad_norm": 0.73828125,
      "learning_rate": 7.062696784554123e-05,
      "loss": 1.0818,
      "step": 6901
    },
    {
      "epoch": 0.8833994624344041,
      "grad_norm": 0.66796875,
      "learning_rate": 7.047401345855442e-05,
      "loss": 1.3091,
      "step": 6902
    },
    {
      "epoch": 0.8835274542429284,
      "grad_norm": 0.96484375,
      "learning_rate": 7.032121882568975e-05,
      "loss": 1.5825,
      "step": 6903
    },
    {
      "epoch": 0.8836554460514527,
      "grad_norm": 0.51171875,
      "learning_rate": 7.016858397320758e-05,
      "loss": 1.1899,
      "step": 6904
    },
    {
      "epoch": 0.8837834378599769,
      "grad_norm": 0.5859375,
      "learning_rate": 7.001610892734067e-05,
      "loss": 1.051,
      "step": 6905
    },
    {
      "epoch": 0.8839114296685012,
      "grad_norm": 0.765625,
      "learning_rate": 6.986379371429408e-05,
      "loss": 1.1902,
      "step": 6906
    },
    {
      "epoch": 0.8840394214770255,
      "grad_norm": 0.62109375,
      "learning_rate": 6.971163836024574e-05,
      "loss": 1.148,
      "step": 6907
    },
    {
      "epoch": 0.8841674132855497,
      "grad_norm": 0.765625,
      "learning_rate": 6.955964289134597e-05,
      "loss": 1.6229,
      "step": 6908
    },
    {
      "epoch": 0.884295405094074,
      "grad_norm": 0.66796875,
      "learning_rate": 6.9407807333718e-05,
      "loss": 1.0453,
      "step": 6909
    },
    {
      "epoch": 0.8844233969025982,
      "grad_norm": 0.625,
      "learning_rate": 6.92561317134568e-05,
      "loss": 1.2309,
      "step": 6910
    },
    {
      "epoch": 0.8845513887111225,
      "grad_norm": 0.73046875,
      "learning_rate": 6.910461605663043e-05,
      "loss": 1.5018,
      "step": 6911
    },
    {
      "epoch": 0.8846793805196468,
      "grad_norm": 0.734375,
      "learning_rate": 6.895326038927941e-05,
      "loss": 1.2526,
      "step": 6912
    },
    {
      "epoch": 0.884807372328171,
      "grad_norm": 0.8203125,
      "learning_rate": 6.880206473741646e-05,
      "loss": 1.8287,
      "step": 6913
    },
    {
      "epoch": 0.8849353641366953,
      "grad_norm": 0.7109375,
      "learning_rate": 6.865102912702725e-05,
      "loss": 1.4217,
      "step": 6914
    },
    {
      "epoch": 0.8850633559452195,
      "grad_norm": 0.703125,
      "learning_rate": 6.850015358406959e-05,
      "loss": 1.4995,
      "step": 6915
    },
    {
      "epoch": 0.8851913477537438,
      "grad_norm": 0.65625,
      "learning_rate": 6.834943813447392e-05,
      "loss": 1.7748,
      "step": 6916
    },
    {
      "epoch": 0.885319339562268,
      "grad_norm": 0.6328125,
      "learning_rate": 6.81988828041431e-05,
      "loss": 1.1814,
      "step": 6917
    },
    {
      "epoch": 0.8854473313707922,
      "grad_norm": 0.640625,
      "learning_rate": 6.804848761895266e-05,
      "loss": 1.0471,
      "step": 6918
    },
    {
      "epoch": 0.8855753231793165,
      "grad_norm": 0.61328125,
      "learning_rate": 6.789825260475014e-05,
      "loss": 1.0389,
      "step": 6919
    },
    {
      "epoch": 0.8857033149878408,
      "grad_norm": 0.59765625,
      "learning_rate": 6.774817778735598e-05,
      "loss": 1.0675,
      "step": 6920
    },
    {
      "epoch": 0.885831306796365,
      "grad_norm": 0.6015625,
      "learning_rate": 6.759826319256312e-05,
      "loss": 1.0725,
      "step": 6921
    },
    {
      "epoch": 0.8859592986048893,
      "grad_norm": 0.64453125,
      "learning_rate": 6.744850884613673e-05,
      "loss": 1.4555,
      "step": 6922
    },
    {
      "epoch": 0.8860872904134135,
      "grad_norm": 0.7265625,
      "learning_rate": 6.729891477381467e-05,
      "loss": 1.3694,
      "step": 6923
    },
    {
      "epoch": 0.8862152822219378,
      "grad_norm": 0.5390625,
      "learning_rate": 6.714948100130669e-05,
      "loss": 0.8964,
      "step": 6924
    },
    {
      "epoch": 0.8863432740304621,
      "grad_norm": 0.7734375,
      "learning_rate": 6.700020755429558e-05,
      "loss": 1.2581,
      "step": 6925
    },
    {
      "epoch": 0.8864712658389863,
      "grad_norm": 0.72265625,
      "learning_rate": 6.685109445843673e-05,
      "loss": 2.0216,
      "step": 6926
    },
    {
      "epoch": 0.8865992576475106,
      "grad_norm": 0.7890625,
      "learning_rate": 6.670214173935751e-05,
      "loss": 1.502,
      "step": 6927
    },
    {
      "epoch": 0.8867272494560348,
      "grad_norm": 0.60546875,
      "learning_rate": 6.655334942265755e-05,
      "loss": 1.1758,
      "step": 6928
    },
    {
      "epoch": 0.8868552412645591,
      "grad_norm": 0.87109375,
      "learning_rate": 6.640471753390954e-05,
      "loss": 1.3899,
      "step": 6929
    },
    {
      "epoch": 0.8869832330730834,
      "grad_norm": 0.6875,
      "learning_rate": 6.62562460986581e-05,
      "loss": 1.5035,
      "step": 6930
    },
    {
      "epoch": 0.8871112248816075,
      "grad_norm": 0.51171875,
      "learning_rate": 6.610793514242074e-05,
      "loss": 0.5682,
      "step": 6931
    },
    {
      "epoch": 0.8872392166901318,
      "grad_norm": 0.69921875,
      "learning_rate": 6.595978469068675e-05,
      "loss": 1.6245,
      "step": 6932
    },
    {
      "epoch": 0.8873672084986561,
      "grad_norm": 0.53125,
      "learning_rate": 6.58117947689183e-05,
      "loss": 0.9848,
      "step": 6933
    },
    {
      "epoch": 0.8874952003071803,
      "grad_norm": 0.62109375,
      "learning_rate": 6.566396540254993e-05,
      "loss": 1.093,
      "step": 6934
    },
    {
      "epoch": 0.8876231921157046,
      "grad_norm": 0.5703125,
      "learning_rate": 6.551629661698843e-05,
      "loss": 1.1916,
      "step": 6935
    },
    {
      "epoch": 0.8877511839242288,
      "grad_norm": 0.55078125,
      "learning_rate": 6.536878843761319e-05,
      "loss": 1.0193,
      "step": 6936
    },
    {
      "epoch": 0.8878791757327531,
      "grad_norm": 0.546875,
      "learning_rate": 6.522144088977577e-05,
      "loss": 1.0361,
      "step": 6937
    },
    {
      "epoch": 0.8880071675412774,
      "grad_norm": 0.62890625,
      "learning_rate": 6.507425399880041e-05,
      "loss": 1.2596,
      "step": 6938
    },
    {
      "epoch": 0.8881351593498016,
      "grad_norm": 0.671875,
      "learning_rate": 6.492722778998328e-05,
      "loss": 1.1934,
      "step": 6939
    },
    {
      "epoch": 0.8882631511583259,
      "grad_norm": 0.6640625,
      "learning_rate": 6.478036228859363e-05,
      "loss": 1.657,
      "step": 6940
    },
    {
      "epoch": 0.8883911429668501,
      "grad_norm": 0.64453125,
      "learning_rate": 6.463365751987226e-05,
      "loss": 1.3214,
      "step": 6941
    },
    {
      "epoch": 0.8885191347753744,
      "grad_norm": 0.55078125,
      "learning_rate": 6.448711350903291e-05,
      "loss": 0.8892,
      "step": 6942
    },
    {
      "epoch": 0.8886471265838987,
      "grad_norm": 0.54296875,
      "learning_rate": 6.434073028126163e-05,
      "loss": 0.6864,
      "step": 6943
    },
    {
      "epoch": 0.8887751183924228,
      "grad_norm": 0.890625,
      "learning_rate": 6.419450786171666e-05,
      "loss": 1.9425,
      "step": 6944
    },
    {
      "epoch": 0.8889031102009471,
      "grad_norm": 0.5,
      "learning_rate": 6.404844627552885e-05,
      "loss": 0.8474,
      "step": 6945
    },
    {
      "epoch": 0.8890311020094714,
      "grad_norm": 0.609375,
      "learning_rate": 6.390254554780107e-05,
      "loss": 1.3625,
      "step": 6946
    },
    {
      "epoch": 0.8891590938179956,
      "grad_norm": 0.6484375,
      "learning_rate": 6.375680570360876e-05,
      "loss": 1.1235,
      "step": 6947
    },
    {
      "epoch": 0.8892870856265199,
      "grad_norm": 0.70703125,
      "learning_rate": 6.361122676799957e-05,
      "loss": 1.2492,
      "step": 6948
    },
    {
      "epoch": 0.8894150774350441,
      "grad_norm": 0.6171875,
      "learning_rate": 6.346580876599394e-05,
      "loss": 1.1474,
      "step": 6949
    },
    {
      "epoch": 0.8895430692435684,
      "grad_norm": 0.87109375,
      "learning_rate": 6.33205517225841e-05,
      "loss": 1.3313,
      "step": 6950
    },
    {
      "epoch": 0.8896710610520927,
      "grad_norm": 0.73046875,
      "learning_rate": 6.317545566273475e-05,
      "loss": 1.3677,
      "step": 6951
    },
    {
      "epoch": 0.8897990528606169,
      "grad_norm": 0.52734375,
      "learning_rate": 6.303052061138314e-05,
      "loss": 0.7653,
      "step": 6952
    },
    {
      "epoch": 0.8899270446691412,
      "grad_norm": 0.7734375,
      "learning_rate": 6.28857465934387e-05,
      "loss": 1.5897,
      "step": 6953
    },
    {
      "epoch": 0.8900550364776655,
      "grad_norm": 0.546875,
      "learning_rate": 6.27411336337832e-05,
      "loss": 1.2967,
      "step": 6954
    },
    {
      "epoch": 0.8901830282861897,
      "grad_norm": 0.73828125,
      "learning_rate": 6.259668175727063e-05,
      "loss": 1.4949,
      "step": 6955
    },
    {
      "epoch": 0.890311020094714,
      "grad_norm": 0.61328125,
      "learning_rate": 6.245239098872746e-05,
      "loss": 0.9481,
      "step": 6956
    },
    {
      "epoch": 0.8904390119032382,
      "grad_norm": 0.71875,
      "learning_rate": 6.230826135295231e-05,
      "loss": 1.4716,
      "step": 6957
    },
    {
      "epoch": 0.8905670037117625,
      "grad_norm": 0.6171875,
      "learning_rate": 6.216429287471626e-05,
      "loss": 1.3941,
      "step": 6958
    },
    {
      "epoch": 0.8906949955202867,
      "grad_norm": 0.7734375,
      "learning_rate": 6.202048557876272e-05,
      "loss": 1.6401,
      "step": 6959
    },
    {
      "epoch": 0.8908229873288109,
      "grad_norm": 0.65234375,
      "learning_rate": 6.187683948980715e-05,
      "loss": 1.2172,
      "step": 6960
    },
    {
      "epoch": 0.8909509791373352,
      "grad_norm": 0.6796875,
      "learning_rate": 6.17333546325376e-05,
      "loss": 1.3049,
      "step": 6961
    },
    {
      "epoch": 0.8910789709458594,
      "grad_norm": 0.578125,
      "learning_rate": 6.159003103161432e-05,
      "loss": 0.8279,
      "step": 6962
    },
    {
      "epoch": 0.8912069627543837,
      "grad_norm": 0.62109375,
      "learning_rate": 6.144686871166949e-05,
      "loss": 1.3914,
      "step": 6963
    },
    {
      "epoch": 0.891334954562908,
      "grad_norm": 0.7734375,
      "learning_rate": 6.130386769730811e-05,
      "loss": 1.5065,
      "step": 6964
    },
    {
      "epoch": 0.8914629463714322,
      "grad_norm": 0.73828125,
      "learning_rate": 6.11610280131072e-05,
      "loss": 1.2295,
      "step": 6965
    },
    {
      "epoch": 0.8915909381799565,
      "grad_norm": 0.89453125,
      "learning_rate": 6.101834968361608e-05,
      "loss": 1.2158,
      "step": 6966
    },
    {
      "epoch": 0.8917189299884808,
      "grad_norm": 0.78125,
      "learning_rate": 6.0875832733356394e-05,
      "loss": 1.4647,
      "step": 6967
    },
    {
      "epoch": 0.891846921797005,
      "grad_norm": 0.57421875,
      "learning_rate": 6.0733477186821764e-05,
      "loss": 1.0427,
      "step": 6968
    },
    {
      "epoch": 0.8919749136055293,
      "grad_norm": 0.640625,
      "learning_rate": 6.05912830684785e-05,
      "loss": 1.2329,
      "step": 6969
    },
    {
      "epoch": 0.8921029054140535,
      "grad_norm": 0.703125,
      "learning_rate": 6.044925040276483e-05,
      "loss": 1.4745,
      "step": 6970
    },
    {
      "epoch": 0.8922308972225778,
      "grad_norm": 0.6875,
      "learning_rate": 6.0307379214091684e-05,
      "loss": 1.4897,
      "step": 6971
    },
    {
      "epoch": 0.892358889031102,
      "grad_norm": 0.58984375,
      "learning_rate": 6.016566952684155e-05,
      "loss": 1.0351,
      "step": 6972
    },
    {
      "epoch": 0.8924868808396262,
      "grad_norm": 0.59375,
      "learning_rate": 6.002412136536972e-05,
      "loss": 0.8937,
      "step": 6973
    },
    {
      "epoch": 0.8926148726481505,
      "grad_norm": 0.62109375,
      "learning_rate": 5.9882734754003523e-05,
      "loss": 1.5508,
      "step": 6974
    },
    {
      "epoch": 0.8927428644566747,
      "grad_norm": 0.6484375,
      "learning_rate": 5.974150971704251e-05,
      "loss": 1.0885,
      "step": 6975
    },
    {
      "epoch": 0.892870856265199,
      "grad_norm": 0.671875,
      "learning_rate": 5.96004462787586e-05,
      "loss": 1.5074,
      "step": 6976
    },
    {
      "epoch": 0.8929988480737233,
      "grad_norm": 0.7421875,
      "learning_rate": 5.945954446339574e-05,
      "loss": 1.4553,
      "step": 6977
    },
    {
      "epoch": 0.8931268398822475,
      "grad_norm": 0.66796875,
      "learning_rate": 5.9318804295170114e-05,
      "loss": 1.755,
      "step": 6978
    },
    {
      "epoch": 0.8932548316907718,
      "grad_norm": 0.6640625,
      "learning_rate": 5.917822579827037e-05,
      "loss": 1.1967,
      "step": 6979
    },
    {
      "epoch": 0.8933828234992961,
      "grad_norm": 0.68359375,
      "learning_rate": 5.9037808996857064e-05,
      "loss": 1.1038,
      "step": 6980
    },
    {
      "epoch": 0.8935108153078203,
      "grad_norm": 0.609375,
      "learning_rate": 5.889755391506313e-05,
      "loss": 1.0043,
      "step": 6981
    },
    {
      "epoch": 0.8936388071163446,
      "grad_norm": 0.96484375,
      "learning_rate": 5.875746057699383e-05,
      "loss": 1.336,
      "step": 6982
    },
    {
      "epoch": 0.8937667989248688,
      "grad_norm": 0.6328125,
      "learning_rate": 5.861752900672623e-05,
      "loss": 1.3724,
      "step": 6983
    },
    {
      "epoch": 0.8938947907333931,
      "grad_norm": 0.7421875,
      "learning_rate": 5.847775922831011e-05,
      "loss": 1.2278,
      "step": 6984
    },
    {
      "epoch": 0.8940227825419174,
      "grad_norm": 0.64453125,
      "learning_rate": 5.8338151265767134e-05,
      "loss": 0.889,
      "step": 6985
    },
    {
      "epoch": 0.8941507743504415,
      "grad_norm": 0.87109375,
      "learning_rate": 5.819870514309089e-05,
      "loss": 2.2294,
      "step": 6986
    },
    {
      "epoch": 0.8942787661589658,
      "grad_norm": 0.6171875,
      "learning_rate": 5.805942088424776e-05,
      "loss": 1.078,
      "step": 6987
    },
    {
      "epoch": 0.89440675796749,
      "grad_norm": 0.76171875,
      "learning_rate": 5.7920298513175815e-05,
      "loss": 1.723,
      "step": 6988
    },
    {
      "epoch": 0.8945347497760143,
      "grad_norm": 0.546875,
      "learning_rate": 5.7781338053785826e-05,
      "loss": 0.9356,
      "step": 6989
    },
    {
      "epoch": 0.8946627415845386,
      "grad_norm": 0.7890625,
      "learning_rate": 5.76425395299599e-05,
      "loss": 1.3908,
      "step": 6990
    },
    {
      "epoch": 0.8947907333930628,
      "grad_norm": 0.69921875,
      "learning_rate": 5.750390296555319e-05,
      "loss": 1.3265,
      "step": 6991
    },
    {
      "epoch": 0.8949187252015871,
      "grad_norm": 0.6953125,
      "learning_rate": 5.736542838439252e-05,
      "loss": 1.7221,
      "step": 6992
    },
    {
      "epoch": 0.8950467170101114,
      "grad_norm": 0.734375,
      "learning_rate": 5.7227115810276956e-05,
      "loss": 1.2575,
      "step": 6993
    },
    {
      "epoch": 0.8951747088186356,
      "grad_norm": 0.671875,
      "learning_rate": 5.708896526697782e-05,
      "loss": 1.4329,
      "step": 6994
    },
    {
      "epoch": 0.8953027006271599,
      "grad_norm": 0.6875,
      "learning_rate": 5.695097677823846e-05,
      "loss": 1.0308,
      "step": 6995
    },
    {
      "epoch": 0.8954306924356841,
      "grad_norm": 0.7265625,
      "learning_rate": 5.681315036777457e-05,
      "loss": 1.1795,
      "step": 6996
    },
    {
      "epoch": 0.8955586842442084,
      "grad_norm": 0.83203125,
      "learning_rate": 5.667548605927364e-05,
      "loss": 1.2797,
      "step": 6997
    },
    {
      "epoch": 0.8956866760527327,
      "grad_norm": 0.63671875,
      "learning_rate": 5.653798387639586e-05,
      "loss": 1.1739,
      "step": 6998
    },
    {
      "epoch": 0.8958146678612569,
      "grad_norm": 0.578125,
      "learning_rate": 5.640064384277288e-05,
      "loss": 0.9278,
      "step": 6999
    },
    {
      "epoch": 0.8959426596697811,
      "grad_norm": 0.76171875,
      "learning_rate": 5.6263465982008933e-05,
      "loss": 2.1875,
      "step": 7000
    },
    {
      "epoch": 0.8960706514783053,
      "grad_norm": 0.56640625,
      "learning_rate": 5.6126450317680265e-05,
      "loss": 1.1753,
      "step": 7001
    },
    {
      "epoch": 0.8961986432868296,
      "grad_norm": 0.56640625,
      "learning_rate": 5.5989596873335156e-05,
      "loss": 0.9284,
      "step": 7002
    },
    {
      "epoch": 0.8963266350953539,
      "grad_norm": 0.73046875,
      "learning_rate": 5.5852905672494235e-05,
      "loss": 2.0911,
      "step": 7003
    },
    {
      "epoch": 0.8964546269038781,
      "grad_norm": 0.58984375,
      "learning_rate": 5.5716376738649933e-05,
      "loss": 1.0411,
      "step": 7004
    },
    {
      "epoch": 0.8965826187124024,
      "grad_norm": 0.703125,
      "learning_rate": 5.5580010095267145e-05,
      "loss": 1.6312,
      "step": 7005
    },
    {
      "epoch": 0.8967106105209267,
      "grad_norm": 0.6875,
      "learning_rate": 5.5443805765782565e-05,
      "loss": 1.6877,
      "step": 7006
    },
    {
      "epoch": 0.8968386023294509,
      "grad_norm": 0.75390625,
      "learning_rate": 5.530776377360536e-05,
      "loss": 1.4825,
      "step": 7007
    },
    {
      "epoch": 0.8969665941379752,
      "grad_norm": 0.58984375,
      "learning_rate": 5.517188414211616e-05,
      "loss": 1.0574,
      "step": 7008
    },
    {
      "epoch": 0.8970945859464994,
      "grad_norm": 0.6953125,
      "learning_rate": 5.5036166894668285e-05,
      "loss": 1.708,
      "step": 7009
    },
    {
      "epoch": 0.8972225777550237,
      "grad_norm": 0.57421875,
      "learning_rate": 5.4900612054586964e-05,
      "loss": 0.6501,
      "step": 7010
    },
    {
      "epoch": 0.897350569563548,
      "grad_norm": 0.59765625,
      "learning_rate": 5.476521964516956e-05,
      "loss": 0.7071,
      "step": 7011
    },
    {
      "epoch": 0.8974785613720722,
      "grad_norm": 0.73828125,
      "learning_rate": 5.462998968968525e-05,
      "loss": 1.2115,
      "step": 7012
    },
    {
      "epoch": 0.8976065531805965,
      "grad_norm": 0.71484375,
      "learning_rate": 5.449492221137564e-05,
      "loss": 1.5498,
      "step": 7013
    },
    {
      "epoch": 0.8977345449891206,
      "grad_norm": 0.69921875,
      "learning_rate": 5.4360017233454296e-05,
      "loss": 1.2326,
      "step": 7014
    },
    {
      "epoch": 0.8978625367976449,
      "grad_norm": 0.69921875,
      "learning_rate": 5.422527477910666e-05,
