{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9996178830722201,
  "eval_steps": 500,
  "global_step": 2289,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 139.4198717367427,
      "learning_rate": 1.4492753623188408e-07,
      "loss": 1.6997,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 139.33991456785594,
      "learning_rate": 2.8985507246376816e-07,
      "loss": 1.6449,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 140.85327682523882,
      "learning_rate": 4.347826086956522e-07,
      "loss": 1.7401,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 156.77414587586205,
      "learning_rate": 5.797101449275363e-07,
      "loss": 1.7159,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 150.06892737579525,
      "learning_rate": 7.246376811594204e-07,
      "loss": 1.8043,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 133.41149724881788,
      "learning_rate": 8.695652173913044e-07,
      "loss": 1.5665,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 55.32413966238607,
      "learning_rate": 1.0144927536231885e-06,
      "loss": 1.3441,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 51.065109819301284,
      "learning_rate": 1.1594202898550726e-06,
      "loss": 1.3609,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 23.14917625314554,
      "learning_rate": 1.3043478260869566e-06,
      "loss": 0.93,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 20.78121816113515,
      "learning_rate": 1.4492753623188408e-06,
      "loss": 1.0166,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 20.293584731312173,
      "learning_rate": 1.5942028985507246e-06,
      "loss": 0.9444,
      "step": 11
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.495428155977057,
      "learning_rate": 1.7391304347826088e-06,
      "loss": 0.7843,
      "step": 12
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.234374747618972,
      "learning_rate": 1.884057971014493e-06,
      "loss": 0.5982,
      "step": 13
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.235539081862138,
      "learning_rate": 2.028985507246377e-06,
      "loss": 0.6129,
      "step": 14
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.417433701075291,
      "learning_rate": 2.173913043478261e-06,
      "loss": 0.6395,
      "step": 15
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.19854476938665,
      "learning_rate": 2.3188405797101453e-06,
      "loss": 0.5731,
      "step": 16
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.221081259426901,
      "learning_rate": 2.4637681159420295e-06,
      "loss": 0.4792,
      "step": 17
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.11080560262814,
      "learning_rate": 2.6086956521739132e-06,
      "loss": 0.4879,
      "step": 18
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.26918292009478,
      "learning_rate": 2.7536231884057974e-06,
      "loss": 0.4466,
      "step": 19
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.095169673937903,
      "learning_rate": 2.8985507246376816e-06,
      "loss": 0.4596,
      "step": 20
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.9350111327845085,
      "learning_rate": 3.043478260869566e-06,
      "loss": 0.4134,
      "step": 21
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.0588988588055654,
      "learning_rate": 3.188405797101449e-06,
      "loss": 0.4919,
      "step": 22
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.570741059196282,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.4817,
      "step": 23
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.032911616898733,
      "learning_rate": 3.4782608695652175e-06,
      "loss": 0.411,
      "step": 24
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.288651627517517,
      "learning_rate": 3.6231884057971017e-06,
      "loss": 0.404,
      "step": 25
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.863204795045959,
      "learning_rate": 3.768115942028986e-06,
      "loss": 0.3929,
      "step": 26
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.4737884634266374,
      "learning_rate": 3.91304347826087e-06,
      "loss": 0.3541,
      "step": 27
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.389554970062527,
      "learning_rate": 4.057971014492754e-06,
      "loss": 0.4217,
      "step": 28
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.9602534223177344,
      "learning_rate": 4.202898550724638e-06,
      "loss": 0.3627,
      "step": 29
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6761612133304475,
      "learning_rate": 4.347826086956522e-06,
      "loss": 0.3793,
      "step": 30
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3945538895805307,
      "learning_rate": 4.492753623188406e-06,
      "loss": 0.2834,
      "step": 31
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.371471022762334,
      "learning_rate": 4.637681159420291e-06,
      "loss": 0.3005,
      "step": 32
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.8195129669839116,
      "learning_rate": 4.782608695652174e-06,
      "loss": 0.4239,
      "step": 33
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.2331691056714615,
      "learning_rate": 4.927536231884059e-06,
      "loss": 0.3048,
      "step": 34
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.51484410384854,
      "learning_rate": 5.072463768115943e-06,
      "loss": 0.3258,
      "step": 35
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4449895343099657,
      "learning_rate": 5.2173913043478265e-06,
      "loss": 0.3101,
      "step": 36
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.365150288291428,
      "learning_rate": 5.362318840579711e-06,
      "loss": 0.3186,
      "step": 37
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5455623026925878,
      "learning_rate": 5.507246376811595e-06,
      "loss": 0.3947,
      "step": 38
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2503045161363198,
      "learning_rate": 5.652173913043479e-06,
      "loss": 0.2697,
      "step": 39
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.467943734249912,
      "learning_rate": 5.797101449275363e-06,
      "loss": 0.2896,
      "step": 40
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.29296485122179,
      "learning_rate": 5.942028985507247e-06,
      "loss": 0.334,
      "step": 41
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3767197256376944,
      "learning_rate": 6.086956521739132e-06,
      "loss": 0.2763,
      "step": 42
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2964716324368526,
      "learning_rate": 6.2318840579710145e-06,
      "loss": 0.2926,
      "step": 43
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3945549899036243,
      "learning_rate": 6.376811594202898e-06,
      "loss": 0.341,
      "step": 44
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.485996452781697,
      "learning_rate": 6.521739130434783e-06,
      "loss": 0.2987,
      "step": 45
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1677437375823008,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.2785,
      "step": 46
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1190587174262476,
      "learning_rate": 6.811594202898551e-06,
      "loss": 0.2815,
      "step": 47
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.155953494340067,
      "learning_rate": 6.956521739130435e-06,
      "loss": 0.3023,
      "step": 48
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.35549223092658,
      "learning_rate": 7.10144927536232e-06,
      "loss": 0.3107,
      "step": 49
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.448419862051946,
      "learning_rate": 7.246376811594203e-06,
      "loss": 0.3178,
      "step": 50
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.028911048161468,
      "learning_rate": 7.391304347826087e-06,
      "loss": 0.3014,
      "step": 51
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.273079911823269,
      "learning_rate": 7.536231884057972e-06,
      "loss": 0.3185,
      "step": 52
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9253286294873813,
      "learning_rate": 7.681159420289856e-06,
      "loss": 0.2292,
      "step": 53
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1197536381461797,
      "learning_rate": 7.82608695652174e-06,
      "loss": 0.2668,
      "step": 54
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.950744161106596,
      "learning_rate": 7.971014492753623e-06,
      "loss": 0.3056,
      "step": 55
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.098833489242608,
      "learning_rate": 8.115942028985508e-06,
      "loss": 0.2666,
      "step": 56
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.085975500575603,
      "learning_rate": 8.260869565217392e-06,
      "loss": 0.2356,
      "step": 57
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0115372813376786,
      "learning_rate": 8.405797101449275e-06,
      "loss": 0.296,
      "step": 58
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.147562961587654,
      "learning_rate": 8.55072463768116e-06,
      "loss": 0.2752,
      "step": 59
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.095814118187103,
      "learning_rate": 8.695652173913044e-06,
      "loss": 0.2076,
      "step": 60
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1284074856869943,
      "learning_rate": 8.840579710144929e-06,
      "loss": 0.297,
      "step": 61
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.061449728554865,
      "learning_rate": 8.985507246376812e-06,
      "loss": 0.4022,
      "step": 62
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1970429387875505,
      "learning_rate": 9.130434782608697e-06,
      "loss": 0.2847,
      "step": 63
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.969310288990886,
      "learning_rate": 9.275362318840581e-06,
      "loss": 0.2647,
      "step": 64
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1598144387754323,
      "learning_rate": 9.420289855072464e-06,
      "loss": 0.2091,
      "step": 65
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0823459233941324,
      "learning_rate": 9.565217391304349e-06,
      "loss": 0.2344,
      "step": 66
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4138807971963194,
      "learning_rate": 9.710144927536233e-06,
      "loss": 0.2313,
      "step": 67
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2233135157477775,
      "learning_rate": 9.855072463768118e-06,
      "loss": 0.3476,
      "step": 68
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0693524741996043,
      "learning_rate": 1e-05,
      "loss": 0.1792,
      "step": 69
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1266101312825896,
      "learning_rate": 9.999994993505624e-06,
      "loss": 0.2489,
      "step": 70
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.846435443248496,
      "learning_rate": 9.99997997403252e-06,
      "loss": 0.2431,
      "step": 71
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0049429126116136,
      "learning_rate": 9.999954941610767e-06,
      "loss": 0.2082,
      "step": 72
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1217300778769435,
      "learning_rate": 9.999919896290497e-06,
      "loss": 0.2291,
      "step": 73
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9384875378858968,
      "learning_rate": 9.999874838141888e-06,
      "loss": 0.2881,
      "step": 74
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1285197612400895,
      "learning_rate": 9.999819767255175e-06,
      "loss": 0.2542,
      "step": 75
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1389967852180654,
      "learning_rate": 9.999754683740643e-06,
      "loss": 0.2504,
      "step": 76
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4355621946888926,
      "learning_rate": 9.999679587728626e-06,
      "loss": 0.2603,
      "step": 77
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.166861637416314,
      "learning_rate": 9.999594479369515e-06,
      "loss": 0.3375,
      "step": 78
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.93378809814654,
      "learning_rate": 9.999499358833745e-06,
      "loss": 0.2143,
      "step": 79
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9797004735933759,
      "learning_rate": 9.999394226311804e-06,
      "loss": 0.2641,
      "step": 80
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9614291150060643,
      "learning_rate": 9.999279082014233e-06,
      "loss": 0.2964,
      "step": 81
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2648160383951463,
      "learning_rate": 9.999153926171617e-06,
      "loss": 0.2082,
      "step": 82
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9176786803199668,
      "learning_rate": 9.999018759034594e-06,
      "loss": 0.2796,
      "step": 83
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.094805409256972,
      "learning_rate": 9.998873580873848e-06,
      "loss": 0.3405,
      "step": 84
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9136009605594537,
      "learning_rate": 9.998718391980113e-06,
      "loss": 0.2575,
      "step": 85
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8636224420613123,
      "learning_rate": 9.998553192664172e-06,
      "loss": 0.2686,
      "step": 86
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1349366502807827,
      "learning_rate": 9.998377983256851e-06,
      "loss": 0.2307,
      "step": 87
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.933929339838491,
      "learning_rate": 9.998192764109023e-06,
      "loss": 0.2566,
      "step": 88
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7667788225603855,
      "learning_rate": 9.99799753559161e-06,
      "loss": 0.2076,
      "step": 89
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9347684426436205,
      "learning_rate": 9.997792298095572e-06,
      "loss": 0.2063,
      "step": 90
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1081554981503747,
      "learning_rate": 9.997577052031922e-06,
      "loss": 0.2323,
      "step": 91
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.90225832477509,
      "learning_rate": 9.99735179783171e-06,
      "loss": 0.241,
      "step": 92
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1344895342679058,
      "learning_rate": 9.997116535946028e-06,
      "loss": 0.305,
      "step": 93
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3817608385029494,
      "learning_rate": 9.99687126684601e-06,
      "loss": 0.1717,
      "step": 94
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8714832327555764,
      "learning_rate": 9.996615991022835e-06,
      "loss": 0.237,
      "step": 95
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8616918082304161,
      "learning_rate": 9.996350708987715e-06,
      "loss": 0.2844,
      "step": 96
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8394764430496113,
      "learning_rate": 9.996075421271905e-06,
      "loss": 0.269,
      "step": 97
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9710355975924323,
      "learning_rate": 9.995790128426693e-06,
      "loss": 0.1853,
      "step": 98
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9117400511784357,
      "learning_rate": 9.99549483102341e-06,
      "loss": 0.2456,
      "step": 99
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8953275584934957,
      "learning_rate": 9.995189529653412e-06,
      "loss": 0.1977,
      "step": 100
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8841521266493073,
      "learning_rate": 9.9948742249281e-06,
      "loss": 0.2887,
      "step": 101
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0255743384630778,
      "learning_rate": 9.994548917478899e-06,
      "loss": 0.2305,
      "step": 102
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9459896540610084,
      "learning_rate": 9.994213607957273e-06,
      "loss": 0.3036,
      "step": 103
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7556695673820621,
      "learning_rate": 9.993868297034709e-06,
      "loss": 0.2073,
      "step": 104
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8006378179616107,
      "learning_rate": 9.993512985402724e-06,
      "loss": 0.2032,
      "step": 105
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8994576396574006,
      "learning_rate": 9.993147673772869e-06,
      "loss": 0.2105,
      "step": 106
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.81256578905194,
      "learning_rate": 9.992772362876714e-06,
      "loss": 0.265,
      "step": 107
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.073093318088847,
      "learning_rate": 9.992387053465857e-06,
      "loss": 0.2578,
      "step": 108
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8909584572861269,
      "learning_rate": 9.991991746311916e-06,
      "loss": 0.254,
      "step": 109
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7873435749825535,
      "learning_rate": 9.991586442206533e-06,
      "loss": 0.2479,
      "step": 110
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.986418132321028,
      "learning_rate": 9.99117114196137e-06,
      "loss": 0.3085,
      "step": 111
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8665368690223054,
      "learning_rate": 9.990745846408102e-06,
      "loss": 0.2246,
      "step": 112
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9874449739968107,
      "learning_rate": 9.99031055639843e-06,
      "loss": 0.3122,
      "step": 113
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8378621802647275,
      "learning_rate": 9.989865272804064e-06,
      "loss": 0.2424,
      "step": 114
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9202149778474653,
      "learning_rate": 9.989409996516726e-06,
      "loss": 0.2532,
      "step": 115
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7399841379088477,
      "learning_rate": 9.98894472844815e-06,
      "loss": 0.214,
      "step": 116
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.60787432334073,
      "learning_rate": 9.988469469530086e-06,
      "loss": 0.1904,
      "step": 117
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.042292963638229,
      "learning_rate": 9.987984220714283e-06,
      "loss": 0.3087,
      "step": 118
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.099962674947596,
      "learning_rate": 9.9874889829725e-06,
      "loss": 0.2337,
      "step": 119
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9729743564105415,
      "learning_rate": 9.986983757296498e-06,
      "loss": 0.2616,
      "step": 120
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8996239446540535,
      "learning_rate": 9.986468544698041e-06,
      "loss": 0.2932,
      "step": 121
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8535961030582497,
      "learning_rate": 9.985943346208893e-06,
      "loss": 0.2533,
      "step": 122
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9219922345356661,
      "learning_rate": 9.985408162880813e-06,
      "loss": 0.2731,
      "step": 123
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1126773114861406,
      "learning_rate": 9.984862995785564e-06,
      "loss": 0.3453,
      "step": 124
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7033928143605557,
      "learning_rate": 9.984307846014891e-06,
      "loss": 0.1779,
      "step": 125
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7268052757244645,
      "learning_rate": 9.983742714680538e-06,
      "loss": 0.2402,
      "step": 126
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8249039220294023,
      "learning_rate": 9.983167602914236e-06,
      "loss": 0.2878,
      "step": 127
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8126168446085251,
      "learning_rate": 9.9825825118677e-06,
      "loss": 0.302,
      "step": 128
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9767527234595357,
      "learning_rate": 9.981987442712634e-06,
      "loss": 0.2523,
      "step": 129
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7872176719115138,
      "learning_rate": 9.981382396640721e-06,
      "loss": 0.22,
      "step": 130
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.013150623421478,
      "learning_rate": 9.980767374863627e-06,
      "loss": 0.2935,
      "step": 131
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.939413602822474,
      "learning_rate": 9.98014237861299e-06,
      "loss": 0.2513,
      "step": 132
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.195468358704904,
      "learning_rate": 9.97950740914043e-06,
      "loss": 0.2961,
      "step": 133
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.003539883191869,
      "learning_rate": 9.978862467717532e-06,
      "loss": 0.3117,
      "step": 134
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8519910991235708,
      "learning_rate": 9.978207555635856e-06,
      "loss": 0.2235,
      "step": 135
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6310468442314447,
      "learning_rate": 9.977542674206927e-06,
      "loss": 0.1886,
      "step": 136
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8083777225091442,
      "learning_rate": 9.976867824762236e-06,
      "loss": 0.2447,
      "step": 137
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.817600674005443,
      "learning_rate": 9.976183008653232e-06,
      "loss": 0.2596,
      "step": 138
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.592336573526578,
      "learning_rate": 9.97548822725133e-06,
      "loss": 0.2421,
      "step": 139
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.695833361068321,
      "learning_rate": 9.974783481947896e-06,
      "loss": 0.1915,
      "step": 140
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6468887228123497,
      "learning_rate": 9.974068774154252e-06,
      "loss": 0.2495,
      "step": 141
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7193803709491033,
      "learning_rate": 9.97334410530167e-06,
      "loss": 0.2414,
      "step": 142
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8923780895765816,
      "learning_rate": 9.972609476841368e-06,
      "loss": 0.2203,
      "step": 143
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8592042733034304,
      "learning_rate": 9.971864890244514e-06,
      "loss": 0.271,
      "step": 144
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8099704815389306,
      "learning_rate": 9.971110347002213e-06,
      "loss": 0.2883,
      "step": 145
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.899369590432476,
      "learning_rate": 9.970345848625516e-06,
      "loss": 0.2644,
      "step": 146
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7296993925877628,
      "learning_rate": 9.9695713966454e-06,
      "loss": 0.2558,
      "step": 147
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.025577644085121,
      "learning_rate": 9.968786992612785e-06,
      "loss": 0.2983,
      "step": 148
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8441039546972464,
      "learning_rate": 9.967992638098517e-06,
      "loss": 0.3305,
      "step": 149
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6509826143866286,
      "learning_rate": 9.967188334693364e-06,
      "loss": 0.2533,
      "step": 150
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5694994683570058,
      "learning_rate": 9.966374084008025e-06,
      "loss": 0.2885,
      "step": 151
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5208801453612446,
      "learning_rate": 9.965549887673121e-06,
      "loss": 0.2328,
      "step": 152
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7956720774897712,
      "learning_rate": 9.964715747339178e-06,
      "loss": 0.2237,
      "step": 153
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7445041197380624,
      "learning_rate": 9.963871664676647e-06,
      "loss": 0.2609,
      "step": 154
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.606029984237376,
      "learning_rate": 9.963017641375888e-06,
      "loss": 0.2103,
      "step": 155
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5629548874945058,
      "learning_rate": 9.962153679147162e-06,
      "loss": 0.1885,
      "step": 156
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5816092256130931,
      "learning_rate": 9.961279779720641e-06,
      "loss": 0.204,
      "step": 157
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6784713415217771,
      "learning_rate": 9.960395944846394e-06,
      "loss": 0.2404,
      "step": 158
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9795500828255852,
      "learning_rate": 9.959502176294384e-06,
      "loss": 0.2874,
      "step": 159
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0182481848201324,
      "learning_rate": 9.958598475854474e-06,
      "loss": 0.2691,
      "step": 160
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6587844923176527,
      "learning_rate": 9.957684845336409e-06,
      "loss": 0.2016,
      "step": 161
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6351761290741111,
      "learning_rate": 9.956761286569824e-06,
      "loss": 0.2599,
      "step": 162
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5558911733395677,
      "learning_rate": 9.955827801404237e-06,
      "loss": 0.2353,
      "step": 163
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7158870035768319,
      "learning_rate": 9.954884391709043e-06,
      "loss": 0.1984,
      "step": 164
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7691974808050333,
      "learning_rate": 9.95393105937351e-06,
      "loss": 0.2968,
      "step": 165
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5694488649212657,
      "learning_rate": 9.952967806306784e-06,
      "loss": 0.2095,
      "step": 166
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8462403965311203,
      "learning_rate": 9.951994634437866e-06,
      "loss": 0.2393,
      "step": 167
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.889576575373559,
      "learning_rate": 9.951011545715637e-06,
      "loss": 0.3,
      "step": 168
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.833481869451497,
      "learning_rate": 9.950018542108818e-06,
      "loss": 0.1899,
      "step": 169
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6955839161717112,
      "learning_rate": 9.949015625606007e-06,
      "loss": 0.211,
      "step": 170
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5061286855184295,
      "learning_rate": 9.948002798215632e-06,
      "loss": 0.265,
      "step": 171
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.842239287773047,
      "learning_rate": 9.946980061965985e-06,
      "loss": 0.2498,
      "step": 172
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9993873859886957,
      "learning_rate": 9.945947418905193e-06,
      "loss": 0.2254,
      "step": 173
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7293596818123296,
      "learning_rate": 9.944904871101227e-06,
      "loss": 0.223,
      "step": 174
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9297517274806189,
      "learning_rate": 9.943852420641888e-06,
      "loss": 0.2865,
      "step": 175
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8243017489776259,
      "learning_rate": 9.942790069634814e-06,
      "loss": 0.2231,
      "step": 176
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7193707099718165,
      "learning_rate": 9.941717820207461e-06,
      "loss": 0.2626,
      "step": 177
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6958192270658277,
      "learning_rate": 9.94063567450712e-06,
      "loss": 0.2554,
      "step": 178
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6378186340215481,
      "learning_rate": 9.939543634700891e-06,
      "loss": 0.2356,
      "step": 179
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7136974626114883,
      "learning_rate": 9.938441702975689e-06,
      "loss": 0.2885,
      "step": 180
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9615875952268773,
      "learning_rate": 9.937329881538242e-06,
      "loss": 0.2778,
      "step": 181
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.60922591268781,
      "learning_rate": 9.936208172615081e-06,
      "loss": 0.2621,
      "step": 182
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3932177033585538,
      "learning_rate": 9.935076578452535e-06,
      "loss": 0.194,
      "step": 183
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7727167274880318,
      "learning_rate": 9.933935101316735e-06,
      "loss": 0.2125,
      "step": 184
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.52947492011434,
      "learning_rate": 9.932783743493599e-06,
      "loss": 0.237,
      "step": 185
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.918734256665052,
      "learning_rate": 9.931622507288835e-06,
      "loss": 0.2208,
      "step": 186
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6857159543647162,
      "learning_rate": 9.930451395027932e-06,
      "loss": 0.2042,
      "step": 187
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6542891696728586,
      "learning_rate": 9.929270409056154e-06,
      "loss": 0.2798,
      "step": 188
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5906438853677463,
      "learning_rate": 9.928079551738542e-06,
      "loss": 0.1983,
      "step": 189
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9177619899212717,
      "learning_rate": 9.926878825459907e-06,
      "loss": 0.3407,
      "step": 190
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.011388901595617,
      "learning_rate": 9.925668232624819e-06,
      "loss": 0.2748,
      "step": 191
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5540272594190097,
      "learning_rate": 9.924447775657607e-06,
      "loss": 0.2103,
      "step": 192
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6802247303634028,
      "learning_rate": 9.923217457002356e-06,
      "loss": 0.2057,
      "step": 193
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.688438362499149,
      "learning_rate": 9.9219772791229e-06,
      "loss": 0.2521,
      "step": 194
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.910856306352271,
      "learning_rate": 9.92072724450282e-06,
      "loss": 0.2509,
      "step": 195
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8892363436881001,
      "learning_rate": 9.919467355645425e-06,
      "loss": 0.2522,
      "step": 196
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6705299678377405,
      "learning_rate": 9.91819761507377e-06,
      "loss": 0.1784,
      "step": 197
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7823552034904797,
      "learning_rate": 9.916918025330635e-06,
      "loss": 0.2406,
      "step": 198
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5031200241806681,
      "learning_rate": 9.915628588978522e-06,
      "loss": 0.1892,
      "step": 199
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5589226541605954,
      "learning_rate": 9.914329308599655e-06,
      "loss": 0.1913,
      "step": 200
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.599254939527351,
      "learning_rate": 9.913020186795967e-06,
      "loss": 0.2101,
      "step": 201
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.607671785991434,
      "learning_rate": 9.911701226189105e-06,
      "loss": 0.2102,
      "step": 202
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9054997375379126,
      "learning_rate": 9.910372429420416e-06,
      "loss": 0.2567,
      "step": 203
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9838587916428339,
      "learning_rate": 9.909033799150947e-06,
      "loss": 0.2471,
      "step": 204
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.618435217185729,
      "learning_rate": 9.907685338061433e-06,
      "loss": 0.1785,
      "step": 205
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6696745817086305,
      "learning_rate": 9.906327048852301e-06,
      "loss": 0.245,
      "step": 206
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.392828950915525,
      "learning_rate": 9.904958934243655e-06,
      "loss": 0.1901,
      "step": 207