      "loss": 2.2636,
      "step": 7015
    },
    {
      "epoch": 0.8979905286061692,
      "grad_norm": 0.80859375,
      "learning_rate": 5.409069487149054e-05,
      "loss": 1.2158,
      "step": 7016
    },
    {
      "epoch": 0.8981185204146934,
      "grad_norm": 0.65625,
      "learning_rate": 5.395627753373567e-05,
      "loss": 0.8902,
      "step": 7017
    },
    {
      "epoch": 0.8982465122232177,
      "grad_norm": 0.5859375,
      "learning_rate": 5.382202278894388e-05,
      "loss": 0.8741,
      "step": 7018
    },
    {
      "epoch": 0.898374504031742,
      "grad_norm": 0.69140625,
      "learning_rate": 5.368793066018896e-05,
      "loss": 1.3564,
      "step": 7019
    },
    {
      "epoch": 0.8985024958402662,
      "grad_norm": 1.1484375,
      "learning_rate": 5.35540011705169e-05,
      "loss": 1.0164,
      "step": 7020
    },
    {
      "epoch": 0.8986304876487905,
      "grad_norm": 0.71875,
      "learning_rate": 5.3420234342945514e-05,
      "loss": 1.6733,
      "step": 7021
    },
    {
      "epoch": 0.8987584794573147,
      "grad_norm": 0.63671875,
      "learning_rate": 5.328663020046487e-05,
      "loss": 1.175,
      "step": 7022
    },
    {
      "epoch": 0.898886471265839,
      "grad_norm": 0.54296875,
      "learning_rate": 5.315318876603714e-05,
      "loss": 0.989,
      "step": 7023
    },
    {
      "epoch": 0.8990144630743633,
      "grad_norm": 0.61328125,
      "learning_rate": 5.30199100625961e-05,
      "loss": 1.0515,
      "step": 7024
    },
    {
      "epoch": 0.8991424548828875,
      "grad_norm": 0.76171875,
      "learning_rate": 5.2886794113048195e-05,
      "loss": 1.4224,
      "step": 7025
    },
    {
      "epoch": 0.8992704466914118,
      "grad_norm": 0.63671875,
      "learning_rate": 5.275384094027125e-05,
      "loss": 1.1901,
      "step": 7026
    },
    {
      "epoch": 0.8993984384999361,
      "grad_norm": 0.62109375,
      "learning_rate": 5.2621050567115744e-05,
      "loss": 1.2595,
      "step": 7027
    },
    {
      "epoch": 0.8995264303084602,
      "grad_norm": 0.57421875,
      "learning_rate": 5.248842301640355e-05,
      "loss": 0.7152,
      "step": 7028
    },
    {
      "epoch": 0.8996544221169845,
      "grad_norm": 0.69921875,
      "learning_rate": 5.235595831092921e-05,
      "loss": 2.0291,
      "step": 7029
    },
    {
      "epoch": 0.8997824139255087,
      "grad_norm": 0.78125,
      "learning_rate": 5.222365647345861e-05,
      "loss": 1.2634,
      "step": 7030
    },
    {
      "epoch": 0.899910405734033,
      "grad_norm": 0.7890625,
      "learning_rate": 5.209151752673003e-05,
      "loss": 1.2086,
      "step": 7031
    },
    {
      "epoch": 0.9000383975425573,
      "grad_norm": 0.486328125,
      "learning_rate": 5.195954149345383e-05,
      "loss": 0.8486,
      "step": 7032
    },
    {
      "epoch": 0.9001663893510815,
      "grad_norm": 0.6328125,
      "learning_rate": 5.182772839631222e-05,
      "loss": 1.3288,
      "step": 7033
    },
    {
      "epoch": 0.9002943811596058,
      "grad_norm": 0.6328125,
      "learning_rate": 5.169607825795919e-05,
      "loss": 1.0289,
      "step": 7034
    },
    {
      "epoch": 0.90042237296813,
      "grad_norm": 0.85546875,
      "learning_rate": 5.156459110102119e-05,
      "loss": 1.2715,
      "step": 7035
    },
    {
      "epoch": 0.9005503647766543,
      "grad_norm": 0.462890625,
      "learning_rate": 5.143326694809636e-05,
      "loss": 0.5315,
      "step": 7036
    },
    {
      "epoch": 0.9006783565851786,
      "grad_norm": 0.78125,
      "learning_rate": 5.1302105821754765e-05,
      "loss": 1.9082,
      "step": 7037
    },
    {
      "epoch": 0.9008063483937028,
      "grad_norm": 0.482421875,
      "learning_rate": 5.117110774453881e-05,
      "loss": 0.7989,
      "step": 7038
    },
    {
      "epoch": 0.9009343402022271,
      "grad_norm": 0.578125,
      "learning_rate": 5.104027273896239e-05,
      "loss": 0.9675,
      "step": 7039
    },
    {
      "epoch": 0.9010623320107514,
      "grad_norm": 0.77734375,
      "learning_rate": 5.090960082751184e-05,
      "loss": 1.6182,
      "step": 7040
    },
    {
      "epoch": 0.9011903238192756,
      "grad_norm": 0.6484375,
      "learning_rate": 5.077909203264508e-05,
      "loss": 1.2184,
      "step": 7041
    },
    {
      "epoch": 0.9013183156277998,
      "grad_norm": 0.6640625,
      "learning_rate": 5.0648746376792396e-05,
      "loss": 1.8113,
      "step": 7042
    },
    {
      "epoch": 0.901446307436324,
      "grad_norm": 0.6484375,
      "learning_rate": 5.051856388235543e-05,
      "loss": 1.2986,
      "step": 7043
    },
    {
      "epoch": 0.9015742992448483,
      "grad_norm": 0.671875,
      "learning_rate": 5.0388544571708494e-05,
      "loss": 1.4754,
      "step": 7044
    },
    {
      "epoch": 0.9017022910533726,
      "grad_norm": 0.66796875,
      "learning_rate": 5.025868846719728e-05,
      "loss": 1.117,
      "step": 7045
    },
    {
      "epoch": 0.9018302828618968,
      "grad_norm": 0.77734375,
      "learning_rate": 5.012899559114004e-05,
      "loss": 2.2323,
      "step": 7046
    },
    {
      "epoch": 0.9019582746704211,
      "grad_norm": 0.83203125,
      "learning_rate": 4.999946596582594e-05,
      "loss": 1.4446,
      "step": 7047
    },
    {
      "epoch": 0.9020862664789453,
      "grad_norm": 0.58203125,
      "learning_rate": 4.987009961351729e-05,
      "loss": 1.1018,
      "step": 7048
    },
    {
      "epoch": 0.9022142582874696,
      "grad_norm": 0.77734375,
      "learning_rate": 4.974089655644776e-05,
      "loss": 1.6634,
      "step": 7049
    },
    {
      "epoch": 0.9023422500959939,
      "grad_norm": 0.78515625,
      "learning_rate": 4.961185681682279e-05,
      "loss": 2.1095,
      "step": 7050
    },
    {
      "epoch": 0.9024702419045181,
      "grad_norm": 0.52734375,
      "learning_rate": 4.94829804168202e-05,
      "loss": 0.5936,
      "step": 7051
    },
    {
      "epoch": 0.9025982337130424,
      "grad_norm": 1.0078125,
      "learning_rate": 4.935426737858917e-05,
      "loss": 2.0331,
      "step": 7052
    },
    {
      "epoch": 0.9027262255215667,
      "grad_norm": 0.58984375,
      "learning_rate": 4.9225717724251326e-05,
      "loss": 0.9377,
      "step": 7053
    },
    {
      "epoch": 0.9028542173300909,
      "grad_norm": 0.75,
      "learning_rate": 4.9097331475900006e-05,
      "loss": 1.5176,
      "step": 7054
    },
    {
      "epoch": 0.9029822091386152,
      "grad_norm": 0.62109375,
      "learning_rate": 4.8969108655600445e-05,
      "loss": 1.2552,
      "step": 7055
    },
    {
      "epoch": 0.9031102009471393,
      "grad_norm": 0.63671875,
      "learning_rate": 4.884104928538979e-05,
      "loss": 0.9139,
      "step": 7056
    },
    {
      "epoch": 0.9032381927556636,
      "grad_norm": 0.578125,
      "learning_rate": 4.871315338727711e-05,
      "loss": 1.2179,
      "step": 7057
    },
    {
      "epoch": 0.9033661845641879,
      "grad_norm": 0.71484375,
      "learning_rate": 4.858542098324348e-05,
      "loss": 1.3109,
      "step": 7058
    },
    {
      "epoch": 0.9034941763727121,
      "grad_norm": 0.69140625,
      "learning_rate": 4.845785209524156e-05,
      "loss": 1.3907,
      "step": 7059
    },
    {
      "epoch": 0.9036221681812364,
      "grad_norm": 0.54296875,
      "learning_rate": 4.8330446745196486e-05,
      "loss": 1.2605,
      "step": 7060
    },
    {
      "epoch": 0.9037501599897606,
      "grad_norm": 0.6953125,
      "learning_rate": 4.8203204955004634e-05,
      "loss": 1.7747,
      "step": 7061
    },
    {
      "epoch": 0.9038781517982849,
      "grad_norm": 0.578125,
      "learning_rate": 4.807612674653472e-05,
      "loss": 1.0932,
      "step": 7062
    },
    {
      "epoch": 0.9040061436068092,
      "grad_norm": 0.7421875,
      "learning_rate": 4.794921214162717e-05,
      "loss": 1.9243,
      "step": 7063
    },
    {
      "epoch": 0.9041341354153334,
      "grad_norm": 0.9921875,
      "learning_rate": 4.782246116209443e-05,
      "loss": 1.9346,
      "step": 7064
    },
    {
      "epoch": 0.9042621272238577,
      "grad_norm": 0.71484375,
      "learning_rate": 4.7695873829720405e-05,
      "loss": 1.2539,
      "step": 7065
    },
    {
      "epoch": 0.904390119032382,
      "grad_norm": 0.609375,
      "learning_rate": 4.7569450166261355e-05,
      "loss": 0.9817,
      "step": 7066
    },
    {
      "epoch": 0.9045181108409062,
      "grad_norm": 0.62890625,
      "learning_rate": 4.744319019344534e-05,
      "loss": 1.3898,
      "step": 7067
    },
    {
      "epoch": 0.9046461026494305,
      "grad_norm": 0.70703125,
      "learning_rate": 4.731709393297212e-05,
      "loss": 1.4314,
      "step": 7068
    },
    {
      "epoch": 0.9047740944579546,
      "grad_norm": 0.68359375,
      "learning_rate": 4.719116140651336e-05,
      "loss": 1.1836,
      "step": 7069
    },
    {
      "epoch": 0.9049020862664789,
      "grad_norm": 0.65234375,
      "learning_rate": 4.706539263571241e-05,
      "loss": 1.3145,
      "step": 7070
    },
    {
      "epoch": 0.9050300780750032,
      "grad_norm": 0.61328125,
      "learning_rate": 4.693978764218499e-05,
      "loss": 1.1618,
      "step": 7071
    },
    {
      "epoch": 0.9051580698835274,
      "grad_norm": 0.5078125,
      "learning_rate": 4.681434644751825e-05,
      "loss": 0.8274,
      "step": 7072
    },
    {
      "epoch": 0.9052860616920517,
      "grad_norm": 0.625,
      "learning_rate": 4.6689069073271415e-05,
      "loss": 1.8399,
      "step": 7073
    },
    {
      "epoch": 0.9054140535005759,
      "grad_norm": 0.7109375,
      "learning_rate": 4.656395554097514e-05,
      "loss": 1.227,
      "step": 7074
    },
    {
      "epoch": 0.9055420453091002,
      "grad_norm": 0.5625,
      "learning_rate": 4.643900587213246e-05,
      "loss": 1.4569,
      "step": 7075
    },
    {
      "epoch": 0.9056700371176245,
      "grad_norm": 0.65234375,
      "learning_rate": 4.631422008821784e-05,
      "loss": 1.3689,
      "step": 7076
    },
    {
      "epoch": 0.9057980289261487,
      "grad_norm": 0.6484375,
      "learning_rate": 4.618959821067792e-05,
      "loss": 1.3101,
      "step": 7077
    },
    {
      "epoch": 0.905926020734673,
      "grad_norm": 0.59765625,
      "learning_rate": 4.606514026093078e-05,
      "loss": 1.1557,
      "step": 7078
    },
    {
      "epoch": 0.9060540125431973,
      "grad_norm": 0.75390625,
      "learning_rate": 4.5940846260366766e-05,
      "loss": 1.8513,
      "step": 7079
    },
    {
      "epoch": 0.9061820043517215,
      "grad_norm": 0.55859375,
      "learning_rate": 4.5816716230347556e-05,
      "loss": 1.1859,
      "step": 7080
    },
    {
      "epoch": 0.9063099961602458,
      "grad_norm": 0.8671875,
      "learning_rate": 4.569275019220709e-05,
      "loss": 1.654,
      "step": 7081
    },
    {
      "epoch": 0.90643798796877,
      "grad_norm": 0.56640625,
      "learning_rate": 4.556894816725099e-05,
      "loss": 0.7372,
      "step": 7082
    },
    {
      "epoch": 0.9065659797772943,
      "grad_norm": 0.68359375,
      "learning_rate": 4.544531017675646e-05,
      "loss": 1.0335,
      "step": 7083
    },
    {
      "epoch": 0.9066939715858185,
      "grad_norm": 0.61328125,
      "learning_rate": 4.532183624197283e-05,
      "loss": 1.2064,
      "step": 7084
    },
    {
      "epoch": 0.9068219633943427,
      "grad_norm": 0.5546875,
      "learning_rate": 4.519852638412103e-05,
      "loss": 1.0515,
      "step": 7085
    },
    {
      "epoch": 0.906949955202867,
      "grad_norm": 0.69921875,
      "learning_rate": 4.507538062439409e-05,
      "loss": 1.5891,
      "step": 7086
    },
    {
      "epoch": 0.9070779470113912,
      "grad_norm": 0.73828125,
      "learning_rate": 4.49523989839562e-05,
      "loss": 1.8384,
      "step": 7087
    },
    {
      "epoch": 0.9072059388199155,
      "grad_norm": 0.6953125,
      "learning_rate": 4.4829581483944004e-05,
      "loss": 1.6506,
      "step": 7088
    },
    {
      "epoch": 0.9073339306284398,
      "grad_norm": 0.69140625,
      "learning_rate": 4.470692814546573e-05,
      "loss": 1.0895,
      "step": 7089
    },
    {
      "epoch": 0.907461922436964,
      "grad_norm": 0.62890625,
      "learning_rate": 4.4584438989601186e-05,
      "loss": 1.0166,
      "step": 7090
    },
    {
      "epoch": 0.9075899142454883,
      "grad_norm": 0.67578125,
      "learning_rate": 4.44621140374023e-05,
      "loss": 1.4948,
      "step": 7091
    },
    {
      "epoch": 0.9077179060540126,
      "grad_norm": 0.71875,
      "learning_rate": 4.433995330989216e-05,
      "loss": 1.5399,
      "step": 7092
    },
    {
      "epoch": 0.9078458978625368,
      "grad_norm": 0.6953125,
      "learning_rate": 4.421795682806662e-05,
      "loss": 1.7591,
      "step": 7093
    },
    {
      "epoch": 0.9079738896710611,
      "grad_norm": 0.53125,
      "learning_rate": 4.409612461289247e-05,
      "loss": 1.1147,
      "step": 7094
    },
    {
      "epoch": 0.9081018814795853,
      "grad_norm": 0.6171875,
      "learning_rate": 4.397445668530886e-05,
      "loss": 1.1263,
      "step": 7095
    },
    {
      "epoch": 0.9082298732881096,
      "grad_norm": 0.6015625,
      "learning_rate": 4.385295306622594e-05,
      "loss": 0.877,
      "step": 7096
    },
    {
      "epoch": 0.9083578650966339,
      "grad_norm": 0.78515625,
      "learning_rate": 4.373161377652624e-05,
      "loss": 0.9216,
      "step": 7097
    },
    {
      "epoch": 0.908485856905158,
      "grad_norm": 0.6640625,
      "learning_rate": 4.3610438837064056e-05,
      "loss": 1.7136,
      "step": 7098
    },
    {
      "epoch": 0.9086138487136823,
      "grad_norm": 0.5859375,
      "learning_rate": 4.34894282686652e-05,
      "loss": 1.0171,
      "step": 7099
    },
    {
      "epoch": 0.9087418405222066,
      "grad_norm": 0.7265625,
      "learning_rate": 4.3368582092127126e-05,
      "loss": 1.5052,
      "step": 7100
    },
    {
      "epoch": 0.9088698323307308,
      "grad_norm": 0.6484375,
      "learning_rate": 4.324790032821935e-05,
      "loss": 1.5504,
      "step": 7101
    },
    {
      "epoch": 0.9089978241392551,
      "grad_norm": 0.7265625,
      "learning_rate": 4.312738299768293e-05,
      "loss": 1.3666,
      "step": 7102
    },
    {
      "epoch": 0.9091258159477793,
      "grad_norm": 0.703125,
      "learning_rate": 4.300703012123075e-05,
      "loss": 1.3106,
      "step": 7103
    },
    {
      "epoch": 0.9092538077563036,
      "grad_norm": 0.765625,
      "learning_rate": 4.2886841719547485e-05,
      "loss": 1.3246,
      "step": 7104
    },
    {
      "epoch": 0.9093817995648279,
      "grad_norm": 0.59375,
      "learning_rate": 4.2766817813289396e-05,
      "loss": 1.3433,
      "step": 7105
    },
    {
      "epoch": 0.9095097913733521,
      "grad_norm": 0.63671875,
      "learning_rate": 4.2646958423084526e-05,
      "loss": 1.127,
      "step": 7106
    },
    {
      "epoch": 0.9096377831818764,
      "grad_norm": 0.66015625,
      "learning_rate": 4.252726356953274e-05,
      "loss": 1.4461,
      "step": 7107
    },
    {
      "epoch": 0.9097657749904006,
      "grad_norm": 1.1171875,
      "learning_rate": 4.240773327320557e-05,
      "loss": 1.9467,
      "step": 7108
    },
    {
      "epoch": 0.9098937667989249,
      "grad_norm": 0.58203125,
      "learning_rate": 4.228836755464593e-05,
      "loss": 0.6659,
      "step": 7109
    },
    {
      "epoch": 0.9100217586074492,
      "grad_norm": 0.60546875,
      "learning_rate": 4.2169166434369076e-05,
      "loss": 1.1636,
      "step": 7110
    },
    {
      "epoch": 0.9101497504159733,
      "grad_norm": 0.52734375,
      "learning_rate": 4.205012993286139e-05,
      "loss": 0.9667,
      "step": 7111
    },
    {
      "epoch": 0.9102777422244976,
      "grad_norm": 0.6328125,
      "learning_rate": 4.1931258070581515e-05,
      "loss": 1.0606,
      "step": 7112
    },
    {
      "epoch": 0.9104057340330219,
      "grad_norm": 0.6796875,
      "learning_rate": 4.181255086795954e-05,
      "loss": 1.3421,
      "step": 7113
    },
    {
      "epoch": 0.9105337258415461,
      "grad_norm": 0.65625,
      "learning_rate": 4.169400834539683e-05,
      "loss": 1.3568,
      "step": 7114
    },
    {
      "epoch": 0.9106617176500704,
      "grad_norm": 0.72265625,
      "learning_rate": 4.157563052326696e-05,
      "loss": 1.5347,
      "step": 7115
    },
    {
      "epoch": 0.9107897094585946,
      "grad_norm": 0.77734375,
      "learning_rate": 4.145741742191533e-05,
      "loss": 1.4018,
      "step": 7116
    },
    {
      "epoch": 0.9109177012671189,
      "grad_norm": 0.78515625,
      "learning_rate": 4.1339369061658694e-05,
      "loss": 1.422,
      "step": 7117
    },
    {
      "epoch": 0.9110456930756432,
      "grad_norm": 0.62890625,
      "learning_rate": 4.122148546278548e-05,
      "loss": 1.0924,
      "step": 7118
    },
    {
      "epoch": 0.9111736848841674,
      "grad_norm": 0.640625,
      "learning_rate": 4.1103766645555927e-05,
      "loss": 1.0768,
      "step": 7119
    },
    {
      "epoch": 0.9113016766926917,
      "grad_norm": 0.60546875,
      "learning_rate": 4.098621263020197e-05,
      "loss": 1.5358,
      "step": 7120
    },
    {
      "epoch": 0.9114296685012159,
      "grad_norm": 0.59765625,
      "learning_rate": 4.0868823436927106e-05,
      "loss": 0.8316,
      "step": 7121
    },
    {
      "epoch": 0.9115576603097402,
      "grad_norm": 0.6484375,
      "learning_rate": 4.0751599085906646e-05,
      "loss": 1.0937,
      "step": 7122
    },
    {
      "epoch": 0.9116856521182645,
      "grad_norm": 0.5546875,
      "learning_rate": 4.063453959728747e-05,
      "loss": 1.0212,
      "step": 7123
    },
    {
      "epoch": 0.9118136439267887,
      "grad_norm": 0.5625,
      "learning_rate": 4.051764499118804e-05,
      "loss": 1.0111,
      "step": 7124
    },
    {
      "epoch": 0.911941635735313,
      "grad_norm": 0.66796875,
      "learning_rate": 4.0400915287698734e-05,
      "loss": 1.2817,
      "step": 7125
    },
    {
      "epoch": 0.9120696275438372,
      "grad_norm": 0.6015625,
      "learning_rate": 4.0284350506881395e-05,
      "loss": 1.0092,
      "step": 7126
    },
    {
      "epoch": 0.9121976193523614,
      "grad_norm": 0.625,
      "learning_rate": 4.0167950668769546e-05,
      "loss": 1.295,
      "step": 7127
    },
    {
      "epoch": 0.9123256111608857,
      "grad_norm": 0.78515625,
      "learning_rate": 4.005171579336853e-05,
      "loss": 0.9912,
      "step": 7128
    },
    {
      "epoch": 0.9124536029694099,
      "grad_norm": 0.70703125,
      "learning_rate": 3.9935645900654904e-05,
      "loss": 1.0945,
      "step": 7129
    },
    {
      "epoch": 0.9125815947779342,
      "grad_norm": 1.203125,
      "learning_rate": 3.981974101057761e-05,
      "loss": 1.066,
      "step": 7130
    },
    {
      "epoch": 0.9127095865864585,
      "grad_norm": 0.640625,
      "learning_rate": 3.9704001143056276e-05,
      "loss": 1.3389,
      "step": 7131
    },
    {
      "epoch": 0.9128375783949827,
      "grad_norm": 0.67578125,
      "learning_rate": 3.9588426317982985e-05,
      "loss": 1.5236,
      "step": 7132
    },
    {
      "epoch": 0.912965570203507,
      "grad_norm": 0.5703125,
      "learning_rate": 3.947301655522106e-05,
      "loss": 0.9985,
      "step": 7133
    },
    {
      "epoch": 0.9130935620120312,
      "grad_norm": 0.7890625,
      "learning_rate": 3.935777187460565e-05,
      "loss": 1.6553,
      "step": 7134
    },
    {
      "epoch": 0.9132215538205555,
      "grad_norm": 0.7734375,
      "learning_rate": 3.924269229594335e-05,
      "loss": 1.2191,
      "step": 7135
    },
    {
      "epoch": 0.9133495456290798,
      "grad_norm": 0.62890625,
      "learning_rate": 3.912777783901234e-05,
      "loss": 1.0344,
      "step": 7136
    },
    {
      "epoch": 0.913477537437604,
      "grad_norm": 0.6484375,
      "learning_rate": 3.901302852356259e-05,
      "loss": 1.1649,
      "step": 7137
    },
    {
      "epoch": 0.9136055292461283,
      "grad_norm": 0.57421875,
      "learning_rate": 3.8898444369315554e-05,
      "loss": 0.9198,
      "step": 7138
    },
    {
      "epoch": 0.9137335210546526,
      "grad_norm": 0.828125,
      "learning_rate": 3.878402539596482e-05,
      "loss": 1.6909,
      "step": 7139
    },
    {
      "epoch": 0.9138615128631767,
      "grad_norm": 0.91015625,
      "learning_rate": 3.866977162317464e-05,
      "loss": 1.7527,
      "step": 7140
    },
    {
      "epoch": 0.913989504671701,
      "grad_norm": 0.490234375,
      "learning_rate": 3.855568307058155e-05,
      "loss": 0.8144,
      "step": 7141
    },
    {
      "epoch": 0.9141174964802252,
      "grad_norm": 0.640625,
      "learning_rate": 3.8441759757793516e-05,
      "loss": 1.291,
      "step": 7142
    },
    {
      "epoch": 0.9142454882887495,
      "grad_norm": 0.62109375,
      "learning_rate": 3.8328001704390104e-05,
      "loss": 1.0755,
      "step": 7143
    },
    {
      "epoch": 0.9143734800972738,
      "grad_norm": 0.6796875,
      "learning_rate": 3.821440892992267e-05,
      "loss": 1.5769,
      "step": 7144
    },
    {
      "epoch": 0.914501471905798,
      "grad_norm": 0.63671875,
      "learning_rate": 3.8100981453913716e-05,
      "loss": 1.4656,
      "step": 7145
    },
    {
      "epoch": 0.9146294637143223,
      "grad_norm": 0.66796875,
      "learning_rate": 3.798771929585754e-05,
      "loss": 0.9973,
      "step": 7146
    },
    {
      "epoch": 0.9147574555228465,
      "grad_norm": 0.6640625,
      "learning_rate": 3.787462247522033e-05,
      "loss": 1.833,
      "step": 7147
    },
    {
      "epoch": 0.9148854473313708,
      "grad_norm": 0.78515625,
      "learning_rate": 3.776169101143945e-05,
      "loss": 1.9395,
      "step": 7148
    },
    {
      "epoch": 0.9150134391398951,
      "grad_norm": 0.58984375,
      "learning_rate": 3.7648924923924135e-05,
      "loss": 1.0631,
      "step": 7149
    },
    {
      "epoch": 0.9151414309484193,
      "grad_norm": 0.53125,
      "learning_rate": 3.753632423205489e-05,
      "loss": 0.8048,
      "step": 7150
    },
    {
      "epoch": 0.9152694227569436,
      "grad_norm": 0.6484375,
      "learning_rate": 3.742388895518411e-05,
      "loss": 0.9815,
      "step": 7151
    },
    {
      "epoch": 0.9153974145654679,
      "grad_norm": 0.58984375,
      "learning_rate": 3.7311619112635584e-05,
      "loss": 0.9152,
      "step": 7152
    },
    {
      "epoch": 0.915525406373992,
      "grad_norm": 0.58984375,
      "learning_rate": 3.7199514723704754e-05,
      "loss": 1.2145,
      "step": 7153
    },
    {
      "epoch": 0.9156533981825163,
      "grad_norm": 0.6328125,
      "learning_rate": 3.708757580765842e-05,
      "loss": 0.8519,
      "step": 7154
    },
    {
      "epoch": 0.9157813899910405,
      "grad_norm": 0.62890625,
      "learning_rate": 3.6975802383735325e-05,
      "loss": 0.9585,
      "step": 7155
    },
    {
      "epoch": 0.9159093817995648,
      "grad_norm": 0.60546875,
      "learning_rate": 3.686419447114531e-05,
      "loss": 1.2383,
      "step": 7156
    },
    {
      "epoch": 0.9160373736080891,
      "grad_norm": 0.63671875,
      "learning_rate": 3.675275208907025e-05,
      "loss": 1.0011,
      "step": 7157
    },
    {
      "epoch": 0.9161653654166133,
      "grad_norm": 0.62890625,
      "learning_rate": 3.664147525666306e-05,
      "loss": 1.2651,
      "step": 7158
    },
    {
      "epoch": 0.9162933572251376,
      "grad_norm": 0.61328125,
      "learning_rate": 3.653036399304865e-05,
      "loss": 1.1003,
      "step": 7159
    },
    {
      "epoch": 0.9164213490336618,
      "grad_norm": 0.61328125,
      "learning_rate": 3.64194183173232e-05,
      "loss": 1.2451,
      "step": 7160
    },
    {
      "epoch": 0.9165493408421861,
      "grad_norm": 0.57421875,
      "learning_rate": 3.630863824855468e-05,
      "loss": 1.0277,
      "step": 7161
    },
    {
      "epoch": 0.9166773326507104,
      "grad_norm": 0.79296875,
      "learning_rate": 3.61980238057823e-05,
      "loss": 1.8214,
      "step": 7162
    },
    {
      "epoch": 0.9168053244592346,
      "grad_norm": 0.703125,
      "learning_rate": 3.608757500801696e-05,
      "loss": 1.7388,
      "step": 7163
    },
    {
      "epoch": 0.9169333162677589,
      "grad_norm": 0.65625,
      "learning_rate": 3.5977291874241035e-05,
      "loss": 1.2028,
      "step": 7164
    },
    {
      "epoch": 0.9170613080762832,
      "grad_norm": 0.69921875,
      "learning_rate": 3.58671744234087e-05,
      "loss": 0.8566,
      "step": 7165
    },
    {
      "epoch": 0.9171892998848074,
      "grad_norm": 0.7109375,
      "learning_rate": 3.575722267444526e-05,
      "loss": 1.4664,
      "step": 7166
    },
    {
      "epoch": 0.9173172916933316,
      "grad_norm": 0.671875,
      "learning_rate": 3.564743664624759e-05,
      "loss": 1.3674,
      "step": 7167
    },
    {
      "epoch": 0.9174452835018558,
      "grad_norm": 0.671875,
      "learning_rate": 3.553781635768438e-05,
      "loss": 1.0362,
      "step": 7168
    },
    {
      "epoch": 0.9175732753103801,
      "grad_norm": 0.77734375,
      "learning_rate": 3.542836182759546e-05,
      "loss": 1.8225,
      "step": 7169
    },
    {
      "epoch": 0.9177012671189044,
      "grad_norm": 0.90234375,
      "learning_rate": 3.531907307479265e-05,
      "loss": 2.0865,
      "step": 7170
    },
    {
      "epoch": 0.9178292589274286,
      "grad_norm": 0.59375,
      "learning_rate": 3.520995011805883e-05,
      "loss": 1.1705,
      "step": 7171
    },
    {
      "epoch": 0.9179572507359529,
      "grad_norm": 0.58984375,
      "learning_rate": 3.5100992976148436e-05,
      "loss": 1.2688,
      "step": 7172
    },
    {
      "epoch": 0.9180852425444771,
      "grad_norm": 0.625,
      "learning_rate": 3.499220166778783e-05,
      "loss": 1.2349,
      "step": 7173
    },
    {
      "epoch": 0.9182132343530014,
      "grad_norm": 0.75390625,
      "learning_rate": 3.488357621167426e-05,
      "loss": 2.2854,
      "step": 7174
    },
    {
      "epoch": 0.9183412261615257,
      "grad_norm": 0.54296875,
      "learning_rate": 3.4775116626477036e-05,
      "loss": 0.7769,
      "step": 7175
    },
    {
      "epoch": 0.9184692179700499,
      "grad_norm": 0.5390625,
      "learning_rate": 3.466682293083645e-05,
      "loss": 1.1432,
      "step": 7176
    },
    {
      "epoch": 0.9185972097785742,
      "grad_norm": 0.494140625,
      "learning_rate": 3.4558695143364736e-05,
      "loss": 0.6112,
      "step": 7177
    },
    {
      "epoch": 0.9187252015870985,
      "grad_norm": 0.8828125,
      "learning_rate": 3.4450733282645254e-05,
      "loss": 2.0639,
      "step": 7178
    },
    {
      "epoch": 0.9188531933956227,
      "grad_norm": 0.65234375,
      "learning_rate": 3.4342937367233154e-05,
      "loss": 1.1411,
      "step": 7179
    },
    {
      "epoch": 0.918981185204147,
      "grad_norm": 0.63671875,
      "learning_rate": 3.423530741565461e-05,
      "loss": 1.2709,
      "step": 7180
    },
    {
      "epoch": 0.9191091770126711,
      "grad_norm": 0.7421875,
      "learning_rate": 3.412784344640796e-05,
      "loss": 1.353,
      "step": 7181
    },
    {
      "epoch": 0.9192371688211954,
      "grad_norm": 0.7109375,
      "learning_rate": 3.4020545477962403e-05,
      "loss": 1.7763,
      "step": 7182
    },
    {
      "epoch": 0.9193651606297197,
      "grad_norm": 0.66015625,
      "learning_rate": 3.3913413528758876e-05,
      "loss": 0.9806,
      "step": 7183
    },
    {
      "epoch": 0.9194931524382439,
      "grad_norm": 0.69921875,
      "learning_rate": 3.3806447617209636e-05,
      "loss": 1.5415,
      "step": 7184
    },
    {
      "epoch": 0.9196211442467682,
      "grad_norm": 0.69921875,
      "learning_rate": 3.369964776169876e-05,
      "loss": 1.8714,
      "step": 7185
    },
    {
      "epoch": 0.9197491360552925,
      "grad_norm": 0.7734375,
      "learning_rate": 3.359301398058146e-05,
      "loss": 1.3952,
      "step": 7186
    },
    {
      "epoch": 0.9198771278638167,
      "grad_norm": 0.5390625,
      "learning_rate": 3.3486546292184284e-05,
      "loss": 0.7393,
      "step": 7187
    },
    {
      "epoch": 0.920005119672341,
      "grad_norm": 0.7890625,
      "learning_rate": 3.338024471480583e-05,
      "loss": 1.7557,
      "step": 7188
    },
    {
      "epoch": 0.9201331114808652,
      "grad_norm": 0.6640625,
      "learning_rate": 3.327410926671526e-05,
      "loss": 1.5267,
      "step": 7189
    },
    {
      "epoch": 0.9202611032893895,
      "grad_norm": 0.6484375,
      "learning_rate": 3.3168139966153974e-05,
      "loss": 1.113,
      "step": 7190
    },
    {
      "epoch": 0.9203890950979138,
      "grad_norm": 0.70703125,
      "learning_rate": 3.306233683133441e-05,
      "loss": 1.3718,
      "step": 7191
    },
    {
      "epoch": 0.920517086906438,
      "grad_norm": 0.7578125,
      "learning_rate": 3.295669988044059e-05,
      "loss": 1.4191,
      "step": 7192
    },
    {
      "epoch": 0.9206450787149623,
      "grad_norm": 0.71875,
      "learning_rate": 3.285122913162797e-05,
      "loss": 1.4476,
      "step": 7193
    },
    {
      "epoch": 0.9207730705234864,
      "grad_norm": 0.53125,
      "learning_rate": 3.274592460302339e-05,
      "loss": 0.8771,
      "step": 7194
    },
    {
      "epoch": 0.9209010623320107,
      "grad_norm": 0.609375,
      "learning_rate": 3.264078631272505e-05,
      "loss": 1.2699,
      "step": 7195
    },
    {
      "epoch": 0.921029054140535,
      "grad_norm": 0.77734375,
      "learning_rate": 3.253581427880281e-05,
      "loss": 1.4665,
      "step": 7196
    },
    {
      "epoch": 0.9211570459490592,
      "grad_norm": 0.76171875,
      "learning_rate": 3.243100851929792e-05,
      "loss": 1.8622,
      "step": 7197
    },
    {
      "epoch": 0.9212850377575835,
      "grad_norm": 0.8125,
      "learning_rate": 3.2326369052222635e-05,
      "loss": 1.002,
      "step": 7198
    },
    {
      "epoch": 0.9214130295661078,
      "grad_norm": 0.71484375,
      "learning_rate": 3.2221895895561013e-05,
      "loss": 1.6169,
      "step": 7199
    },
    {
      "epoch": 0.921541021374632,
      "grad_norm": 0.734375,
      "learning_rate": 3.2117589067268696e-05,
      "loss": 1.7447,
      "step": 7200
    },
    {
      "epoch": 0.9216690131831563,
      "grad_norm": 0.6171875,
      "learning_rate": 3.201344858527233e-05,
      "loss": 1.1782,
      "step": 7201
    },
    {
      "epoch": 0.9217970049916805,
      "grad_norm": 0.6875,
      "learning_rate": 3.190947446747017e-05,
      "loss": 1.745,
      "step": 7202
    },
    {
      "epoch": 0.9219249968002048,
      "grad_norm": 0.6015625,
      "learning_rate": 3.18056667317318e-05,
      "loss": 1.0928,
      "step": 7203
    },
    {
      "epoch": 0.9220529886087291,
      "grad_norm": 0.63671875,
      "learning_rate": 3.170202539589839e-05,
      "loss": 1.2122,
      "step": 7204
    },
    {
      "epoch": 0.9221809804172533,
      "grad_norm": 0.61328125,
      "learning_rate": 3.159855047778226e-05,
      "loss": 1.0202,
      "step": 7205
    },
    {
      "epoch": 0.9223089722257776,
      "grad_norm": 0.65234375,
      "learning_rate": 3.149524199516729e-05,
      "loss": 1.2713,
      "step": 7206
    },
    {
      "epoch": 0.9224369640343018,
      "grad_norm": 0.7265625,
      "learning_rate": 3.1392099965808716e-05,
      "loss": 1.431,
      "step": 7207
    },
    {
      "epoch": 0.922564955842826,
      "grad_norm": 0.84765625,
      "learning_rate": 3.128912440743325e-05,
      "loss": 1.7244,
      "step": 7208
    },
    {
      "epoch": 0.9226929476513503,
      "grad_norm": 0.5703125,
      "learning_rate": 3.118631533773886e-05,
      "loss": 1.4054,
      "step": 7209
    },
    {
      "epoch": 0.9228209394598745,
      "grad_norm": 0.6015625,
      "learning_rate": 3.108367277439505e-05,
      "loss": 1.1894,
      "step": 7210
    },
    {
      "epoch": 0.9229489312683988,
      "grad_norm": 0.80859375,
      "learning_rate": 3.098119673504241e-05,
      "loss": 2.2067,
      "step": 7211
    },
    {
      "epoch": 0.9230769230769231,
      "grad_norm": 0.482421875,
      "learning_rate": 3.0878887237293154e-05,
      "loss": 0.6942,
      "step": 7212
    },
    {
      "epoch": 0.9232049148854473,
      "grad_norm": 0.62109375,
      "learning_rate": 3.0776744298730895e-05,
      "loss": 1.5675,
      "step": 7213
    },
    {
      "epoch": 0.9233329066939716,
      "grad_norm": 0.70703125,
      "learning_rate": 3.06747679369106e-05,
      "loss": 1.175,
      "step": 7214
    },
    {
      "epoch": 0.9234608985024958,
      "grad_norm": 0.69140625,
      "learning_rate": 3.0572958169358456e-05,
      "loss": 1.5389,
      "step": 7215
    },
    {
      "epoch": 0.9235888903110201,
      "grad_norm": 0.69140625,
      "learning_rate": 3.0471315013572143e-05,
      "loss": 1.9574,
      "step": 7216
    },
    {
      "epoch": 0.9237168821195444,
      "grad_norm": 0.57421875,
      "learning_rate": 3.0369838487020796e-05,
      "loss": 0.9763,
      "step": 7217
    },
    {
      "epoch": 0.9238448739280686,
      "grad_norm": 0.451171875,
      "learning_rate": 3.0268528607144685e-05,
      "loss": 0.6109,
      "step": 7218
    },
    {
      "epoch": 0.9239728657365929,
      "grad_norm": 0.76953125,
      "learning_rate": 3.016738539135566e-05,
      "loss": 1.5177,
      "step": 7219
    },
    {
      "epoch": 0.9241008575451171,
      "grad_norm": 0.52734375,
      "learning_rate": 3.00664088570366e-05,
      "loss": 1.0007,
      "step": 7220
    },
    {
      "epoch": 0.9242288493536414,
      "grad_norm": 0.9296875,
      "learning_rate": 2.9965599021542278e-05,
      "loss": 1.9076,
      "step": 7221
    },
    {
      "epoch": 0.9243568411621657,
      "grad_norm": 0.68359375,
      "learning_rate": 2.9864955902198177e-05,
      "loss": 1.1823,
      "step": 7222
    },
    {
      "epoch": 0.9244848329706898,
      "grad_norm": 0.5390625,
      "learning_rate": 2.976447951630179e-05,
      "loss": 0.9145,
      "step": 7223
    },
    {
      "epoch": 0.9246128247792141,
      "grad_norm": 0.6875,
      "learning_rate": 2.9664169881121305e-05,
      "loss": 0.8746,
      "step": 7224
    },
    {
      "epoch": 0.9247408165877384,
      "grad_norm": 0.4921875,
      "learning_rate": 2.956402701389671e-05,
      "loss": 0.8362,
      "step": 7225
    },
    {
      "epoch": 0.9248688083962626,
      "grad_norm": 0.78125,
      "learning_rate": 2.9464050931839236e-05,
      "loss": 1.098,
      "step": 7226
    },
    {
      "epoch": 0.9249968002047869,
      "grad_norm": 0.6171875,
      "learning_rate": 2.936424165213125e-05,
      "loss": 1.0328,
      "step": 7227
    },
    {
      "epoch": 0.9251247920133111,
      "grad_norm": 0.796875,
      "learning_rate": 2.9264599191926588e-05,
      "loss": 1.4256,
      "step": 7228
    },
    {
      "epoch": 0.9252527838218354,
      "grad_norm": 0.63671875,
      "learning_rate": 2.916512356835066e-05,
      "loss": 0.73,
      "step": 7229
    },
    {
      "epoch": 0.9253807756303597,
      "grad_norm": 0.54296875,
      "learning_rate": 2.9065814798499678e-05,
      "loss": 1.2118,
      "step": 7230
    },
    {
      "epoch": 0.9255087674388839,
      "grad_norm": 0.59765625,
      "learning_rate": 2.896667289944166e-05,
      "loss": 1.2119,
      "step": 7231
    },
    {
      "epoch": 0.9256367592474082,
      "grad_norm": 0.67578125,
      "learning_rate": 2.886769788821575e-05,
      "loss": 1.5002,
      "step": 7232
    },
    {
      "epoch": 0.9257647510559324,
      "grad_norm": 0.6171875,
      "learning_rate": 2.876888978183234e-05,
      "loss": 1.0247,
      "step": 7233
    },
    {
      "epoch": 0.9258927428644567,
      "grad_norm": 0.7734375,
      "learning_rate": 2.8670248597273075e-05,
      "loss": 1.9285,
      "step": 7234
    },
    {
      "epoch": 0.926020734672981,
      "grad_norm": 0.7109375,
      "learning_rate": 2.8571774351491275e-05,
      "loss": 1.2041,
      "step": 7235
    },
    {
      "epoch": 0.9261487264815051,
      "grad_norm": 0.6015625,
      "learning_rate": 2.8473467061411184e-05,
      "loss": 1.2101,
      "step": 7236
    },
    {
      "epoch": 0.9262767182900294,
      "grad_norm": 0.625,
      "learning_rate": 2.837532674392862e-05,
      "loss": 1.1438,
      "step": 7237
    },
    {
      "epoch": 0.9264047100985537,
      "grad_norm": 0.640625,
      "learning_rate": 2.8277353415910422e-05,
      "loss": 1.1685,
      "step": 7238
    },
    {
      "epoch": 0.9265327019070779,
      "grad_norm": 0.498046875,
      "learning_rate": 2.8179547094195124e-05,
      "loss": 0.9125,
      "step": 7239
    },
    {
      "epoch": 0.9266606937156022,
      "grad_norm": 0.7109375,
      "learning_rate": 2.8081907795592164e-05,
      "loss": 1.4172,
      "step": 7240
    },
    {
      "epoch": 0.9267886855241264,
      "grad_norm": 0.74609375,
      "learning_rate": 2.798443553688257e-05,
      "loss": 1.3335,
      "step": 7241
    },
    {
      "epoch": 0.9269166773326507,
      "grad_norm": 0.69140625,
      "learning_rate": 2.7887130334818376e-05,
      "loss": 1.7512,
      "step": 7242
    },
    {
      "epoch": 0.927044669141175,
      "grad_norm": 0.625,
      "learning_rate": 2.7789992206123104e-05,
      "loss": 1.2921,