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7425617229040802,
      "learning_rate": 9.903580996975282e-06,
      "loss": 0.2031,
      "step": 208
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7024515481299036,
      "learning_rate": 9.902193239806634e-06,
      "loss": 0.2452,
      "step": 209
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8752799472272037,
      "learning_rate": 9.90079566551683e-06,
      "loss": 0.3282,
      "step": 210
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6643000395191165,
      "learning_rate": 9.899388276904653e-06,
      "loss": 0.2306,
      "step": 211
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7509065082431765,
      "learning_rate": 9.897971076788532e-06,
      "loss": 0.2931,
      "step": 212
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7728757959793047,
      "learning_rate": 9.89654406800655e-06,
      "loss": 0.3039,
      "step": 213
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7110524517234067,
      "learning_rate": 9.895107253416434e-06,
      "loss": 0.1863,
      "step": 214
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6659483396194743,
      "learning_rate": 9.893660635895542e-06,
      "loss": 0.2504,
      "step": 215
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7612907689482133,
      "learning_rate": 9.892204218340866e-06,
      "loss": 0.2439,
      "step": 216
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8411317160250582,
      "learning_rate": 9.890738003669029e-06,
      "loss": 0.2561,
      "step": 217
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.805749728571466,
      "learning_rate": 9.889261994816266e-06,
      "loss": 0.2071,
      "step": 218
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6853321853638796,
      "learning_rate": 9.887776194738433e-06,
      "loss": 0.2128,
      "step": 219
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5482874508601405,
      "learning_rate": 9.886280606410984e-06,
      "loss": 0.2319,
      "step": 220
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.452393028060986,
      "learning_rate": 9.884775232828985e-06,
      "loss": 0.2525,
      "step": 221
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7059631816849607,
      "learning_rate": 9.883260077007092e-06,
      "loss": 0.257,
      "step": 222
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8961711394062148,
      "learning_rate": 9.881735141979554e-06,
      "loss": 0.2242,
      "step": 223
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.511797586590514,
      "learning_rate": 9.8802004308002e-06,
      "loss": 0.1955,
      "step": 224
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.020830762016121,
      "learning_rate": 9.878655946542443e-06,
      "loss": 0.2533,
      "step": 225
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.490093453288554,
      "learning_rate": 9.87710169229926e-06,
      "loss": 0.2268,
      "step": 226
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5241386828590486,
      "learning_rate": 9.8755376711832e-06,
      "loss": 0.2712,
      "step": 227
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4676769564980552,
      "learning_rate": 9.873963886326365e-06,
      "loss": 0.2028,
      "step": 228
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5490892678052868,
      "learning_rate": 9.872380340880416e-06,
      "loss": 0.2524,
      "step": 229
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8076810808784352,
      "learning_rate": 9.870787038016556e-06,
      "loss": 0.2445,
      "step": 230
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7702079989725992,
      "learning_rate": 9.869183980925531e-06,
      "loss": 0.2267,
      "step": 231
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5827799833243492,
      "learning_rate": 9.867571172817617e-06,
      "loss": 0.2341,
      "step": 232
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6890438081814851,
      "learning_rate": 9.865948616922623e-06,
      "loss": 0.2194,
      "step": 233
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5534654973198792,
      "learning_rate": 9.864316316489873e-06,
      "loss": 0.252,
      "step": 234
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6984847172522834,
      "learning_rate": 9.862674274788212e-06,
      "loss": 0.2389,
      "step": 235
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4146273621741656,
      "learning_rate": 9.861022495105983e-06,
      "loss": 0.2055,
      "step": 236
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5801528825703548,
      "learning_rate": 9.85936098075104e-06,
      "loss": 0.1868,
      "step": 237
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6221926246008802,
      "learning_rate": 9.85768973505073e-06,
      "loss": 0.2573,
      "step": 238
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9249325469936667,
      "learning_rate": 9.856008761351882e-06,
      "loss": 0.2566,
      "step": 239
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6423687415384367,
      "learning_rate": 9.85431806302081e-06,
      "loss": 0.2386,
      "step": 240
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4577016130010658,
      "learning_rate": 9.85261764344331e-06,
      "loss": 0.2094,
      "step": 241
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7351838100029997,
      "learning_rate": 9.850907506024627e-06,
      "loss": 0.2944,
      "step": 242
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5601422493747312,
      "learning_rate": 9.849187654189486e-06,
      "loss": 0.2379,
      "step": 243
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5714162832120715,
      "learning_rate": 9.847458091382057e-06,
      "loss": 0.2005,
      "step": 244
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5697871677047917,
      "learning_rate": 9.845718821065957e-06,
      "loss": 0.2566,
      "step": 245
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5102120184361798,
      "learning_rate": 9.843969846724247e-06,
      "loss": 0.1732,
      "step": 246
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.483560136569595,
      "learning_rate": 9.842211171859418e-06,
      "loss": 0.2132,
      "step": 247
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7759452695784435,
      "learning_rate": 9.840442799993389e-06,
      "loss": 0.247,
      "step": 248
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5476992328742043,
      "learning_rate": 9.838664734667496e-06,
      "loss": 0.2211,
      "step": 249
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6055377966651656,
      "learning_rate": 9.83687697944249e-06,
      "loss": 0.2122,
      "step": 250
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.426691716044699,
      "learning_rate": 9.835079537898523e-06,
      "loss": 0.247,
      "step": 251
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8403146717694454,
      "learning_rate": 9.833272413635151e-06,
      "loss": 0.3107,
      "step": 252
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.507774473011047,
      "learning_rate": 9.831455610271315e-06,
      "loss": 0.1847,
      "step": 253
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6024148285530597,
      "learning_rate": 9.829629131445342e-06,
      "loss": 0.2527,
      "step": 254
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8236735021931736,
      "learning_rate": 9.827792980814934e-06,
      "loss": 0.2614,
      "step": 255
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7890537183306798,
      "learning_rate": 9.825947162057163e-06,
      "loss": 0.3064,
      "step": 256
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4578155708280904,
      "learning_rate": 9.82409167886846e-06,
      "loss": 0.2033,
      "step": 257
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.400653468398114,
      "learning_rate": 9.822226534964613e-06,
      "loss": 0.2221,
      "step": 258
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6482222748034765,
      "learning_rate": 9.820351734080754e-06,
      "loss": 0.2127,
      "step": 259
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.941686744440311,
      "learning_rate": 9.818467279971356e-06,
      "loss": 0.245,
      "step": 260
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6207662512796566,
      "learning_rate": 9.81657317641022e-06,
      "loss": 0.2462,
      "step": 261
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5481345450610977,
      "learning_rate": 9.814669427190477e-06,
      "loss": 0.249,
      "step": 262
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5105082248961013,
      "learning_rate": 9.81275603612457e-06,
      "loss": 0.262,
      "step": 263
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6921816759404873,
      "learning_rate": 9.810833007044247e-06,
      "loss": 0.2927,
      "step": 264
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5305113672462503,
      "learning_rate": 9.808900343800568e-06,
      "loss": 0.2675,
      "step": 265
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8443591826135717,
      "learning_rate": 9.80695805026388e-06,
      "loss": 0.2422,
      "step": 266
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5540972427238597,
      "learning_rate": 9.80500613032381e-06,
      "loss": 0.2564,
      "step": 267
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5828540373065454,
      "learning_rate": 9.803044587889273e-06,
      "loss": 0.2197,
      "step": 268
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8756640547240244,
      "learning_rate": 9.801073426888447e-06,
      "loss": 0.2666,
      "step": 269
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7568825434743032,
      "learning_rate": 9.799092651268778e-06,
      "loss": 0.2786,
      "step": 270
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.64276675772922,
      "learning_rate": 9.79710226499696e-06,
      "loss": 0.2742,
      "step": 271
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5478724388242864,
      "learning_rate": 9.795102272058937e-06,
      "loss": 0.1734,
      "step": 272
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4594403524402133,
      "learning_rate": 9.79309267645989e-06,
      "loss": 0.2751,
      "step": 273
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.385664159559033,
      "learning_rate": 9.791073482224229e-06,
      "loss": 0.214,
      "step": 274
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.556877224496772,
      "learning_rate": 9.789044693395593e-06,
      "loss": 0.2366,
      "step": 275
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4746668304576422,
      "learning_rate": 9.787006314036824e-06,
      "loss": 0.176,
      "step": 276
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.695223060467979,
      "learning_rate": 9.784958348229978e-06,
      "loss": 0.2389,
      "step": 277
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5498600280886436,
      "learning_rate": 9.78290080007631e-06,
      "loss": 0.2596,
      "step": 278
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.549148724637892,
      "learning_rate": 9.780833673696255e-06,
      "loss": 0.2922,
      "step": 279
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5787235571468978,
      "learning_rate": 9.778756973229442e-06,
      "loss": 0.2372,
      "step": 280
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5919788581978302,
      "learning_rate": 9.776670702834662e-06,
      "loss": 0.1911,
      "step": 281
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4505685924002085,
      "learning_rate": 9.774574866689878e-06,
      "loss": 0.2443,
      "step": 282
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8192221749846271,
      "learning_rate": 9.772469468992206e-06,
      "loss": 0.2568,
      "step": 283
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.776596900664368,
      "learning_rate": 9.77035451395791e-06,
      "loss": 0.2662,
      "step": 284
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4852638805110547,
      "learning_rate": 9.768230005822394e-06,
      "loss": 0.2209,
      "step": 285
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5365354066452879,
      "learning_rate": 9.766095948840195e-06,
      "loss": 0.1966,
      "step": 286
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7190820431168954,
      "learning_rate": 9.76395234728497e-06,
      "loss": 0.2428,
      "step": 287
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.658293792797876,
      "learning_rate": 9.761799205449492e-06,
      "loss": 0.2865,
      "step": 288
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.608018631883266,
      "learning_rate": 9.759636527645633e-06,
      "loss": 0.2476,
      "step": 289
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.542229788527783,
      "learning_rate": 9.757464318204374e-06,
      "loss": 0.2037,
      "step": 290
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.626492917245043,
      "learning_rate": 9.755282581475769e-06,
      "loss": 0.2422,
      "step": 291
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4615167276869756,
      "learning_rate": 9.753091321828964e-06,
      "loss": 0.2122,
      "step": 292
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4971359026612043,
      "learning_rate": 9.750890543652169e-06,
      "loss": 0.2239,
      "step": 293
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6942148725321633,
      "learning_rate": 9.74868025135266e-06,
      "loss": 0.2118,
      "step": 294
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4778342242823457,
      "learning_rate": 9.74646044935676e-06,
      "loss": 0.1894,
      "step": 295
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.540507327647611,
      "learning_rate": 9.744231142109842e-06,
      "loss": 0.2415,
      "step": 296
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8228984668542727,
      "learning_rate": 9.741992334076309e-06,
      "loss": 0.2229,
      "step": 297
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6109652584506715,
      "learning_rate": 9.739744029739596e-06,
      "loss": 0.2537,
      "step": 298
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.568603539374847,
      "learning_rate": 9.737486233602149e-06,
      "loss": 0.2318,
      "step": 299
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5758412007903408,
      "learning_rate": 9.735218950185428e-06,
      "loss": 0.3298,
      "step": 300
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5763809174455317,
      "learning_rate": 9.73294218402989e-06,
      "loss": 0.2376,
      "step": 301
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5765041704611509,
      "learning_rate": 9.730655939694979e-06,
      "loss": 0.2658,
      "step": 302
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5911403616302842,
      "learning_rate": 9.728360221759125e-06,
      "loss": 0.2371,
      "step": 303
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6405027215437908,
      "learning_rate": 9.726055034819726e-06,
      "loss": 0.2761,
      "step": 304
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5464740698265413,
      "learning_rate": 9.723740383493146e-06,
      "loss": 0.2268,
      "step": 305
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.815793464071285,
      "learning_rate": 9.721416272414698e-06,
      "loss": 0.2471,
      "step": 306
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5898529995535668,
      "learning_rate": 9.719082706238645e-06,
      "loss": 0.2568,
      "step": 307
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.733520815976427,
      "learning_rate": 9.716739689638179e-06,
      "loss": 0.2705,
      "step": 308
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.472231038470541,
      "learning_rate": 9.714387227305422e-06,
      "loss": 0.2371,
      "step": 309
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6913352523277525,
      "learning_rate": 9.712025323951406e-06,
      "loss": 0.2633,
      "step": 310
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.497735994513071,
      "learning_rate": 9.709653984306077e-06,
      "loss": 0.1872,
      "step": 311
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6104610220845144,
      "learning_rate": 9.70727321311827e-06,
      "loss": 0.1891,
      "step": 312
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7441601861636489,
      "learning_rate": 9.70488301515572e-06,
      "loss": 0.2777,
      "step": 313
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.67453385444093,
      "learning_rate": 9.702483395205023e-06,
      "loss": 0.219,
      "step": 314
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.65202648115151,
      "learning_rate": 9.700074358071658e-06,
      "loss": 0.3121,
      "step": 315
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5776154301290874,
      "learning_rate": 9.697655908579958e-06,
      "loss": 0.252,
      "step": 316
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4292945399774923,
      "learning_rate": 9.695228051573098e-06,
      "loss": 0.1895,
      "step": 317
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5367720435132135,
      "learning_rate": 9.692790791913107e-06,
      "loss": 0.238,
      "step": 318
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6048007076804263,
      "learning_rate": 9.69034413448083e-06,
      "loss": 0.1973,
      "step": 319
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.668080889425604,
      "learning_rate": 9.68788808417594e-06,
      "loss": 0.2462,
      "step": 320
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5814267296937652,
      "learning_rate": 9.68542264591692e-06,
      "loss": 0.237,
      "step": 321
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6461533387973244,
      "learning_rate": 9.682947824641045e-06,
      "loss": 0.2591,
      "step": 322
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6406699368803233,
      "learning_rate": 9.680463625304393e-06,
      "loss": 0.2455,
      "step": 323
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.650354535642039,
      "learning_rate": 9.677970052881811e-06,
      "loss": 0.233,
      "step": 324
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6636766953310282,
      "learning_rate": 9.675467112366925e-06,
      "loss": 0.2161,
      "step": 325
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4778548830279765,
      "learning_rate": 9.672954808772118e-06,
      "loss": 0.2375,
      "step": 326
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3804376203855266,
      "learning_rate": 9.670433147128522e-06,
      "loss": 0.2318,
      "step": 327
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5011301838669684,
      "learning_rate": 9.667902132486009e-06,
      "loss": 0.2053,
      "step": 328
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7388419254605094,
      "learning_rate": 9.665361769913187e-06,
      "loss": 0.315,
      "step": 329
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7934732605547576,
      "learning_rate": 9.66281206449738e-06,
      "loss": 0.1679,
      "step": 330
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4004567839089588,
      "learning_rate": 9.660253021344618e-06,
      "loss": 0.2248,
      "step": 331
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6711529021252833,
      "learning_rate": 9.65768464557964e-06,
      "loss": 0.2313,
      "step": 332
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4505012824749275,
      "learning_rate": 9.65510694234587e-06,
      "loss": 0.2704,
      "step": 333
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5839901754602568,
      "learning_rate": 9.652519916805406e-06,
      "loss": 0.1724,
      "step": 334
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.490295735351521,
      "learning_rate": 9.649923574139023e-06,
      "loss": 0.2897,
      "step": 335
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5280744717505252,
      "learning_rate": 9.64731791954615e-06,
      "loss": 0.2363,
      "step": 336
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2751373187679589,
      "learning_rate": 9.644702958244865e-06,
      "loss": 0.1933,
      "step": 337
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7390218644590267,
      "learning_rate": 9.642078695471882e-06,
      "loss": 0.221,
      "step": 338
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4541310625569643,
      "learning_rate": 9.639445136482549e-06,
      "loss": 0.2374,
      "step": 339
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8360909647923214,
      "learning_rate": 9.636802286550817e-06,
      "loss": 0.2377,
      "step": 340
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8689183470654973,
      "learning_rate": 9.634150150969259e-06,
      "loss": 0.3194,
      "step": 341
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3941854055655307,
      "learning_rate": 9.631488735049033e-06,
      "loss": 0.2503,
      "step": 342
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.403305498131308,
      "learning_rate": 9.628818044119884e-06,
      "loss": 0.2317,
      "step": 343
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4778320850746833,
      "learning_rate": 9.62613808353013e-06,
      "loss": 0.234,
      "step": 344
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5623470987119688,
      "learning_rate": 9.623448858646658e-06,
      "loss": 0.2432,
      "step": 345
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6057801120806232,
      "learning_rate": 9.620750374854901e-06,
      "loss": 0.2376,
      "step": 346
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5037846559180295,
      "learning_rate": 9.618042637558838e-06,
      "loss": 0.1928,
      "step": 347
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.839561411274333,
      "learning_rate": 9.615325652180976e-06,
      "loss": 0.264,
      "step": 348
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5533748657839157,
      "learning_rate": 9.612599424162344e-06,
      "loss": 0.2195,
      "step": 349
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4719019520886187,
      "learning_rate": 9.609863958962483e-06,
      "loss": 0.2365,
      "step": 350
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3825994663368102,
      "learning_rate": 9.607119262059426e-06,
      "loss": 0.2541,
      "step": 351
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6787497469187875,
      "learning_rate": 9.604365338949697e-06,
      "loss": 0.3119,
      "step": 352
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.209418088821381,
      "learning_rate": 9.601602195148299e-06,
      "loss": 0.2168,
      "step": 353
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.415301571936414,
      "learning_rate": 9.598829836188694e-06,
      "loss": 0.2298,
      "step": 354
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3828686981945217,
      "learning_rate": 9.596048267622806e-06,
      "loss": 0.2444,
      "step": 355
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7409572131802111,
      "learning_rate": 9.593257495020993e-06,
      "loss": 0.2733,
      "step": 356
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4257890365884398,
      "learning_rate": 9.590457523972055e-06,
      "loss": 0.2717,
      "step": 357
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5175557120291865,
      "learning_rate": 9.587648360083205e-06,
      "loss": 0.2052,
      "step": 358
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5032730657493178,
      "learning_rate": 9.584830008980068e-06,
      "loss": 0.2198,
      "step": 359
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5353824603144022,
      "learning_rate": 9.582002476306668e-06,
      "loss": 0.2508,
      "step": 360
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7470166631839108,
      "learning_rate": 9.579165767725416e-06,
      "loss": 0.2119,
      "step": 361
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.548731055819908,
      "learning_rate": 9.576319888917099e-06,
      "loss": 0.2348,
      "step": 362
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5117674606695963,
      "learning_rate": 9.573464845580864e-06,
      "loss": 0.2566,
      "step": 363
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.466401108099996,
      "learning_rate": 9.570600643434217e-06,
      "loss": 0.232,
      "step": 364
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4133835406568989,
      "learning_rate": 9.567727288213005e-06,
      "loss": 0.2094,
      "step": 365
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3639676543406065,
      "learning_rate": 9.5648447856714e-06,
      "loss": 0.201,
      "step": 366
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4618074844645295,
      "learning_rate": 9.561953141581895e-06,
      "loss": 0.2024,
      "step": 367
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7091668182817024,
      "learning_rate": 9.559052361735288e-06,
      "loss": 0.235,
      "step": 368
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7191493729654088,
      "learning_rate": 9.55614245194068e-06,
      "loss": 0.2266,
      "step": 369
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7096006402012796,
      "learning_rate": 9.553223418025443e-06,
      "loss": 0.2887,
      "step": 370
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6674844056050355,
      "learning_rate": 9.550295265835235e-06,
      "loss": 0.2671,
      "step": 371
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4918941764314504,
      "learning_rate": 9.547358001233959e-06,
      "loss": 0.2446,
      "step": 372
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5391760035519246,
      "learning_rate": 9.544411630103782e-06,
      "loss": 0.188,
      "step": 373
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6394389129573728,
      "learning_rate": 9.541456158345094e-06,
      "loss": 0.2483,
      "step": 374
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5298340014086473,
      "learning_rate": 9.538491591876522e-06,
      "loss": 0.2413,
      "step": 375
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6375127285493074,
      "learning_rate": 9.535517936634895e-06,
      "loss": 0.2316,
      "step": 376
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5412065008680225,
      "learning_rate": 9.53253519857525e-06,
      "loss": 0.2456,
      "step": 377
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4493994028669326,
      "learning_rate": 9.529543383670814e-06,
      "loss": 0.2068,
      "step": 378
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5818953343100728,
      "learning_rate": 9.526542497912984e-06,
      "loss": 0.28,
      "step": 379
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4206555085064627,
      "learning_rate": 9.523532547311332e-06,
      "loss": 0.1757,
      "step": 380
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.519391836852476,
      "learning_rate": 9.520513537893574e-06,
      "loss": 0.2619,
      "step": 381
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4319682018284974,
      "learning_rate": 9.517485475705574e-06,
      "loss": 0.2268,
      "step": 382
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7114213131872955,
      "learning_rate": 9.514448366811322e-06,
      "loss": 0.2863,
      "step": 383
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3932509621392448,
      "learning_rate": 9.511402217292927e-06,
      "loss": 0.1727,
      "step": 384
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4084607964337519,
      "learning_rate": 9.508347033250597e-06,
      "loss": 0.2003,
      "step": 385
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5340005200550662,
      "learning_rate": 9.505282820802642e-06,
      "loss": 0.2512,
      "step": 386
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6589176915904997,
      "learning_rate": 9.502209586085444e-06,
      "loss": 0.2573,
      "step": 387
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3080735843566078,
      "learning_rate": 9.499127335253456e-06,
      "loss": 0.1604,
      "step": 388
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5161112071023042,
      "learning_rate": 9.496036074479184e-06,
      "loss": 0.2493,
      "step": 389
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6295635156425756,
      "learning_rate": 9.492935809953185e-06,
      "loss": 0.2574,
      "step": 390
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3446026691756845,
      "learning_rate": 9.48982654788404e-06,
      "loss": 0.1912,
      "step": 391
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6706306655135041,
      "learning_rate": 9.486708294498348e-06,
      "loss": 0.2313,
      "step": 392
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4109264570902729,
      "learning_rate": 9.48358105604072e-06,
      "loss": 0.1965,
      "step": 393
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5712397464305596,
      "learning_rate": 9.480444838773753e-06,
      "loss": 0.2989,
      "step": 394
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4160028233725483,
      "learning_rate": 9.47729964897803e-06,
      "loss": 0.2297,
      "step": 395
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5279122113176997,
      "learning_rate": 9.474145492952102e-06,
      "loss": 0.2507,
      "step": 396
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.389122358295883,
      "learning_rate": 9.470982377012475e-06,
      "loss": 0.1953,
      "step": 397
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5127482023676166,
      "learning_rate": 9.467810307493595e-06,
      "loss": 0.2589,
      "step": 398
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5603042381807397,
      "learning_rate": 9.464629290747844e-06,
      "loss": 0.2584,
      "step": 399
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4939225980738748,
      "learning_rate": 9.461439333145516e-06,
      "loss": 0.1932,
      "step": 400
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4512788619391497,
      "learning_rate": 9.458240441074818e-06,
      "loss": 0.2012,
      "step": 401
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4104749410700608,
      "learning_rate": 9.45503262094184e-06,
      "loss": 0.1912,
      "step": 402
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.622723842049091,
      "learning_rate": 9.451815879170556e-06,
      "loss": 0.2598,
      "step": 403
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6057041778363608,
      "learning_rate": 9.448590222202808e-06,
      "loss": 0.2709,
      "step": 404
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5639351763471907,
      "learning_rate": 9.445355656498284e-06,
      "loss": 0.2655,
      "step": 405
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4406161039636565,
      "learning_rate": 9.442112188534526e-06,
      "loss": 0.2773,
      "step": 406
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.482003614160717,
      "learning_rate": 9.438859824806889e-06,
      "loss": 0.1925,
      "step": 407
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4830945136071942,
      "learning_rate": 9.435598571828552e-06,
      "loss": 0.2106,
      "step": 408
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4701803474299193,
      "learning_rate": 9.432328436130493e-06,
      "loss": 0.2196,
      "step": 409
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4563734835601387,
      "learning_rate": 9.429049424261478e-06,
      "loss": 0.1879,
      "step": 410
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5854456511722363,
      "learning_rate": 9.425761542788049e-06,
      "loss": 0.2177,
      "step": 411
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3240550464245295,
      "learning_rate": 9.422464798294509e-06,
      "loss": 0.179,
      "step": 412
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4795638152880777,
      "learning_rate": 9.419159197382913e-06,
      "loss": 0.2019,
      "step": 413