      "step": 7243
    },
    {
      "epoch": 0.9271726609496992,
      "grad_norm": 0.765625,
      "learning_rate": 2.7693021167491507e-05,
      "loss": 1.2562,
      "step": 7244
    },
    {
      "epoch": 0.9273006527582235,
      "grad_norm": 0.71875,
      "learning_rate": 2.7596217235589917e-05,
      "loss": 1.1057,
      "step": 7245
    },
    {
      "epoch": 0.9274286445667477,
      "grad_norm": 0.6796875,
      "learning_rate": 2.749958042705514e-05,
      "loss": 1.1457,
      "step": 7246
    },
    {
      "epoch": 0.927556636375272,
      "grad_norm": 0.65625,
      "learning_rate": 2.740311075849611e-05,
      "loss": 1.0665,
      "step": 7247
    },
    {
      "epoch": 0.9276846281837963,
      "grad_norm": 0.6875,
      "learning_rate": 2.7306808246492676e-05,
      "loss": 1.6068,
      "step": 7248
    },
    {
      "epoch": 0.9278126199923205,
      "grad_norm": 0.59375,
      "learning_rate": 2.7210672907595935e-05,
      "loss": 0.9735,
      "step": 7249
    },
    {
      "epoch": 0.9279406118008448,
      "grad_norm": 0.59765625,
      "learning_rate": 2.7114704758328334e-05,
      "loss": 0.8537,
      "step": 7250
    },
    {
      "epoch": 0.928068603609369,
      "grad_norm": 0.765625,
      "learning_rate": 2.701890381518357e-05,
      "loss": 1.1325,
      "step": 7251
    },
    {
      "epoch": 0.9281965954178932,
      "grad_norm": 0.7578125,
      "learning_rate": 2.6923270094626474e-05,
      "loss": 1.3095,
      "step": 7252
    },
    {
      "epoch": 0.9283245872264175,
      "grad_norm": 0.578125,
      "learning_rate": 2.6827803613093448e-05,
      "loss": 0.7333,
      "step": 7253
    },
    {
      "epoch": 0.9284525790349417,
      "grad_norm": 0.58203125,
      "learning_rate": 2.6732504386991928e-05,
      "loss": 1.2308,
      "step": 7254
    },
    {
      "epoch": 0.928580570843466,
      "grad_norm": 0.50390625,
      "learning_rate": 2.663737243270048e-05,
      "loss": 0.7533,
      "step": 7255
    },
    {
      "epoch": 0.9287085626519903,
      "grad_norm": 0.80078125,
      "learning_rate": 2.654240776656913e-05,
      "loss": 1.5704,
      "step": 7256
    },
    {
      "epoch": 0.9288365544605145,
      "grad_norm": 0.5546875,
      "learning_rate": 2.6447610404919054e-05,
      "loss": 0.9897,
      "step": 7257
    },
    {
      "epoch": 0.9289645462690388,
      "grad_norm": 0.5390625,
      "learning_rate": 2.6352980364042877e-05,
      "loss": 1.0538,
      "step": 7258
    },
    {
      "epoch": 0.9290925380775631,
      "grad_norm": 0.64453125,
      "learning_rate": 2.625851766020426e-05,
      "loss": 1.3485,
      "step": 7259
    },
    {
      "epoch": 0.9292205298860873,
      "grad_norm": 0.58984375,
      "learning_rate": 2.6164222309637996e-05,
      "loss": 1.1208,
      "step": 7260
    },
    {
      "epoch": 0.9293485216946116,
      "grad_norm": 0.87890625,
      "learning_rate": 2.607009432855034e-05,
      "loss": 1.5964,
      "step": 7261
    },
    {
      "epoch": 0.9294765135031358,
      "grad_norm": 0.625,
      "learning_rate": 2.59761337331188e-05,
      "loss": 1.0107,
      "step": 7262
    },
    {
      "epoch": 0.9296045053116601,
      "grad_norm": 0.51953125,
      "learning_rate": 2.588234053949212e-05,
      "loss": 1.0516,
      "step": 7263
    },
    {
      "epoch": 0.9297324971201844,
      "grad_norm": 0.55859375,
      "learning_rate": 2.5788714763789965e-05,
      "loss": 1.1149,
      "step": 7264
    },
    {
      "epoch": 0.9298604889287085,
      "grad_norm": 0.6796875,
      "learning_rate": 2.5695256422103464e-05,
      "loss": 0.9339,
      "step": 7265
    },
    {
      "epoch": 0.9299884807372328,
      "grad_norm": 0.921875,
      "learning_rate": 2.5601965530494987e-05,
      "loss": 1.6543,
      "step": 7266
    },
    {
      "epoch": 0.930116472545757,
      "grad_norm": 0.63671875,
      "learning_rate": 2.5508842104998265e-05,
      "loss": 0.9969,
      "step": 7267
    },
    {
      "epoch": 0.9302444643542813,
      "grad_norm": 0.671875,
      "learning_rate": 2.541588616161772e-05,
      "loss": 1.4386,
      "step": 7268
    },
    {
      "epoch": 0.9303724561628056,
      "grad_norm": 0.62890625,
      "learning_rate": 2.532309771632968e-05,
      "loss": 1.3202,
      "step": 7269
    },
    {
      "epoch": 0.9305004479713298,
      "grad_norm": 0.546875,
      "learning_rate": 2.523047678508106e-05,
      "loss": 0.8114,
      "step": 7270
    },
    {
      "epoch": 0.9306284397798541,
      "grad_norm": 0.828125,
      "learning_rate": 2.5138023383790454e-05,
      "loss": 1.8844,
      "step": 7271
    },
    {
      "epoch": 0.9307564315883784,
      "grad_norm": 0.6328125,
      "learning_rate": 2.504573752834738e-05,
      "loss": 1.7221,
      "step": 7272
    },
    {
      "epoch": 0.9308844233969026,
      "grad_norm": 0.70703125,
      "learning_rate": 2.4953619234612812e-05,
      "loss": 1.0227,
      "step": 7273
    },
    {
      "epoch": 0.9310124152054269,
      "grad_norm": 0.6171875,
      "learning_rate": 2.4861668518418535e-05,
      "loss": 1.0563,
      "step": 7274
    },
    {
      "epoch": 0.9311404070139511,
      "grad_norm": 0.53515625,
      "learning_rate": 2.4769885395568015e-05,
      "loss": 1.0697,
      "step": 7275
    },
    {
      "epoch": 0.9312683988224754,
      "grad_norm": 0.7578125,
      "learning_rate": 2.4678269881835636e-05,
      "loss": 1.2335,
      "step": 7276
    },
    {
      "epoch": 0.9313963906309997,
      "grad_norm": 0.69921875,
      "learning_rate": 2.45868219929668e-05,
      "loss": 1.7821,
      "step": 7277
    },
    {
      "epoch": 0.9315243824395238,
      "grad_norm": 0.78125,
      "learning_rate": 2.4495541744678497e-05,
      "loss": 1.5068,
      "step": 7278
    },
    {
      "epoch": 0.9316523742480481,
      "grad_norm": 0.85546875,
      "learning_rate": 2.4404429152658724e-05,
      "loss": 1.8967,
      "step": 7279
    },
    {
      "epoch": 0.9317803660565723,
      "grad_norm": 0.73046875,
      "learning_rate": 2.431348423256652e-05,
      "loss": 1.7277,
      "step": 7280
    },
    {
      "epoch": 0.9319083578650966,
      "grad_norm": 0.65234375,
      "learning_rate": 2.4222707000032374e-05,
      "loss": 1.261,
      "step": 7281
    },
    {
      "epoch": 0.9320363496736209,
      "grad_norm": 0.71875,
      "learning_rate": 2.41320974706577e-05,
      "loss": 1.3236,
      "step": 7282
    },
    {
      "epoch": 0.9321643414821451,
      "grad_norm": 0.75,
      "learning_rate": 2.404165566001537e-05,
      "loss": 1.6904,
      "step": 7283
    },
    {
      "epoch": 0.9322923332906694,
      "grad_norm": 0.58203125,
      "learning_rate": 2.3951381583649178e-05,
      "loss": 1.1405,
      "step": 7284
    },
    {
      "epoch": 0.9324203250991937,
      "grad_norm": 0.75390625,
      "learning_rate": 2.3861275257074265e-05,
      "loss": 1.5847,
      "step": 7285
    },
    {
      "epoch": 0.9325483169077179,
      "grad_norm": 0.72265625,
      "learning_rate": 2.3771336695776802e-05,
      "loss": 1.9625,
      "step": 7286
    },
    {
      "epoch": 0.9326763087162422,
      "grad_norm": 0.73828125,
      "learning_rate": 2.3681565915214086e-05,
      "loss": 1.105,
      "step": 7287
    },
    {
      "epoch": 0.9328043005247664,
      "grad_norm": 0.5625,
      "learning_rate": 2.3591962930814894e-05,
      "loss": 1.0677,
      "step": 7288
    },
    {
      "epoch": 0.9329322923332907,
      "grad_norm": 0.609375,
      "learning_rate": 2.3502527757978896e-05,
      "loss": 1.3722,
      "step": 7289
    },
    {
      "epoch": 0.933060284141815,
      "grad_norm": 0.65625,
      "learning_rate": 2.341326041207692e-05,
      "loss": 1.2119,
      "step": 7290
    },
    {
      "epoch": 0.9331882759503392,
      "grad_norm": 0.546875,
      "learning_rate": 2.3324160908451018e-05,
      "loss": 1.0004,
      "step": 7291
    },
    {
      "epoch": 0.9333162677588634,
      "grad_norm": 0.5859375,
      "learning_rate": 2.3235229262414392e-05,
      "loss": 0.9334,
      "step": 7292
    },
    {
      "epoch": 0.9334442595673876,
      "grad_norm": 0.5703125,
      "learning_rate": 2.3146465489251367e-05,
      "loss": 1.0628,
      "step": 7293
    },
    {
      "epoch": 0.9335722513759119,
      "grad_norm": 0.61328125,
      "learning_rate": 2.3057869604217407e-05,
      "loss": 1.3144,
      "step": 7294
    },
    {
      "epoch": 0.9337002431844362,
      "grad_norm": 0.78125,
      "learning_rate": 2.296944162253922e-05,
      "loss": 1.5947,
      "step": 7295
    },
    {
      "epoch": 0.9338282349929604,
      "grad_norm": 0.7421875,
      "learning_rate": 2.2881181559414655e-05,
      "loss": 1.5796,
      "step": 7296
    },
    {
      "epoch": 0.9339562268014847,
      "grad_norm": 0.60546875,
      "learning_rate": 2.2793089430012458e-05,
      "loss": 1.0058,
      "step": 7297
    },
    {
      "epoch": 0.934084218610009,
      "grad_norm": 0.6328125,
      "learning_rate": 2.270516524947297e-05,
      "loss": 1.4262,
      "step": 7298
    },
    {
      "epoch": 0.9342122104185332,
      "grad_norm": 0.609375,
      "learning_rate": 2.26174090329071e-05,
      "loss": 1.121,
      "step": 7299
    },
    {
      "epoch": 0.9343402022270575,
      "grad_norm": 0.62109375,
      "learning_rate": 2.2529820795397226e-05,
      "loss": 1.0257,
      "step": 7300
    },
    {
      "epoch": 0.9344681940355817,
      "grad_norm": 0.6484375,
      "learning_rate": 2.2442400551996868e-05,
      "loss": 0.9734,
      "step": 7301
    },
    {
      "epoch": 0.934596185844106,
      "grad_norm": 0.71484375,
      "learning_rate": 2.2355148317730668e-05,
      "loss": 1.4949,
      "step": 7302
    },
    {
      "epoch": 0.9347241776526303,
      "grad_norm": 0.65234375,
      "learning_rate": 2.22680641075943e-05,
      "loss": 1.3556,
      "step": 7303
    },
    {
      "epoch": 0.9348521694611545,
      "grad_norm": 0.59375,
      "learning_rate": 2.2181147936554567e-05,
      "loss": 1.427,
      "step": 7304
    },
    {
      "epoch": 0.9349801612696788,
      "grad_norm": 0.62109375,
      "learning_rate": 2.2094399819549303e-05,
      "loss": 1.1091,
      "step": 7305
    },
    {
      "epoch": 0.9351081530782029,
      "grad_norm": 0.65234375,
      "learning_rate": 2.20078197714878e-05,
      "loss": 1.0525,
      "step": 7306
    },
    {
      "epoch": 0.9352361448867272,
      "grad_norm": 0.59765625,
      "learning_rate": 2.1921407807250272e-05,
      "loss": 0.9384,
      "step": 7307
    },
    {
      "epoch": 0.9353641366952515,
      "grad_norm": 0.63671875,
      "learning_rate": 2.183516394168783e-05,
      "loss": 1.1675,
      "step": 7308
    },
    {
      "epoch": 0.9354921285037757,
      "grad_norm": 0.75,
      "learning_rate": 2.1749088189622845e-05,
      "loss": 1.7883,
      "step": 7309
    },
    {
      "epoch": 0.9356201203123,
      "grad_norm": 0.640625,
      "learning_rate": 2.166318056584904e-05,
      "loss": 1.2717,
      "step": 7310
    },
    {
      "epoch": 0.9357481121208243,
      "grad_norm": 0.640625,
      "learning_rate": 2.1577441085130934e-05,
      "loss": 1.1216,
      "step": 7311
    },
    {
      "epoch": 0.9358761039293485,
      "grad_norm": 0.78125,
      "learning_rate": 2.1491869762204297e-05,
      "loss": 1.3331,
      "step": 7312
    },
    {
      "epoch": 0.9360040957378728,
      "grad_norm": 0.6640625,
      "learning_rate": 2.140646661177581e-05,
      "loss": 1.4323,
      "step": 7313
    },
    {
      "epoch": 0.936132087546397,
      "grad_norm": 0.5234375,
      "learning_rate": 2.132123164852351e-05,
      "loss": 0.8994,
      "step": 7314
    },
    {
      "epoch": 0.9362600793549213,
      "grad_norm": 0.890625,
      "learning_rate": 2.1236164887096342e-05,
      "loss": 1.9351,
      "step": 7315
    },
    {
      "epoch": 0.9363880711634456,
      "grad_norm": 0.71875,
      "learning_rate": 2.1151266342114506e-05,
      "loss": 1.6955,
      "step": 7316
    },
    {
      "epoch": 0.9365160629719698,
      "grad_norm": 0.6953125,
      "learning_rate": 2.1066536028169105e-05,
      "loss": 1.6772,
      "step": 7317
    },
    {
      "epoch": 0.9366440547804941,
      "grad_norm": 0.7578125,
      "learning_rate": 2.0981973959822377e-05,
      "loss": 1.3778,
      "step": 7318
    },
    {
      "epoch": 0.9367720465890182,
      "grad_norm": 0.55078125,
      "learning_rate": 2.0897580151607808e-05,
      "loss": 1.0524,
      "step": 7319
    },
    {
      "epoch": 0.9369000383975425,
      "grad_norm": 0.609375,
      "learning_rate": 2.08133546180298e-05,
      "loss": 1.3963,
      "step": 7320
    },
    {
      "epoch": 0.9370280302060668,
      "grad_norm": 0.72265625,
      "learning_rate": 2.0729297373563993e-05,
      "loss": 1.6554,
      "step": 7321
    },
    {
      "epoch": 0.937156022014591,
      "grad_norm": 0.6640625,
      "learning_rate": 2.064540843265672e-05,
      "loss": 1.1962,
      "step": 7322
    },
    {
      "epoch": 0.9372840138231153,
      "grad_norm": 0.5546875,
      "learning_rate": 2.0561687809725782e-05,
      "loss": 1.0622,
      "step": 7323
    },
    {
      "epoch": 0.9374120056316396,
      "grad_norm": 0.7109375,
      "learning_rate": 2.047813551916e-05,
      "loss": 1.5639,
      "step": 7324
    },
    {
      "epoch": 0.9375399974401638,
      "grad_norm": 0.55859375,
      "learning_rate": 2.0394751575319116e-05,
      "loss": 1.258,
      "step": 7325
    },
    {
      "epoch": 0.9376679892486881,
      "grad_norm": 0.7578125,
      "learning_rate": 2.0311535992533993e-05,
      "loss": 1.4774,
      "step": 7326
    },
    {
      "epoch": 0.9377959810572123,
      "grad_norm": 0.5859375,
      "learning_rate": 2.0228488785106634e-05,
      "loss": 1.0753,
      "step": 7327
    },
    {
      "epoch": 0.9379239728657366,
      "grad_norm": 0.7734375,
      "learning_rate": 2.0145609967309963e-05,
      "loss": 1.2936,
      "step": 7328
    },
    {
      "epoch": 0.9380519646742609,
      "grad_norm": 0.8359375,
      "learning_rate": 2.0062899553388357e-05,
      "loss": 1.3096,
      "step": 7329
    },
    {
      "epoch": 0.9381799564827851,
      "grad_norm": 0.70703125,
      "learning_rate": 1.998035755755656e-05,
      "loss": 1.4639,
      "step": 7330
    },
    {
      "epoch": 0.9383079482913094,
      "grad_norm": 0.70703125,
      "learning_rate": 1.989798399400089e-05,
      "loss": 1.472,
      "step": 7331
    },
    {
      "epoch": 0.9384359400998337,
      "grad_norm": 0.6484375,
      "learning_rate": 1.9815778876878577e-05,
      "loss": 1.1644,
      "step": 7332
    },
    {
      "epoch": 0.9385639319083579,
      "grad_norm": 0.6796875,
      "learning_rate": 1.9733742220317873e-05,
      "loss": 1.2643,
      "step": 7333
    },
    {
      "epoch": 0.9386919237168821,
      "grad_norm": 0.6875,
      "learning_rate": 1.9651874038418283e-05,
      "loss": 1.1888,
      "step": 7334
    },
    {
      "epoch": 0.9388199155254063,
      "grad_norm": 0.67578125,
      "learning_rate": 1.9570174345249992e-05,
      "loss": 1.6116,
      "step": 7335
    },
    {
      "epoch": 0.9389479073339306,
      "grad_norm": 0.66015625,
      "learning_rate": 1.9488643154854434e-05,
      "loss": 0.9107,
      "step": 7336
    },
    {
      "epoch": 0.9390758991424549,
      "grad_norm": 0.59375,
      "learning_rate": 1.9407280481244183e-05,
      "loss": 1.271,
      "step": 7337
    },
    {
      "epoch": 0.9392038909509791,
      "grad_norm": 0.7109375,
      "learning_rate": 1.9326086338402605e-05,
      "loss": 1.63,
      "step": 7338
    },
    {
      "epoch": 0.9393318827595034,
      "grad_norm": 0.63671875,
      "learning_rate": 1.9245060740284316e-05,
      "loss": 0.8548,
      "step": 7339
    },
    {
      "epoch": 0.9394598745680276,
      "grad_norm": 0.625,
      "learning_rate": 1.916420370081484e-05,
      "loss": 1.5296,
      "step": 7340
    },
    {
      "epoch": 0.9395878663765519,
      "grad_norm": 0.546875,
      "learning_rate": 1.9083515233890736e-05,
      "loss": 1.0537,
      "step": 7341
    },
    {
      "epoch": 0.9397158581850762,
      "grad_norm": 0.74609375,
      "learning_rate": 1.9002995353379683e-05,
      "loss": 1.3953,
      "step": 7342
    },
    {
      "epoch": 0.9398438499936004,
      "grad_norm": 0.5546875,
      "learning_rate": 1.89226440731205e-05,
      "loss": 1.1903,
      "step": 7343
    },
    {
      "epoch": 0.9399718418021247,
      "grad_norm": 0.7265625,
      "learning_rate": 1.8842461406922473e-05,
      "loss": 0.9246,
      "step": 7344
    },
    {
      "epoch": 0.940099833610649,
      "grad_norm": 0.72265625,
      "learning_rate": 1.876244736856658e-05,
      "loss": 1.4625,
      "step": 7345
    },
    {
      "epoch": 0.9402278254191732,
      "grad_norm": 0.6328125,
      "learning_rate": 1.868260197180438e-05,
      "loss": 1.2463,
      "step": 7346
    },
    {
      "epoch": 0.9403558172276975,
      "grad_norm": 0.80859375,
      "learning_rate": 1.860292523035878e-05,
      "loss": 1.2823,
      "step": 7347
    },
    {
      "epoch": 0.9404838090362216,
      "grad_norm": 0.54296875,
      "learning_rate": 1.852341715792327e-05,
      "loss": 0.837,
      "step": 7348
    },
    {
      "epoch": 0.9406118008447459,
      "grad_norm": 0.6640625,
      "learning_rate": 1.8444077768162815e-05,
      "loss": 1.3919,
      "step": 7349
    },
    {
      "epoch": 0.9407397926532702,
      "grad_norm": 0.57421875,
      "learning_rate": 1.8364907074713056e-05,
      "loss": 0.9697,
      "step": 7350
    },
    {
      "epoch": 0.9408677844617944,
      "grad_norm": 0.65234375,
      "learning_rate": 1.828590509118089e-05,
      "loss": 1.4825,
      "step": 7351
    },
    {
      "epoch": 0.9409957762703187,
      "grad_norm": 0.6328125,
      "learning_rate": 1.8207071831144006e-05,
      "loss": 1.1794,
      "step": 7352
    },
    {
      "epoch": 0.9411237680788429,
      "grad_norm": 0.474609375,
      "learning_rate": 1.8128407308151128e-05,
      "loss": 0.7091,
      "step": 7353
    },
    {
      "epoch": 0.9412517598873672,
      "grad_norm": 0.58984375,
      "learning_rate": 1.8049911535722106e-05,
      "loss": 0.8835,
      "step": 7354
    },
    {
      "epoch": 0.9413797516958915,
      "grad_norm": 0.60546875,
      "learning_rate": 1.7971584527347594e-05,
      "loss": 1.2695,
      "step": 7355
    },
    {
      "epoch": 0.9415077435044157,
      "grad_norm": 0.58984375,
      "learning_rate": 1.7893426296489712e-05,
      "loss": 0.9409,
      "step": 7356
    },
    {
      "epoch": 0.94163573531294,
      "grad_norm": 0.6171875,
      "learning_rate": 1.7815436856580714e-05,
      "loss": 1.3932,
      "step": 7357
    },
    {
      "epoch": 0.9417637271214643,
      "grad_norm": 0.80859375,
      "learning_rate": 1.7737616221024653e-05,
      "loss": 1.8897,
      "step": 7358
    },
    {
      "epoch": 0.9418917189299885,
      "grad_norm": 0.5546875,
      "learning_rate": 1.7659964403196284e-05,
      "loss": 0.8283,
      "step": 7359
    },
    {
      "epoch": 0.9420197107385128,
      "grad_norm": 0.6640625,
      "learning_rate": 1.7582481416441144e-05,
      "loss": 1.6753,
      "step": 7360
    },
    {
      "epoch": 0.942147702547037,
      "grad_norm": 0.5546875,
      "learning_rate": 1.750516727407614e-05,
      "loss": 0.8635,
      "step": 7361
    },
    {
      "epoch": 0.9422756943555612,
      "grad_norm": 0.9140625,
      "learning_rate": 1.7428021989388866e-05,
      "loss": 1.3136,
      "step": 7362
    },
    {
      "epoch": 0.9424036861640855,
      "grad_norm": 0.75,
      "learning_rate": 1.7351045575638048e-05,
      "loss": 1.3276,
      "step": 7363
    },
    {
      "epoch": 0.9425316779726097,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7274238046053214e-05,
      "loss": 1.709,
      "step": 7364
    },
    {
      "epoch": 0.942659669781134,
      "grad_norm": 0.984375,
      "learning_rate": 1.719759941383514e-05,
      "loss": 1.2921,
      "step": 7365
    },
    {
      "epoch": 0.9427876615896582,
      "grad_norm": 0.66015625,
      "learning_rate": 1.712112969215529e-05,
      "loss": 1.4301,
      "step": 7366
    },
    {
      "epoch": 0.9429156533981825,
      "grad_norm": 0.6796875,
      "learning_rate": 1.704482889415637e-05,
      "loss": 1.4726,
      "step": 7367
    },
    {
      "epoch": 0.9430436452067068,
      "grad_norm": 0.515625,
      "learning_rate": 1.6968697032951673e-05,
      "loss": 0.9482,
      "step": 7368
    },
    {
      "epoch": 0.943171637015231,
      "grad_norm": 0.53515625,
      "learning_rate": 1.6892734121625954e-05,
      "loss": 0.857,
      "step": 7369
    },
    {
      "epoch": 0.9432996288237553,
      "grad_norm": 0.55078125,
      "learning_rate": 1.681694017323454e-05,
      "loss": 0.8758,
      "step": 7370
    },
    {
      "epoch": 0.9434276206322796,
      "grad_norm": 0.9921875,
      "learning_rate": 1.6741315200803796e-05,
      "loss": 1.6523,
      "step": 7371
    },
    {
      "epoch": 0.9435556124408038,
      "grad_norm": 0.5546875,
      "learning_rate": 1.6665859217331104e-05,
      "loss": 0.6927,
      "step": 7372
    },
    {
      "epoch": 0.9436836042493281,
      "grad_norm": 1.28125,
      "learning_rate": 1.6590572235784862e-05,
      "loss": 0.9053,
      "step": 7373
    },
    {
      "epoch": 0.9438115960578523,
      "grad_norm": 0.67578125,
      "learning_rate": 1.6515454269104613e-05,
      "loss": 1.1728,
      "step": 7374
    },
    {
      "epoch": 0.9439395878663766,
      "grad_norm": 0.5859375,
      "learning_rate": 1.6440505330200138e-05,
      "loss": 1.245,
      "step": 7375
    },
    {
      "epoch": 0.9440675796749008,
      "grad_norm": 0.671875,
      "learning_rate": 1.6365725431952915e-05,
      "loss": 1.2265,
      "step": 7376
    },
    {
      "epoch": 0.944195571483425,
      "grad_norm": 0.75,
      "learning_rate": 1.6291114587214883e-05,
      "loss": 1.226,
      "step": 7377
    },
    {
      "epoch": 0.9443235632919493,
      "grad_norm": 0.71484375,
      "learning_rate": 1.621667280880945e-05,
      "loss": 1.9332,
      "step": 7378
    },
    {
      "epoch": 0.9444515551004735,
      "grad_norm": 0.7265625,
      "learning_rate": 1.6142400109530274e-05,
      "loss": 1.3824,
      "step": 7379
    },
    {
      "epoch": 0.9445795469089978,
      "grad_norm": 0.6875,
      "learning_rate": 1.6068296502142587e-05,
      "loss": 1.3149,
      "step": 7380
    },
    {
      "epoch": 0.9447075387175221,
      "grad_norm": 0.6328125,
      "learning_rate": 1.59943619993822e-05,
      "loss": 1.1857,
      "step": 7381
    },
    {
      "epoch": 0.9448355305260463,
      "grad_norm": 0.54296875,
      "learning_rate": 1.592059661395584e-05,
      "loss": 0.9296,
      "step": 7382
    },
    {
      "epoch": 0.9449635223345706,
      "grad_norm": 0.66015625,
      "learning_rate": 1.5847000358541473e-05,
      "loss": 1.3669,
      "step": 7383
    },
    {
      "epoch": 0.9450915141430949,
      "grad_norm": 0.5546875,
      "learning_rate": 1.5773573245787766e-05,
      "loss": 1.0403,
      "step": 7384
    },
    {
      "epoch": 0.9452195059516191,
      "grad_norm": 0.68359375,
      "learning_rate": 1.570031528831428e-05,
      "loss": 1.3391,
      "step": 7385
    },
    {
      "epoch": 0.9453474977601434,
      "grad_norm": 0.8359375,
      "learning_rate": 1.5627226498711623e-05,
      "loss": 1.3486,
      "step": 7386
    },
    {
      "epoch": 0.9454754895686676,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5554306889541513e-05,
      "loss": 1.1609,
      "step": 7387
    },
    {
      "epoch": 0.9456034813771919,
      "grad_norm": 0.609375,
      "learning_rate": 1.5481556473335933e-05,
      "loss": 1.0777,
      "step": 7388
    },
    {
      "epoch": 0.9457314731857162,
      "grad_norm": 0.66015625,
      "learning_rate": 1.5408975262598546e-05,
      "loss": 1.4696,
      "step": 7389
    },
    {
      "epoch": 0.9458594649942403,
      "grad_norm": 0.6875,
      "learning_rate": 1.5336563269803372e-05,
      "loss": 1.1545,
      "step": 7390
    },
    {
      "epoch": 0.9459874568027646,
      "grad_norm": 0.59375,
      "learning_rate": 1.52643205073959e-05,
      "loss": 1.0167,
      "step": 7391
    },
    {
      "epoch": 0.9461154486112888,
      "grad_norm": 0.7578125,
      "learning_rate": 1.519224698779198e-05,
      "loss": 1.4186,
      "step": 7392
    },
    {
      "epoch": 0.9462434404198131,
      "grad_norm": 0.58984375,
      "learning_rate": 1.51203427233787e-05,
      "loss": 1.2135,
      "step": 7393
    },
    {
      "epoch": 0.9463714322283374,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5048607726513951e-05,
      "loss": 1.2881,
      "step": 7394
    },
    {
      "epoch": 0.9464994240368616,
      "grad_norm": 0.6953125,
      "learning_rate": 1.4977042009526431e-05,
      "loss": 1.4367,
      "step": 7395
    },
    {
      "epoch": 0.9466274158453859,
      "grad_norm": 1.015625,
      "learning_rate": 1.4905645584716188e-05,
      "loss": 1.3082,
      "step": 7396
    },
    {
      "epoch": 0.9467554076539102,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4834418464353627e-05,
      "loss": 1.0316,
      "step": 7397
    },
    {
      "epoch": 0.9468833994624344,
      "grad_norm": 0.65234375,
      "learning_rate": 1.4763360660680292e-05,
      "loss": 1.3126,
      "step": 7398
    },
    {
      "epoch": 0.9470113912709587,
      "grad_norm": 0.57421875,
      "learning_rate": 1.4692472185908635e-05,
      "loss": 0.9804,
      "step": 7399
    },
    {
      "epoch": 0.9471393830794829,
      "grad_norm": 0.74609375,
      "learning_rate": 1.462175305222202e-05,
      "loss": 1.5542,
      "step": 7400
    },
    {
      "epoch": 0.9472673748880072,
      "grad_norm": 0.78125,
      "learning_rate": 1.4551203271774726e-05,
      "loss": 1.4889,
      "step": 7401
    },
    {
      "epoch": 0.9473953666965315,
      "grad_norm": 0.69921875,
      "learning_rate": 1.4480822856691722e-05,
      "loss": 1.1303,
      "step": 7402
    },
    {
      "epoch": 0.9475233585050556,
      "grad_norm": 0.71484375,
      "learning_rate": 1.4410611819069219e-05,
      "loss": 0.4942,
      "step": 7403
    },
    {
      "epoch": 0.9476513503135799,
      "grad_norm": 0.68359375,
      "learning_rate": 1.43405701709739e-05,
      "loss": 1.6651,
      "step": 7404
    },
    {
      "epoch": 0.9477793421221042,
      "grad_norm": 0.625,
      "learning_rate": 1.4270697924443799e-05,
      "loss": 1.2995,
      "step": 7405
    },
    {
      "epoch": 0.9479073339306284,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4200995091487424e-05,
      "loss": 1.0481,
      "step": 7406
    },
    {
      "epoch": 0.9480353257391527,
      "grad_norm": 0.62890625,
      "learning_rate": 1.4131461684084412e-05,
      "loss": 1.2847,
      "step": 7407
    },
    {
      "epoch": 0.9481633175476769,
      "grad_norm": 0.49609375,
      "learning_rate": 1.4062097714185207e-05,
      "loss": 1.0193,
      "step": 7408
    },
    {
      "epoch": 0.9482913093562012,
      "grad_norm": 0.609375,
      "learning_rate": 1.3992903193711159e-05,
      "loss": 1.0927,
      "step": 7409
    },
    {
      "epoch": 0.9484193011647255,
      "grad_norm": 0.64453125,
      "learning_rate": 1.3923878134554313e-05,
      "loss": 1.1384,
      "step": 7410
    },
    {
      "epoch": 0.9485472929732497,
      "grad_norm": 0.70703125,
      "learning_rate": 1.3855022548577845e-05,
      "loss": 1.4291,
      "step": 7411
    },
    {
      "epoch": 0.948675284781774,
      "grad_norm": 0.86328125,
      "learning_rate": 1.3786336447615844e-05,
      "loss": 1.344,
      "step": 7412
    },
    {
      "epoch": 0.9488032765902982,
      "grad_norm": 0.65234375,
      "learning_rate": 1.371781984347298e-05,
      "loss": 1.3381,
      "step": 7413
    },
    {
      "epoch": 0.9489312683988225,
      "grad_norm": 0.53515625,
      "learning_rate": 1.364947274792483e-05,
      "loss": 1.1402,
      "step": 7414
    },
    {
      "epoch": 0.9490592602073468,
      "grad_norm": 0.60546875,
      "learning_rate": 1.3581295172718222e-05,
      "loss": 0.7361,
      "step": 7415
    },
    {
      "epoch": 0.949187252015871,
      "grad_norm": 0.71484375,
      "learning_rate": 1.351328712957034e-05,
      "loss": 1.0652,
      "step": 7416
    },
    {
      "epoch": 0.9493152438243952,
      "grad_norm": 0.53125,
      "learning_rate": 1.344544863016961e-05,
      "loss": 0.9984,
      "step": 7417
    },
    {
      "epoch": 0.9494432356329195,
      "grad_norm": 0.65625,
      "learning_rate": 1.3377779686175151e-05,
      "loss": 1.1771,
      "step": 7418
    },
    {
      "epoch": 0.9495712274414437,
      "grad_norm": 0.59765625,
      "learning_rate": 1.3310280309216993e-05,
      "loss": 1.1088,
      "step": 7419
    },
    {
      "epoch": 0.949699219249968,
      "grad_norm": 0.6875,
      "learning_rate": 1.3242950510895967e-05,
      "loss": 1.6848,
      "step": 7420
    },
    {
      "epoch": 0.9498272110584922,
      "grad_norm": 0.57421875,
      "learning_rate": 1.3175790302783709e-05,
      "loss": 0.849,
      "step": 7421
    },
    {
      "epoch": 0.9499552028670165,
      "grad_norm": 0.734375,
      "learning_rate": 1.3108799696422979e-05,
      "loss": 1.6725,
      "step": 7422
    },
    {
      "epoch": 0.9500831946755408,
      "grad_norm": 0.5859375,
      "learning_rate": 1.304197870332713e-05,
      "loss": 0.9146,
      "step": 7423
    },
    {
      "epoch": 0.950211186484065,
      "grad_norm": 0.5546875,
      "learning_rate": 1.2975327334980302e-05,
      "loss": 1.0658,
      "step": 7424
    },
    {
      "epoch": 0.9503391782925893,
      "grad_norm": 0.62890625,
      "learning_rate": 1.290884560283767e-05,
      "loss": 1.2802,
      "step": 7425
    },
    {
      "epoch": 0.9504671701011135,
      "grad_norm": 0.66015625,
      "learning_rate": 1.2842533518325317e-05,
      "loss": 1.38,
      "step": 7426
    },
    {
      "epoch": 0.9505951619096378,
      "grad_norm": 0.67578125,
      "learning_rate": 1.27763910928399e-05,
      "loss": 1.5536,
      "step": 7427
    },
    {
      "epoch": 0.9507231537181621,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2710418337749108e-05,
      "loss": 1.3478,
      "step": 7428
    },
    {
      "epoch": 0.9508511455266863,
      "grad_norm": 0.6875,
      "learning_rate": 1.2644615264391535e-05,
      "loss": 1.4781,
      "step": 7429
    },
    {
      "epoch": 0.9509791373352106,
      "grad_norm": 0.59375,
      "learning_rate": 1.2578981884076469e-05,
      "loss": 1.089,
      "step": 7430
    },
    {
      "epoch": 0.9511071291437349,
      "grad_norm": 0.734375,
      "learning_rate": 1.2513518208083997e-05,
      "loss": 1.6568,
      "step": 7431
    },
    {
      "epoch": 0.951235120952259,
      "grad_norm": 0.70703125,
      "learning_rate": 1.2448224247665118e-05,
      "loss": 1.2663,
      "step": 7432
    },
    {
      "epoch": 0.9513631127607833,
      "grad_norm": 0.8046875,
      "learning_rate": 1.2383100014041748e-05,
      "loss": 1.4813,
      "step": 7433
    },
    {
      "epoch": 0.9514911045693075,
      "grad_norm": 0.46875,
      "learning_rate": 1.2318145518406376e-05,
      "loss": 0.6611,
      "step": 7434
    },
    {
      "epoch": 0.9516190963778318,
      "grad_norm": 0.69140625,
      "learning_rate": 1.2253360771922739e-05,
      "loss": 1.6714,
      "step": 7435
    },
    {
      "epoch": 0.9517470881863561,
      "grad_norm": 0.890625,
      "learning_rate": 1.218874578572493e-05,
      "loss": 1.8971,
      "step": 7436
    },
    {
      "epoch": 0.9518750799948803,
      "grad_norm": 0.703125,
      "learning_rate": 1.2124300570918179e-05,
      "loss": 0.9184,
      "step": 7437
    },
    {
      "epoch": 0.9520030718034046,
      "grad_norm": 0.68359375,
      "learning_rate": 1.2060025138578402e-05,
      "loss": 1.44,
      "step": 7438
    },
    {
      "epoch": 0.9521310636119288,
      "grad_norm": 0.6484375,
      "learning_rate": 1.1995919499752427e-05,
      "loss": 1.1189,
      "step": 7439
    },
    {
      "epoch": 0.9522590554204531,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1931983665457668e-05,
      "loss": 1.3124,
      "step": 7440
    },
    {
      "epoch": 0.9523870472289774,
      "grad_norm": 0.69921875,
      "learning_rate": 1.1868217646682888e-05,
      "loss": 1.2806,
      "step": 7441
    },
    {
      "epoch": 0.9525150390375016,
      "grad_norm": 0.80859375,
      "learning_rate": 1.1804621454386987e-05,
      "loss": 1.3877,
      "step": 7442
    },
    {
      "epoch": 0.9526430308460259,
      "grad_norm": 0.9296875,
      "learning_rate": 1.1741195099500113e-05,
      "loss": 1.3672,
      "step": 7443
    },
    {
      "epoch": 0.9527710226545502,
      "grad_norm": 0.6171875,
      "learning_rate": 1.167793859292321e-05,
      "loss": 1.005,
      "step": 7444
    },
    {
      "epoch": 0.9528990144630743,
      "grad_norm": 0.64453125,
      "learning_rate": 1.1614851945527804e-05,
      "loss": 1.4338,
      "step": 7445
    },
    {
      "epoch": 0.9530270062715986,
      "grad_norm": 0.4765625,
      "learning_rate": 1.1551935168156336e-05,
      "loss": 0.6883,
      "step": 7446
    },
    {
      "epoch": 0.9531549980801228,
      "grad_norm": 0.52734375,
      "learning_rate": 1.1489188271622154e-05,
      "loss": 0.8097,
      "step": 7447
    },
    {
      "epoch": 0.9532829898886471,
      "grad_norm": 0.73046875,
      "learning_rate": 1.1426611266709408e-05,
      "loss": 1.5806,
      "step": 7448
    },
    {
      "epoch": 0.9534109816971714,
      "grad_norm": 0.55859375,
      "learning_rate": 1.136420416417272e-05,
      "loss": 1.115,
      "step": 7449
    },
    {
      "epoch": 0.9535389735056956,
      "grad_norm": 0.671875,
      "learning_rate": 1.1301966974738065e-05,
      "loss": 1.4782,
      "step": 7450
    },
    {
      "epoch": 0.9536669653142199,
      "grad_norm": 0.61328125,
      "learning_rate": 1.1239899709101775e-05,
      "loss": 1.2902,
      "step": 7451
    },
    {
      "epoch": 0.9537949571227441,
      "grad_norm": 0.6171875,
      "learning_rate": 1.1178002377930984e-05,
      "loss": 0.9425,
      "step": 7452
    },
    {
      "epoch": 0.9539229489312684,
      "grad_norm": 0.6875,
      "learning_rate": 1.1116274991864073e-05,
      "loss": 1.3138,
      "step": 7453
    },
    {
      "epoch": 0.9540509407397927,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1054717561509553e-05,
      "loss": 1.771,
      "step": 7454
    },
    {
      "epoch": 0.9541789325483169,
      "grad_norm": 0.59765625,
      "learning_rate": 1.0993330097447296e-05,
      "loss": 1.1265,
      "step": 7455
    },
    {
      "epoch": 0.9543069243568412,
      "grad_norm": 0.7265625,
      "learning_rate": 1.093211261022764e-05,
      "loss": 1.0729,
      "step": 7456
    },
    {
      "epoch": 0.9544349161653655,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0871065110371724e-05,
      "loss": 1.2602,
      "step": 7457
    },
    {
      "epoch": 0.9545629079738897,
      "grad_norm": 0.609375,
      "learning_rate": 1.0810187608371824e-05,
      "loss": 1.1589,
      "step": 7458
    },
    {
      "epoch": 0.954690899782414,
      "grad_norm": 0.60546875,
      "learning_rate": 1.0749480114690458e-05,
      "loss": 1.0745,
      "step": 7459
    },
    {
      "epoch": 0.9548188915909381,
      "grad_norm": 0.60546875,
      "learning_rate": 1.0688942639761278e-05,
      "loss": 1.3958,
      "step": 7460
    },
    {
      "epoch": 0.9549468833994624,
      "grad_norm": 0.58203125,
      "learning_rate": 1.0628575193988633e-05,
      "loss": 1.0572,
      "step": 7461
    },
    {
      "epoch": 0.9550748752079867,
      "grad_norm": 0.65234375,
      "learning_rate": 1.0568377787747664e-05,
      "loss": 1.4697,
      "step": 7462
    },
    {
      "epoch": 0.9552028670165109,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0508350431384206e-05,
      "loss": 1.1572,
      "step": 7463
    },
    {
      "epoch": 0.9553308588250352,
      "grad_norm": 0.63671875,
      "learning_rate": 1.044849313521501e-05,
      "loss": 1.2422,
      "step": 7464
    },
    {
      "epoch": 0.9554588506335594,
      "grad_norm": 0.65625,
      "learning_rate": 1.038880590952762e-05,
      "loss": 1.2621,
      "step": 7465
    },
    {
      "epoch": 0.9555868424420837,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0329288764580058e-05,
      "loss": 1.5431,
      "step": 7466
    },
    {
      "epoch": 0.955714834250608,
      "grad_norm": 0.76953125,
      "learning_rate": 1.026994171060136e-05,
      "loss": 0.9013,
      "step": 7467
    },
    {
      "epoch": 0.9558428260591322,
      "grad_norm": 0.625,
      "learning_rate": 1.0210764757791369e-05,
      "loss": 1.171,
      "step": 7468
    },
    {
      "epoch": 0.9559708178676565,
      "grad_norm": 0.703125,
      "learning_rate": 1.0151757916320504e-05,
      "loss": 1.003,
      "step": 7469
    },
    {
      "epoch": 0.9560988096761808,
      "grad_norm": 0.67578125,
      "learning_rate": 1.0092921196330096e-05,
      "loss": 1.1393,
      "step": 7470
    },
    {
      "epoch": 0.956226801484705,
      "grad_norm": 0.6796875,
      "learning_rate": 1.0034254607932169e-05,
      "loss": 1.2497,
      "step": 7471
    },
    {
      "epoch": 0.9563547932932293,