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5584870174752594,
      "learning_rate": 9.415844746673047e-06,
      "loss": 0.2223,
      "step": 414
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5324351348312155,
      "learning_rate": 9.412521452802425e-06,
      "loss": 0.2689,
      "step": 415
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4934672302188814,
      "learning_rate": 9.409189322426267e-06,
      "loss": 0.1934,
      "step": 416
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4822623785304663,
      "learning_rate": 9.40584836221749e-06,
      "loss": 0.3176,
      "step": 417
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.42640328997091,
      "learning_rate": 9.402498578866695e-06,
      "loss": 0.2732,
      "step": 418
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8896472608003578,
      "learning_rate": 9.399139979082148e-06,
      "loss": 0.2532,
      "step": 419
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4889950781319525,
      "learning_rate": 9.395772569589774e-06,
      "loss": 0.1939,
      "step": 420
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.450481827966664,
      "learning_rate": 9.39239635713314e-06,
      "loss": 0.2424,
      "step": 421
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4075117803921127,
      "learning_rate": 9.389011348473442e-06,
      "loss": 0.2197,
      "step": 422
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6116196410991632,
      "learning_rate": 9.38561755038949e-06,
      "loss": 0.2448,
      "step": 423
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6357737981825757,
      "learning_rate": 9.382214969677697e-06,
      "loss": 0.2342,
      "step": 424
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2877903020350792,
      "learning_rate": 9.37880361315206e-06,
      "loss": 0.1888,
      "step": 425
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4288295677078346,
      "learning_rate": 9.375383487644161e-06,
      "loss": 0.2591,
      "step": 426
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5913788732095884,
      "learning_rate": 9.37195460000313e-06,
      "loss": 0.196,
      "step": 427
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3302447862020053,
      "learning_rate": 9.368516957095652e-06,
      "loss": 0.2424,
      "step": 428
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4361017838825691,
      "learning_rate": 9.365070565805941e-06,
      "loss": 0.2482,
      "step": 429
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.522359840769197,
      "learning_rate": 9.361615433035736e-06,
      "loss": 0.229,
      "step": 430
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6655519690507983,
      "learning_rate": 9.358151565704275e-06,
      "loss": 0.2142,
      "step": 431
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3058555692857647,
      "learning_rate": 9.35467897074829e-06,
      "loss": 0.1703,
      "step": 432
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3498431525356038,
      "learning_rate": 9.351197655121995e-06,
      "loss": 0.1959,
      "step": 433
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4452728848185246,
      "learning_rate": 9.347707625797062e-06,
      "loss": 0.1917,
      "step": 434
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.385450331467578,
      "learning_rate": 9.34420888976262e-06,
      "loss": 0.19,
      "step": 435
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5215718289691529,
      "learning_rate": 9.340701454025224e-06,
      "loss": 0.2597,
      "step": 436
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6067917616674485,
      "learning_rate": 9.337185325608864e-06,
      "loss": 0.2257,
      "step": 437
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4238322566494623,
      "learning_rate": 9.333660511554926e-06,
      "loss": 0.2128,
      "step": 438
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4422417268125751,
      "learning_rate": 9.330127018922195e-06,
      "loss": 0.2161,
      "step": 439
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4661740860937005,
      "learning_rate": 9.326584854786835e-06,
      "loss": 0.2723,
      "step": 440
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4715208909938047,
      "learning_rate": 9.323034026242378e-06,
      "loss": 0.2611,
      "step": 441
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8227165502957963,
      "learning_rate": 9.319474540399704e-06,
      "loss": 0.2661,
      "step": 442
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5268808310836521,
      "learning_rate": 9.315906404387032e-06,
      "loss": 0.2373,
      "step": 443
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3383138504755894,
      "learning_rate": 9.312329625349903e-06,
      "loss": 0.2112,
      "step": 444
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6289036559740029,
      "learning_rate": 9.308744210451167e-06,
      "loss": 0.2435,
      "step": 445
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3520431598418725,
      "learning_rate": 9.305150166870966e-06,
      "loss": 0.2294,
      "step": 446
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3826848188548615,
      "learning_rate": 9.301547501806725e-06,
      "loss": 0.2276,
      "step": 447
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4762182006522009,
      "learning_rate": 9.297936222473134e-06,
      "loss": 0.2193,
      "step": 448
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3703114974071575,
      "learning_rate": 9.294316336102132e-06,
      "loss": 0.1883,
      "step": 449
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4223013085222818,
      "learning_rate": 9.290687849942893e-06,
      "loss": 0.2401,
      "step": 450
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.460279499582394,
      "learning_rate": 9.287050771261818e-06,
      "loss": 0.2133,
      "step": 451
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3093680558460008,
      "learning_rate": 9.283405107342512e-06,
      "loss": 0.1795,
      "step": 452
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6416072082917015,
      "learning_rate": 9.279750865485772e-06,
      "loss": 0.1988,
      "step": 453
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.45952594875811,
      "learning_rate": 9.276088053009578e-06,
      "loss": 0.2262,
      "step": 454
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5969280368204715,
      "learning_rate": 9.272416677249067e-06,
      "loss": 0.208,
      "step": 455
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3873719018584991,
      "learning_rate": 9.268736745556528e-06,
      "loss": 0.2523,
      "step": 456
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5176889248084804,
      "learning_rate": 9.265048265301385e-06,
      "loss": 0.2323,
      "step": 457
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7226857001416893,
      "learning_rate": 9.261351243870178e-06,
      "loss": 0.2826,
      "step": 458
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4378458315355833,
      "learning_rate": 9.257645688666557e-06,
      "loss": 0.2024,
      "step": 459
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3003342490760494,
      "learning_rate": 9.253931607111257e-06,
      "loss": 0.2036,
      "step": 460
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.488023206590022,
      "learning_rate": 9.25020900664209e-06,
      "loss": 0.2131,
      "step": 461
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5794863785100612,
      "learning_rate": 9.246477894713925e-06,
      "loss": 0.2619,
      "step": 462
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7113714802976556,
      "learning_rate": 9.242738278798682e-06,
      "loss": 0.2472,
      "step": 463
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5200597136838736,
      "learning_rate": 9.238990166385304e-06,
      "loss": 0.2181,
      "step": 464
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4527479716527656,
      "learning_rate": 9.235233564979756e-06,
      "loss": 0.2041,
      "step": 465
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3376893897525808,
      "learning_rate": 9.231468482104994e-06,
      "loss": 0.1836,
      "step": 466
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.377348154320528,
      "learning_rate": 9.227694925300972e-06,
      "loss": 0.2622,
      "step": 467
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.328030048422613,
      "learning_rate": 9.223912902124601e-06,
      "loss": 0.207,
      "step": 468
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3839733977783981,
      "learning_rate": 9.220122420149753e-06,
      "loss": 0.2053,
      "step": 469
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5416529827464887,
      "learning_rate": 9.216323486967239e-06,
      "loss": 0.2253,
      "step": 470
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.521793668307093,
      "learning_rate": 9.212516110184794e-06,
      "loss": 0.2957,
      "step": 471
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5351333200294155,
      "learning_rate": 9.208700297427063e-06,
      "loss": 0.263,
      "step": 472
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4386751658299517,
      "learning_rate": 9.204876056335584e-06,
      "loss": 0.2405,
      "step": 473
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4464802786301674,
      "learning_rate": 9.201043394568773e-06,
      "loss": 0.2363,
      "step": 474
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.478749653510334,
      "learning_rate": 9.197202319801908e-06,
      "loss": 0.1761,
      "step": 475
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3130495995917975,
      "learning_rate": 9.193352839727122e-06,
      "loss": 0.1936,
      "step": 476
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.292374955534055,
      "learning_rate": 9.18949496205337e-06,
      "loss": 0.1526,
      "step": 477
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3373226602720405,
      "learning_rate": 9.18562869450643e-06,
      "loss": 0.1688,
      "step": 478
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4110247268909608,
      "learning_rate": 9.181754044828882e-06,
      "loss": 0.249,
      "step": 479
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.464939718406061,
      "learning_rate": 9.17787102078009e-06,
      "loss": 0.2708,
      "step": 480
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4268394745278077,
      "learning_rate": 9.17397963013619e-06,
      "loss": 0.2008,
      "step": 481
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3873666986351585,
      "learning_rate": 9.170079880690072e-06,
      "loss": 0.1905,
      "step": 482
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.64211667238452,
      "learning_rate": 9.166171780251365e-06,
      "loss": 0.2237,
      "step": 483
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.485913533445401,
      "learning_rate": 9.162255336646422e-06,
      "loss": 0.1818,
      "step": 484
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.52000115524981,
      "learning_rate": 9.158330557718304e-06,
      "loss": 0.2309,
      "step": 485
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3832204971096536,
      "learning_rate": 9.154397451326766e-06,
      "loss": 0.1787,
      "step": 486
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2939110495908381,
      "learning_rate": 9.150456025348236e-06,
      "loss": 0.1614,
      "step": 487
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4486254815920727,
      "learning_rate": 9.146506287675805e-06,
      "loss": 0.2295,
      "step": 488
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6891619622658325,
      "learning_rate": 9.142548246219212e-06,
      "loss": 0.2251,
      "step": 489
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4730548588545926,
      "learning_rate": 9.138581908904818e-06,
      "loss": 0.2257,
      "step": 490
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3685587147578577,
      "learning_rate": 9.134607283675601e-06,
      "loss": 0.179,
      "step": 491
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3884063136472495,
      "learning_rate": 9.13062437849114e-06,
      "loss": 0.1826,
      "step": 492
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3256434796054912,
      "learning_rate": 9.12663320132759e-06,
      "loss": 0.1717,
      "step": 493
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.537852054952254,
      "learning_rate": 9.122633760177674e-06,
      "loss": 0.2276,
      "step": 494
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4887073121407226,
      "learning_rate": 9.118626063050661e-06,
      "loss": 0.2578,
      "step": 495
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0510188133900424,
      "learning_rate": 9.114610117972362e-06,
      "loss": 0.2027,
      "step": 496
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5009237085412057,
      "learning_rate": 9.110585932985095e-06,
      "loss": 0.2336,
      "step": 497
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3855660320294005,
      "learning_rate": 9.106553516147683e-06,
      "loss": 0.2184,
      "step": 498
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4445535602291382,
      "learning_rate": 9.102512875535439e-06,
      "loss": 0.2154,
      "step": 499
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4790540621036417,
      "learning_rate": 9.098464019240137e-06,
      "loss": 0.289,
      "step": 500
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.494673071409777,
      "learning_rate": 9.09440695537001e-06,
      "loss": 0.2257,
      "step": 501
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3516641899436668,
      "learning_rate": 9.090341692049725e-06,
      "loss": 0.1807,
      "step": 502
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.401363923755623,
      "learning_rate": 9.086268237420368e-06,
      "loss": 0.2052,
      "step": 503
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6956062140369625,
      "learning_rate": 9.082186599639429e-06,
      "loss": 0.2499,
      "step": 504
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3818894835135351,
      "learning_rate": 9.07809678688079e-06,
      "loss": 0.1724,
      "step": 505
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.510440599680523,
      "learning_rate": 9.073998807334698e-06,
      "loss": 0.2447,
      "step": 506
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3441478934087392,
      "learning_rate": 9.069892669207757e-06,
      "loss": 0.1622,
      "step": 507
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3690649408775901,
      "learning_rate": 9.065778380722914e-06,
      "loss": 0.265,
      "step": 508
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4015667609867264,
      "learning_rate": 9.06165595011943e-06,
      "loss": 0.1565,
      "step": 509
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.544176367025118,
      "learning_rate": 9.057525385652877e-06,
      "loss": 0.2189,
      "step": 510
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3172868440059742,
      "learning_rate": 9.053386695595114e-06,
      "loss": 0.1681,
      "step": 511
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6211020332642505,
      "learning_rate": 9.049239888234271e-06,
      "loss": 0.2411,
      "step": 512
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3138309023203834,
      "learning_rate": 9.045084971874738e-06,
      "loss": 0.2379,
      "step": 513
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7781226440436388,
      "learning_rate": 9.040921954837139e-06,
      "loss": 0.2701,
      "step": 514
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.52813094339364,
      "learning_rate": 9.036750845458323e-06,
      "loss": 0.2466,
      "step": 515
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4763097729408088,
      "learning_rate": 9.032571652091341e-06,
      "loss": 0.2042,
      "step": 516
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5187956158035698,
      "learning_rate": 9.028384383105444e-06,
      "loss": 0.203,
      "step": 517
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.428791217043001,
      "learning_rate": 9.02418904688604e-06,
      "loss": 0.201,
      "step": 518
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4867798602637328,
      "learning_rate": 9.019985651834703e-06,
      "loss": 0.2523,
      "step": 519
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1322218138520423,
      "learning_rate": 9.015774206369144e-06,
      "loss": 0.1469,
      "step": 520
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.333761014399461,
      "learning_rate": 9.01155471892319e-06,
      "loss": 0.1622,
      "step": 521
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.280477900497724,
      "learning_rate": 9.007327197946782e-06,
      "loss": 0.1203,
      "step": 522
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3714166705255406,
      "learning_rate": 9.00309165190594e-06,
      "loss": 0.2692,
      "step": 523
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7047689667392163,
      "learning_rate": 8.99884808928276e-06,
      "loss": 0.2546,
      "step": 524
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.392631906894928,
      "learning_rate": 8.994596518575393e-06,
      "loss": 0.26,
      "step": 525
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4471947871893625,
      "learning_rate": 8.990336948298023e-06,
      "loss": 0.2451,
      "step": 526
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4820591565015668,
      "learning_rate": 8.986069386980855e-06,
      "loss": 0.2351,
      "step": 527
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3668958535531748,
      "learning_rate": 8.9817938431701e-06,
      "loss": 0.2025,
      "step": 528
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.239383936011545,
      "learning_rate": 8.97751032542795e-06,
      "loss": 0.2238,
      "step": 529
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2387351577463739,
      "learning_rate": 8.97321884233257e-06,
      "loss": 0.1794,
      "step": 530
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2649773588982762,
      "learning_rate": 8.968919402478076e-06,
      "loss": 0.2031,
      "step": 531
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.390580030031749,
      "learning_rate": 8.964612014474512e-06,
      "loss": 0.2648,
      "step": 532
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4172884852155057,
      "learning_rate": 8.960296686947846e-06,
      "loss": 0.1712,
      "step": 533
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.477014991880758,
      "learning_rate": 8.955973428539943e-06,
      "loss": 0.1892,
      "step": 534
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6096493975225268,
      "learning_rate": 8.95164224790855e-06,
      "loss": 0.2558,
      "step": 535
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.253223851073495,
      "learning_rate": 8.947303153727283e-06,
      "loss": 0.1869,
      "step": 536
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.516177980491247,
      "learning_rate": 8.942956154685596e-06,
      "loss": 0.2325,
      "step": 537
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4935450392826626,
      "learning_rate": 8.938601259488784e-06,
      "loss": 0.1948,
      "step": 538
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6544447632157222,
      "learning_rate": 8.93423847685795e-06,
      "loss": 0.275,
      "step": 539
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3475723951124685,
      "learning_rate": 8.929867815529992e-06,
      "loss": 0.2409,
      "step": 540
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4398380932529322,
      "learning_rate": 8.925489284257587e-06,
      "loss": 0.2822,
      "step": 541
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.22467963537104,
      "learning_rate": 8.921102891809171e-06,
      "loss": 0.182,
      "step": 542
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.418099137416899,
      "learning_rate": 8.916708646968924e-06,
      "loss": 0.2119,
      "step": 543
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7745506011141308,
      "learning_rate": 8.91230655853675e-06,
      "loss": 0.2808,
      "step": 544
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4704867474650054,
      "learning_rate": 8.907896635328262e-06,
      "loss": 0.2581,
      "step": 545
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.426929218611027,
      "learning_rate": 8.903478886174763e-06,
      "loss": 0.2044,
      "step": 546
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6346500723311421,
      "learning_rate": 8.899053319923228e-06,
      "loss": 0.2311,
      "step": 547
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5254779141488817,
      "learning_rate": 8.894619945436285e-06,
      "loss": 0.2324,
      "step": 548
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.390141451159735,
      "learning_rate": 8.890178771592198e-06,
      "loss": 0.2101,
      "step": 549
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5137147413539822,
      "learning_rate": 8.885729807284855e-06,
      "loss": 0.2666,
      "step": 550
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5992075375437078,
      "learning_rate": 8.881273061423741e-06,
      "loss": 0.2369,
      "step": 551
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5157623393441042,
      "learning_rate": 8.876808542933924e-06,
      "loss": 0.2175,
      "step": 552
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3544926315171755,
      "learning_rate": 8.87233626075604e-06,
      "loss": 0.213,
      "step": 553
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3450726336356116,
      "learning_rate": 8.86785622384627e-06,
      "loss": 0.2229,
      "step": 554
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3189835170053836,
      "learning_rate": 8.863368441176326e-06,
      "loss": 0.2046,
      "step": 555
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6143826872608258,
      "learning_rate": 8.858872921733433e-06,
      "loss": 0.2693,
      "step": 556
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3707595574646678,
      "learning_rate": 8.854369674520306e-06,
      "loss": 0.2147,
      "step": 557
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3931348241155537,
      "learning_rate": 8.84985870855514e-06,
      "loss": 0.2677,
      "step": 558
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4301627854580041,
      "learning_rate": 8.845340032871584e-06,
      "loss": 0.2526,
      "step": 559
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3686454720557961,
      "learning_rate": 8.840813656518728e-06,
      "loss": 0.232,
      "step": 560
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3484771265472866,
      "learning_rate": 8.836279588561084e-06,
      "loss": 0.2171,
      "step": 561
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4702187404449876,
      "learning_rate": 8.831737838078562e-06,
      "loss": 0.2207,
      "step": 562
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.512952602840696,
      "learning_rate": 8.827188414166468e-06,
      "loss": 0.256,
      "step": 563
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3183802927162898,
      "learning_rate": 8.822631325935463e-06,
      "loss": 0.2336,
      "step": 564
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3389213962464033,
      "learning_rate": 8.818066582511564e-06,
      "loss": 0.2103,
      "step": 565
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.756811759937168,
      "learning_rate": 8.813494193036115e-06,
      "loss": 0.2917,
      "step": 566
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3391171881789696,
      "learning_rate": 8.808914166665773e-06,
      "loss": 0.3015,
      "step": 567
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4641852361631678,
      "learning_rate": 8.80432651257249e-06,
      "loss": 0.1995,
      "step": 568
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7523861092907933,
      "learning_rate": 8.799731239943488e-06,
      "loss": 0.2779,
      "step": 569
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4350200915149074,
      "learning_rate": 8.795128357981253e-06,
      "loss": 0.2053,
      "step": 570
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.521125733223817,
      "learning_rate": 8.790517875903506e-06,
      "loss": 0.2337,
      "step": 571
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5603991427599475,
      "learning_rate": 8.785899802943186e-06,
      "loss": 0.2227,
      "step": 572
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5987797078172366,
      "learning_rate": 8.781274148348438e-06,
      "loss": 0.2239,
      "step": 573
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6995450134395165,
      "learning_rate": 8.776640921382585e-06,
      "loss": 0.2059,
      "step": 574
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4590948525617484,
      "learning_rate": 8.772000131324117e-06,
      "loss": 0.2012,
      "step": 575
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8756516684667974,
      "learning_rate": 8.767351787466674e-06,
      "loss": 0.2731,
      "step": 576
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6337684071579361,
      "learning_rate": 8.762695899119012e-06,
      "loss": 0.2637,
      "step": 577
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.426426043928621,
      "learning_rate": 8.75803247560501e-06,
      "loss": 0.1912,
      "step": 578
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4274954897432452,
      "learning_rate": 8.753361526263622e-06,
      "loss": 0.2291,
      "step": 579
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4466275303203555,
      "learning_rate": 8.748683060448886e-06,
      "loss": 0.2271,
      "step": 580
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3678499461800329,
      "learning_rate": 8.743997087529886e-06,
      "loss": 0.2007,
      "step": 581
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.58604762161168,
      "learning_rate": 8.739303616890739e-06,
      "loss": 0.2693,
      "step": 582
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5007446272140708,
      "learning_rate": 8.734602657930582e-06,
      "loss": 0.2513,
      "step": 583
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6137784519504867,
      "learning_rate": 8.729894220063542e-06,
      "loss": 0.2557,
      "step": 584
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3316179656059433,
      "learning_rate": 8.725178312718727e-06,
      "loss": 0.2645,
      "step": 585
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.373239756700817,
      "learning_rate": 8.720454945340202e-06,
      "loss": 0.2204,
      "step": 586
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3929985331825467,
      "learning_rate": 8.715724127386971e-06,
      "loss": 0.2371,
      "step": 587
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5236381177395464,
      "learning_rate": 8.710985868332964e-06,
      "loss": 0.2873,
      "step": 588
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4558060384958431,
      "learning_rate": 8.706240177667003e-06,
      "loss": 0.2362,
      "step": 589
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.474440192326182,
      "learning_rate": 8.701487064892797e-06,
      "loss": 0.2068,
      "step": 590
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4359942276867248,
      "learning_rate": 8.696726539528924e-06,
      "loss": 0.2196,
      "step": 591
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4407613140317779,
      "learning_rate": 8.691958611108795e-06,
      "loss": 0.2364,
      "step": 592
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5001155206261125,
      "learning_rate": 8.687183289180659e-06,
      "loss": 0.2955,
      "step": 593
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4462255823452517,
      "learning_rate": 8.682400583307562e-06,
      "loss": 0.2288,
      "step": 594
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2990507202704602,
      "learning_rate": 8.67761050306734e-06,
      "loss": 0.1955,
      "step": 595
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6343828091549593,
      "learning_rate": 8.672813058052596e-06,
      "loss": 0.2569,
      "step": 596
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5272181435694299,
      "learning_rate": 8.668008257870684e-06,
      "loss": 0.1815,
      "step": 597
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2719611862312488,
      "learning_rate": 8.663196112143685e-06,
      "loss": 0.1681,
      "step": 598
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.477862462388859,
      "learning_rate": 8.658376630508391e-06,
      "loss": 0.2379,
      "step": 599
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5530847793459945,
      "learning_rate": 8.65354982261629e-06,
      "loss": 0.2234,
      "step": 600
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3634831082618888,
      "learning_rate": 8.648715698133529e-06,
      "loss": 0.2086,
      "step": 601
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.790231320205886,
      "learning_rate": 8.64387426674092e-06,
      "loss": 0.2471,
      "step": 602
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.626053483481875,
      "learning_rate": 8.639025538133899e-06,
      "loss": 0.1795,
      "step": 603
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4893603850804764,
      "learning_rate": 8.634169522022522e-06,
      "loss": 0.188,
      "step": 604
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.463849105036166,
      "learning_rate": 8.629306228131435e-06,
      "loss": 0.2741,
      "step": 605
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2887117493516982,
      "learning_rate": 8.62443566619986e-06,
      "loss": 0.2399,
      "step": 606
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5708907150175158,
      "learning_rate": 8.619557845981571e-06,
      "loss": 0.2188,
      "step": 607
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.680640566178907,
      "learning_rate": 8.614672777244885e-06,
      "loss": 0.2812,
      "step": 608
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5567567347324964,
      "learning_rate": 8.609780469772623e-06,
      "loss": 0.2491,
      "step": 609
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3799000446273013,
      "learning_rate": 8.604880933362113e-06,
      "loss": 0.3289,
      "step": 610
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5603584955184422,
      "learning_rate": 8.599974177825155e-06,
      "loss": 0.2182,
      "step": 611
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.357643697323284,
      "learning_rate": 8.595060212988006e-06,
      "loss": 0.2097,
      "step": 612
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2939742537377787,
      "learning_rate": 8.590139048691361e-06,
      "loss": 0.1584,
      "step": 613
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.352224161937632,
      "learning_rate": 8.585210694790333e-06,
      "loss": 0.2326,
      "step": 614
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3725174581917323,
      "learning_rate": 8.580275161154432e-06,
      "loss": 0.278,
      "step": 615
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2108941051469937,
      "learning_rate": 8.575332457667546e-06,
      "loss": 0.2102,
      "step": 616
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.710239797140787,
      "learning_rate": 8.570382594227923e-06,
      "loss": 0.227,
      "step": 617
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2724486568098108,
      "learning_rate": 8.56542558074815e-06,
      "loss": 0.1676,
      "step": 618
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3901329270987368,
      "learning_rate": 8.56046142715513e-06,
      "loss": 0.2673,
      "step": 619
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3664144302389138,
      "learning_rate": 8.555490143390061e-06,
      "loss": 0.1888,
      "step": 620
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4395567666044238,
      "learning_rate": 8.550511739408428e-06,
      "loss": 0.2454,
      "step": 621
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.489471683019921,
      "learning_rate": 8.545526225179976e-06,
      "loss": 0.263,
      "step": 622
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4438851859216042,
      "learning_rate": 8.540533610688679e-06,
      "loss": 0.2135,
      "step": 623
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2461360781352666,
      "learning_rate": 8.535533905932739e-06,
      "loss": 0.2189,
      "step": 624
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.582640736563393,
      "learning_rate": 8.53052712092455e-06,
      "loss": 0.2313,
      "step": 625
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.298247273950771,
      "learning_rate": 8.52551326569069e-06,
      "loss": 0.1872,
      "step": 626
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4158888733387451,
      "learning_rate": 8.520492350271895e-06,
      "loss": 0.2217,
      "step": 627
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3842064366426572,
      "learning_rate": 8.51546438472304e-06,
      "loss": 0.2403,
      "step": 628
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4273720586369432,
      "learning_rate": 8.510429379113114e-06,
      "loss": 0.1833,
      "step": 629
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3852112257177571,
      "learning_rate": 8.50538734352521e-06,
      "loss": 0.1719,
      "step": 630
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3359848110502495,
      "learning_rate": 8.500338288056496e-06,
      "loss": 0.252,
      "step": 631
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3818308080887078,
      "learning_rate": 8.495282222818202e-06,
      "loss": 0.2074,
      "step": 632
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6633757429507163,
      "learning_rate": 8.490219157935589e-06,
      "loss": 0.2903,
      "step": 633
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2836738986211325,
      "learning_rate": 8.485149103547943e-06,
      "loss": 0.2176,
      "step": 634
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3324413629512293,
      "learning_rate": 8.480072069808541e-06,
      "loss": 0.2579,
      "step": 635
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4652855234094093,
      "learning_rate": 8.47498806688464e-06,
      "loss": 0.2113,
      "step": 636
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2482060322878645,
      "learning_rate": 8.469897104957454e-06,
      "loss": 0.1851,
      "step": 637
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.401871707640103,
      "learning_rate": 8.464799194222129e-06,
      "loss": 0.2685,
      "step": 638
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.36749902946082,
      "learning_rate": 8.459694344887732e-06,
      "loss": 0.2653,
      "step": 639
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3346571229442254,
      "learning_rate": 8.454582567177223e-06,
      "loss": 0.236,
      "step": 640
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6088332277950657,
      "learning_rate": 8.449463871327434e-06,
      "loss": 0.2472,
      "step": 641
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5310067221319397,
      "learning_rate": 8.444338267589057e-06,
      "loss": 0.2184,
      "step": 642
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2386101724724117,
      "learning_rate": 8.439205766226613e-06,
      "loss": 0.1936,
      "step": 643
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3236667030023652,
      "learning_rate": 8.434066377518437e-06,
      "loss": 0.2103,
      "step": 644
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5371329803597205,
      "learning_rate": 8.428920111756658e-06,
      "loss": 0.1492,
      "step": 645
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3670953932584726,
      "learning_rate": 8.423766979247176e-06,
      "loss": 0.1813,
      "step": 646
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4243872316589663,
      "learning_rate": 8.418606990309645e-06,
      "loss": 0.2058,
      "step": 647
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5221492745472092,
      "learning_rate": 8.413440155277444e-06,
      "loss": 0.2407,
      "step": 648
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3247634678692481,
      "learning_rate": 8.408266484497664e-06,
      "loss": 0.2087,
      "step": 649
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.444992028080282,
      "learning_rate": 8.403085988331092e-06,
      "loss": 0.2684,
      "step": 650
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.163278152418018,
      "learning_rate": 8.397898677152173e-06,
      "loss": 0.1562,
      "step": 651
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5709072384550469,
      "learning_rate": 8.392704561349006e-06,
      "loss": 0.2739,
      "step": 652
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3346099595918441,
      "learning_rate": 8.387503651323317e-06,
      "loss": 0.2305,
      "step": 653
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6191594941519132,
      "learning_rate": 8.382295957490435e-06,
      "loss": 0.2284,
      "step": 654
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3713856796283928,
      "learning_rate": 8.377081490279278e-06,
      "loss": 0.2499,
      "step": 655
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2741902840621007,
      "learning_rate": 8.371860260132323e-06,
      "loss": 0.2098,
      "step": 656
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1607011917765278,
      "learning_rate": 8.366632277505598e-06,
      "loss": 0.2042,
      "step": 657
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.496801887731441,
      "learning_rate": 8.361397552868645e-06,
      "loss": 0.2033,
      "step": 658
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2150204690389046,
      "learning_rate": 8.356156096704516e-06,
      "loss": 0.2115,
      "step": 659
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.483861856763154,
      "learning_rate": 8.350907919509734e-06,
      "loss": 0.2194,
      "step": 660
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6036628608049506,
      "learning_rate": 8.345653031794292e-06,
      "loss": 0.2253,
      "step": 661
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.336983940519561,
      "learning_rate": 8.340391444081615e-06,
      "loss": 0.2283,
      "step": 662
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1996540699125506,
      "learning_rate": 8.335123166908544e-06,
      "loss": 0.21,
      "step": 663
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5850166399002181,
      "learning_rate": 8.329848210825322e-06,
      "loss": 0.3125,
      "step": 664
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3811410730624993,
      "learning_rate": 8.324566586395563e-06,
      "loss": 0.2989,
      "step": 665
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1920556570388823,
      "learning_rate": 8.319278304196237e-06,
      "loss": 0.1755,
      "step": 666
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3716304596667481,
      "learning_rate": 8.313983374817643e-06,
      "loss": 0.2121,
      "step": 667
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4043977851110934,
      "learning_rate": 8.3086818088634e-06,
      "loss": 0.2306,
      "step": 668
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3772088189879543,
      "learning_rate": 8.303373616950408e-06,
      "loss": 0.219,
      "step": 669
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3786756560321909,
      "learning_rate": 8.298058809708841e-06,
      "loss": 0.1799,
      "step": 670
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3667646529667434,
      "learning_rate": 8.29273739778212e-06,
      "loss": 0.2006,
      "step": 671
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.393770100131567,
      "learning_rate": 8.287409391826895e-06,
      "loss": 0.2361,
      "step": 672
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4170721951560306,
      "learning_rate": 8.282074802513016e-06,
      "loss": 0.1748,
      "step": 673
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.611868316813894,
      "learning_rate": 8.27673364052352e-06,
      "loss": 0.1615,
      "step": 674
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3698482019301315,
      "learning_rate": 8.271385916554605e-06,
      "loss": 0.2537,
      "step": 675
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3867740756889253,
      "learning_rate": 8.266031641315612e-06,
      "loss": 0.1849,
      "step": 676
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.41540694061278,
      "learning_rate": 8.260670825529002e-06,
      "loss": 0.2836,
      "step": 677
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.556600199446318,
      "learning_rate": 8.255303479930332e-06,
      "loss": 0.1671,
      "step": 678
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5051898421646384,
      "learning_rate": 8.249929615268234e-06,
      "loss": 0.241,
      "step": 679
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4005233848917849,
      "learning_rate": 8.2445492423044e-06,
      "loss": 0.1934,
      "step": 680
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2697584178996018,
      "learning_rate": 8.239162371813552e-06,
      "loss": 0.1838,
      "step": 681
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3992528734100595,
      "learning_rate": 8.233769014583423e-06,
      "loss": 0.2508,
      "step": 682
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2562904229089487,
      "learning_rate": 8.22836918141474e-06,
      "loss": 0.2063,
      "step": 683
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.212029012609275,
      "learning_rate": 8.222962883121196e-06,
      "loss": 0.1928,
      "step": 684
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5282219998105193,
      "learning_rate": 8.217550130529433e-06,
      "loss": 0.2301,
      "step": 685
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5348906076130424,
      "learning_rate": 8.212130934479015e-06,
      "loss": 0.2146,
      "step": 686
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3960440387655684,
      "learning_rate": 8.206705305822414e-06,
      "loss": 0.2218,
      "step": 687
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3547547253144268,
      "learning_rate": 8.201273255424979e-06,
      "loss": 0.223,
      "step": 688
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.169654835491014,
      "learning_rate": 8.195834794164925e-06,
      "loss": 0.1896,
      "step": 689
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3345360022685602,
      "learning_rate": 8.190389932933301e-06,
      "loss": 0.2335,
      "step": 690
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4611679265619821,
      "learning_rate": 8.184938682633973e-06,
      "loss": 0.229,
      "step": 691
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.316272346486635,
      "learning_rate": 8.179481054183605e-06,
      "loss": 0.1869,
      "step": 692
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4150052275808762,
      "learning_rate": 8.17401705851163e-06,
      "loss": 0.2531,
      "step": 693
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3683161070722536,
      "learning_rate": 8.168546706560231e-06,
      "loss": 0.1994,
      "step": 694
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3557998765153303,
      "learning_rate": 8.163070009284326e-06,
      "loss": 0.2055,
      "step": 695
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5673366815832006,
      "learning_rate": 8.157586977651534e-06,
      "loss": 0.2565,
      "step": 696
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4477938225405609,
      "learning_rate": 8.152097622642167e-06,
      "loss": 0.2115,
      "step": 697
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.416901510789192,
      "learning_rate": 8.146601955249187e-06,
      "loss": 0.2273,
      "step": 698
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2676507164459283,
      "learning_rate": 8.141099986478212e-06,
      "loss": 0.1706,
      "step": 699
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3817970342989552,
      "learning_rate": 8.135591727347469e-06,
      "loss": 0.2049,
      "step": 700
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5470218974230898,
      "learning_rate": 8.130077188887786e-06,
      "loss": 0.2572,
      "step": 701
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4559211326418209,
      "learning_rate": 8.124556382142566e-06,
      "loss": 0.2067,
      "step": 702
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2157045771320043,
      "learning_rate": 8.119029318167763e-06,
      "loss": 0.187,
      "step": 703
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2329996324468269,
      "learning_rate": 8.113496008031863e-06,
      "loss": 0.2228,
      "step": 704
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5178348160910475,
      "learning_rate": 8.107956462815862e-06,
      "loss": 0.2009,
      "step": 705
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5173873513651819,
      "learning_rate": 8.102410693613239e-06,
      "loss": 0.2014,
      "step": 706
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.489907120379038,
      "learning_rate": 8.096858711529938e-06,
      "loss": 0.2719,
      "step": 707
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4567138581782364,
      "learning_rate": 8.09130052768435e-06,
      "loss": 0.2128,
      "step": 708
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2980709742861036,
      "learning_rate": 8.085736153207277e-06,
      "loss": 0.1904,
      "step": 709
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3571883763520458,
      "learning_rate": 8.080165599241924e-06,
      "loss": 0.2407,
      "step": 710
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1911361545501362,
      "learning_rate": 8.074588876943872e-06,
      "loss": 0.2072,
      "step": 711
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1838458317453298,
      "learning_rate": 8.069005997481053e-06,
      "loss": 0.1658,
      "step": 712
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5582428751104815,
      "learning_rate": 8.063416972033725e-06,
      "loss": 0.2369,
      "step": 713
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.448625711875147,
      "learning_rate": 8.057821811794457e-06,
      "loss": 0.157,
      "step": 714
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2716160004897243,
      "learning_rate": 8.05222052796811e-06,
      "loss": 0.2111,
      "step": 715
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3995510125962969,
      "learning_rate": 8.046613131771798e-06,
      "loss": 0.2498,
      "step": 716
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3641109780833356,
      "learning_rate": 8.040999634434883e-06,
      "loss": 0.2199,
      "step": 717
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3794839143240507,
      "learning_rate": 8.03538004719894e-06,
      "loss": 0.1908,
      "step": 718
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3057267158424555,
      "learning_rate": 8.029754381317741e-06,
      "loss": 0.1842,
      "step": 719
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3923453087047304,
      "learning_rate": 8.024122648057234e-06,
      "loss": 0.1957,
      "step": 720
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5532217563214035,
      "learning_rate": 8.018484858695514e-06,
      "loss": 0.2741,
      "step": 721
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.325026429692661,
      "learning_rate": 8.012841024522808e-06,
      "loss": 0.1495,
      "step": 722
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.247342343578845,
      "learning_rate": 8.00719115684144e-06,
      "loss": 0.1731,
      "step": 723
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.387492032748767,
      "learning_rate": 8.001535266965829e-06,
      "loss": 0.216,
      "step": 724
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.258047240172986,
      "learning_rate": 7.995873366222442e-06,
      "loss": 0.2106,
      "step": 725
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2376523647373756,
      "learning_rate": 7.990205465949792e-06,
      "loss": 0.2135,
      "step": 726
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.320741951217798,
      "learning_rate": 7.9845315774984e-06,
      "loss": 0.2374,
      "step": 727
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4560663794587922,
      "learning_rate": 7.978851712230786e-06,
      "loss": 0.2299,
      "step": 728
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4521643552431327,
      "learning_rate": 7.973165881521435e-06,
      "loss": 0.2355,
      "step": 729
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4006171230804598,
      "learning_rate": 7.967474096756775e-06,
      "loss": 0.2197,
      "step": 730
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.507141723188566,
      "learning_rate": 7.961776369335164e-06,
      "loss": 0.2896,
      "step": 731
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4914995816615797,
      "learning_rate": 7.95607271066686e-06,
      "loss": 0.2602,
      "step": 732
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3329366216089469,
      "learning_rate": 7.950363132173992e-06,
      "loss": 0.1887,
      "step": 733
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4285367600401817,
      "learning_rate": 7.944647645290555e-06,
      "loss": 0.2435,
      "step": 734
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4639622759501603,
      "learning_rate": 7.938926261462366e-06,
      "loss": 0.2354,
      "step": 735
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.406914507111704,
      "learning_rate": 7.933198992147057e-06,
      "loss": 0.2283,
      "step": 736
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.352524782194941,
      "learning_rate": 7.927465848814045e-06,
      "loss": 0.1762,
      "step": 737
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2926057440903231,
      "learning_rate": 7.921726842944507e-06,
      "loss": 0.1615,
      "step": 738
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2708612867474203,
      "learning_rate": 7.915981986031367e-06,
      "loss": 0.204,
      "step": 739
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2212537082585693,
      "learning_rate": 7.910231289579263e-06,
      "loss": 0.258,
      "step": 740
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2771436530706395,
      "learning_rate": 7.90447476510452e-06,
      "loss": 0.196,
      "step": 741
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2355504096336085,
      "learning_rate": 7.898712424135149e-06,
      "loss": 0.1992,
      "step": 742
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.177458432212328,
      "learning_rate": 7.892944278210796e-06,
      "loss": 0.217,
      "step": 743
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3862630829060882,
      "learning_rate": 7.887170338882742e-06,
      "loss": 0.1791,
      "step": 744
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3319303555612236,
      "learning_rate": 7.88139061771386e-06,
      "loss": 0.1962,
      "step": 745
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4507491692660812,
      "learning_rate": 7.875605126278608e-06,
      "loss": 0.1872,
      "step": 746
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2430331871698628,
      "learning_rate": 7.869813876162997e-06,
      "loss": 0.1866,
      "step": 747
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2995531508364535,
      "learning_rate": 7.864016878964575e-06,
      "loss": 0.1899,
      "step": 748
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3078329799604256,
      "learning_rate": 7.858214146292394e-06,
      "loss": 0.1757,
      "step": 749
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3047445776266866,
      "learning_rate": 7.852405689766993e-06,
      "loss": 0.2181,
      "step": 750
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3308517720943847,
      "learning_rate": 7.846591521020373e-06,
      "loss": 0.2048,
      "step": 751
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4039366985601598,
      "learning_rate": 7.840771651695976e-06,
      "loss": 0.215,
      "step": 752
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3657435711286086,
      "learning_rate": 7.834946093448658e-06,
      "loss": 0.1756,
      "step": 753
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4611894783227624,
      "learning_rate": 7.829114857944672e-06,
      "loss": 0.2541,
      "step": 754
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4139932401804687,
      "learning_rate": 7.823277956861635e-06,
      "loss": 0.259,
      "step": 755
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.532238194112718,
      "learning_rate": 7.817435401888514e-06,
      "loss": 0.24,
      "step": 756
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3891906740814532,
      "learning_rate": 7.811587204725593e-06,
      "loss": 0.2748,
      "step": 757
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.471111353057393,
      "learning_rate": 7.80573337708446e-06,
      "loss": 0.2106,
      "step": 758
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5228818865512705,
      "learning_rate": 7.799873930687979e-06,
      "loss": 0.2073,
      "step": 759
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3745767459016986,
      "learning_rate": 7.79400887727026e-06,
      "loss": 0.1692,
      "step": 760
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5322550876566754,
      "learning_rate": 7.78813822857665e-06,
      "loss": 0.264,
      "step": 761
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.418393327162263,
      "learning_rate": 7.782261996363694e-06,
      "loss": 0.1741,
      "step": 762
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2811034536146098,
      "learning_rate": 7.776380192399122e-06,
      "loss": 0.1725,
      "step": 763
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3347838722533258,
      "learning_rate": 7.770492828461824e-06,
      "loss": 0.2237,
      "step": 764
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.421321909294947,
      "learning_rate": 7.764599916341817e-06,
      "loss": 0.2574,
      "step": 765
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3218077152909116,
      "learning_rate": 7.75870146784024e-06,
      "loss": 0.1858,
      "step": 766
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3988594272407506,
      "learning_rate": 7.752797494769307e-06,
      "loss": 0.1903,
      "step": 767
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2782222163985342,
      "learning_rate": 7.7468880089523e-06,
      "loss": 0.2312,
      "step": 768
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1725275871240384,
      "learning_rate": 7.74097302222355e-06,
      "loss": 0.2292,
      "step": 769
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2758452553661437,
      "learning_rate": 7.73505254642839e-06,
      "loss": 0.163,
      "step": 770
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5081159087823988,
      "learning_rate": 7.729126593423151e-06,
      "loss": 0.2171,
      "step": 771
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.441948885735313,
      "learning_rate": 7.723195175075136e-06,
      "loss": 0.2075,
      "step": 772
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.249236549103611,
      "learning_rate": 7.71725830326259e-06,
      "loss": 0.1723,
      "step": 773
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.252850385307966,
      "learning_rate": 7.711315989874677e-06,
      "loss": 0.1943,
      "step": 774
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.309042785336002,
      "learning_rate": 7.70536824681146e-06,
      "loss": 0.2143,
      "step": 775
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5430730413688167,
      "learning_rate": 7.69941508598388e-06,
      "loss": 0.2463,
      "step": 776
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1197288566876964,
      "learning_rate": 7.69345651931372e-06,
      "loss": 0.1579,
      "step": 777
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3136830619007525,
      "learning_rate": 7.687492558733593e-06,
      "loss": 0.2241,
      "step": 778
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4278482318409749,
      "learning_rate": 7.681523216186912e-06,
      "loss": 0.1895,
      "step": 779
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4393853219636548,
      "learning_rate": 7.675548503627872e-06,
      "loss": 0.2004,
      "step": 780
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5035154525934205,
      "learning_rate": 7.669568433021418e-06,
      "loss": 0.267,
      "step": 781
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3653881467665585,
      "learning_rate": 7.663583016343224e-06,
      "loss": 0.1982,
      "step": 782
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3885275635900107,
      "learning_rate": 7.65759226557967e-06,
      "loss": 0.2355,
      "step": 783
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3865687695686477,
      "learning_rate": 7.651596192727826e-06,
      "loss": 0.1796,
      "step": 784
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4259253024713197,
      "learning_rate": 7.64559480979541e-06,
      "loss": 0.2088,
      "step": 785
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.309929603530691,
      "learning_rate": 7.63958812880078e-06,
      "loss": 0.2463,
      "step": 786
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0959827464521774,
      "learning_rate": 7.633576161772898e-06,
      "loss": 0.1867,
      "step": 787
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2576566740110757,
      "learning_rate": 7.627558920751321e-06,
      "loss": 0.1826,
      "step": 788
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.521474732031726,
      "learning_rate": 7.621536417786159e-06,
      "loss": 0.2286,
      "step": 789
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2854727615656365,
      "learning_rate": 7.615508664938065e-06,
      "loss": 0.1429,
      "step": 790
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.557769004812575,
      "learning_rate": 7.6094756742782e-06,
      "loss": 0.2188,
      "step": 791
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5516005715057117,
      "learning_rate": 7.6034374578882205e-06,
      "loss": 0.2016,
      "step": 792
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.398260339328202,
      "learning_rate": 7.597394027860243e-06,
      "loss": 0.2657,
      "step": 793
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.373157473032203,
      "learning_rate": 7.5913453962968296e-06,
      "loss": 0.1878,
      "step": 794
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3989741899826071,
      "learning_rate": 7.585291575310952e-06,
      "loss": 0.2196,
      "step": 795
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3717688158519465,
      "learning_rate": 7.579232577025984e-06,
      "loss": 0.2276,
      "step": 796
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.344270537734827,
      "learning_rate": 7.5731684135756566e-06,
      "loss": 0.1816,
      "step": 797
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4271568218543567,
      "learning_rate": 7.567099097104053e-06,
      "loss": 0.2307,
      "step": 798
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.658342824135686,
      "learning_rate": 7.5610246397655715e-06,
      "loss": 0.2753,
      "step": 799
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.447051197299827,
      "learning_rate": 7.5549450537249096e-06,
      "loss": 0.2137,
      "step": 800
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.447163791327868,
      "learning_rate": 7.548860351157028e-06,
      "loss": 0.236,
      "step": 801
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4198141270519447,
      "learning_rate": 7.542770544247141e-06,
      "loss": 0.2482,
      "step": 802
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5461940626544008,
      "learning_rate": 7.536675645190682e-06,
      "loss": 0.2799,
      "step": 803
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.503577634204635,
      "learning_rate": 7.530575666193283e-06,
      "loss": 0.262,
      "step": 804
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4417802939519906,
      "learning_rate": 7.524470619470745e-06,
      "loss": 0.2278,
      "step": 805
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4114081264608496,
      "learning_rate": 7.518360517249024e-06,
      "loss": 0.2353,
      "step": 806
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4983461922563195,
      "learning_rate": 7.512245371764197e-06,
      "loss": 0.2128,
      "step": 807
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4423302094584296,
      "learning_rate": 7.506125195262437e-06,
      "loss": 0.26,
      "step": 808
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3551920866250788,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.2492,
      "step": 809
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4264059353805185,
      "learning_rate": 7.493869798243188e-06,
      "loss": 0.2404,
      "step": 810
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3970535827817137,
      "learning_rate": 7.487734602268326e-06,
      "loss": 0.2224,
      "step": 811
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4125040020338544,
      "learning_rate": 7.4815944243617455e-06,
      "loss": 0.2441,
      "step": 812
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3950412226646816,
      "learning_rate": 7.475449276819753e-06,
      "loss": 0.2328,
      "step": 813
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7065064283598488,
      "learning_rate": 7.469299171948608e-06,
      "loss": 0.2495,
      "step": 814
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3313469346873519,
      "learning_rate": 7.463144122064494e-06,
      "loss": 0.2522,
      "step": 815
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.303781743672682,
      "learning_rate": 7.456984139493502e-06,
      "loss": 0.2507,
      "step": 816
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.304559319805089,
      "learning_rate": 7.4508192365716005e-06,
      "loss": 0.2438,
      "step": 817
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.403317074750053,
      "learning_rate": 7.444649425644606e-06,
      "loss": 0.207,
      "step": 818
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3638632246301536,
      "learning_rate": 7.438474719068174e-06,
      "loss": 0.1735,
      "step": 819
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5414993725809585,
      "learning_rate": 7.432295129207753e-06,
      "loss": 0.2356,
      "step": 820
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3825428755002764,
      "learning_rate": 7.426110668438578e-06,
      "loss": 0.1627,
      "step": 821
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2701589783276548,
      "learning_rate": 7.419921349145635e-06,
      "loss": 0.1779,
      "step": 822
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3653096727709395,
      "learning_rate": 7.413727183723642e-06,
      "loss": 0.2547,
      "step": 823
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3650444445233314,
      "learning_rate": 7.4075281845770196e-06,
      "loss": 0.2412,
      "step": 824
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3200273792937927,
      "learning_rate": 7.401324364119872e-06,
      "loss": 0.1995,
      "step": 825
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2761731257523867,