      "grad_norm": 0.55078125,
      "learning_rate": 9.975758161209547e-06,
      "loss": 0.8546,
      "step": 7472
    },
    {
      "epoch": 0.9564827851017534,
      "grad_norm": 0.62109375,
      "learning_rate": 9.917431866215742e-06,
      "loss": 1.0306,
      "step": 7473
    },
    {
      "epoch": 0.9566107769102777,
      "grad_norm": 0.5859375,
      "learning_rate": 9.85927573297518e-06,
      "loss": 1.2612,
      "step": 7474
    },
    {
      "epoch": 0.956738768718802,
      "grad_norm": 0.703125,
      "learning_rate": 9.801289771482979e-06,
      "loss": 1.4966,
      "step": 7475
    },
    {
      "epoch": 0.9568667605273262,
      "grad_norm": 0.62890625,
      "learning_rate": 9.743473991704721e-06,
      "loss": 1.4056,
      "step": 7476
    },
    {
      "epoch": 0.9569947523358505,
      "grad_norm": 0.734375,
      "learning_rate": 9.685828403577125e-06,
      "loss": 1.8793,
      "step": 7477
    },
    {
      "epoch": 0.9571227441443748,
      "grad_norm": 0.75390625,
      "learning_rate": 9.628353017007595e-06,
      "loss": 1.4277,
      "step": 7478
    },
    {
      "epoch": 0.957250735952899,
      "grad_norm": 0.734375,
      "learning_rate": 9.571047841874126e-06,
      "loss": 1.3427,
      "step": 7479
    },
    {
      "epoch": 0.9573787277614233,
      "grad_norm": 0.59375,
      "learning_rate": 9.513912888025612e-06,
      "loss": 0.7426,
      "step": 7480
    },
    {
      "epoch": 0.9575067195699475,
      "grad_norm": 0.80859375,
      "learning_rate": 9.456948165281421e-06,
      "loss": 2.5019,
      "step": 7481
    },
    {
      "epoch": 0.9576347113784718,
      "grad_norm": 0.56640625,
      "learning_rate": 9.400153683432167e-06,
      "loss": 0.903,
      "step": 7482
    },
    {
      "epoch": 0.9577627031869961,
      "grad_norm": 0.74609375,
      "learning_rate": 9.343529452238597e-06,
      "loss": 1.8025,
      "step": 7483
    },
    {
      "epoch": 0.9578906949955203,
      "grad_norm": 0.71484375,
      "learning_rate": 9.287075481432705e-06,
      "loss": 1.797,
      "step": 7484
    },
    {
      "epoch": 0.9580186868040446,
      "grad_norm": 0.6953125,
      "learning_rate": 9.230791780717063e-06,
      "loss": 1.0188,
      "step": 7485
    },
    {
      "epoch": 0.9581466786125687,
      "grad_norm": 0.57421875,
      "learning_rate": 9.17467835976471e-06,
      "loss": 1.282,
      "step": 7486
    },
    {
      "epoch": 0.958274670421093,
      "grad_norm": 0.5859375,
      "learning_rate": 9.118735228219821e-06,
      "loss": 1.1854,
      "step": 7487
    },
    {
      "epoch": 0.9584026622296173,
      "grad_norm": 0.62890625,
      "learning_rate": 9.062962395697149e-06,
      "loss": 1.2838,
      "step": 7488
    },
    {
      "epoch": 0.9585306540381415,
      "grad_norm": 0.640625,
      "learning_rate": 9.00735987178214e-06,
      "loss": 1.1681,
      "step": 7489
    },
    {
      "epoch": 0.9586586458466658,
      "grad_norm": 0.8359375,
      "learning_rate": 8.951927666030813e-06,
      "loss": 2.0416,
      "step": 7490
    },
    {
      "epoch": 0.9587866376551901,
      "grad_norm": 0.796875,
      "learning_rate": 8.896665787970326e-06,
      "loss": 2.0226,
      "step": 7491
    },
    {
      "epoch": 0.9589146294637143,
      "grad_norm": 0.6484375,
      "learning_rate": 8.841574247098194e-06,
      "loss": 1.2248,
      "step": 7492
    },
    {
      "epoch": 0.9590426212722386,
      "grad_norm": 0.67578125,
      "learning_rate": 8.786653052882842e-06,
      "loss": 1.4571,
      "step": 7493
    },
    {
      "epoch": 0.9591706130807628,
      "grad_norm": 0.66796875,
      "learning_rate": 8.731902214763387e-06,
      "loss": 1.0433,
      "step": 7494
    },
    {
      "epoch": 0.9592986048892871,
      "grad_norm": 0.6953125,
      "learning_rate": 8.677321742149524e-06,
      "loss": 1.2215,
      "step": 7495
    },
    {
      "epoch": 0.9594265966978114,
      "grad_norm": 0.65625,
      "learning_rate": 8.622911644421972e-06,
      "loss": 1.0617,
      "step": 7496
    },
    {
      "epoch": 0.9595545885063356,
      "grad_norm": 0.609375,
      "learning_rate": 8.568671930931915e-06,
      "loss": 1.2573,
      "step": 7497
    },
    {
      "epoch": 0.9596825803148599,
      "grad_norm": 0.5390625,
      "learning_rate": 8.514602611001343e-06,
      "loss": 1.0874,
      "step": 7498
    },
    {
      "epoch": 0.959810572123384,
      "grad_norm": 0.6328125,
      "learning_rate": 8.460703693922933e-06,
      "loss": 1.0882,
      "step": 7499
    },
    {
      "epoch": 0.9599385639319084,
      "grad_norm": 0.85546875,
      "learning_rate": 8.406975188959943e-06,
      "loss": 2.3958,
      "step": 7500
    },
    {
      "epoch": 0.9600665557404326,
      "grad_norm": 0.75,
      "learning_rate": 8.353417105346762e-06,
      "loss": 1.4024,
      "step": 7501
    },
    {
      "epoch": 0.9601945475489568,
      "grad_norm": 0.703125,
      "learning_rate": 8.300029452288027e-06,
      "loss": 1.9261,
      "step": 7502
    },
    {
      "epoch": 0.9603225393574811,
      "grad_norm": 0.6015625,
      "learning_rate": 8.2468122389594e-06,
      "loss": 0.9134,
      "step": 7503
    },
    {
      "epoch": 0.9604505311660054,
      "grad_norm": 0.765625,
      "learning_rate": 8.193765474506899e-06,
      "loss": 1.4316,
      "step": 7504
    },
    {
      "epoch": 0.9605785229745296,
      "grad_norm": 0.70703125,
      "learning_rate": 8.140889168047671e-06,
      "loss": 1.4356,
      "step": 7505
    },
    {
      "epoch": 0.9607065147830539,
      "grad_norm": 0.57421875,
      "learning_rate": 8.088183328669341e-06,
      "loss": 1.0342,
      "step": 7506
    },
    {
      "epoch": 0.9608345065915781,
      "grad_norm": 0.53125,
      "learning_rate": 8.035647965430215e-06,
      "loss": 1.2967,
      "step": 7507
    },
    {
      "epoch": 0.9609624984001024,
      "grad_norm": 0.66796875,
      "learning_rate": 7.983283087359406e-06,
      "loss": 1.4675,
      "step": 7508
    },
    {
      "epoch": 0.9610904902086267,
      "grad_norm": 0.640625,
      "learning_rate": 7.931088703456602e-06,
      "loss": 1.3239,
      "step": 7509
    },
    {
      "epoch": 0.9612184820171509,
      "grad_norm": 0.703125,
      "learning_rate": 7.879064822692294e-06,
      "loss": 1.1895,
      "step": 7510
    },
    {
      "epoch": 0.9613464738256752,
      "grad_norm": 0.796875,
      "learning_rate": 7.827211454007666e-06,
      "loss": 2.0297,
      "step": 7511
    },
    {
      "epoch": 0.9614744656341994,
      "grad_norm": 0.56640625,
      "learning_rate": 7.775528606314365e-06,
      "loss": 1.1003,
      "step": 7512
    },
    {
      "epoch": 0.9616024574427237,
      "grad_norm": 0.65625,
      "learning_rate": 7.724016288495173e-06,
      "loss": 0.8079,
      "step": 7513
    },
    {
      "epoch": 0.961730449251248,
      "grad_norm": 0.65234375,
      "learning_rate": 7.672674509403232e-06,
      "loss": 1.1731,
      "step": 7514
    },
    {
      "epoch": 0.9618584410597721,
      "grad_norm": 0.490234375,
      "learning_rate": 7.621503277862485e-06,
      "loss": 0.8523,
      "step": 7515
    },
    {
      "epoch": 0.9619864328682964,
      "grad_norm": 0.71875,
      "learning_rate": 7.570502602667451e-06,
      "loss": 1.0342,
      "step": 7516
    },
    {
      "epoch": 0.9621144246768207,
      "grad_norm": 0.83984375,
      "learning_rate": 7.5196724925833405e-06,
      "loss": 1.5056,
      "step": 7517
    },
    {
      "epoch": 0.9622424164853449,
      "grad_norm": 0.73046875,
      "learning_rate": 7.469012956346388e-06,
      "loss": 1.2152,
      "step": 7518
    },
    {
      "epoch": 0.9623704082938692,
      "grad_norm": 0.62890625,
      "learning_rate": 7.418524002663296e-06,
      "loss": 1.3152,
      "step": 7519
    },
    {
      "epoch": 0.9624984001023934,
      "grad_norm": 0.5703125,
      "learning_rate": 7.368205640211012e-06,
      "loss": 0.9833,
      "step": 7520
    },
    {
      "epoch": 0.9626263919109177,
      "grad_norm": 0.6015625,
      "learning_rate": 7.31805787763784e-06,
      "loss": 1.009,
      "step": 7521
    },
    {
      "epoch": 0.962754383719442,
      "grad_norm": 0.5625,
      "learning_rate": 7.268080723562553e-06,
      "loss": 1.1865,
      "step": 7522
    },
    {
      "epoch": 0.9628823755279662,
      "grad_norm": 0.7421875,
      "learning_rate": 7.218274186574281e-06,
      "loss": 1.3723,
      "step": 7523
    },
    {
      "epoch": 0.9630103673364905,
      "grad_norm": 0.625,
      "learning_rate": 7.168638275233397e-06,
      "loss": 1.4144,
      "step": 7524
    },
    {
      "epoch": 0.9631383591450147,
      "grad_norm": 0.86328125,
      "learning_rate": 7.119172998070411e-06,
      "loss": 1.1453,
      "step": 7525
    },
    {
      "epoch": 0.963266350953539,
      "grad_norm": 0.625,
      "learning_rate": 7.069878363586746e-06,
      "loss": 1.4703,
      "step": 7526
    },
    {
      "epoch": 0.9633943427620633,
      "grad_norm": 0.6484375,
      "learning_rate": 7.020754380254624e-06,
      "loss": 1.1549,
      "step": 7527
    },
    {
      "epoch": 0.9635223345705874,
      "grad_norm": 0.65625,
      "learning_rate": 6.971801056516624e-06,
      "loss": 1.2478,
      "step": 7528
    },
    {
      "epoch": 0.9636503263791117,
      "grad_norm": 0.54296875,
      "learning_rate": 6.923018400786241e-06,
      "loss": 0.7673,
      "step": 7529
    },
    {
      "epoch": 0.963778318187636,
      "grad_norm": 0.828125,
      "learning_rate": 6.8744064214476545e-06,
      "loss": 1.5444,
      "step": 7530
    },
    {
      "epoch": 0.9639063099961602,
      "grad_norm": 0.6484375,
      "learning_rate": 6.8259651268555154e-06,
      "loss": 1.1685,
      "step": 7531
    },
    {
      "epoch": 0.9640343018046845,
      "grad_norm": 0.66015625,
      "learning_rate": 6.777694525335276e-06,
      "loss": 1.1252,
      "step": 7532
    },
    {
      "epoch": 0.9641622936132087,
      "grad_norm": 0.66015625,
      "learning_rate": 6.729594625183078e-06,
      "loss": 0.9325,
      "step": 7533
    },
    {
      "epoch": 0.964290285421733,
      "grad_norm": 0.68359375,
      "learning_rate": 6.6816654346655296e-06,
      "loss": 1.3637,
      "step": 7534
    },
    {
      "epoch": 0.9644182772302573,
      "grad_norm": 0.62109375,
      "learning_rate": 6.633906962020153e-06,
      "loss": 1.158,
      "step": 7535
    },
    {
      "epoch": 0.9645462690387815,
      "grad_norm": 0.64453125,
      "learning_rate": 6.586319215454939e-06,
      "loss": 1.2123,
      "step": 7536
    },
    {
      "epoch": 0.9646742608473058,
      "grad_norm": 0.640625,
      "learning_rate": 6.538902203148789e-06,
      "loss": 1.4784,
      "step": 7537
    },
    {
      "epoch": 0.96480225265583,
      "grad_norm": 0.71484375,
      "learning_rate": 6.491655933250962e-06,
      "loss": 1.1367,
      "step": 7538
    },
    {
      "epoch": 0.9649302444643543,
      "grad_norm": 0.6171875,
      "learning_rate": 6.4445804138814065e-06,
      "loss": 1.3062,
      "step": 7539
    },
    {
      "epoch": 0.9650582362728786,
      "grad_norm": 0.64453125,
      "learning_rate": 6.3976756531308745e-06,
      "loss": 1.4394,
      "step": 7540
    },
    {
      "epoch": 0.9651862280814028,
      "grad_norm": 0.64453125,
      "learning_rate": 6.350941659060805e-06,
      "loss": 1.8225,
      "step": 7541
    },
    {
      "epoch": 0.965314219889927,
      "grad_norm": 0.76953125,
      "learning_rate": 6.304378439703106e-06,
      "loss": 1.1458,
      "step": 7542
    },
    {
      "epoch": 0.9654422116984513,
      "grad_norm": 0.6640625,
      "learning_rate": 6.257986003060489e-06,
      "loss": 1.2515,
      "step": 7543
    },
    {
      "epoch": 0.9655702035069755,
      "grad_norm": 0.609375,
      "learning_rate": 6.211764357106131e-06,
      "loss": 1.1473,
      "step": 7544
    },
    {
      "epoch": 0.9656981953154998,
      "grad_norm": 0.69140625,
      "learning_rate": 6.16571350978401e-06,
      "loss": 1.4845,
      "step": 7545
    },
    {
      "epoch": 0.965826187124024,
      "grad_norm": 0.765625,
      "learning_rate": 6.119833469008906e-06,
      "loss": 1.0006,
      "step": 7546
    },
    {
      "epoch": 0.9659541789325483,
      "grad_norm": 0.72265625,
      "learning_rate": 6.074124242665735e-06,
      "loss": 1.511,
      "step": 7547
    },
    {
      "epoch": 0.9660821707410726,
      "grad_norm": 0.671875,
      "learning_rate": 6.028585838610656e-06,
      "loss": 1.1538,
      "step": 7548
    },
    {
      "epoch": 0.9662101625495968,
      "grad_norm": 0.60546875,
      "learning_rate": 5.983218264669965e-06,
      "loss": 1.3544,
      "step": 7549
    },
    {
      "epoch": 0.9663381543581211,
      "grad_norm": 0.69921875,
      "learning_rate": 5.938021528640869e-06,
      "loss": 2.0538,
      "step": 7550
    },
    {
      "epoch": 0.9664661461666454,
      "grad_norm": 0.80859375,
      "learning_rate": 5.892995638291155e-06,
      "loss": 1.7865,
      "step": 7551
    },
    {
      "epoch": 0.9665941379751696,
      "grad_norm": 0.546875,
      "learning_rate": 5.848140601359408e-06,
      "loss": 0.8754,
      "step": 7552
    },
    {
      "epoch": 0.9667221297836939,
      "grad_norm": 0.53515625,
      "learning_rate": 5.8034564255544654e-06,
      "loss": 1.0543,
      "step": 7553
    },
    {
      "epoch": 0.9668501215922181,
      "grad_norm": 0.6953125,
      "learning_rate": 5.758943118556181e-06,
      "loss": 1.474,
      "step": 7554
    },
    {
      "epoch": 0.9669781134007424,
      "grad_norm": 0.71484375,
      "learning_rate": 5.714600688014882e-06,
      "loss": 1.2033,
      "step": 7555
    },
    {
      "epoch": 0.9671061052092667,
      "grad_norm": 0.66015625,
      "learning_rate": 5.67042914155147e-06,
      "loss": 1.1382,
      "step": 7556
    },
    {
      "epoch": 0.9672340970177908,
      "grad_norm": 0.74609375,
      "learning_rate": 5.62642848675754e-06,
      "loss": 1.4167,
      "step": 7557
    },
    {
      "epoch": 0.9673620888263151,
      "grad_norm": 0.7421875,
      "learning_rate": 5.582598731195376e-06,
      "loss": 1.4513,
      "step": 7558
    },
    {
      "epoch": 0.9674900806348393,
      "grad_norm": 0.58984375,
      "learning_rate": 5.538939882397842e-06,
      "loss": 1.0617,
      "step": 7559
    },
    {
      "epoch": 0.9676180724433636,
      "grad_norm": 0.6171875,
      "learning_rate": 5.4954519478682685e-06,
      "loss": 1.3555,
      "step": 7560
    },
    {
      "epoch": 0.9677460642518879,
      "grad_norm": 0.61328125,
      "learning_rate": 5.452134935080899e-06,
      "loss": 1.2107,
      "step": 7561
    },
    {
      "epoch": 0.9678740560604121,
      "grad_norm": 0.7734375,
      "learning_rate": 5.408988851480445e-06,
      "loss": 1.4609,
      "step": 7562
    },
    {
      "epoch": 0.9680020478689364,
      "grad_norm": 0.53515625,
      "learning_rate": 5.366013704482309e-06,
      "loss": 0.8075,
      "step": 7563
    },
    {
      "epoch": 0.9681300396774607,
      "grad_norm": 0.9375,
      "learning_rate": 5.3232095014723595e-06,
      "loss": 1.8681,
      "step": 7564
    },
    {
      "epoch": 0.9682580314859849,
      "grad_norm": 0.67578125,
      "learning_rate": 5.280576249807268e-06,
      "loss": 1.6119,
      "step": 7565
    },
    {
      "epoch": 0.9683860232945092,
      "grad_norm": 0.69140625,
      "learning_rate": 5.238113956814283e-06,
      "loss": 1.4864,
      "step": 7566
    },
    {
      "epoch": 0.9685140151030334,
      "grad_norm": 0.63671875,
      "learning_rate": 5.195822629791125e-06,
      "loss": 1.4004,
      "step": 7567
    },
    {
      "epoch": 0.9686420069115577,
      "grad_norm": 0.6796875,
      "learning_rate": 5.153702276006311e-06,
      "loss": 1.2327,
      "step": 7568
    },
    {
      "epoch": 0.968769998720082,
      "grad_norm": 0.703125,
      "learning_rate": 5.111752902698941e-06,
      "loss": 1.2283,
      "step": 7569
    },
    {
      "epoch": 0.9688979905286061,
      "grad_norm": 0.62890625,
      "learning_rate": 5.06997451707858e-06,
      "loss": 1.386,
      "step": 7570
    },
    {
      "epoch": 0.9690259823371304,
      "grad_norm": 0.640625,
      "learning_rate": 5.028367126325595e-06,
      "loss": 0.9887,
      "step": 7571
    },
    {
      "epoch": 0.9691539741456546,
      "grad_norm": 0.62109375,
      "learning_rate": 4.986930737590933e-06,
      "loss": 1.3942,
      "step": 7572
    },
    {
      "epoch": 0.9692819659541789,
      "grad_norm": 0.73828125,
      "learning_rate": 4.9456653579961204e-06,
      "loss": 1.2143,
      "step": 7573
    },
    {
      "epoch": 0.9694099577627032,
      "grad_norm": 0.796875,
      "learning_rate": 4.90457099463315e-06,
      "loss": 1.3556,
      "step": 7574
    },
    {
      "epoch": 0.9695379495712274,
      "grad_norm": 0.64453125,
      "learning_rate": 4.863647654564929e-06,
      "loss": 1.3877,
      "step": 7575
    },
    {
      "epoch": 0.9696659413797517,
      "grad_norm": 0.447265625,
      "learning_rate": 4.822895344824607e-06,
      "loss": 0.5977,
      "step": 7576
    },
    {
      "epoch": 0.969793933188276,
      "grad_norm": 0.55859375,
      "learning_rate": 4.782314072416361e-06,
      "loss": 0.9316,
      "step": 7577
    },
    {
      "epoch": 0.9699219249968002,
      "grad_norm": 0.73828125,
      "learning_rate": 4.741903844314499e-06,
      "loss": 2.2112,
      "step": 7578
    },
    {
      "epoch": 0.9700499168053245,
      "grad_norm": 0.77734375,
      "learning_rate": 4.701664667464245e-06,
      "loss": 1.6217,
      "step": 7579
    },
    {
      "epoch": 0.9701779086138487,
      "grad_norm": 0.46484375,
      "learning_rate": 4.66159654878151e-06,
      "loss": 0.6803,
      "step": 7580
    },
    {
      "epoch": 0.970305900422373,
      "grad_norm": 0.59765625,
      "learning_rate": 4.621699495152454e-06,
      "loss": 0.9801,
      "step": 7581
    },
    {
      "epoch": 0.9704338922308973,
      "grad_norm": 0.62109375,
      "learning_rate": 4.581973513434145e-06,
      "loss": 0.8224,
      "step": 7582
    },
    {
      "epoch": 0.9705618840394215,
      "grad_norm": 0.60546875,
      "learning_rate": 4.542418610454124e-06,
      "loss": 0.8478,
      "step": 7583
    },
    {
      "epoch": 0.9706898758479457,
      "grad_norm": 0.59765625,
      "learning_rate": 4.503034793010508e-06,
      "loss": 0.8711,
      "step": 7584
    },
    {
      "epoch": 0.9708178676564699,
      "grad_norm": 0.69140625,
      "learning_rate": 4.463822067871992e-06,
      "loss": 1.5242,
      "step": 7585
    },
    {
      "epoch": 0.9709458594649942,
      "grad_norm": 0.5859375,