      "learning_rate": 7.395115734775953e-06,
      "loss": 0.1597,
      "step": 826
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4847234243419185,
      "learning_rate": 7.388902308978652e-06,
      "loss": 0.2532,
      "step": 827
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.293889422035715,
      "learning_rate": 7.382684099170959e-06,
      "loss": 0.2221,
      "step": 828
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.606609567803644,
      "learning_rate": 7.37646111780545e-06,
      "loss": 0.236,
      "step": 829
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4413538959088792,
      "learning_rate": 7.370233377344252e-06,
      "loss": 0.2054,
      "step": 830
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3677715993811395,
      "learning_rate": 7.364000890259024e-06,
      "loss": 0.1812,
      "step": 831
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.38971266906721,
      "learning_rate": 7.357763669030931e-06,
      "loss": 0.2069,
      "step": 832
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3536684037952609,
      "learning_rate": 7.351521726150619e-06,
      "loss": 0.2176,
      "step": 833
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2514603042723516,
      "learning_rate": 7.3452750741181855e-06,
      "loss": 0.2367,
      "step": 834
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3149773684826296,
      "learning_rate": 7.339023725443166e-06,
      "loss": 0.2433,
      "step": 835
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1830894667824519,
      "learning_rate": 7.332767692644493e-06,
      "loss": 0.1812,
      "step": 836
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4253456705564744,
      "learning_rate": 7.326506988250488e-06,
      "loss": 0.2261,
      "step": 837
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3010417543613921,
      "learning_rate": 7.320241624798821e-06,
      "loss": 0.2456,
      "step": 838
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.318411084904983,
      "learning_rate": 7.313971614836496e-06,
      "loss": 0.2364,
      "step": 839
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3309508227436395,
      "learning_rate": 7.307696970919819e-06,
      "loss": 0.2162,
      "step": 840
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.451262328108976,
      "learning_rate": 7.3014177056143795e-06,
      "loss": 0.2286,
      "step": 841
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4132363322491819,
      "learning_rate": 7.295133831495021e-06,
      "loss": 0.2224,
      "step": 842
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3422202374296976,
      "learning_rate": 7.288845361145812e-06,
      "loss": 0.2283,
      "step": 843
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.400198869241597,
      "learning_rate": 7.282552307160033e-06,
      "loss": 0.2729,
      "step": 844
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5177700276992794,
      "learning_rate": 7.276254682140137e-06,
      "loss": 0.2406,
      "step": 845
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4587490719723648,
      "learning_rate": 7.269952498697734e-06,
      "loss": 0.238,
      "step": 846
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3695748299743316,
      "learning_rate": 7.2636457694535655e-06,
      "loss": 0.2075,
      "step": 847
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3263189567387381,
      "learning_rate": 7.2573345070374675e-06,
      "loss": 0.1537,
      "step": 848
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2617125005835852,
      "learning_rate": 7.251018724088367e-06,
      "loss": 0.1971,
      "step": 849
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2405228114784828,
      "learning_rate": 7.244698433254231e-06,
      "loss": 0.1982,
      "step": 850
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4030999115735678,
      "learning_rate": 7.238373647192063e-06,
      "loss": 0.1807,
      "step": 851
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3426393611502285,
      "learning_rate": 7.232044378567864e-06,
      "loss": 0.2658,
      "step": 852
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4108374443492102,
      "learning_rate": 7.225710640056614e-06,
      "loss": 0.2293,
      "step": 853
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3507491087112569,
      "learning_rate": 7.2193724443422405e-06,
      "loss": 0.282,
      "step": 854
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2370695678775512,
      "learning_rate": 7.213029804117604e-06,
      "loss": 0.1902,
      "step": 855
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2951988864691606,
      "learning_rate": 7.206682732084458e-06,
      "loss": 0.2693,
      "step": 856
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4169400565005463,
      "learning_rate": 7.200331240953439e-06,
      "loss": 0.2041,
      "step": 857
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2952183276570757,
      "learning_rate": 7.193975343444024e-06,
      "loss": 0.1851,
      "step": 858
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2376712966110353,
      "learning_rate": 7.187615052284522e-06,
      "loss": 0.1876,
      "step": 859
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.187089855906794,
      "learning_rate": 7.181250380212037e-06,
      "loss": 0.189,
      "step": 860
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3838847004811112,
      "learning_rate": 7.174881339972448e-06,
      "loss": 0.1931,
      "step": 861
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6060691837519847,
      "learning_rate": 7.1685079443203784e-06,
      "loss": 0.2836,
      "step": 862
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.415682549873026,
      "learning_rate": 7.16213020601918e-06,
      "loss": 0.2421,
      "step": 863
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.304366469414161,
      "learning_rate": 7.155748137840892e-06,
      "loss": 0.1967,
      "step": 864
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1749446935654673,
      "learning_rate": 7.149361752566233e-06,
      "loss": 0.185,
      "step": 865
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.559504729824471,
      "learning_rate": 7.142971062984563e-06,
      "loss": 0.2331,
      "step": 866
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3282818876695377,
      "learning_rate": 7.136576081893863e-06,
      "loss": 0.201,
      "step": 867
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3835335756864908,
      "learning_rate": 7.130176822100707e-06,
      "loss": 0.1718,
      "step": 868
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4664550193556876,
      "learning_rate": 7.12377329642024e-06,
      "loss": 0.2317,
      "step": 869
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4838357237284525,
      "learning_rate": 7.117365517676146e-06,
      "loss": 0.2349,
      "step": 870
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3384048489822704,
      "learning_rate": 7.110953498700629e-06,
      "loss": 0.2315,
      "step": 871
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4244395913147376,
      "learning_rate": 7.104537252334383e-06,
      "loss": 0.1937,
      "step": 872
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.468754252364756,
      "learning_rate": 7.09811679142657e-06,
      "loss": 0.2167,
      "step": 873
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3254855168395066,
      "learning_rate": 7.09169212883479e-06,
      "loss": 0.1969,
      "step": 874
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2720992680574736,
      "learning_rate": 7.085263277425058e-06,
      "loss": 0.1564,
      "step": 875
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2182161750912075,
      "learning_rate": 7.078830250071778e-06,
      "loss": 0.207,
      "step": 876
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4203255602032099,
      "learning_rate": 7.072393059657716e-06,
      "loss": 0.1717,
      "step": 877
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.439337564245391,
      "learning_rate": 7.065951719073973e-06,
      "loss": 0.2909,
      "step": 878
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.327919062913578,
      "learning_rate": 7.059506241219964e-06,
      "loss": 0.1747,
      "step": 879
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4153244084240877,
      "learning_rate": 7.053056639003389e-06,
      "loss": 0.1802,
      "step": 880
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3151194336970664,
      "learning_rate": 7.046602925340209e-06,
      "loss": 0.1805,
      "step": 881
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4498788815473398,
      "learning_rate": 7.0401451131546125e-06,
      "loss": 0.2343,
      "step": 882
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.542826197998622,
      "learning_rate": 7.033683215379002e-06,
      "loss": 0.2282,
      "step": 883
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2079274929436847,
      "learning_rate": 7.027217244953958e-06,
      "loss": 0.1367,
      "step": 884
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6854301339218132,
      "learning_rate": 7.020747214828221e-06,
      "loss": 0.1844,
      "step": 885
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5687779779670346,
      "learning_rate": 7.014273137958656e-06,
      "loss": 0.1988,
      "step": 886
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4023740756602707,
      "learning_rate": 7.007795027310236e-06,
      "loss": 0.1887,
      "step": 887
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3353420395036886,
      "learning_rate": 7.001312895856011e-06,
      "loss": 0.1875,
      "step": 888
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4599554362108553,
      "learning_rate": 6.994826756577082e-06,
      "loss": 0.2102,
      "step": 889
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4139054110892626,
      "learning_rate": 6.988336622462578e-06,
      "loss": 0.2461,
      "step": 890
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3760342248520492,
      "learning_rate": 6.981842506509626e-06,
      "loss": 0.202,
      "step": 891
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3984718782636694,
      "learning_rate": 6.975344421723329e-06,
      "loss": 0.1727,
      "step": 892
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3975840783288527,
      "learning_rate": 6.968842381116736e-06,
      "loss": 0.2296,
      "step": 893
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3246657337948222,
      "learning_rate": 6.962336397710819e-06,
      "loss": 0.2329,
      "step": 894
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3092293519976206,
      "learning_rate": 6.955826484534447e-06,
      "loss": 0.1417,
      "step": 895
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2361433900127223,
      "learning_rate": 6.949312654624355e-06,
      "loss": 0.1948,
      "step": 896
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4962923784603999,
      "learning_rate": 6.942794921025127e-06,
      "loss": 0.243,
      "step": 897
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2527141187856905,
      "learning_rate": 6.93627329678916e-06,
      "loss": 0.1957,
      "step": 898
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3747947945500099,
      "learning_rate": 6.9297477949766445e-06,
      "loss": 0.1912,
      "step": 899
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.35062912213523,
      "learning_rate": 6.9232184286555345e-06,
      "loss": 0.1742,
      "step": 900
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4251094800835216,
      "learning_rate": 6.916685210901527e-06,
      "loss": 0.2469,
      "step": 901
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4812818750364831,
      "learning_rate": 6.910148154798026e-06,
      "loss": 0.2482,
      "step": 902
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3155905493682785,
      "learning_rate": 6.903607273436128e-06,
      "loss": 0.2724,
      "step": 903
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.228899441427879,
      "learning_rate": 6.897062579914587e-06,
      "loss": 0.2043,
      "step": 904
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3096848656288718,
      "learning_rate": 6.890514087339789e-06,
      "loss": 0.1715,
      "step": 905
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.474291946270809,
      "learning_rate": 6.883961808825732e-06,
      "loss": 0.2578,
      "step": 906
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2825359395278226,
      "learning_rate": 6.8774057574939956e-06,
      "loss": 0.2083,
      "step": 907
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6743252401929913,
      "learning_rate": 6.870845946473711e-06,
      "loss": 0.1345,
      "step": 908
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5952203276730124,
      "learning_rate": 6.864282388901544e-06,
      "loss": 0.2222,
      "step": 909
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6198873145242847,
      "learning_rate": 6.857715097921657e-06,
      "loss": 0.2305,
      "step": 910
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3905343595892241,
      "learning_rate": 6.851144086685695e-06,
      "loss": 0.2457,
      "step": 911
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3113288681248079,
      "learning_rate": 6.844569368352748e-06,
      "loss": 0.2695,
      "step": 912
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2866551554052297,
      "learning_rate": 6.837990956089334e-06,
      "loss": 0.1984,
      "step": 913
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3777475004003925,
      "learning_rate": 6.831408863069364e-06,
      "loss": 0.1834,
      "step": 914
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.422825145390839,
      "learning_rate": 6.824823102474127e-06,
      "loss": 0.1941,
      "step": 915
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2290268838104004,
      "learning_rate": 6.818233687492248e-06,
      "loss": 0.1879,
      "step": 916
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2626323229209946,
      "learning_rate": 6.811640631319677e-06,
      "loss": 0.1842,
      "step": 917
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.42619360243098,
      "learning_rate": 6.805043947159652e-06,
      "loss": 0.2539,
      "step": 918
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4012430312969997,
      "learning_rate": 6.79844364822268e-06,
      "loss": 0.2408,
      "step": 919
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2694340807609577,
      "learning_rate": 6.7918397477265e-06,
      "loss": 0.2358,
      "step": 920
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.535830374315427,
      "learning_rate": 6.785232258896078e-06,
      "loss": 0.2265,
      "step": 921
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2295362629845854,
      "learning_rate": 6.778621194963545e-06,
      "loss": 0.1799,
      "step": 922
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2048118537897166,
      "learning_rate": 6.77200656916821e-06,
      "loss": 0.1928,
      "step": 923
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4331278729911339,
      "learning_rate": 6.765388394756504e-06,
      "loss": 0.2323,
      "step": 924
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3937627511183834,
      "learning_rate": 6.758766684981972e-06,
      "loss": 0.2135,
      "step": 925
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3291120381538082,
      "learning_rate": 6.752141453105231e-06,
      "loss": 0.1772,
      "step": 926
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5750154343062712,
      "learning_rate": 6.745512712393958e-06,
      "loss": 0.241,
      "step": 927
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4706333140490864,
      "learning_rate": 6.7388804761228535e-06,
      "loss": 0.1918,
      "step": 928
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3090960560806397,
      "learning_rate": 6.732244757573619e-06,
      "loss": 0.168,
      "step": 929
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.202957890793412,
      "learning_rate": 6.725605570034928e-06,
      "loss": 0.1734,
      "step": 930
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.323812969947997,
      "learning_rate": 6.718962926802407e-06,
      "loss": 0.2116,
      "step": 931
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3959507320968478,
      "learning_rate": 6.712316841178592e-06,
      "loss": 0.2556,
      "step": 932
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4179085935422193,
      "learning_rate": 6.705667326472926e-06,
      "loss": 0.2085,
      "step": 933
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2895927006571708,
      "learning_rate": 6.699014396001707e-06,
      "loss": 0.1957,
      "step": 934
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.374807291221726,
      "learning_rate": 6.69235806308808e-06,
      "loss": 0.201,
      "step": 935
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.435053929129893,
      "learning_rate": 6.685698341062003e-06,
      "loss": 0.2086,
      "step": 936
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2459406865751526,
      "learning_rate": 6.67903524326022e-06,
      "loss": 0.153,
      "step": 937
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.248666691821859,
      "learning_rate": 6.672368783026235e-06,
      "loss": 0.1953,
      "step": 938
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.441916382180111,
      "learning_rate": 6.665698973710289e-06,
      "loss": 0.2038,
      "step": 939
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1531313447952394,
      "learning_rate": 6.659025828669323e-06,
      "loss": 0.1702,
      "step": 940
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.510460686690441,
      "learning_rate": 6.652349361266967e-06,
      "loss": 0.2744,
      "step": 941
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5458530103937502,
      "learning_rate": 6.645669584873495e-06,
      "loss": 0.2476,
      "step": 942
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4125572107460347,
      "learning_rate": 6.638986512865814e-06,
      "loss": 0.1614,
      "step": 943
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4171431624345558,
      "learning_rate": 6.632300158627427e-06,
      "loss": 0.1691,
      "step": 944
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4245608645502448,
      "learning_rate": 6.625610535548418e-06,
      "loss": 0.2231,
      "step": 945
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6483054628190996,
      "learning_rate": 6.6189176570254045e-06,
      "loss": 0.1902,
      "step": 946
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3351587204047244,
      "learning_rate": 6.612221536461534e-06,
      "loss": 0.1927,
      "step": 947
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1271837581099662,
      "learning_rate": 6.605522187266441e-06,
      "loss": 0.1947,
      "step": 948
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.433055876856815,
      "learning_rate": 6.598819622856227e-06,
      "loss": 0.232,
      "step": 949
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2512037369611528,
      "learning_rate": 6.592113856653433e-06,
      "loss": 0.1677,
      "step": 950
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.400575459498579,
      "learning_rate": 6.585404902087011e-06,
      "loss": 0.2004,
      "step": 951
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1321745849747413,
      "learning_rate": 6.578692772592298e-06,
      "loss": 0.1745,
      "step": 952
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.333423370029414,
      "learning_rate": 6.571977481610989e-06,
      "loss": 0.2163,
      "step": 953
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2160222487447323,
      "learning_rate": 6.565259042591112e-06,
      "loss": 0.1835,
      "step": 954
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2575725646428098,
      "learning_rate": 6.558537468986998e-06,
      "loss": 0.213,
      "step": 955
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1706333797705184,
      "learning_rate": 6.551812774259253e-06,
      "loss": 0.185,
      "step": 956
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.385609133563116,
      "learning_rate": 6.545084971874738e-06,
      "loss": 0.229,
      "step": 957
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3139472298850812,
      "learning_rate": 6.538354075306532e-06,
      "loss": 0.1768,
      "step": 958
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2644383210227716,
      "learning_rate": 6.531620098033919e-06,
      "loss": 0.2132,
      "step": 959
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3203164727684147,
      "learning_rate": 6.524883053542339e-06,
      "loss": 0.2268,
      "step": 960
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4065908144865422,
      "learning_rate": 6.518142955323389e-06,
      "loss": 0.1687,
      "step": 961
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0727785978940498,
      "learning_rate": 6.511399816874769e-06,
      "loss": 0.1651,
      "step": 962
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2218949754261463,
      "learning_rate": 6.504653651700278e-06,
      "loss": 0.1998,
      "step": 963
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2668743511790241,
      "learning_rate": 6.497904473309766e-06,
      "loss": 0.1894,
      "step": 964
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2503446412737045,
      "learning_rate": 6.4911522952191275e-06,
      "loss": 0.176,
      "step": 965
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.406786495292246,
      "learning_rate": 6.4843971309502554e-06,
      "loss": 0.2151,
      "step": 966
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.459268935787368,
      "learning_rate": 6.477638994031028e-06,
      "loss": 0.2561,
      "step": 967
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5400168517031434,
      "learning_rate": 6.470877897995275e-06,
      "loss": 0.2682,
      "step": 968
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1950041346288842,
      "learning_rate": 6.464113856382752e-06,
      "loss": 0.2688,
      "step": 969
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.22185160130978,
      "learning_rate": 6.457346882739114e-06,
      "loss": 0.1781,
      "step": 970
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2973507930785495,
      "learning_rate": 6.450576990615888e-06,
      "loss": 0.2454,
      "step": 971
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.413521599512616,
      "learning_rate": 6.443804193570441e-06,
      "loss": 0.2113,
      "step": 972
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5585101021044911,
      "learning_rate": 6.437028505165965e-06,
      "loss": 0.239,
      "step": 973
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3937399659910088,
      "learning_rate": 6.430249938971438e-06,
      "loss": 0.2155,
      "step": 974
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3166317931974736,
      "learning_rate": 6.423468508561599e-06,
      "loss": 0.2165,
      "step": 975
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3229831087909942,
      "learning_rate": 6.416684227516927e-06,
      "loss": 0.1344,
      "step": 976
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2595894286379854,
      "learning_rate": 6.409897109423607e-06,
      "loss": 0.181,
      "step": 977
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4568577009784,
      "learning_rate": 6.403107167873509e-06,
      "loss": 0.2417,
      "step": 978
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1166534394967986,
      "learning_rate": 6.396314416464151e-06,
      "loss": 0.2184,
      "step": 979
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6035043774268725,
      "learning_rate": 6.389518868798683e-06,
      "loss": 0.2757,
      "step": 980
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3749978128601927,
      "learning_rate": 6.382720538485856e-06,
      "loss": 0.2196,
      "step": 981
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2994404241882802,
      "learning_rate": 6.375919439139988e-06,
      "loss": 0.1856,
      "step": 982
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3503223535010804,
      "learning_rate": 6.3691155843809475e-06,
      "loss": 0.2207,
      "step": 983
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.300529006189319,
      "learning_rate": 6.3623089878341146e-06,
      "loss": 0.1843,
      "step": 984
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.391361181381471,
      "learning_rate": 6.355499663130369e-06,
      "loss": 0.2529,
      "step": 985
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3008448832598953,
      "learning_rate": 6.348687623906047e-06,
      "loss": 0.2004,
      "step": 986
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.252436946034745,
      "learning_rate": 6.341872883802923e-06,
      "loss": 0.1711,
      "step": 987
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.344314815281958,
      "learning_rate": 6.33505545646818e-06,
      "loss": 0.1842,
      "step": 988
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4069362206447198,
      "learning_rate": 6.328235355554382e-06,
      "loss": 0.2325,
      "step": 989
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4575802862513068,
      "learning_rate": 6.32141259471945e-06,
      "loss": 0.2017,
      "step": 990
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2767894832711713,
      "learning_rate": 6.314587187626628e-06,
      "loss": 0.1929,
      "step": 991
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2721159627806684,
      "learning_rate": 6.307759147944461e-06,
      "loss": 0.2346,
      "step": 992
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2814755786845258,
      "learning_rate": 6.3009284893467655e-06,
      "loss": 0.22,
      "step": 993
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3481269448672986,
      "learning_rate": 6.294095225512604e-06,
      "loss": 0.2307,
      "step": 994
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.223049951224015,
      "learning_rate": 6.287259370126255e-06,
      "loss": 0.1877,
      "step": 995
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.378264773015318,
      "learning_rate": 6.280420936877187e-06,
      "loss": 0.2559,
      "step": 996
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2794777426210389,
      "learning_rate": 6.273579939460032e-06,
      "loss": 0.2042,
      "step": 997
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3722174885539362,
      "learning_rate": 6.266736391574553e-06,
      "loss": 0.2541,
      "step": 998
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4297179148983328,
      "learning_rate": 6.259890306925627e-06,
      "loss": 0.23,
      "step": 999
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3626635839162478,
      "learning_rate": 6.253041699223206e-06,
      "loss": 0.2064,
      "step": 1000
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2626605690528634,
      "learning_rate": 6.246190582182296e-06,
      "loss": 0.1797,
      "step": 1001
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2555168700386738,
      "learning_rate": 6.239336969522931e-06,
      "loss": 0.169,
      "step": 1002
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2138287577557905,
      "learning_rate": 6.232480874970137e-06,
      "loss": 0.2315,
      "step": 1003
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2088924674566768,
      "learning_rate": 6.225622312253916e-06,
      "loss": 0.2197,
      "step": 1004
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.390858823232818,
      "learning_rate": 6.218761295109209e-06,
      "loss": 0.2376,
      "step": 1005
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4848947978697165,
      "learning_rate": 6.211897837275873e-06,
      "loss": 0.2693,
      "step": 1006
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3116530789725065,
      "learning_rate": 6.205031952498656e-06,
      "loss": 0.2173,
      "step": 1007
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.407213614017502,
      "learning_rate": 6.198163654527161e-06,
      "loss": 0.2161,
      "step": 1008
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2762069904019602,
      "learning_rate": 6.191292957115825e-06,
      "loss": 0.1934,
      "step": 1009
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.36550579612325,
      "learning_rate": 6.184419874023895e-06,
      "loss": 0.1764,
      "step": 1010
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2423785549466102,
      "learning_rate": 6.177544419015388e-06,
      "loss": 0.2368,
      "step": 1011
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3864889787302974,
      "learning_rate": 6.170666605859077e-06,
      "loss": 0.2384,
      "step": 1012
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4109031937415972,
      "learning_rate": 6.163786448328455e-06,
      "loss": 0.1807,
      "step": 1013
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.34953931394327,
      "learning_rate": 6.156903960201709e-06,
      "loss": 0.2259,
      "step": 1014
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3282422423014215,
      "learning_rate": 6.150019155261694e-06,
      "loss": 0.19,
      "step": 1015
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3738280095049007,
      "learning_rate": 6.1431320472959054e-06,
      "loss": 0.1784,
      "step": 1016
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2678775970665264,
      "learning_rate": 6.136242650096451e-06,
      "loss": 0.2135,
      "step": 1017
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.118469559232694,
      "learning_rate": 6.12935097746002e-06,
      "loss": 0.1448,
      "step": 1018
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3521707311593438,
      "learning_rate": 6.122457043187863e-06,
      "loss": 0.1972,
      "step": 1019
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4400050955097827,
      "learning_rate": 6.115560861085757e-06,
      "loss": 0.2308,
      "step": 1020
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.297349305930591,
      "learning_rate": 6.108662444963979e-06,
      "loss": 0.2375,
      "step": 1021
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3529639914266198,
      "learning_rate": 6.101761808637283e-06,
      "loss": 0.2128,
      "step": 1022
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.293384082648,
      "learning_rate": 6.094858965924866e-06,
      "loss": 0.1655,
      "step": 1023
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3841572113646585,
      "learning_rate": 6.087953930650349e-06,
      "loss": 0.1697,
      "step": 1024
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.259045630752443,
      "learning_rate": 6.081046716641735e-06,
      "loss": 0.1816,
      "step": 1025
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1295595554503277,
      "learning_rate": 6.0741373377314005e-06,
      "loss": 0.1305,
      "step": 1026
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2161429861521613,
      "learning_rate": 6.0672258077560475e-06,
      "loss": 0.179,
      "step": 1027
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.377742210005644,
      "learning_rate": 6.060312140556694e-06,
      "loss": 0.2127,
      "step": 1028
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2080365110578117,
      "learning_rate": 6.053396349978632e-06,
      "loss": 0.1769,
      "step": 1029
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2006247728312411,
      "learning_rate": 6.046478449871409e-06,
      "loss": 0.1619,
      "step": 1030
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2484425342875418,
      "learning_rate": 6.039558454088796e-06,
      "loss": 0.1986,
      "step": 1031
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.335644541114283,
      "learning_rate": 6.032636376488763e-06,
      "loss": 0.1903,
      "step": 1032
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.304662073452436,
      "learning_rate": 6.025712230933443e-06,
      "loss": 0.1996,
      "step": 1033
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3541082313027561,
      "learning_rate": 6.018786031289119e-06,
      "loss": 0.162,
      "step": 1034
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3797656204116304,
      "learning_rate": 6.0118577914261786e-06,
      "loss": 0.2021,
      "step": 1035
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2916179136787698,
      "learning_rate": 6.004927525219103e-06,
      "loss": 0.2613,
      "step": 1036
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5066060682782267,
      "learning_rate": 5.997995246546424e-06,
      "loss": 0.2266,
      "step": 1037
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5931932831656301,
      "learning_rate": 5.9910609692907105e-06,
      "loss": 0.2197,
      "step": 1038
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2513720130987007,
      "learning_rate": 5.984124707338528e-06,
      "loss": 0.1811,
      "step": 1039
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4406035262383015,
      "learning_rate": 5.977186474580421e-06,
      "loss": 0.1887,
      "step": 1040
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.37021530401625,
      "learning_rate": 5.970246284910877e-06,
      "loss": 0.1552,
      "step": 1041
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2893012238793995,
      "learning_rate": 5.963304152228307e-06,
      "loss": 0.2392,
      "step": 1042
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3939555530060102,
      "learning_rate": 5.9563600904350074e-06,
      "loss": 0.2016,
      "step": 1043
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5314318754932554,
      "learning_rate": 5.949414113437142e-06,
      "loss": 0.2179,
      "step": 1044
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.26284189826104,
      "learning_rate": 5.942466235144708e-06,
      "loss": 0.1913,
      "step": 1045
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3244397404198143,
      "learning_rate": 5.935516469471514e-06,
      "loss": 0.233,
      "step": 1046
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2888286055484741,
      "learning_rate": 5.9285648303351404e-06,
      "loss": 0.215,
      "step": 1047
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4720355878776252,
      "learning_rate": 5.921611331656928e-06,
      "loss": 0.2185,
      "step": 1048
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.305319598888082,
      "learning_rate": 5.914655987361934e-06,
      "loss": 0.2595,
      "step": 1049
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.561753526950287,
      "learning_rate": 5.907698811378919e-06,
      "loss": 0.2694,
      "step": 1050
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.234583485017962,
      "learning_rate": 5.900739817640304e-06,
      "loss": 0.2106,
      "step": 1051
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3936138056936864,
      "learning_rate": 5.893779020082159e-06,
      "loss": 0.1565,
      "step": 1052
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2669825517797841,
      "learning_rate": 5.886816432644155e-06,
      "loss": 0.2659,
      "step": 1053
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3868037768515489,
      "learning_rate": 5.8798520692695605e-06,
      "loss": 0.1608,
      "step": 1054
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4335541749786689,
      "learning_rate": 5.872885943905191e-06,
      "loss": 0.2341,
      "step": 1055
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4971370118924232,
      "learning_rate": 5.865918070501394e-06,
      "loss": 0.1862,
      "step": 1056
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2961554237274757,
      "learning_rate": 5.858948463012015e-06,
      "loss": 0.1859,
      "step": 1057
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3308495998139263,
      "learning_rate": 5.851977135394379e-06,
      "loss": 0.2529,
      "step": 1058
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3026388851780029,
      "learning_rate": 5.8450041016092465e-06,
      "loss": 0.2007,
      "step": 1059
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3717720024994167,
      "learning_rate": 5.838029375620801e-06,
      "loss": 0.2153,
      "step": 1060
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.28885930673361,
      "learning_rate": 5.831052971396613e-06,
      "loss": 0.1437,
      "step": 1061
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4389339911753112,
      "learning_rate": 5.824074902907614e-06,
      "loss": 0.2386,
      "step": 1062
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.210586990358928,
      "learning_rate": 5.817095184128067e-06,
      "loss": 0.2408,
      "step": 1063
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3863913772232812,
      "learning_rate": 5.810113829035544e-06,
      "loss": 0.1886,
      "step": 1064
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5042474726781643,
      "learning_rate": 5.803130851610887e-06,
      "loss": 0.1753,
      "step": 1065
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4902446641373097,
      "learning_rate": 5.796146265838195e-06,
      "loss": 0.2015,
      "step": 1066
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.614880362377743,
      "learning_rate": 5.78916008570478e-06,
      "loss": 0.1822,
      "step": 1067
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4880625638072245,
      "learning_rate": 5.782172325201155e-06,
      "loss": 0.2894,
      "step": 1068
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7248197885393834,
      "learning_rate": 5.77518299832099e-06,
      "loss": 0.2016,
      "step": 1069
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3083114355047367,
      "learning_rate": 5.768192119061096e-06,
      "loss": 0.2218,
      "step": 1070
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4376884809454964,
      "learning_rate": 5.761199701421392e-06,
      "loss": 0.2197,
      "step": 1071
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6553714485126876,
      "learning_rate": 5.754205759404879e-06,
      "loss": 0.2057,
      "step": 1072
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2830311216066679,
      "learning_rate": 5.747210307017609e-06,
      "loss": 0.2009,
      "step": 1073
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3599612396898333,
      "learning_rate": 5.740213358268658e-06,
      "loss": 0.1852,
      "step": 1074
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3172916716324503,
      "learning_rate": 5.733214927170101e-06,
      "loss": 0.1693,
      "step": 1075
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2718828768383705,
      "learning_rate": 5.726215027736981e-06,
      "loss": 0.1942,
      "step": 1076
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.403096744401289,
      "learning_rate": 5.719213673987277e-06,
      "loss": 0.1898,
      "step": 1077
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3753461414303925,
      "learning_rate": 5.712210879941891e-06,
      "loss": 0.2278,
      "step": 1078
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3125446609926752,
      "learning_rate": 5.705206659624597e-06,
      "loss": 0.2025,
      "step": 1079
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3929648645753496,
      "learning_rate": 5.698201027062034e-06,
      "loss": 0.235,
      "step": 1080
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2748903690073212,
      "learning_rate": 5.6911939962836625e-06,
      "loss": 0.2507,
      "step": 1081
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3376491775771409,
      "learning_rate": 5.684185581321752e-06,
      "loss": 0.2635,
      "step": 1082
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2948862351754635,
      "learning_rate": 5.6771757962113325e-06,
      "loss": 0.1581,
      "step": 1083
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.359875552844881,
      "learning_rate": 5.670164654990189e-06,
      "loss": 0.1695,
      "step": 1084
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.417266049093337,
      "learning_rate": 5.663152171698813e-06,
      "loss": 0.1831,
      "step": 1085
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2483952346967957,
      "learning_rate": 5.656138360380391e-06,
      "loss": 0.1776,
      "step": 1086
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.137438732524122,
      "learning_rate": 5.649123235080762e-06,
      "loss": 0.1344,
      "step": 1087
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3463548885424366,
      "learning_rate": 5.642106809848405e-06,
      "loss": 0.1754,
      "step": 1088
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2602935826657062,
      "learning_rate": 5.635089098734394e-06,
      "loss": 0.2292,
      "step": 1089
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2600789768907814,
      "learning_rate": 5.628070115792384e-06,
      "loss": 0.2018,
      "step": 1090
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.157795367689486,
      "learning_rate": 5.621049875078571e-06,
      "loss": 0.1833,
      "step": 1091
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4555713664256213,
      "learning_rate": 5.614028390651676e-06,
      "loss": 0.2096,
      "step": 1092
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3702522943638107,
      "learning_rate": 5.6070056765729065e-06,
      "loss": 0.1883,
      "step": 1093
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2342098654336007,
      "learning_rate": 5.599981746905935e-06,
      "loss": 0.1383,
      "step": 1094
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3698646496716078,
      "learning_rate": 5.592956615716867e-06,
      "loss": 0.2177,
      "step": 1095
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.271227389764084,
      "learning_rate": 5.585930297074214e-06,
      "loss": 0.1778,
      "step": 1096
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2902655505958023,
      "learning_rate": 5.578902805048866e-06,
      "loss": 0.1759,
      "step": 1097
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4619952460405543,
      "learning_rate": 5.571874153714063e-06,
      "loss": 0.2465,
      "step": 1098
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1766112430893356,
      "learning_rate": 5.564844357145365e-06,
      "loss": 0.1366,
      "step": 1099
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3838713196929502,
      "learning_rate": 5.55781342942063e-06,
      "loss": 0.2639,
      "step": 1100
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4287205826684801,
      "learning_rate": 5.550781384619974e-06,
      "loss": 0.2681,
      "step": 1101
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5850098427959278,
      "learning_rate": 5.543748236825756e-06,
      "loss": 0.1938,
      "step": 1102
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.381962089416708,
      "learning_rate": 5.536714000122543e-06,
      "loss": 0.2491,
      "step": 1103
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2483033014132234,
      "learning_rate": 5.529678688597081e-06,
      "loss": 0.1942,
      "step": 1104
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3676854049661042,
      "learning_rate": 5.522642316338268e-06,
      "loss": 0.1906,
      "step": 1105
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4332870295554874,
      "learning_rate": 5.515604897437129e-06,
      "loss": 0.2126,
      "step": 1106
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5722145947249386,
      "learning_rate": 5.50856644598678e-06,
      "loss": 0.1808,
      "step": 1107
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.329617480903326,
      "learning_rate": 5.501526976082413e-06,
      "loss": 0.1807,
      "step": 1108
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3112761333495386,
      "learning_rate": 5.49448650182125e-06,
      "loss": 0.1657,
      "step": 1109
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2295526866574191,
      "learning_rate": 5.487445037302531e-06,
      "loss": 0.1986,
      "step": 1110
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.269961340582299,
      "learning_rate": 5.480402596627477e-06,
      "loss": 0.1742,
      "step": 1111
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1489717414096594,
      "learning_rate": 5.473359193899265e-06,
      "loss": 0.183,
      "step": 1112
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3322795169514083,
      "learning_rate": 5.466314843222993e-06,
      "loss": 0.2612,
      "step": 1113
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1536604716934897,
      "learning_rate": 5.459269558705667e-06,
      "loss": 0.1929,
      "step": 1114
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2603947041577264,
      "learning_rate": 5.452223354456154e-06,
      "loss": 0.2124,
      "step": 1115
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.282084950817454,
      "learning_rate": 5.445176244585171e-06,
      "loss": 0.2211,
      "step": 1116
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.477189589758439,
      "learning_rate": 5.438128243205241e-06,
      "loss": 0.2705,
      "step": 1117
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2202535014738483,
      "learning_rate": 5.431079364430677e-06,
      "loss": 0.1898,
      "step": 1118
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5438161181873038,
      "learning_rate": 5.4240296223775465e-06,
      "loss": 0.203,
      "step": 1119
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4051526562058965,
      "learning_rate": 5.416979031163649e-06,
      "loss": 0.2179,
      "step": 1120
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.319949913645062,
      "learning_rate": 5.409927604908481e-06,
      "loss": 0.2114,
      "step": 1121
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4234170083626385,
      "learning_rate": 5.402875357733215e-06,
      "loss": 0.2213,
      "step": 1122
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3775425271666644,
      "learning_rate": 5.395822303760662e-06,
      "loss": 0.2301,
      "step": 1123
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1984238386862482,
      "learning_rate": 5.388768457115254e-06,
      "loss": 0.174,
      "step": 1124
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3929736911798811,
      "learning_rate": 5.381713831923008e-06,
      "loss": 0.1918,
      "step": 1125
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.31029435035214,
      "learning_rate": 5.3746584423115025e-06,
      "loss": 0.209,
      "step": 1126
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5432524382231085,
      "learning_rate": 5.367602302409841e-06,
      "loss": 0.224,
      "step": 1127
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1546305591430965,
      "learning_rate": 5.360545426348638e-06,
      "loss": 0.1797,
      "step": 1128
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2763521225054755,
      "learning_rate": 5.353487828259973e-06,
      "loss": 0.2063,
      "step": 1129
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4968897835867763,
      "learning_rate": 5.346429522277381e-06,
      "loss": 0.1748,
      "step": 1130
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3065261207796763,
      "learning_rate": 5.339370522535805e-06,
      "loss": 0.2494,
      "step": 1131
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1705317763949739,
      "learning_rate": 5.3323108431715845e-06,
      "loss": 0.1501,
      "step": 1132
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5389797818267816,
      "learning_rate": 5.325250498322417e-06,
      "loss": 0.2742,
      "step": 1133
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3337101225546337,
      "learning_rate": 5.318189502127332e-06,
      "loss": 0.1936,
      "step": 1134
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5966888594371405,
      "learning_rate": 5.311127868726667e-06,
      "loss": 0.2835,
      "step": 1135
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.291714019431373,
      "learning_rate": 5.304065612262033e-06,
      "loss": 0.2081,
      "step": 1136
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2880696143426826,
      "learning_rate": 5.297002746876284e-06,
      "loss": 0.2157,
      "step": 1137
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4213655205757878,
      "learning_rate": 5.289939286713505e-06,
      "loss": 0.2536,
      "step": 1138
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4869778961209745,
      "learning_rate": 5.282875245918963e-06,
      "loss": 0.2028,
      "step": 1139
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.214846930954013,
      "learning_rate": 5.275810638639088e-06,
      "loss": 0.2047,
      "step": 1140
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2401506175979147,
      "learning_rate": 5.26874547902145e-06,
      "loss": 0.1559,
      "step": 1141
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2320565614266155,
      "learning_rate": 5.2616797812147205e-06,
      "loss": 0.1935,
      "step": 1142
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.329929865051522,
      "learning_rate": 5.2546135593686484e-06,
      "loss": 0.2273,
      "step": 1143
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3486194194000323,
      "learning_rate": 5.247546827634035e-06,
      "loss": 0.1843,
      "step": 1144
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2206890305756675,
      "learning_rate": 5.240479600162704e-06,
      "loss": 0.1911,
      "step": 1145
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4869755214857652,
      "learning_rate": 5.233411891107464e-06,
      "loss": 0.1949,
      "step": 1146
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1732039037936886,
      "learning_rate": 5.226343714622096e-06,
      "loss": 0.2009,
      "step": 1147
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2686764443631402,
      "learning_rate": 5.219275084861315e-06,
      "loss": 0.2139,
      "step": 1148
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4645597876510081,
      "learning_rate": 5.212206015980742e-06,
      "loss": 0.2249,
      "step": 1149
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4614559673443825,
      "learning_rate": 5.2051365221368785e-06,
      "loss": 0.2185,
      "step": 1150
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3330413405232628,
      "learning_rate": 5.198066617487078e-06,
      "loss": 0.1819,
      "step": 1151
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3317502172224185,
      "learning_rate": 5.190996316189515e-06,
      "loss": 0.2364,
      "step": 1152
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3530333153306635,
      "learning_rate": 5.1839256324031575e-06,
      "loss": 0.2267,
      "step": 1153
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3146235813581102,
      "learning_rate": 5.176854580287744e-06,
      "loss": 0.2484,
      "step": 1154
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4166842558006283,
      "learning_rate": 5.169783174003744e-06,
      "loss": 0.2514,
      "step": 1155
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3249591779125163,
      "learning_rate": 5.1627114277123435e-06,
      "loss": 0.1885,
      "step": 1156
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3100409920800586,
      "learning_rate": 5.155639355575402e-06,
      "loss": 0.1912,
      "step": 1157
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3289983640814642,
      "learning_rate": 5.14856697175544e-06,
      "loss": 0.1803,
      "step": 1158
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2687167656151361,
      "learning_rate": 5.141494290415592e-06,
      "loss": 0.1652,
      "step": 1159
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3190049612930022,
      "learning_rate": 5.1344213257195965e-06,
      "loss": 0.2553,
      "step": 1160
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3842237906383486,
      "learning_rate": 5.127348091831755e-06,
      "loss": 0.1936,
      "step": 1161
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.375347688318873,
      "learning_rate": 5.120274602916915e-06,
      "loss": 0.1839,
      "step": 1162
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4190142570773765,
      "learning_rate": 5.1132008731404225e-06,
      "loss": 0.1859,
      "step": 1163
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.269754231994527,
      "learning_rate": 5.106126916668118e-06,
      "loss": 0.2052,
      "step": 1164
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2858854194582579,
      "learning_rate": 5.099052747666289e-06,
      "loss": 0.2195,
      "step": 1165
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3582451900833044,
      "learning_rate": 5.09197838030165e-06,
      "loss": 0.1785,
      "step": 1166
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.223594421450724,
      "learning_rate": 5.084903828741312e-06,
      "loss": 0.2148,
      "step": 1167
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3965268829457347,
      "learning_rate": 5.0778291071527596e-06,
      "loss": 0.2336,
      "step": 1168
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3771994929034908,
      "learning_rate": 5.070754229703811e-06,
      "loss": 0.1789,
      "step": 1169
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4528083462258161,
      "learning_rate": 5.063679210562602e-06,
      "loss": 0.1859,
      "step": 1170
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2044929030006213,
      "learning_rate": 5.056604063897549e-06,
      "loss": 0.1571,
      "step": 1171
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.507125428600444,
      "learning_rate": 5.049528803877324e-06,
      "loss": 0.1778,
      "step": 1172
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1960140172074518,
      "learning_rate": 5.042453444670829e-06,
      "loss": 0.232,
      "step": 1173
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3749945349876849,
      "learning_rate": 5.0353780004471605e-06,
      "loss": 0.1627,
      "step": 1174
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5169639793802006,
      "learning_rate": 5.028302485375588e-06,
      "loss": 0.1797,
      "step": 1175
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2448030866224917,
      "learning_rate": 5.021226913625522e-06,
      "loss": 0.2002,
      "step": 1176
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.442270297753889,
      "learning_rate": 5.0141512993664864e-06,
      "loss": 0.2414,
      "step": 1177
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3232084623085585,
      "learning_rate": 5.007075656768091e-06,
      "loss": 0.2226,
      "step": 1178
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5328321499428696,
      "learning_rate": 5e-06,
      "loss": 0.235,
      "step": 1179
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4217847548788505,
      "learning_rate": 4.99292434323191e-06,
      "loss": 0.2334,
      "step": 1180
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2767820879026897,
      "learning_rate": 4.985848700633515e-06,
      "loss": 0.192,
      "step": 1181
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.296160910452887,
      "learning_rate": 4.97877308637448e-06,
      "loss": 0.2374,
      "step": 1182
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.371293687594327,
      "learning_rate": 4.971697514624414e-06,
      "loss": 0.2544,
      "step": 1183
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2933664808084602,
      "learning_rate": 4.964621999552841e-06,
      "loss": 0.2061,
      "step": 1184
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2586368633545983,
      "learning_rate": 4.957546555329174e-06,
      "loss": 0.2006,
      "step": 1185
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.164471064771334,
      "learning_rate": 4.950471196122677e-06,
      "loss": 0.1779,
      "step": 1186
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3528424157031929,
      "learning_rate": 4.943395936102453e-06,
      "loss": 0.2166,
      "step": 1187
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.649529610779581,
      "learning_rate": 4.9363207894374e-06,
      "loss": 0.2268,
      "step": 1188
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.324702948948129,
      "learning_rate": 4.929245770296191e-06,
      "loss": 0.2232,
      "step": 1189
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2362720030093413,
      "learning_rate": 4.922170892847242e-06,
      "loss": 0.1594,
      "step": 1190
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.271998068853122,
      "learning_rate": 4.91509617125869e-06,
      "loss": 0.2367,
      "step": 1191
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.053579778498583,
      "learning_rate": 4.908021619698352e-06,
      "loss": 0.1554,
      "step": 1192
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3263651808582366,
      "learning_rate": 4.900947252333712e-06,
      "loss": 0.2422,
      "step": 1193
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5045713251664807,
      "learning_rate": 4.8938730833318825e-06,
      "loss": 0.2491,
      "step": 1194
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2882713430191854,
      "learning_rate": 4.8867991268595775e-06,
      "loss": 0.2261,
      "step": 1195
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.247882917399944,
      "learning_rate": 4.879725397083087e-06,
      "loss": 0.1839,
      "step": 1196
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1823444576195228,
      "learning_rate": 4.872651908168244e-06,
      "loss": 0.1976,
      "step": 1197
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2346993717543047,
      "learning_rate": 4.865578674280406e-06,
      "loss": 0.2105,
      "step": 1198
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.193689263382415,
      "learning_rate": 4.85850570958441e-06,
      "loss": 0.2163,
      "step": 1199
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1711178098411712,
      "learning_rate": 4.851433028244562e-06,
      "loss": 0.1798,
      "step": 1200
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3969714401327957,
      "learning_rate": 4.844360644424599e-06,
      "loss": 0.2333,
      "step": 1201
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4059247808950446,
      "learning_rate": 4.837288572287658e-06,
      "loss": 0.1575,
      "step": 1202
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0829695089154399,
      "learning_rate": 4.830216825996257e-06,
      "loss": 0.1555,
      "step": 1203
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2149350718676735,
      "learning_rate": 4.8231454197122575e-06,
      "loss": 0.1774,
      "step": 1204
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2787210648450191,
      "learning_rate": 4.816074367596843e-06,
      "loss": 0.1699,
      "step": 1205
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.220950269889546,
      "learning_rate": 4.809003683810487e-06,
      "loss": 0.1835,
      "step": 1206
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3592819535233427,
      "learning_rate": 4.801933382512923e-06,
      "loss": 0.2469,
      "step": 1207
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1902053240749786,
      "learning_rate": 4.794863477863122e-06,
      "loss": 0.2199,
      "step": 1208
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2598391316524418,
      "learning_rate": 4.78779398401926e-06,
      "loss": 0.1915,
      "step": 1209
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2750215364265103,
      "learning_rate": 4.780724915138686e-06,
      "loss": 0.2059,
      "step": 1210
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5006518212500126,
      "learning_rate": 4.7736562853779066e-06,
      "loss": 0.2162,
      "step": 1211
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3432317677225754,
      "learning_rate": 4.766588108892538e-06,
      "loss": 0.1727,
      "step": 1212
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.295052761275213,
      "learning_rate": 4.7595203998373e-06,
      "loss": 0.2239,
      "step": 1213
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2960612694739602,
      "learning_rate": 4.752453172365966e-06,
      "loss": 0.2152,
      "step": 1214
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3373487852287387,
      "learning_rate": 4.745386440631354e-06,
      "loss": 0.2068,
      "step": 1215
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2850998656494224,
      "learning_rate": 4.738320218785281e-06,
      "loss": 0.1935,
      "step": 1216
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3441441343483904,
      "learning_rate": 4.73125452097855e-06,
      "loss": 0.2374,
      "step": 1217
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.371547674736108,
      "learning_rate": 4.7241893613609126e-06,
      "loss": 0.2226,
      "step": 1218
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2397016787101467,
      "learning_rate": 4.717124754081038e-06,
      "loss": 0.2243,
      "step": 1219
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2808996416993494,
      "learning_rate": 4.710060713286496e-06,
      "loss": 0.2115,
      "step": 1220
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.286719348507916,
      "learning_rate": 4.702997253123716e-06,
      "loss": 0.1739,
      "step": 1221
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.385705933323673,
      "learning_rate": 4.6959343877379695e-06,
      "loss": 0.2201,
      "step": 1222
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3936241638848261,
      "learning_rate": 4.688872131273334e-06,
      "loss": 0.181,
      "step": 1223
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.407659754065257,
      "learning_rate": 4.6818104978726685e-06,
      "loss": 0.203,
      "step": 1224
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1773144816706962,
      "learning_rate": 4.674749501677585e-06,
      "loss": 0.1635,
      "step": 1225
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.11137363384617,
      "learning_rate": 4.667689156828417e-06,
      "loss": 0.2556,
      "step": 1226
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.351173334411585,
      "learning_rate": 4.660629477464197e-06,
      "loss": 0.2123,
      "step": 1227
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1985080885602033,
      "learning_rate": 4.653570477722621e-06,
      "loss": 0.1421,
      "step": 1228
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.36663175619096,
      "learning_rate": 4.646512171740028e-06,
      "loss": 0.1883,
      "step": 1229
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.240391105854636,
      "learning_rate": 4.639454573651364e-06,
      "loss": 0.1819,
      "step": 1230
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2212185270127085,
      "learning_rate": 4.6323976975901595e-06,
      "loss": 0.1934,
      "step": 1231
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.298529027798293,
      "learning_rate": 4.625341557688499e-06,
      "loss": 0.2046,
      "step": 1232
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.422232700542969,
      "learning_rate": 4.618286168076993e-06,
      "loss": 0.2223,
      "step": 1233
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4367707083359111,
      "learning_rate": 4.611231542884747e-06,
      "loss": 0.239,
      "step": 1234
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3955244716818809,
      "learning_rate": 4.6041776962393405e-06,
      "loss": 0.1861,
      "step": 1235
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3934362181246036,