      "learning_rate": 4.424780441778076e-06,
      "loss": 0.8084,
      "step": 7586
    },
    {
      "epoch": 0.9710738512735185,
      "grad_norm": 0.7890625,
      "learning_rate": 4.385909921438503e-06,
      "loss": 1.9207,
      "step": 7587
    },
    {
      "epoch": 0.9712018430820427,
      "grad_norm": 0.58203125,
      "learning_rate": 4.347210513534039e-06,
      "loss": 0.9794,
      "step": 7588
    },
    {
      "epoch": 0.971329834890567,
      "grad_norm": 0.73046875,
      "learning_rate": 4.308682224715588e-06,
      "loss": 1.5718,
      "step": 7589
    },
    {
      "epoch": 0.9714578266990913,
      "grad_norm": 0.875,
      "learning_rate": 4.2703250616049625e-06,
      "loss": 1.1393,
      "step": 7590
    },
    {
      "epoch": 0.9715858185076155,
      "grad_norm": 0.79296875,
      "learning_rate": 4.232139030794447e-06,
      "loss": 1.324,
      "step": 7591
    },
    {
      "epoch": 0.9717138103161398,
      "grad_norm": 0.609375,
      "learning_rate": 4.194124138847011e-06,
      "loss": 1.1397,
      "step": 7592
    },
    {
      "epoch": 0.971841802124664,
      "grad_norm": 0.66015625,
      "learning_rate": 4.156280392295986e-06,
      "loss": 1.2762,
      "step": 7593
    },
    {
      "epoch": 0.9719697939331883,
      "grad_norm": 0.65625,
      "learning_rate": 4.118607797645502e-06,
      "loss": 1.0449,
      "step": 7594
    },
    {
      "epoch": 0.9720977857417126,
      "grad_norm": 0.8984375,
      "learning_rate": 4.081106361370157e-06,
      "loss": 1.7073,
      "step": 7595
    },
    {
      "epoch": 0.9722257775502368,
      "grad_norm": 0.6875,
      "learning_rate": 4.043776089915241e-06,
      "loss": 1.1832,
      "step": 7596
    },
    {
      "epoch": 0.972353769358761,
      "grad_norm": 0.69921875,
      "learning_rate": 4.006616989696621e-06,
      "loss": 1.3325,
      "step": 7597
    },
    {
      "epoch": 0.9724817611672852,
      "grad_norm": 0.7421875,
      "learning_rate": 3.969629067100522e-06,
      "loss": 1.5695,
      "step": 7598
    },
    {
      "epoch": 0.9726097529758095,
      "grad_norm": 0.6171875,
      "learning_rate": 3.932812328484081e-06,
      "loss": 1.1152,
      "step": 7599
    },
    {
      "epoch": 0.9727377447843338,
      "grad_norm": 0.69140625,
      "learning_rate": 3.896166780174792e-06,
      "loss": 1.1303,
      "step": 7600
    },
    {
      "epoch": 0.972865736592858,
      "grad_norm": 0.5390625,
      "learning_rate": 3.85969242847084e-06,
      "loss": 1.0784,
      "step": 7601
    },
    {
      "epoch": 0.9729937284013823,
      "grad_norm": 0.6328125,
      "learning_rate": 3.8233892796407655e-06,
      "loss": 1.1425,
      "step": 7602
    },
    {
      "epoch": 0.9731217202099066,
      "grad_norm": 0.625,
      "learning_rate": 3.787257339924133e-06,
      "loss": 1.5591,
      "step": 7603
    },
    {
      "epoch": 0.9732497120184308,
      "grad_norm": 0.59375,
      "learning_rate": 3.7512966155305307e-06,
      "loss": 1.3087,
      "step": 7604
    },
    {
      "epoch": 0.9733777038269551,
      "grad_norm": 0.625,
      "learning_rate": 3.715507112640459e-06,
      "loss": 1.5297,
      "step": 7605
    },
    {
      "epoch": 0.9735056956354793,
      "grad_norm": 0.84375,
      "learning_rate": 3.679888837404999e-06,
      "loss": 1.2543,
      "step": 7606
    },
    {
      "epoch": 0.9736336874440036,
      "grad_norm": 0.6875,
      "learning_rate": 3.6444417959456965e-06,
      "loss": 1.3066,
      "step": 7607
    },
    {
      "epoch": 0.9737616792525279,
      "grad_norm": 0.60546875,
      "learning_rate": 3.6091659943546796e-06,
      "loss": 1.2266,
      "step": 7608
    },
    {
      "epoch": 0.9738896710610521,
      "grad_norm": 0.65234375,
      "learning_rate": 3.5740614386947647e-06,
      "loss": 1.2107,
      "step": 7609
    },
    {
      "epoch": 0.9740176628695764,
      "grad_norm": 0.79296875,
      "learning_rate": 3.5391281349991257e-06,
      "loss": 1.4656,
      "step": 7610
    },
    {
      "epoch": 0.9741456546781005,
      "grad_norm": 0.56640625,
      "learning_rate": 3.5043660892716265e-06,
      "loss": 1.0906,
      "step": 7611
    },
    {
      "epoch": 0.9742736464866248,
      "grad_norm": 0.65234375,
      "learning_rate": 3.469775307486711e-06,
      "loss": 1.2089,
      "step": 7612
    },
    {
      "epoch": 0.9744016382951491,
      "grad_norm": 0.64453125,
      "learning_rate": 3.435355795589401e-06,
      "loss": 1.0741,
      "step": 7613
    },
    {
      "epoch": 0.9745296301036733,
      "grad_norm": 0.59375,
      "learning_rate": 3.401107559495187e-06,
      "loss": 0.9924,
      "step": 7614
    },
    {
      "epoch": 0.9746576219121976,
      "grad_norm": 0.578125,
      "learning_rate": 3.367030605090249e-06,
      "loss": 1.1738,
      "step": 7615
    },
    {
      "epoch": 0.9747856137207219,
      "grad_norm": 0.62109375,
      "learning_rate": 3.3331249382312357e-06,
      "loss": 1.1016,
      "step": 7616
    },
    {
      "epoch": 0.9749136055292461,
      "grad_norm": 0.59765625,
      "learning_rate": 3.299390564745264e-06,
      "loss": 0.8735,
      "step": 7617
    },
    {
      "epoch": 0.9750415973377704,
      "grad_norm": 0.67578125,
      "learning_rate": 3.265827490430362e-06,
      "loss": 1.2482,
      "step": 7618
    },
    {
      "epoch": 0.9751695891462946,
      "grad_norm": 0.68359375,
      "learning_rate": 3.232435721054805e-06,
      "loss": 1.1521,
      "step": 7619
    },
    {
      "epoch": 0.9752975809548189,
      "grad_norm": 0.69140625,
      "learning_rate": 3.1992152623574465e-06,
      "loss": 1.1189,
      "step": 7620
    },
    {
      "epoch": 0.9754255727633432,
      "grad_norm": 0.6484375,
      "learning_rate": 3.166166120047831e-06,
      "loss": 1.3359,
      "step": 7621
    },
    {
      "epoch": 0.9755535645718674,
      "grad_norm": 0.60546875,
      "learning_rate": 3.133288299805859e-06,
      "loss": 1.242,
      "step": 7622
    },
    {
      "epoch": 0.9756815563803917,
      "grad_norm": 0.7265625,
      "learning_rate": 3.1005818072822324e-06,
      "loss": 1.5072,
      "step": 7623
    },
    {
      "epoch": 0.9758095481889159,
      "grad_norm": 0.62890625,
      "learning_rate": 3.068046648098122e-06,
      "loss": 1.1459,
      "step": 7624
    },
    {
      "epoch": 0.9759375399974402,
      "grad_norm": 0.64453125,
      "learning_rate": 3.0356828278451653e-06,
      "loss": 0.814,
      "step": 7625
    },
    {
      "epoch": 0.9760655318059644,
      "grad_norm": 0.60546875,
      "learning_rate": 3.0034903520856916e-06,
      "loss": 1.0374,
      "step": 7626
    },
    {
      "epoch": 0.9761935236144886,
      "grad_norm": 0.80859375,
      "learning_rate": 2.971469226352275e-06,
      "loss": 1.3171,
      "step": 7627
    },
    {
      "epoch": 0.9763215154230129,
      "grad_norm": 0.625,
      "learning_rate": 2.9396194561485125e-06,
      "loss": 1.0373,
      "step": 7628
    },
    {
      "epoch": 0.9764495072315372,
      "grad_norm": 0.65234375,
      "learning_rate": 2.907941046948248e-06,
      "loss": 1.21,
      "step": 7629
    },
    {
      "epoch": 0.9765774990400614,
      "grad_norm": 0.76171875,
      "learning_rate": 2.8764340041957936e-06,
      "loss": 1.5873,
      "step": 7630
    },
    {
      "epoch": 0.9767054908485857,
      "grad_norm": 0.64453125,
      "learning_rate": 2.845098333306262e-06,
      "loss": 1.2388,
      "step": 7631
    },
    {
      "epoch": 0.9768334826571099,
      "grad_norm": 0.69140625,
      "learning_rate": 2.8139340396652336e-06,
      "loss": 1.0252,
      "step": 7632
    },
    {
      "epoch": 0.9769614744656342,
      "grad_norm": 0.69140625,
      "learning_rate": 2.7829411286287574e-06,
      "loss": 1.8253,
      "step": 7633
    },
    {
      "epoch": 0.9770894662741585,
      "grad_norm": 0.671875,
      "learning_rate": 2.752119605523351e-06,
      "loss": 1.2156,
      "step": 7634
    },
    {
      "epoch": 0.9772174580826827,
      "grad_norm": 0.73046875,
      "learning_rate": 2.7214694756464433e-06,
      "loss": 1.7647,
      "step": 7635
    },
    {
      "epoch": 0.977345449891207,
      "grad_norm": 0.73828125,
      "learning_rate": 2.690990744265487e-06,
      "loss": 1.158,
      "step": 7636
    },
    {
      "epoch": 0.9774734416997313,
      "grad_norm": 0.59375,
      "learning_rate": 2.6606834166188475e-06,
      "loss": 1.2144,
      "step": 7637
    },
    {
      "epoch": 0.9776014335082555,
      "grad_norm": 0.78515625,
      "learning_rate": 2.6305474979154697e-06,
      "loss": 0.916,
      "step": 7638
    },
    {
      "epoch": 0.9777294253167798,
      "grad_norm": 0.72265625,
      "learning_rate": 2.600582993334544e-06,
      "loss": 1.4087,
      "step": 7639
    },
    {
      "epoch": 0.9778574171253039,
      "grad_norm": 0.53515625,
      "learning_rate": 2.57078990802595e-06,
      "loss": 0.919,
      "step": 7640
    },
    {
      "epoch": 0.9779854089338282,
      "grad_norm": 0.75,
      "learning_rate": 2.541168247110148e-06,
      "loss": 1.4588,
      "step": 7641
    },
    {
      "epoch": 0.9781134007423525,
      "grad_norm": 0.8203125,
      "learning_rate": 2.511718015678177e-06,
      "loss": 2.0011,
      "step": 7642
    },
    {
      "epoch": 0.9782413925508767,
      "grad_norm": 0.58984375,
      "learning_rate": 2.4824392187914323e-06,
      "loss": 0.886,
      "step": 7643
    },
    {
      "epoch": 0.978369384359401,
      "grad_norm": 0.5859375,
      "learning_rate": 2.453331861482e-06,
      "loss": 1.1801,
      "step": 7644
    },
    {
      "epoch": 0.9784973761679252,
      "grad_norm": 0.73046875,
      "learning_rate": 2.4243959487524335e-06,
      "loss": 1.9849,
      "step": 7645
    },
    {
      "epoch": 0.9786253679764495,
      "grad_norm": 0.74609375,
      "learning_rate": 2.3956314855758665e-06,
      "loss": 1.471,
      "step": 7646
    },
    {
      "epoch": 0.9787533597849738,
      "grad_norm": 0.77734375,
      "learning_rate": 2.3670384768958997e-06,
      "loss": 1.4344,
      "step": 7647
    },
    {
      "epoch": 0.978881351593498,
      "grad_norm": 0.7578125,
      "learning_rate": 2.3386169276267134e-06,
      "loss": 1.5816,
      "step": 7648
    },
    {
      "epoch": 0.9790093434020223,
      "grad_norm": 0.6640625,
      "learning_rate": 2.3103668426530668e-06,
      "loss": 1.2392,
      "step": 7649
    },
    {
      "epoch": 0.9791373352105466,
      "grad_norm": 0.5,
      "learning_rate": 2.282288226830187e-06,
      "loss": 0.9088,
      "step": 7650
    },
    {
      "epoch": 0.9792653270190708,
      "grad_norm": 0.9296875,
      "learning_rate": 2.254381084983659e-06,
      "loss": 2.5579,
      "step": 7651
    },
    {
      "epoch": 0.9793933188275951,
      "grad_norm": 0.57421875,
      "learning_rate": 2.2266454219100894e-06,
      "loss": 0.8467,
      "step": 7652
    },
    {
      "epoch": 0.9795213106361192,
      "grad_norm": 0.62890625,
      "learning_rate": 2.199081242376e-06,
      "loss": 1.029,
      "step": 7653
    },
    {
      "epoch": 0.9796493024446435,
      "grad_norm": 0.7734375,
      "learning_rate": 2.171688551118933e-06,
      "loss": 1.5759,
      "step": 7654
    },
    {
      "epoch": 0.9797772942531678,
      "grad_norm": 0.66015625,
      "learning_rate": 2.1444673528466794e-06,
      "loss": 1.3067,
      "step": 7655
    },
    {
      "epoch": 0.979905286061692,
      "grad_norm": 0.7734375,
      "learning_rate": 2.1174176522377184e-06,
      "loss": 1.496,
      "step": 7656
    },
    {
      "epoch": 0.9800332778702163,
      "grad_norm": 0.75390625,
      "learning_rate": 2.090539453940998e-06,
      "loss": 1.2532,
      "step": 7657
    },
    {
      "epoch": 0.9801612696787405,
      "grad_norm": 0.58984375,
      "learning_rate": 2.0638327625758237e-06,
      "loss": 0.8665,
      "step": 7658
    },
    {
      "epoch": 0.9802892614872648,
      "grad_norm": 0.77734375,
      "learning_rate": 2.0372975827321895e-06,
      "loss": 1.1649,
      "step": 7659
    },
    {
      "epoch": 0.9804172532957891,
      "grad_norm": 0.70703125,
      "learning_rate": 2.0109339189707807e-06,
      "loss": 1.1436,
      "step": 7660
    },
    {
      "epoch": 0.9805452451043133,
      "grad_norm": 0.66796875,
      "learning_rate": 1.9847417758225294e-06,
      "loss": 1.4215,
      "step": 7661
    },
    {
      "epoch": 0.9806732369128376,
      "grad_norm": 0.59765625,
      "learning_rate": 1.958721157788834e-06,
      "loss": 1.1823,
      "step": 7662
    },
    {
      "epoch": 0.9808012287213619,
      "grad_norm": 0.58203125,
      "learning_rate": 1.9328720693420065e-06,
      "loss": 1.0413,
      "step": 7663
    },
    {
      "epoch": 0.9809292205298861,
      "grad_norm": 0.71875,
      "learning_rate": 1.907194514924493e-06,
      "loss": 1.5131,
      "step": 7664
    },
    {
      "epoch": 0.9810572123384104,
      "grad_norm": 0.6875,
      "learning_rate": 1.8816884989494298e-06,
      "loss": 1.3872,
      "step": 7665
    },
    {
      "epoch": 0.9811852041469346,
      "grad_norm": 0.62890625,
      "learning_rate": 1.856354025800422e-06,
      "loss": 1.5878,
      "step": 7666
    },
    {
      "epoch": 0.9813131959554589,
      "grad_norm": 0.796875,
      "learning_rate": 1.8311910998315417e-06,
      "loss": 1.1224,
      "step": 7667
    },
    {
      "epoch": 0.9814411877639831,
      "grad_norm": 0.59375,
      "learning_rate": 1.806199725367552e-06,
      "loss": 1.0522,
      "step": 7668
    },
    {
      "epoch": 0.9815691795725073,
      "grad_norm": 0.59765625,
      "learning_rate": 1.781379906703573e-06,
      "loss": 0.942,
      "step": 7669
    },
    {
      "epoch": 0.9816971713810316,
      "grad_norm": 0.6328125,
      "learning_rate": 1.7567316481054142e-06,
      "loss": 1.2008,
      "step": 7670
    },
    {
      "epoch": 0.9818251631895558,
      "grad_norm": 0.6484375,
      "learning_rate": 1.7322549538091315e-06,
      "loss": 1.5894,
      "step": 7671
    },
    {
      "epoch": 0.9819531549980801,
      "grad_norm": 0.68359375,
      "learning_rate": 1.7079498280213602e-06,
      "loss": 1.4512,
      "step": 7672
    },
    {
      "epoch": 0.9820811468066044,
      "grad_norm": 0.65625,
      "learning_rate": 1.6838162749196472e-06,
      "loss": 1.3655,
      "step": 7673
    },
    {
      "epoch": 0.9822091386151286,
      "grad_norm": 0.66015625,
      "learning_rate": 1.659854298651453e-06,
      "loss": 1.5504,
      "step": 7674
    },
    {
      "epoch": 0.9823371304236529,
      "grad_norm": 0.828125,
      "learning_rate": 1.636063903335039e-06,
      "loss": 1.672,
      "step": 7675
    },
    {
      "epoch": 0.9824651222321772,
      "grad_norm": 0.8828125,
      "learning_rate": 1.6124450930593558e-06,
      "loss": 2.3579,
      "step": 7676
    },
    {
      "epoch": 0.9825931140407014,
      "grad_norm": 0.640625,
      "learning_rate": 1.5889978718836019e-06,
      "loss": 1.1299,
      "step": 7677
    },
    {
      "epoch": 0.9827211058492257,
      "grad_norm": 0.55078125,
      "learning_rate": 1.565722243837553e-06,
      "loss": 0.8458,
      "step": 7678
    },
    {
      "epoch": 0.9828490976577499,
      "grad_norm": 0.78515625,
      "learning_rate": 1.542618212921454e-06,
      "loss": 1.0402,
      "step": 7679
    },
    {
      "epoch": 0.9829770894662742,
      "grad_norm": 0.6875,
      "learning_rate": 1.5196857831061284e-06,
      "loss": 1.3139,
      "step": 7680
    },
    {
      "epoch": 0.9831050812747985,
      "grad_norm": 0.67578125,
      "learning_rate": 1.4969249583328682e-06,
      "loss": 1.4203,
      "step": 7681
    },
    {
      "epoch": 0.9832330730833226,
      "grad_norm": 1.1015625,
      "learning_rate": 1.474335742513655e-06,
      "loss": 1.2201,
      "step": 7682
    },
    {
      "epoch": 0.9833610648918469,
      "grad_norm": 0.734375,
      "learning_rate": 1.4519181395306059e-06,
      "loss": 1.3624,
      "step": 7683
    },
    {
      "epoch": 0.9834890567003711,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4296721532366385e-06,
      "loss": 1.3955,
      "step": 7684
    },
    {
      "epoch": 0.9836170485088954,
      "grad_norm": 0.56640625,
      "learning_rate": 1.4075977874550282e-06,
      "loss": 0.723,
      "step": 7685
    },
    {
      "epoch": 0.9837450403174197,
      "grad_norm": 0.671875,
      "learning_rate": 1.3856950459796292e-06,
      "loss": 1.215,
      "step": 7686
    },
    {
      "epoch": 0.9838730321259439,
      "grad_norm": 0.81640625,
      "learning_rate": 1.3639639325748743e-06,
      "loss": 1.7912,
      "step": 7687
    },
    {
      "epoch": 0.9840010239344682,
      "grad_norm": 0.76953125,
      "learning_rate": 1.342404450975554e-06,
      "loss": 1.465,
      "step": 7688
    },
    {
      "epoch": 0.9841290157429925,
      "grad_norm": 0.68359375,
      "learning_rate": 1.3210166048870375e-06,
      "loss": 1.5159,
      "step": 7689
    },
    {
      "epoch": 0.9842570075515167,
      "grad_norm": 0.59375,
      "learning_rate": 1.2998003979850516e-06,
      "loss": 1.5011,
      "step": 7690
    },
    {
      "epoch": 0.984384999360041,
      "grad_norm": 0.5234375,
      "learning_rate": 1.2787558339161232e-06,
      "loss": 0.748,
      "step": 7691
    },
    {
      "epoch": 0.9845129911685652,
      "grad_norm": 0.5625,
      "learning_rate": 1.2578829162970262e-06,
      "loss": 1.0139,
      "step": 7692
    },
    {
      "epoch": 0.9846409829770895,
      "grad_norm": 0.640625,
      "learning_rate": 1.2371816487150022e-06,
      "loss": 1.0715,
      "step": 7693
    },
    {
      "epoch": 0.9847689747856138,
      "grad_norm": 0.61328125,
      "learning_rate": 1.216652034728094e-06,
      "loss": 1.0503,
      "step": 7694
    },
    {
      "epoch": 0.984896966594138,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1962940778644794e-06,
      "loss": 1.0798,
      "step": 7695
    },
    {
      "epoch": 0.9850249584026622,
      "grad_norm": 0.5625,
      "learning_rate": 1.176107781623026e-06,
      "loss": 1.108,
      "step": 7696
    },
    {
      "epoch": 0.9851529502111864,
      "grad_norm": 0.70703125,
      "learning_rate": 1.1560931494732918e-06,
      "loss": 1.4611,
      "step": 7697
    },
    {
      "epoch": 0.9852809420197107,
      "grad_norm": 0.58984375,
      "learning_rate": 1.136250184854859e-06,
      "loss": 1.3569,
      "step": 7698
    },
    {
      "epoch": 0.985408933828235,
      "grad_norm": 0.765625,
      "learning_rate": 1.1165788911781105e-06,
      "loss": 1.629,
      "step": 7699
    },
    {
      "epoch": 0.9855369256367592,
      "grad_norm": 0.7578125,
      "learning_rate": 1.0970792718240086e-06,