      "learning_rate": 4.597124642266788e-06,
      "loss": 0.2301,
      "step": 1236
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1866788215691455,
      "learning_rate": 4.590072395091521e-06,
      "loss": 0.1672,
      "step": 1237
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3896454620535752,
      "learning_rate": 4.583020968836353e-06,
      "loss": 0.1901,
      "step": 1238
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2492337270011518,
      "learning_rate": 4.575970377622456e-06,
      "loss": 0.21,
      "step": 1239
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0210925745935835,
      "learning_rate": 4.568920635569325e-06,
      "loss": 0.1448,
      "step": 1240
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.448707749785248,
      "learning_rate": 4.56187175679476e-06,
      "loss": 0.1888,
      "step": 1241
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4358767719977998,
      "learning_rate": 4.554823755414829e-06,
      "loss": 0.2562,
      "step": 1242
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5176269240628988,
      "learning_rate": 4.547776645543846e-06,
      "loss": 0.1865,
      "step": 1243
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1732110491575551,
      "learning_rate": 4.540730441294334e-06,
      "loss": 0.2018,
      "step": 1244
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4366418366117182,
      "learning_rate": 4.533685156777008e-06,
      "loss": 0.2236,
      "step": 1245
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0924110968718916,
      "learning_rate": 4.526640806100737e-06,
      "loss": 0.1377,
      "step": 1246
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.409197113702048,
      "learning_rate": 4.519597403372524e-06,
      "loss": 0.2583,
      "step": 1247
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.335577930434069,
      "learning_rate": 4.5125549626974695e-06,
      "loss": 0.2158,
      "step": 1248
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2869868233690072,
      "learning_rate": 4.505513498178752e-06,
      "loss": 0.194,
      "step": 1249
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2964856591423934,
      "learning_rate": 4.4984730239175885e-06,
      "loss": 0.1918,
      "step": 1250
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2671486443996731,
      "learning_rate": 4.491433554013221e-06,
      "loss": 0.2357,
      "step": 1251
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4041629528222057,
      "learning_rate": 4.484395102562873e-06,
      "loss": 0.2384,
      "step": 1252
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3824481275480318,
      "learning_rate": 4.477357683661734e-06,
      "loss": 0.1448,
      "step": 1253
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2454454060455329,
      "learning_rate": 4.47032131140292e-06,
      "loss": 0.2094,
      "step": 1254
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1477209055733375,
      "learning_rate": 4.463285999877458e-06,
      "loss": 0.1513,
      "step": 1255
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.318269524231253,
      "learning_rate": 4.456251763174245e-06,
      "loss": 0.2009,
      "step": 1256
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.268043569343281,
      "learning_rate": 4.449218615380029e-06,
      "loss": 0.1485,
      "step": 1257
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4477914325756833,
      "learning_rate": 4.442186570579373e-06,
      "loss": 0.2382,
      "step": 1258
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.335627967872205,
      "learning_rate": 4.435155642854637e-06,
      "loss": 0.2044,
      "step": 1259
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.580431989539783,
      "learning_rate": 4.42812584628594e-06,
      "loss": 0.2193,
      "step": 1260
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3179344181661816,
      "learning_rate": 4.421097194951137e-06,
      "loss": 0.2155,
      "step": 1261
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.264916886966482,
      "learning_rate": 4.414069702925789e-06,
      "loss": 0.2292,
      "step": 1262
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0367251836878693,
      "learning_rate": 4.407043384283136e-06,
      "loss": 0.1836,
      "step": 1263
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3527390073396588,
      "learning_rate": 4.400018253094065e-06,
      "loss": 0.2478,
      "step": 1264
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2330046288083567,
      "learning_rate": 4.3929943234270935e-06,
      "loss": 0.1744,
      "step": 1265
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1905360255392465,
      "learning_rate": 4.385971609348324e-06,
      "loss": 0.2083,
      "step": 1266
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3776985544252993,
      "learning_rate": 4.37895012492143e-06,
      "loss": 0.2314,
      "step": 1267
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3177542912237152,
      "learning_rate": 4.371929884207617e-06,
      "loss": 0.2132,
      "step": 1268
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2481944226027388,
      "learning_rate": 4.364910901265607e-06,
      "loss": 0.1842,
      "step": 1269
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3028162140603463,
      "learning_rate": 4.3578931901515955e-06,
      "loss": 0.2176,
      "step": 1270
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1566309703106457,
      "learning_rate": 4.350876764919238e-06,
      "loss": 0.1531,
      "step": 1271
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1566589366348403,
      "learning_rate": 4.343861639619611e-06,
      "loss": 0.22,
      "step": 1272
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.35030780401301,
      "learning_rate": 4.336847828301189e-06,
      "loss": 0.2372,
      "step": 1273
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3528759052982553,
      "learning_rate": 4.329835345009813e-06,
      "loss": 0.1781,
      "step": 1274
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3680041339159879,
      "learning_rate": 4.322824203788669e-06,
      "loss": 0.2346,
      "step": 1275
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3643276254326984,
      "learning_rate": 4.3158144186782504e-06,
      "loss": 0.2174,
      "step": 1276
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3521852355380721,
      "learning_rate": 4.308806003716338e-06,
      "loss": 0.1642,
      "step": 1277
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2010216746808722,
      "learning_rate": 4.301798972937968e-06,
      "loss": 0.1705,
      "step": 1278
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1219591740380859,
      "learning_rate": 4.294793340375405e-06,
      "loss": 0.1744,
      "step": 1279
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.324331456282741,
      "learning_rate": 4.287789120058111e-06,
      "loss": 0.2041,
      "step": 1280
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.559651863279569,
      "learning_rate": 4.280786326012724e-06,
      "loss": 0.2366,
      "step": 1281
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4471526047634125,
      "learning_rate": 4.273784972263021e-06,
      "loss": 0.2367,
      "step": 1282
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3109666900250943,
      "learning_rate": 4.266785072829901e-06,
      "loss": 0.2077,
      "step": 1283
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3384163317982667,
      "learning_rate": 4.259786641731344e-06,
      "loss": 0.1893,
      "step": 1284
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.275234284953657,
      "learning_rate": 4.252789692982394e-06,
      "loss": 0.2212,
      "step": 1285
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1642433988725918,
      "learning_rate": 4.245794240595123e-06,
      "loss": 0.176,
      "step": 1286
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3041986358355633,
      "learning_rate": 4.23880029857861e-06,
      "loss": 0.1799,
      "step": 1287
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3878439220912189,
      "learning_rate": 4.231807880938904e-06,
      "loss": 0.2138,
      "step": 1288
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1134543988941141,
      "learning_rate": 4.224817001679011e-06,
      "loss": 0.1689,
      "step": 1289
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2266822878473613,
      "learning_rate": 4.217827674798845e-06,
      "loss": 0.2161,
      "step": 1290
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2441423613827889,
      "learning_rate": 4.210839914295219e-06,
      "loss": 0.191,
      "step": 1291
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2552638386150163,
      "learning_rate": 4.203853734161806e-06,
      "loss": 0.1479,
      "step": 1292
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1887052053174338,
      "learning_rate": 4.196869148389114e-06,
      "loss": 0.1305,
      "step": 1293
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2479035631211828,
      "learning_rate": 4.189886170964458e-06,
      "loss": 0.2019,
      "step": 1294
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2711098568582913,
      "learning_rate": 4.1829048158719344e-06,
      "loss": 0.2406,
      "step": 1295
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.348144624981435,
      "learning_rate": 4.1759250970923876e-06,
      "loss": 0.1787,
      "step": 1296
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.362507648797723,
      "learning_rate": 4.168947028603389e-06,
      "loss": 0.1674,
      "step": 1297
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3439522843598979,
      "learning_rate": 4.161970624379201e-06,
      "loss": 0.2034,
      "step": 1298
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3997027013692942,
      "learning_rate": 4.154995898390756e-06,
      "loss": 0.2104,
      "step": 1299
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1728615572151593,
      "learning_rate": 4.1480228646056225e-06,
      "loss": 0.1856,
      "step": 1300
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.319258480460535,
      "learning_rate": 4.141051536987986e-06,
      "loss": 0.2692,
      "step": 1301
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.12920537597461,
      "learning_rate": 4.1340819294986076e-06,
      "loss": 0.1498,
      "step": 1302
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3392065215815057,
      "learning_rate": 4.127114056094811e-06,
      "loss": 0.1863,
      "step": 1303
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.365298385234084,
      "learning_rate": 4.12014793073044e-06,
      "loss": 0.2642,
      "step": 1304
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2981930598354552,
      "learning_rate": 4.113183567355846e-06,
      "loss": 0.1898,
      "step": 1305
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3305582302414878,
      "learning_rate": 4.106220979917845e-06,
      "loss": 0.2139,
      "step": 1306
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6802806874844598,
      "learning_rate": 4.099260182359698e-06,
      "loss": 0.2609,
      "step": 1307
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.133035302317451,
      "learning_rate": 4.092301188621084e-06,
      "loss": 0.1582,
      "step": 1308
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.377973322852135,
      "learning_rate": 4.085344012638067e-06,
      "loss": 0.2815,
      "step": 1309
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.394852339703651,
      "learning_rate": 4.078388668343075e-06,
      "loss": 0.2257,
      "step": 1310
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4016293227892547,
      "learning_rate": 4.071435169664861e-06,
      "loss": 0.2259,
      "step": 1311
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2575986923148101,
      "learning_rate": 4.064483530528487e-06,
      "loss": 0.1895,
      "step": 1312
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3006995078689532,
      "learning_rate": 4.057533764855291e-06,
      "loss": 0.1906,
      "step": 1313
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.229654485427334,
      "learning_rate": 4.050585886562858e-06,
      "loss": 0.132,
      "step": 1314
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.148057482278972,
      "learning_rate": 4.043639909564993e-06,
      "loss": 0.1719,
      "step": 1315
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1702202558678116,
      "learning_rate": 4.036695847771694e-06,
      "loss": 0.2188,
      "step": 1316
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0873876885899878,
      "learning_rate": 4.029753715089124e-06,
      "loss": 0.2024,
      "step": 1317
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2224484100936954,
      "learning_rate": 4.02281352541958e-06,
      "loss": 0.1943,
      "step": 1318
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2362683204327554,
      "learning_rate": 4.015875292661474e-06,
      "loss": 0.2167,
      "step": 1319
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2120123524674071,
      "learning_rate": 4.008939030709291e-06,
      "loss": 0.2015,
      "step": 1320
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3129756390584693,
      "learning_rate": 4.002004753453577e-06,
      "loss": 0.2111,
      "step": 1321
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2784792947040615,
      "learning_rate": 3.995072474780899e-06,
      "loss": 0.1839,
      "step": 1322
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3774066403441536,
      "learning_rate": 3.988142208573822e-06,
      "loss": 0.2292,
      "step": 1323
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4912066634047365,
      "learning_rate": 3.981213968710882e-06,
      "loss": 0.2098,
      "step": 1324
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2528425225215098,
      "learning_rate": 3.9742877690665575e-06,
      "loss": 0.2381,
      "step": 1325
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.7238289116183587,
      "learning_rate": 3.967363623511238e-06,
      "loss": 0.1886,
      "step": 1326
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5629662335287706,
      "learning_rate": 3.960441545911205e-06,
      "loss": 0.194,
      "step": 1327
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.264298461743461,
      "learning_rate": 3.953521550128591e-06,
      "loss": 0.2233,
      "step": 1328
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2641909221342218,
      "learning_rate": 3.94660365002137e-06,
      "loss": 0.1878,
      "step": 1329
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.323671045575653,
      "learning_rate": 3.939687859443308e-06,
      "loss": 0.2009,
      "step": 1330
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.170937866115427,
      "learning_rate": 3.932774192243953e-06,
      "loss": 0.1739,
      "step": 1331
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3175472497027856,
      "learning_rate": 3.925862662268602e-06,
      "loss": 0.2486,
      "step": 1332
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2996275138607745,
      "learning_rate": 3.918953283358266e-06,
      "loss": 0.1705,
      "step": 1333
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2336433677165068,
      "learning_rate": 3.912046069349654e-06,
      "loss": 0.1614,
      "step": 1334
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4109118070253943,
      "learning_rate": 3.905141034075135e-06,
      "loss": 0.2478,
      "step": 1335
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3658340121146173,
      "learning_rate": 3.898238191362718e-06,
      "loss": 0.2156,
      "step": 1336
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3646144001532317,
      "learning_rate": 3.891337555036022e-06,
      "loss": 0.2182,
      "step": 1337
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1973446703340562,
      "learning_rate": 3.884439138914243e-06,
      "loss": 0.2065,
      "step": 1338
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2471302280895864,
      "learning_rate": 3.877542956812137e-06,
      "loss": 0.2249,
      "step": 1339
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2407537570841907,
      "learning_rate": 3.870649022539979e-06,
      "loss": 0.1962,
      "step": 1340
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3589476688416975,
      "learning_rate": 3.863757349903551e-06,
      "loss": 0.2332,
      "step": 1341
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1716304576895935,
      "learning_rate": 3.856867952704095e-06,
      "loss": 0.1683,
      "step": 1342
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3140806586129077,
      "learning_rate": 3.849980844738308e-06,
      "loss": 0.1837,
      "step": 1343
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.241880899315725,
      "learning_rate": 3.843096039798293e-06,
      "loss": 0.1996,
      "step": 1344
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.256411351350607,
      "learning_rate": 3.836213551671547e-06,
      "loss": 0.2833,
      "step": 1345
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4127420093632737,
      "learning_rate": 3.829333394140924e-06,
      "loss": 0.194,
      "step": 1346
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5513629147254815,
      "learning_rate": 3.822455580984613e-06,
      "loss": 0.2174,
      "step": 1347
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1198481627860173,
      "learning_rate": 3.8155801259761065e-06,
      "loss": 0.1867,
      "step": 1348
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3718633341501756,
      "learning_rate": 3.808707042884176e-06,
      "loss": 0.1756,
      "step": 1349
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2318784708608947,
      "learning_rate": 3.8018363454728415e-06,
      "loss": 0.2089,
      "step": 1350
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.183425810194176,
      "learning_rate": 3.794968047501345e-06,
      "loss": 0.1814,
      "step": 1351
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2984199318509742,
      "learning_rate": 3.7881021627241283e-06,
      "loss": 0.2037,
      "step": 1352
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3107357032020133,
      "learning_rate": 3.781238704890793e-06,
      "loss": 0.2009,
      "step": 1353
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3125057519952357,
      "learning_rate": 3.7743776877460864e-06,
      "loss": 0.1931,
      "step": 1354
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1956806519843668,
      "learning_rate": 3.767519125029865e-06,
      "loss": 0.1725,
      "step": 1355
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4256522049741422,
      "learning_rate": 3.760663030477072e-06,
      "loss": 0.1958,
      "step": 1356
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2651935769261793,
      "learning_rate": 3.7538094178177054e-06,
      "loss": 0.2022,
      "step": 1357
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4037678492051389,
      "learning_rate": 3.7469583007767964e-06,
      "loss": 0.205,
      "step": 1358
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0715655128175907,
      "learning_rate": 3.7401096930743753e-06,
      "loss": 0.1502,
      "step": 1359
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3765417733740442,
      "learning_rate": 3.733263608425447e-06,
      "loss": 0.182,
      "step": 1360
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.33714030993215,
      "learning_rate": 3.72642006053997e-06,
      "loss": 0.1905,
      "step": 1361
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3562356729422995,
      "learning_rate": 3.7195790631228133e-06,
      "loss": 0.235,
      "step": 1362
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2606401065745845,
      "learning_rate": 3.7127406298737458e-06,
      "loss": 0.1281,
      "step": 1363
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2734971692079726,
      "learning_rate": 3.705904774487396e-06,
      "loss": 0.2043,
      "step": 1364
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.523000750668205,
      "learning_rate": 3.6990715106532353e-06,
      "loss": 0.2378,
      "step": 1365
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2997971713516345,
      "learning_rate": 3.69224085205554e-06,
      "loss": 0.1614,
      "step": 1366
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2510069373827237,
      "learning_rate": 3.6854128123733736e-06,
      "loss": 0.2313,
      "step": 1367
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1973454120986184,
      "learning_rate": 3.6785874052805515e-06,
      "loss": 0.2021,
      "step": 1368
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4069429409544867,
      "learning_rate": 3.6717646444456196e-06,
      "loss": 0.2024,
      "step": 1369
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0758547229218747,
      "learning_rate": 3.664944543531822e-06,
      "loss": 0.194,
      "step": 1370
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3641016922875757,
      "learning_rate": 3.658127116197079e-06,
      "loss": 0.184,
      "step": 1371
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.332695198273679,
      "learning_rate": 3.651312376093955e-06,
      "loss": 0.1717,
      "step": 1372
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3798175293475183,
      "learning_rate": 3.6445003368696317e-06,
      "loss": 0.2033,
      "step": 1373
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.326106382661649,
      "learning_rate": 3.6376910121658867e-06,
      "loss": 0.2345,
      "step": 1374
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6057139994203207,
      "learning_rate": 3.6308844156190546e-06,
      "loss": 0.3448,
      "step": 1375
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2783279348552083,
      "learning_rate": 3.6240805608600137e-06,
      "loss": 0.1953,
      "step": 1376
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3145444323371391,
      "learning_rate": 3.6172794615141448e-06,
      "loss": 0.1998,
      "step": 1377
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.199987352309149,
      "learning_rate": 3.6104811312013175e-06,
      "loss": 0.1727,
      "step": 1378
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2227199243588878,
      "learning_rate": 3.60368558353585e-06,
      "loss": 0.2125,
      "step": 1379
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2178278723030733,
      "learning_rate": 3.596892832126494e-06,
      "loss": 0.208,
      "step": 1380
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5135814223391075,
      "learning_rate": 3.590102890576394e-06,
      "loss": 0.2298,
      "step": 1381
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.020188075025786,
      "learning_rate": 3.5833157724830757e-06,
      "loss": 0.1623,
      "step": 1382
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1045459004546505,
      "learning_rate": 3.576531491438403e-06,
      "loss": 0.1517,
      "step": 1383
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.374056632439514,
      "learning_rate": 3.569750061028565e-06,
      "loss": 0.2103,
      "step": 1384
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2060056004078046,
      "learning_rate": 3.5629714948340354e-06,
      "loss": 0.1994,
      "step": 1385
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1650989960880198,
      "learning_rate": 3.5561958064295585e-06,
      "loss": 0.1315,
      "step": 1386
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.329556087120069,
      "learning_rate": 3.5494230093841144e-06,
      "loss": 0.1785,
      "step": 1387
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2011316689292646,
      "learning_rate": 3.5426531172608858e-06,
      "loss": 0.2241,
      "step": 1388
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3415869856873444,
      "learning_rate": 3.5358861436172487e-06,
      "loss": 0.2052,
      "step": 1389
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3362648361963942,
      "learning_rate": 3.5291221020047257e-06,
      "loss": 0.1894,
      "step": 1390
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2160191259466908,
      "learning_rate": 3.5223610059689733e-06,
      "loss": 0.2131,
      "step": 1391
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2646723392181325,
      "learning_rate": 3.515602869049746e-06,
      "loss": 0.1764,
      "step": 1392
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1435080021163557,
      "learning_rate": 3.5088477047808738e-06,
      "loss": 0.2024,
      "step": 1393
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3465309610106366,
      "learning_rate": 3.5020955266902344e-06,
      "loss": 0.2043,
      "step": 1394
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2819144073146063,
      "learning_rate": 3.495346348299724e-06,
      "loss": 0.1572,
      "step": 1395
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2654250366639688,
      "learning_rate": 3.4886001831252324e-06,
      "loss": 0.1723,
      "step": 1396
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2156847940845559,
      "learning_rate": 3.481857044676613e-06,
      "loss": 0.1725,
      "step": 1397
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.355301441050942,
      "learning_rate": 3.475116946457662e-06,
      "loss": 0.2016,
      "step": 1398
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2386361153799725,
      "learning_rate": 3.4683799019660834e-06,
      "loss": 0.1887,
      "step": 1399
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4193092661139155,
      "learning_rate": 3.4616459246934686e-06,
      "loss": 0.2459,
      "step": 1400
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1661492124914714,
      "learning_rate": 3.4549150281252635e-06,
      "loss": 0.1694,
      "step": 1401
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2825941471304396,
      "learning_rate": 3.448187225740749e-06,
      "loss": 0.2144,
      "step": 1402
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.226493355065849,
      "learning_rate": 3.4414625310130038e-06,
      "loss": 0.2076,
      "step": 1403
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.21827887900747,
      "learning_rate": 3.4347409574088896e-06,
      "loss": 0.2669,
      "step": 1404
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5007310544271917,
      "learning_rate": 3.4280225183890124e-06,
      "loss": 0.2379,
      "step": 1405
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3865656312934698,
      "learning_rate": 3.4213072274077047e-06,
      "loss": 0.2104,
      "step": 1406
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2263162102532643,
      "learning_rate": 3.4145950979129916e-06,
      "loss": 0.1939,
      "step": 1407
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4523697934815434,
      "learning_rate": 3.4078861433465693e-06,
      "loss": 0.2752,
      "step": 1408
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4218474093268039,
      "learning_rate": 3.401180377143774e-06,
      "loss": 0.2465,
      "step": 1409
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3193313466458427,
      "learning_rate": 3.39447781273356e-06,
      "loss": 0.1767,
      "step": 1410
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3961695758205974,
      "learning_rate": 3.3877784635384668e-06,
      "loss": 0.1962,
      "step": 1411
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3683681307661881,
      "learning_rate": 3.3810823429745955e-06,
      "loss": 0.2027,
      "step": 1412
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4471266208570692,
      "learning_rate": 3.374389464451583e-06,
      "loss": 0.2126,
      "step": 1413
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2809791919492433,
      "learning_rate": 3.3676998413725726e-06,
      "loss": 0.1991,
      "step": 1414
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4441758759398051,
      "learning_rate": 3.361013487134188e-06,
      "loss": 0.2612,
      "step": 1415
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.406647265987341,
      "learning_rate": 3.354330415126508e-06,
      "loss": 0.2204,
      "step": 1416
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8118172408070277,
      "learning_rate": 3.3476506387330355e-06,
      "loss": 0.1611,
      "step": 1417
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2444454426432983,
      "learning_rate": 3.3409741713306777e-06,
      "loss": 0.192,
      "step": 1418
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2532219813605918,
      "learning_rate": 3.3343010262897125e-06,
      "loss": 0.2066,
      "step": 1419
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4210434022510938,
      "learning_rate": 3.327631216973766e-06,
      "loss": 0.2381,
      "step": 1420
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1218340783832956,
      "learning_rate": 3.320964756739781e-06,
      "loss": 0.155,
      "step": 1421
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2396478951821581,
      "learning_rate": 3.3143016589379984e-06,
      "loss": 0.2505,
      "step": 1422
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.297707203422161,
      "learning_rate": 3.3076419369119216e-06,
      "loss": 0.2312,
      "step": 1423
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4568547695918166,
      "learning_rate": 3.300985603998296e-06,
      "loss": 0.2575,
      "step": 1424
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.347311071459135,
      "learning_rate": 3.2943326735270766e-06,
      "loss": 0.1817,
      "step": 1425
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3721351008499636,
      "learning_rate": 3.2876831588214093e-06,
      "loss": 0.2088,
      "step": 1426
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4984995189985546,
      "learning_rate": 3.2810370731975956e-06,
      "loss": 0.2361,
      "step": 1427
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6507696328004235,
      "learning_rate": 3.2743944299650733e-06,
      "loss": 0.1641,
      "step": 1428
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2893621963446062,
      "learning_rate": 3.2677552424263836e-06,
      "loss": 0.2148,
      "step": 1429
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6453670250270993,
      "learning_rate": 3.2611195238771494e-06,
      "loss": 0.3061,
      "step": 1430
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3885888851832415,
      "learning_rate": 3.254487287606044e-06,
      "loss": 0.2051,
      "step": 1431
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1535114051689985,
      "learning_rate": 3.247858546894772e-06,
      "loss": 0.1593,
      "step": 1432
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4362954232919511,
      "learning_rate": 3.2412333150180298e-06,
      "loss": 0.2377,
      "step": 1433
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4246377870816747,
      "learning_rate": 3.234611605243496e-06,
      "loss": 0.1883,
      "step": 1434
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1756470904464194,
      "learning_rate": 3.2279934308317905e-06,
      "loss": 0.169,
      "step": 1435
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2162470757937789,
      "learning_rate": 3.2213788050364557e-06,
      "loss": 0.1697,
      "step": 1436
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2844015900559014,
      "learning_rate": 3.2147677411039236e-06,
      "loss": 0.2332,
      "step": 1437
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1280220215946617,
      "learning_rate": 3.2081602522734987e-06,
      "loss": 0.1275,
      "step": 1438
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2480297166489425,
      "learning_rate": 3.2015563517773214e-06,
      "loss": 0.199,
      "step": 1439
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2680540751592033,