      "loss": 1.2674,
      "step": 7700
    },
    {
      "epoch": 0.9856649174452835,
      "grad_norm": 0.5546875,
      "learning_rate": 1.0777513301437613e-06,
      "loss": 0.7952,
      "step": 7701
    },
    {
      "epoch": 0.9857929092538078,
      "grad_norm": 0.703125,
      "learning_rate": 1.0585950694591561e-06,
      "loss": 1.1816,
      "step": 7702
    },
    {
      "epoch": 0.985920901062332,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0396104930626705e-06,
      "loss": 1.4056,
      "step": 7703
    },
    {
      "epoch": 0.9860488928708563,
      "grad_norm": 0.8359375,
      "learning_rate": 1.0207976042169165e-06,
      "loss": 1.2592,
      "step": 7704
    },
    {
      "epoch": 0.9861768846793805,
      "grad_norm": 0.703125,
      "learning_rate": 1.0021564061554189e-06,
      "loss": 1.5596,
      "step": 7705
    },
    {
      "epoch": 0.9863048764879048,
      "grad_norm": 0.72265625,
      "learning_rate": 9.836869020817263e-07,
      "loss": 1.1633,
      "step": 7706
    },
    {
      "epoch": 0.9864328682964291,
      "grad_norm": 0.71875,
      "learning_rate": 9.653890951701882e-07,
      "loss": 1.0803,
      "step": 7707
    },
    {
      "epoch": 0.9865608601049533,
      "grad_norm": 1.0078125,
      "learning_rate": 9.472629885657335e-07,
      "loss": 1.5093,
      "step": 7708
    },
    {
      "epoch": 0.9866888519134775,
      "grad_norm": 0.6015625,
      "learning_rate": 9.29308585383537e-07,
      "loss": 0.9156,
      "step": 7709
    },
    {
      "epoch": 0.9868168437220018,
      "grad_norm": 0.70703125,
      "learning_rate": 9.115258887093525e-07,
      "loss": 1.6845,
      "step": 7710
    },
    {
      "epoch": 0.986944835530526,
      "grad_norm": 0.609375,
      "learning_rate": 8.939149015992909e-07,
      "loss": 0.9045,
      "step": 7711
    },
    {
      "epoch": 0.9870728273390503,
      "grad_norm": 0.74609375,
      "learning_rate": 8.764756270803753e-07,
      "loss": 1.1633,
      "step": 7712
    },
    {
      "epoch": 0.9872008191475745,
      "grad_norm": 0.609375,
      "learning_rate": 8.592080681495418e-07,
      "loss": 1.0892,
      "step": 7713
    },
    {
      "epoch": 0.9873288109560988,
      "grad_norm": 0.5,
      "learning_rate": 8.421122277746385e-07,
      "loss": 0.8027,
      "step": 7714
    },
    {
      "epoch": 0.9874568027646231,
      "grad_norm": 0.703125,
      "learning_rate": 8.251881088938706e-07,
      "loss": 1.3451,
      "step": 7715
    },
    {
      "epoch": 0.9875847945731473,
      "grad_norm": 0.70703125,
      "learning_rate": 8.084357144159116e-07,
      "loss": 1.3319,
      "step": 7716
    },
    {
      "epoch": 0.9877127863816716,
      "grad_norm": 0.7421875,
      "learning_rate": 7.918550472199026e-07,
      "loss": 1.203,
      "step": 7717
    },
    {
      "epoch": 0.9878407781901958,
      "grad_norm": 0.5546875,
      "learning_rate": 7.754461101554533e-07,
      "loss": 1.029,
      "step": 7718
    },
    {
      "epoch": 0.9879687699987201,
      "grad_norm": 0.53125,
      "learning_rate": 7.592089060427521e-07,
      "loss": 0.9089,
      "step": 7719
    },
    {
      "epoch": 0.9880967618072444,
      "grad_norm": 0.53515625,
      "learning_rate": 7.431434376725665e-07,
      "loss": 1.0938,
      "step": 7720
    },
    {
      "epoch": 0.9882247536157686,
      "grad_norm": 0.62109375,
      "learning_rate": 7.272497078057994e-07,
      "loss": 1.1817,
      "step": 7721
    },
    {
      "epoch": 0.9883527454242929,
      "grad_norm": 0.609375,
      "learning_rate": 7.115277191741543e-07,
      "loss": 0.9498,
      "step": 7722
    },
    {
      "epoch": 0.9884807372328172,
      "grad_norm": 0.76171875,
      "learning_rate": 6.95977474479692e-07,
      "loss": 1.6569,
      "step": 7723
    },
    {
      "epoch": 0.9886087290413413,
      "grad_norm": 0.6015625,
      "learning_rate": 6.805989763949416e-07,
      "loss": 0.6282,
      "step": 7724
    },
    {
      "epoch": 0.9887367208498656,
      "grad_norm": 0.67578125,
      "learning_rate": 6.653922275630109e-07,
      "loss": 1.7301,
      "step": 7725
    },
    {
      "epoch": 0.9888647126583898,
      "grad_norm": 0.85546875,
      "learning_rate": 6.503572305973648e-07,
      "loss": 1.6303,
      "step": 7726
    },
    {
      "epoch": 0.9889927044669141,
      "grad_norm": 0.8984375,
      "learning_rate": 6.354939880820476e-07,
      "loss": 2.0322,
      "step": 7727
    },
    {
      "epoch": 0.9891206962754384,
      "grad_norm": 0.6328125,
      "learning_rate": 6.208025025714603e-07,
      "loss": 1.3237,
      "step": 7728
    },
    {
      "epoch": 0.9892486880839626,
      "grad_norm": 0.75390625,
      "learning_rate": 6.062827765906942e-07,
      "loss": 1.4071,
      "step": 7729
    },
    {
      "epoch": 0.9893766798924869,
      "grad_norm": 0.56640625,
      "learning_rate": 5.919348126351976e-07,
      "loss": 0.8781,
      "step": 7730
    },
    {
      "epoch": 0.9895046717010111,
      "grad_norm": 0.6640625,
      "learning_rate": 5.777586131707757e-07,
      "loss": 1.3403,
      "step": 7731
    },
    {
      "epoch": 0.9896326635095354,
      "grad_norm": 0.71484375,
      "learning_rate": 5.637541806340352e-07,
      "loss": 1.5385,
      "step": 7732
    },
    {
      "epoch": 0.9897606553180597,
      "grad_norm": 0.73046875,
      "learning_rate": 5.499215174317174e-07,
      "loss": 1.7873,
      "step": 7733
    },
    {
      "epoch": 0.9898886471265839,
      "grad_norm": 0.875,
      "learning_rate": 5.362606259411429e-07,
      "loss": 1.147,
      "step": 7734
    },
    {
      "epoch": 0.9900166389351082,
      "grad_norm": 0.65234375,
      "learning_rate": 5.227715085103224e-07,
      "loss": 1.2078,
      "step": 7735
    },
    {
      "epoch": 0.9901446307436325,
      "grad_norm": 0.671875,
      "learning_rate": 5.094541674574016e-07,
      "loss": 1.6193,
      "step": 7736
    },
    {
      "epoch": 0.9902726225521566,
      "grad_norm": 0.9296875,
      "learning_rate": 4.963086050713273e-07,
      "loss": 1.3844,
      "step": 7737
    },
    {
      "epoch": 0.9904006143606809,
      "grad_norm": 0.61328125,
      "learning_rate": 4.833348236114032e-07,
      "loss": 0.877,
      "step": 7738
    },
    {
      "epoch": 0.9905286061692051,
      "grad_norm": 0.70703125,
      "learning_rate": 4.7053282530717946e-07,
      "loss": 1.6303,
      "step": 7739
    },
    {
      "epoch": 0.9906565979777294,
      "grad_norm": 0.73828125,
      "learning_rate": 4.579026123591179e-07,
      "loss": 1.504,
      "step": 7740
    },
    {
      "epoch": 0.9907845897862537,
      "grad_norm": 0.67578125,
      "learning_rate": 4.4544418693770464e-07,
      "loss": 1.5675,
      "step": 7741
    },
    {
      "epoch": 0.9909125815947779,
      "grad_norm": 0.6484375,
      "learning_rate": 4.331575511843378e-07,
      "loss": 1.6593,
      "step": 7742
    },
    {
      "epoch": 0.9910405734033022,
      "grad_norm": 0.66796875,
      "learning_rate": 4.210427072105505e-07,
      "loss": 1.5741,
      "step": 7743
    },
    {
      "epoch": 0.9911685652118264,
      "grad_norm": 0.53515625,
      "learning_rate": 4.090996570984551e-07,
      "loss": 0.8151,
      "step": 7744
    },
    {
      "epoch": 0.9912965570203507,
      "grad_norm": 0.5546875,
      "learning_rate": 3.973284029007429e-07,
      "loss": 0.9477,
      "step": 7745
    },
    {
      "epoch": 0.991424548828875,
      "grad_norm": 0.95703125,
      "learning_rate": 3.857289466404623e-07,
      "loss": 1.8974,
      "step": 7746
    },
    {
      "epoch": 0.9915525406373992,
      "grad_norm": 0.62109375,
      "learning_rate": 3.743012903112408e-07,
      "loss": 1.2364,
      "step": 7747
    },
    {
      "epoch": 0.9916805324459235,
      "grad_norm": 0.64453125,
      "learning_rate": 3.630454358769519e-07,
      "loss": 1.2693,
      "step": 7748
    },
    {
      "epoch": 0.9918085242544478,
      "grad_norm": 0.6171875,
      "learning_rate": 3.519613852721593e-07,
      "loss": 1.0317,
      "step": 7749
    },
    {
      "epoch": 0.991936516062972,
      "grad_norm": 0.609375,
      "learning_rate": 3.4104914040178346e-07,
      "loss": 1.1233,
      "step": 7750
    },
    {
      "epoch": 0.9920645078714962,
      "grad_norm": 0.8125,
      "learning_rate": 3.3030870314143534e-07,
      "loss": 1.7189,
      "step": 7751
    },
    {
      "epoch": 0.9921924996800204,
      "grad_norm": 0.6875,
      "learning_rate": 3.197400753369717e-07,
      "loss": 1.2214,
      "step": 7752
    },
    {
      "epoch": 0.9923204914885447,
      "grad_norm": 0.609375,
      "learning_rate": 3.0934325880460634e-07,
      "loss": 1.2246,
      "step": 7753
    },
    {
      "epoch": 0.992448483297069,
      "grad_norm": 0.703125,
      "learning_rate": 2.991182553314653e-07,
      "loss": 1.1971,
      "step": 7754
    },
    {
      "epoch": 0.9925764751055932,
      "grad_norm": 0.5546875,
      "learning_rate": 2.8906506667469856e-07,
      "loss": 1.147,
      "step": 7755
    },
    {
      "epoch": 0.9927044669141175,
      "grad_norm": 0.6640625,
      "learning_rate": 2.791836945621462e-07,
      "loss": 1.5582,
      "step": 7756
    },
    {
      "epoch": 0.9928324587226417,
      "grad_norm": 0.625,
      "learning_rate": 2.6947414069222743e-07,
      "loss": 1.1727,
      "step": 7757
    },
    {
      "epoch": 0.992960450531166,
      "grad_norm": 0.67578125,
      "learning_rate": 2.5993640673338534e-07,
      "loss": 1.5483,
      "step": 7758
    },
    {
      "epoch": 0.9930884423396903,
      "grad_norm": 0.6796875,
      "learning_rate": 2.505704943251974e-07,
      "loss": 1.3965,
      "step": 7759
    },
    {
      "epoch": 0.9932164341482145,
      "grad_norm": 0.5546875,
      "learning_rate": 2.413764050770428e-07,
      "loss": 1.0938,
      "step": 7760
    },
    {
      "epoch": 0.9933444259567388,
      "grad_norm": 0.61328125,
      "learning_rate": 2.3235414056921312e-07,
      "loss": 1.2474,
      "step": 7761
    },
    {
      "epoch": 0.9934724177652631,
      "grad_norm": 0.55078125,
      "learning_rate": 2.2350370235235672e-07,
      "loss": 0.69,
      "step": 7762
    },
    {
      "epoch": 0.9936004095737873,
      "grad_norm": 0.69921875,
      "learning_rate": 2.1482509194747923e-07,
      "loss": 1.1803,
      "step": 7763
    },
    {
      "epoch": 0.9937284013823116,
      "grad_norm": 0.671875,
      "learning_rate": 2.0631831084616526e-07,
      "loss": 1.0686,
      "step": 7764
    },
    {
      "epoch": 0.9938563931908357,
      "grad_norm": 0.71875,
      "learning_rate": 1.979833605104675e-07,
      "loss": 1.6951,
      "step": 7765
    },
    {
      "epoch": 0.99398438499936,
      "grad_norm": 0.6875,
      "learning_rate": 1.8982024237290672e-07,
      "loss": 1.3374,
      "step": 7766
    },
    {
      "epoch": 0.9941123768078843,
      "grad_norm": 0.6796875,
      "learning_rate": 1.8182895783636076e-07,
      "loss": 1.4796,
      "step": 7767
    },
    {
      "epoch": 0.9942403686164085,
      "grad_norm": 0.71875,
      "learning_rate": 1.7400950827439754e-07,
      "loss": 1.5825,
      "step": 7768
    },
    {
      "epoch": 0.9943683604249328,
      "grad_norm": 0.6328125,
      "learning_rate": 1.66361895030831e-07,
      "loss": 1.0009,
      "step": 7769
    },
    {
      "epoch": 0.994496352233457,
      "grad_norm": 0.58203125,
      "learning_rate": 1.5888611941994313e-07,
      "loss": 0.8854,
      "step": 7770
    },
    {
      "epoch": 0.9946243440419813,
      "grad_norm": 0.46484375,
      "learning_rate": 1.515821827267061e-07,
      "loss": 0.5965,
      "step": 7771
    },
    {
      "epoch": 0.9947523358505056,
      "grad_norm": 0.72265625,
      "learning_rate": 1.4445008620644905e-07,
      "loss": 1.3185,
      "step": 7772
    },
    {
      "epoch": 0.9948803276590298,
      "grad_norm": 0.640625,
      "learning_rate": 1.3748983108474723e-07,
      "loss": 1.0678,
      "step": 7773
    },
    {
      "epoch": 0.9950083194675541,
      "grad_norm": 0.59375,
      "learning_rate": 1.307014185579769e-07,
      "loss": 1.1014,
      "step": 7774
    },
    {
      "epoch": 0.9951363112760784,
      "grad_norm": 0.6015625,
      "learning_rate": 1.2408484979276047e-07,
      "loss": 1.1888,
      "step": 7775
    },
    {
      "epoch": 0.9952643030846026,
      "grad_norm": 0.6171875,
      "learning_rate": 1.1764012592641038e-07,
      "loss": 1.3292,
      "step": 7776
    },
    {
      "epoch": 0.9953922948931269,
      "grad_norm": 0.74609375,
      "learning_rate": 1.113672480663741e-07,
      "loss": 1.3634,
      "step": 7777
    },
    {
      "epoch": 0.995520286701651,
      "grad_norm": 0.64453125,
      "learning_rate": 1.0526621729090025e-07,
      "loss": 1.1016,
      "step": 7778
    },
    {
      "epoch": 0.9956482785101753,
      "grad_norm": 0.65625,
      "learning_rate": 9.933703464837241e-08,
      "loss": 1.0018,
      "step": 7779
    },
    {
      "epoch": 0.9957762703186996,
      "grad_norm": 0.6484375,
      "learning_rate": 9.357970115808633e-08,
      "loss": 1.1054,
      "step": 7780
    },
    {
      "epoch": 0.9959042621272238,
      "grad_norm": 0.6328125,
      "learning_rate": 8.799421780925077e-08,
      "loss": 1.1555,
      "step": 7781
    },
    {
      "epoch": 0.9960322539357481,
      "grad_norm": 0.65625,
      "learning_rate": 8.258058556187553e-08,
      "loss": 1.4344,
      "step": 7782
    },
    {
      "epoch": 0.9961602457442724,
      "grad_norm": 0.70703125,
      "learning_rate": 7.733880534654958e-08,
      "loss": 1.5572,
      "step": 7783
    },
    {
      "epoch": 0.9962882375527966,
      "grad_norm": 0.71484375,
      "learning_rate": 7.226887806399685e-08,
      "loss": 1.4662,
      "step": 7784
    },
    {
      "epoch": 0.9964162293613209,
      "grad_norm": 0.71484375,
      "learning_rate": 6.737080458574241e-08,
      "loss": 1.5294,
      "step": 7785
    },
    {
      "epoch": 0.9965442211698451,
      "grad_norm": 0.4921875,
      "learning_rate": 6.264458575333532e-08,
      "loss": 0.6154,
      "step": 7786
    },
    {
      "epoch": 0.9966722129783694,
      "grad_norm": 0.76171875,
      "learning_rate": 5.809022237923678e-08,
      "loss": 1.5279,
      "step": 7787
    },
    {
      "epoch": 0.9968002047868937,
      "grad_norm": 0.84765625,
      "learning_rate": 5.370771524615403e-08,
      "loss": 0.9083,
      "step": 7788
    },
    {
      "epoch": 0.9969281965954179,
      "grad_norm": 0.59375,
      "learning_rate": 4.94970651073734e-08,
      "loss": 0.8643,
      "step": 7789
    },
    {
      "epoch": 0.9970561884039422,
      "grad_norm": 0.53515625,
      "learning_rate": 4.545827268642722e-08,
      "loss": 1.049,
      "step": 7790
    },
    {
      "epoch": 0.9971841802124664,
      "grad_norm": 0.62109375,
      "learning_rate": 4.159133867742693e-08,
      "loss": 1.2817,
      "step": 7791
    },
    {
      "epoch": 0.9973121720209907,
      "grad_norm": 0.67578125,
      "learning_rate": 3.7896263745063056e-08,
      "loss": 1.9225,
      "step": 7792
    },
    {
      "epoch": 0.997440163829515,
      "grad_norm": 0.60546875,
      "learning_rate": 3.4373048524383164e-08,
      "loss": 1.3978,
      "step": 7793
    },
    {
      "epoch": 0.9975681556380391,
      "grad_norm": 0.6015625,
      "learning_rate": 3.10216936209029e-08,
      "loss": 0.9134,
      "step": 7794
    },
    {
      "epoch": 0.9976961474465634,
      "grad_norm": 0.89453125,
      "learning_rate": 2.7842199610605967e-08,
      "loss": 1.1647,
      "step": 7795
    },
    {
      "epoch": 0.9978241392550877,
      "grad_norm": 0.68359375,
      "learning_rate": 2.4834567039944133e-08,
      "loss": 1.2581,
      "step": 7796
    },
    {
      "epoch": 0.9979521310636119,
      "grad_norm": 0.5546875,
      "learning_rate": 2.1998796425837243e-08,
      "loss": 1.0091,
      "step": 7797
    },
    {
      "epoch": 0.9980801228721362,
      "grad_norm": 0.451171875,
      "learning_rate": 1.9334888255562177e-08,
      "loss": 0.5595,
      "step": 7798
    },
    {
      "epoch": 0.9982081146806604,
      "grad_norm": 0.828125,
      "learning_rate": 1.6842842987085937e-08,
      "loss": 1.862,
      "step": 7799
    },
    {
      "epoch": 0.9983361064891847,
      "grad_norm": 0.609375,
      "learning_rate": 1.4522661048621544e-08,
      "loss": 1.0089,
      "step": 7800
    },
    {
      "epoch": 0.998464098297709,
      "grad_norm": 0.80859375,
      "learning_rate": 1.2374342839072128e-08,
      "loss": 1.3608,
      "step": 7801
    },
    {
      "epoch": 0.9985920901062332,
      "grad_norm": 0.61328125,
      "learning_rate": 1.0397888727364801e-08,
      "loss": 0.9756,
      "step": 7802
    },
    {
      "epoch": 0.9987200819147575,
      "grad_norm": 0.61328125,
      "learning_rate": 8.593299053449854e-09,
      "loss": 1.2206,
      "step": 7803
    },
    {
      "epoch": 0.9988480737232817,
      "grad_norm": 0.51953125,
      "learning_rate": 6.960574127412578e-09,
      "loss": 0.9246,
      "step": 7804
    },
    {
      "epoch": 0.998976065531806,
      "grad_norm": 0.69921875,
      "learning_rate": 5.499714229917352e-09,
      "loss": 1.182,
      "step": 7805
    },
    {
      "epoch": 0.9991040573403303,
      "grad_norm": 0.6171875,
      "learning_rate": 4.210719611874581e-09,
      "loss": 1.1942,
      "step": 7806
    },
    {
      "epoch": 0.9992320491488544,
      "grad_norm": 0.4609375,
      "learning_rate": 3.0935904948847792e-09,
      "loss": 0.8916,
      "step": 7807
    },
    {
      "epoch": 0.9993600409573787,
      "grad_norm": 0.51171875,
      "learning_rate": 2.1483270710165315e-09,
      "loss": 0.7018,
      "step": 7808
    },
    {
      "epoch": 0.999488032765903,
      "grad_norm": 0.765625,
      "learning_rate": 1.3749295026954656e-09,
      "loss": 1.4432,
      "step": 7809
    },
    {
      "epoch": 0.9996160245744272,
      "grad_norm": 0.63671875,
      "learning_rate": 7.733979228152777e-10,
      "loss": 1.2541,
      "step": 7810
    },
    {
      "epoch": 0.9997440163829515,
      "grad_norm": 0.6796875,
      "learning_rate": 3.437324347377313e-10,
      "loss": 1.3286,
      "step": 7811
    },
    {
      "epoch": 0.9998720081914757,
      "grad_norm": 0.6171875,
      "learning_rate": 8.593311240367995e-11,
      "loss": 0.7655,
      "step": 7812
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0,
      "loss": 0.9618,
      "step": 7813
    }
  ],
  "logging_steps": 1,
  "max_steps": 7813,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 15000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.7044895314634342e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}