      "learning_rate": 3.1949560528403487e-06,
      "loss": 0.2124,
      "step": 1440
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3455422157241383,
      "learning_rate": 3.1883593686803244e-06,
      "loss": 0.1908,
      "step": 1441
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1112677523124124,
      "learning_rate": 3.1817663125077534e-06,
      "loss": 0.1549,
      "step": 1442
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3605765871962512,
      "learning_rate": 3.1751768975258745e-06,
      "loss": 0.1563,
      "step": 1443
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2780186605488122,
      "learning_rate": 3.1685911369306364e-06,
      "loss": 0.2255,
      "step": 1444
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1754059319630803,
      "learning_rate": 3.162009043910668e-06,
      "loss": 0.1793,
      "step": 1445
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2662807306067683,
      "learning_rate": 3.155430631647254e-06,
      "loss": 0.1828,
      "step": 1446
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3029492103900417,
      "learning_rate": 3.1488559133143065e-06,
      "loss": 0.1872,
      "step": 1447
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4992642626822525,
      "learning_rate": 3.1422849020783442e-06,
      "loss": 0.2298,
      "step": 1448
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4299304639416517,
      "learning_rate": 3.1357176110984578e-06,
      "loss": 0.2197,
      "step": 1449
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.249821309654601,
      "learning_rate": 3.1291540535262905e-06,
      "loss": 0.1714,
      "step": 1450
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2881739252274313,
      "learning_rate": 3.122594242506006e-06,
      "loss": 0.2237,
      "step": 1451
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2954060961712976,
      "learning_rate": 3.11603819117427e-06,
      "loss": 0.1831,
      "step": 1452
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2170542052141484,
      "learning_rate": 3.1094859126602132e-06,
      "loss": 0.1552,
      "step": 1453
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2572386156960724,
      "learning_rate": 3.1029374200854167e-06,
      "loss": 0.2261,
      "step": 1454
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2587381185680728,
      "learning_rate": 3.096392726563874e-06,
      "loss": 0.1262,
      "step": 1455
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3280510689334857,
      "learning_rate": 3.089851845201976e-06,
      "loss": 0.2046,
      "step": 1456
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2210461710538305,
      "learning_rate": 3.0833147890984735e-06,
      "loss": 0.1451,
      "step": 1457
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.267655049262667,
      "learning_rate": 3.0767815713444655e-06,
      "loss": 0.1907,
      "step": 1458
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2630106062538877,
      "learning_rate": 3.070252205023356e-06,
      "loss": 0.198,
      "step": 1459
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.114056989046875,
      "learning_rate": 3.063726703210841e-06,
      "loss": 0.1328,
      "step": 1460
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.433593105494717,
      "learning_rate": 3.0572050789748732e-06,
      "loss": 0.1832,
      "step": 1461
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.264387773828275,
      "learning_rate": 3.050687345375645e-06,
      "loss": 0.199,
      "step": 1462
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4215434293231508,
      "learning_rate": 3.044173515465554e-06,
      "loss": 0.19,
      "step": 1463
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2851449055664976,
      "learning_rate": 3.0376636022891813e-06,
      "loss": 0.2581,
      "step": 1464
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2600668584935557,
      "learning_rate": 3.031157618883265e-06,
      "loss": 0.1852,
      "step": 1465
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.251836430847772,
      "learning_rate": 3.0246555782766723e-06,
      "loss": 0.2459,
      "step": 1466
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5813852711106064,
      "learning_rate": 3.0181574934903745e-06,
      "loss": 0.2272,
      "step": 1467
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.338967806468377,
      "learning_rate": 3.0116633775374237e-06,
      "loss": 0.2136,
      "step": 1468
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3306417470243233,
      "learning_rate": 3.0051732434229185e-06,
      "loss": 0.2151,
      "step": 1469
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2649991749569602,
      "learning_rate": 2.998687104143991e-06,
      "loss": 0.219,
      "step": 1470
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2115026920698146,
      "learning_rate": 2.992204972689765e-06,
      "loss": 0.2149,
      "step": 1471
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3026018865202522,
      "learning_rate": 2.985726862041346e-06,
      "loss": 0.1692,
      "step": 1472
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.282687158482048,
      "learning_rate": 2.9792527851717805e-06,
      "loss": 0.2,
      "step": 1473
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4086185068862265,
      "learning_rate": 2.972782755046043e-06,
      "loss": 0.2325,
      "step": 1474
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2332493032135712,
      "learning_rate": 2.966316784621e-06,
      "loss": 0.194,
      "step": 1475
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3559028284515502,
      "learning_rate": 2.9598548868453904e-06,
      "loss": 0.2157,
      "step": 1476
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2037785233036147,
      "learning_rate": 2.9533970746597935e-06,
      "loss": 0.1725,
      "step": 1477
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3481302584557981,
      "learning_rate": 2.9469433609966125e-06,
      "loss": 0.1948,
      "step": 1478
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2251039643613812,
      "learning_rate": 2.9404937587800374e-06,
      "loss": 0.148,
      "step": 1479
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3787804519334248,
      "learning_rate": 2.9340482809260294e-06,
      "loss": 0.2604,
      "step": 1480
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4263769099623163,
      "learning_rate": 2.9276069403422846e-06,
      "loss": 0.2373,
      "step": 1481
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.239133068588298,
      "learning_rate": 2.9211697499282223e-06,
      "loss": 0.1658,
      "step": 1482
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4641140413835623,
      "learning_rate": 2.9147367225749425e-06,
      "loss": 0.2464,
      "step": 1483
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.218064945987518,
      "learning_rate": 2.90830787116521e-06,
      "loss": 0.2004,
      "step": 1484
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4055498547227177,
      "learning_rate": 2.90188320857343e-06,
      "loss": 0.2225,
      "step": 1485
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2059722318469435,
      "learning_rate": 2.895462747665617e-06,
      "loss": 0.1696,
      "step": 1486
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.190128445302428,
      "learning_rate": 2.889046501299373e-06,
      "loss": 0.1953,
      "step": 1487
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1810316504310576,
      "learning_rate": 2.882634482323856e-06,
      "loss": 0.1704,
      "step": 1488
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1459266446443013,
      "learning_rate": 2.8762267035797607e-06,
      "loss": 0.1866,
      "step": 1489
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.111006722992572,
      "learning_rate": 2.8698231778992925e-06,
      "loss": 0.1949,
      "step": 1490
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2263522078718097,
      "learning_rate": 2.8634239181061384e-06,
      "loss": 0.1955,
      "step": 1491
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4587825451307317,
      "learning_rate": 2.857028937015438e-06,
      "loss": 0.2538,
      "step": 1492
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1146079428836115,
      "learning_rate": 2.850638247433768e-06,
      "loss": 0.1774,
      "step": 1493
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1618743674222205,
      "learning_rate": 2.8442518621591085e-06,
      "loss": 0.1619,
      "step": 1494
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3762872418658534,
      "learning_rate": 2.837869793980823e-06,
      "loss": 0.1729,
      "step": 1495
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.348196765668643,
      "learning_rate": 2.831492055679622e-06,
      "loss": 0.1772,
      "step": 1496
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.192359093536242,
      "learning_rate": 2.8251186600275533e-06,
      "loss": 0.1506,
      "step": 1497
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4169485963398527,
      "learning_rate": 2.8187496197879637e-06,
      "loss": 0.2223,
      "step": 1498
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1844083523521791,
      "learning_rate": 2.8123849477154808e-06,
      "loss": 0.2056,
      "step": 1499
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3962512508263627,
      "learning_rate": 2.806024656555978e-06,
      "loss": 0.2448,
      "step": 1500
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3754647249907457,
      "learning_rate": 2.799668759046564e-06,
      "loss": 0.212,
      "step": 1501
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.182188887923924,
      "learning_rate": 2.7933172679155427e-06,
      "loss": 0.2067,
      "step": 1502
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4982057840043894,
      "learning_rate": 2.786970195882398e-06,
      "loss": 0.2057,
      "step": 1503
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.265216187440717,
      "learning_rate": 2.7806275556577624e-06,
      "loss": 0.182,
      "step": 1504
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2226180926386627,
      "learning_rate": 2.7742893599433875e-06,
      "loss": 0.2134,
      "step": 1505
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2172138501473073,
      "learning_rate": 2.7679556214321357e-06,
      "loss": 0.145,
      "step": 1506
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4468366246114626,
      "learning_rate": 2.761626352807938e-06,
      "loss": 0.1776,
      "step": 1507
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1080119615402002,
      "learning_rate": 2.7553015667457694e-06,
      "loss": 0.1742,
      "step": 1508
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2094861020002234,
      "learning_rate": 2.748981275911633e-06,
      "loss": 0.1942,
      "step": 1509
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3731415427306441,
      "learning_rate": 2.7426654929625307e-06,
      "loss": 0.1748,
      "step": 1510
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.230484189313801,
      "learning_rate": 2.736354230546437e-06,
      "loss": 0.1633,
      "step": 1511
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.126823282814556,
      "learning_rate": 2.7300475013022666e-06,
      "loss": 0.1852,
      "step": 1512
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4024281113325017,
      "learning_rate": 2.7237453178598645e-06,
      "loss": 0.2164,
      "step": 1513
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3414608608565823,
      "learning_rate": 2.7174476928399685e-06,
      "loss": 0.247,
      "step": 1514
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1742732791077357,
      "learning_rate": 2.71115463885419e-06,
      "loss": 0.1429,
      "step": 1515
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3298323371876781,
      "learning_rate": 2.7048661685049814e-06,
      "loss": 0.199,
      "step": 1516
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1962610723434501,
      "learning_rate": 2.698582294385621e-06,
      "loss": 0.2013,
      "step": 1517
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3406074912616162,
      "learning_rate": 2.6923030290801817e-06,
      "loss": 0.1904,
      "step": 1518
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3381029337760781,
      "learning_rate": 2.6860283851635067e-06,
      "loss": 0.2053,
      "step": 1519
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.31648925839185,
      "learning_rate": 2.679758375201181e-06,
      "loss": 0.1931,
      "step": 1520
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.413226800036924,
      "learning_rate": 2.6734930117495133e-06,
      "loss": 0.193,
      "step": 1521
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3834856112421878,
      "learning_rate": 2.6672323073555074e-06,
      "loss": 0.1956,
      "step": 1522
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4313285374620024,
      "learning_rate": 2.6609762745568358e-06,
      "loss": 0.2053,
      "step": 1523
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.204926121100622,
      "learning_rate": 2.6547249258818162e-06,
      "loss": 0.1673,
      "step": 1524
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3609838897485942,
      "learning_rate": 2.6484782738493843e-06,
      "loss": 0.1961,
      "step": 1525
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.114596514885314,
      "learning_rate": 2.6422363309690702e-06,
      "loss": 0.1598,
      "step": 1526
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4206313769111083,
      "learning_rate": 2.6359991097409767e-06,
      "loss": 0.179,
      "step": 1527
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0693837157690416,
      "learning_rate": 2.6297666226557505e-06,
      "loss": 0.164,
      "step": 1528
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3359519606032353,
      "learning_rate": 2.6235388821945497e-06,
      "loss": 0.2175,
      "step": 1529
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3154766692692579,
      "learning_rate": 2.6173159008290407e-06,
      "loss": 0.2195,
      "step": 1530
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.143527117112643,
      "learning_rate": 2.61109769102135e-06,
      "loss": 0.1803,
      "step": 1531
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4103037526823197,
      "learning_rate": 2.604884265224048e-06,
      "loss": 0.1805,
      "step": 1532
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3527943921190024,
      "learning_rate": 2.598675635880129e-06,
      "loss": 0.1938,
      "step": 1533
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.36322894810282,
      "learning_rate": 2.59247181542298e-06,
      "loss": 0.1864,
      "step": 1534
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8390896614129641,
      "learning_rate": 2.5862728162763595e-06,
      "loss": 0.2108,
      "step": 1535
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3534760913809385,
      "learning_rate": 2.580078650854366e-06,
      "loss": 0.2086,
      "step": 1536
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2326160350409059,
      "learning_rate": 2.573889331561423e-06,
      "loss": 0.1548,
      "step": 1537
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.217865425484855,
      "learning_rate": 2.5677048707922482e-06,
      "loss": 0.1234,
      "step": 1538
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0833147508052479,
      "learning_rate": 2.5615252809318287e-06,
      "loss": 0.1729,
      "step": 1539
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3983416596009157,
      "learning_rate": 2.555350574355395e-06,
      "loss": 0.1586,
      "step": 1540
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2141640671269762,
      "learning_rate": 2.5491807634284016e-06,
      "loss": 0.1442,
      "step": 1541
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.314441094018191,
      "learning_rate": 2.543015860506499e-06,
      "loss": 0.2464,
      "step": 1542
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2537286846944438,
      "learning_rate": 2.536855877935507e-06,
      "loss": 0.1723,
      "step": 1543
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3321232553813735,
      "learning_rate": 2.5307008280513956e-06,
      "loss": 0.1746,
      "step": 1544
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1524422914256487,
      "learning_rate": 2.524550723180249e-06,
      "loss": 0.1734,
      "step": 1545
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.464959032167817,
      "learning_rate": 2.5184055756382566e-06,
      "loss": 0.1941,
      "step": 1546
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1586162056139726,
      "learning_rate": 2.5122653977316758e-06,
      "loss": 0.1754,
      "step": 1547
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3150510051392181,
      "learning_rate": 2.5061302017568153e-06,
      "loss": 0.1881,
      "step": 1548
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3919208471763755,
      "learning_rate": 2.5000000000000015e-06,
      "loss": 0.1971,
      "step": 1549
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1067747786195663,
      "learning_rate": 2.4938748047375637e-06,
      "loss": 0.1466,
      "step": 1550
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2388498071982776,
      "learning_rate": 2.487754628235805e-06,
      "loss": 0.2658,
      "step": 1551
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2573620714653113,
      "learning_rate": 2.481639482750979e-06,
      "loss": 0.1766,
      "step": 1552
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1549861272823607,
      "learning_rate": 2.4755293805292572e-06,
      "loss": 0.1768,
      "step": 1553
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1129317990144694,
      "learning_rate": 2.469424333806718e-06,
      "loss": 0.1576,
      "step": 1554
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3772402891852373,
      "learning_rate": 2.4633243548093195e-06,
      "loss": 0.1945,
      "step": 1555
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1200145050684465,
      "learning_rate": 2.4572294557528604e-06,
      "loss": 0.1907,
      "step": 1556
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3771926175568139,
      "learning_rate": 2.4511396488429727e-06,
      "loss": 0.2562,
      "step": 1557
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4220978715095414,
      "learning_rate": 2.4450549462750917e-06,
      "loss": 0.2038,
      "step": 1558
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4148931317097544,
      "learning_rate": 2.4389753602344298e-06,
      "loss": 0.2095,
      "step": 1559
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3110944219631426,
      "learning_rate": 2.4329009028959484e-06,
      "loss": 0.1909,
      "step": 1560
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4640805161847184,
      "learning_rate": 2.4268315864243447e-06,
      "loss": 0.2691,
      "step": 1561
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4551900672097386,
      "learning_rate": 2.420767422974018e-06,
      "loss": 0.2062,
      "step": 1562
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.090213178014344,
      "learning_rate": 2.4147084246890478e-06,
      "loss": 0.1508,
      "step": 1563
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4221786836196353,
      "learning_rate": 2.4086546037031734e-06,
      "loss": 0.1866,
      "step": 1564
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2951105736507607,
      "learning_rate": 2.402605972139759e-06,
      "loss": 0.1926,
      "step": 1565
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.34709306421326,
      "learning_rate": 2.396562542111781e-06,
      "loss": 0.2216,
      "step": 1566
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2205208335351407,
      "learning_rate": 2.3905243257218007e-06,
      "loss": 0.1683,
      "step": 1567
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3256508315247146,
      "learning_rate": 2.3844913350619382e-06,
      "loss": 0.223,
      "step": 1568
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2993636616828514,
      "learning_rate": 2.3784635822138424e-06,
      "loss": 0.2405,
      "step": 1569
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4814789320333952,
      "learning_rate": 2.37244107924868e-06,
      "loss": 0.205,
      "step": 1570
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1464631935125638,
      "learning_rate": 2.366423838227102e-06,
      "loss": 0.1992,
      "step": 1571
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1137860297668924,
      "learning_rate": 2.3604118711992237e-06,
      "loss": 0.1147,
      "step": 1572
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1870192278021667,
      "learning_rate": 2.354405190204592e-06,
      "loss": 0.2165,
      "step": 1573
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3732770998454131,
      "learning_rate": 2.348403807272176e-06,
      "loss": 0.1863,
      "step": 1574
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3812006211053385,
      "learning_rate": 2.342407734420331e-06,
      "loss": 0.1904,
      "step": 1575
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0757071707839247,
      "learning_rate": 2.3364169836567805e-06,
      "loss": 0.1543,
      "step": 1576
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3482410863734917,
      "learning_rate": 2.3304315669785853e-06,
      "loss": 0.1971,
      "step": 1577
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4785656950210635,
      "learning_rate": 2.3244514963721276e-06,
      "loss": 0.2507,
      "step": 1578
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2195850606285321,
      "learning_rate": 2.318476783813088e-06,
      "loss": 0.1735,
      "step": 1579
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1416743043707183,
      "learning_rate": 2.3125074412664085e-06,
      "loss": 0.1936,
      "step": 1580
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3133909931804728,
      "learning_rate": 2.306543480686281e-06,
      "loss": 0.2262,
      "step": 1581
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2092723759889163,
      "learning_rate": 2.30058491401612e-06,
      "loss": 0.1577,
      "step": 1582
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.204151625026336,
      "learning_rate": 2.294631753188541e-06,
      "loss": 0.1528,
      "step": 1583
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2778215868031244,
      "learning_rate": 2.288684010125325e-06,
      "loss": 0.1724,
      "step": 1584
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.332787333858535,
      "learning_rate": 2.2827416967374115e-06,
      "loss": 0.1612,
      "step": 1585
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2917696906382747,
      "learning_rate": 2.2768048249248648e-06,
      "loss": 0.1734,
      "step": 1586
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0464468815626873,
      "learning_rate": 2.270873406576849e-06,
      "loss": 0.1283,
      "step": 1587
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2439505043267318,
      "learning_rate": 2.2649474535716128e-06,
      "loss": 0.2519,
      "step": 1588
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4265648413942513,
      "learning_rate": 2.2590269777764516e-06,
      "loss": 0.2078,
      "step": 1589
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1587515212014068,
      "learning_rate": 2.2531119910476994e-06,
      "loss": 0.1264,
      "step": 1590
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2241558835960682,
      "learning_rate": 2.247202505230695e-06,
      "loss": 0.2109,
      "step": 1591
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.540470561976236,
      "learning_rate": 2.241298532159763e-06,
      "loss": 0.2057,
      "step": 1592
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1820027918143539,
      "learning_rate": 2.2354000836581834e-06,
      "loss": 0.1909,
      "step": 1593
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9841144597440228,
      "learning_rate": 2.229507171538178e-06,
      "loss": 0.1068,
      "step": 1594
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.329395750161705,
      "learning_rate": 2.223619807600878e-06,
      "loss": 0.1788,
      "step": 1595
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2011887452946637,
      "learning_rate": 2.217738003636309e-06,
      "loss": 0.1784,
      "step": 1596
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.337453260796749,
      "learning_rate": 2.2118617714233524e-06,
      "loss": 0.2101,
      "step": 1597
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2532877785323306,
      "learning_rate": 2.205991122729742e-06,
      "loss": 0.2097,
      "step": 1598
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3410772599744056,
      "learning_rate": 2.2001260693120236e-06,
      "loss": 0.2019,
      "step": 1599
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.225523949869891,
      "learning_rate": 2.1942666229155424e-06,
      "loss": 0.1807,
      "step": 1600
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.282685906531524,
      "learning_rate": 2.1884127952744095e-06,
      "loss": 0.1897,
      "step": 1601
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4205975960020327,
      "learning_rate": 2.1825645981114862e-06,
      "loss": 0.2316,
      "step": 1602
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1473295549226996,
      "learning_rate": 2.1767220431383652e-06,
      "loss": 0.148,
      "step": 1603
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1526821985177467,
      "learning_rate": 2.1708851420553277e-06,
      "loss": 0.1667,
      "step": 1604
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3938776703016664,
      "learning_rate": 2.1650539065513415e-06,
      "loss": 0.2074,
      "step": 1605
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1949534424826795,
      "learning_rate": 2.159228348304025e-06,
      "loss": 0.1671,
      "step": 1606
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.304911503739554,
      "learning_rate": 2.153408478979628e-06,
      "loss": 0.2173,
      "step": 1607
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.075224352024975,
      "learning_rate": 2.147594310233009e-06,
      "loss": 0.1335,
      "step": 1608
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2883650144248329,
      "learning_rate": 2.141785853707607e-06,
      "loss": 0.1889,
      "step": 1609
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3809761524828457,
      "learning_rate": 2.1359831210354247e-06,
      "loss": 0.2195,
      "step": 1610
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2098003811350004,
      "learning_rate": 2.130186123837002e-06,
      "loss": 0.1613,
      "step": 1611
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2360891836349555,
      "learning_rate": 2.124394873721394e-06,
      "loss": 0.1785,
      "step": 1612
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1703282797647914,
      "learning_rate": 2.118609382286142e-06,
      "loss": 0.1468,
      "step": 1613
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4661057824467234,
      "learning_rate": 2.1128296611172593e-06,
      "loss": 0.2308,
      "step": 1614
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2859100838791562,
      "learning_rate": 2.107055721789203e-06,
      "loss": 0.1667,
      "step": 1615
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.147303367939438,
      "learning_rate": 2.101287575864853e-06,
      "loss": 0.1604,
      "step": 1616
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8713567380715364,
      "learning_rate": 2.0955252348954806e-06,
      "loss": 0.2448,
      "step": 1617
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.25873132658056,
      "learning_rate": 2.08976871042074e-06,
      "loss": 0.2031,
      "step": 1618
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1810464782591867,
      "learning_rate": 2.0840180139686333e-06,
      "loss": 0.1921,
      "step": 1619
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.198312151785253,
      "learning_rate": 2.0782731570554946e-06,
      "loss": 0.1632,
      "step": 1620
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2718709107460855,
      "learning_rate": 2.0725341511859576e-06,
      "loss": 0.1989,
      "step": 1621
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2664603555199443,
      "learning_rate": 2.066801007852944e-06,
      "loss": 0.2005,
      "step": 1622
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2299862629589735,
      "learning_rate": 2.061073738537635e-06,
      "loss": 0.1903,
      "step": 1623
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.511684159877978,
      "learning_rate": 2.0553523547094473e-06,
      "loss": 0.2338,
      "step": 1624
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1351314917037933,
      "learning_rate": 2.0496368678260094e-06,
      "loss": 0.1927,
      "step": 1625
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3043899180975447,
      "learning_rate": 2.043927289333141e-06,
      "loss": 0.1983,
      "step": 1626
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.310538999668118,
      "learning_rate": 2.038223630664835e-06,
      "loss": 0.1796,
      "step": 1627
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5694402343295186,
      "learning_rate": 2.0325259032432264e-06,
      "loss": 0.2402,
      "step": 1628
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2864307944790432,
      "learning_rate": 2.0268341184785674e-06,
      "loss": 0.1882,
      "step": 1629
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1312153753684355,
      "learning_rate": 2.021148287769214e-06,
      "loss": 0.1606,
      "step": 1630
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5613669611091883,
      "learning_rate": 2.015468422501599e-06,
      "loss": 0.1807,
      "step": 1631
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2957503317173324,
      "learning_rate": 2.00979453405021e-06,
      "loss": 0.1767,
      "step": 1632
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2250222837259352,
      "learning_rate": 2.004126633777559e-06,
      "loss": 0.1855,
      "step": 1633
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.356652599507279,
      "learning_rate": 1.998464733034172e-06,
      "loss": 0.1822,
      "step": 1634
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.328650795090311,
      "learning_rate": 1.992808843158559e-06,
      "loss": 0.2778,
      "step": 1635
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2026323912055994,
      "learning_rate": 1.987158975477194e-06,
      "loss": 0.2073,
      "step": 1636
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.202114365828637,
      "learning_rate": 1.9815151413044865e-06,
      "loss": 0.1892,
      "step": 1637
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3668031708738289,
      "learning_rate": 1.9758773519427664e-06,
      "loss": 0.2218,
      "step": 1638
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1280468740747516,
      "learning_rate": 1.9702456186822595e-06,
      "loss": 0.1392,
      "step": 1639
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1633563181611384,
      "learning_rate": 1.9646199528010625e-06,
      "loss": 0.1695,
      "step": 1640
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4391410612738154,
      "learning_rate": 1.959000365565119e-06,
      "loss": 0.2639,
      "step": 1641
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3460051666652448,
      "learning_rate": 1.9533868682282026e-06,
      "loss": 0.2212,
      "step": 1642
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2344765827630582,
      "learning_rate": 1.947779472031891e-06,
      "loss": 0.206,
      "step": 1643
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1263362890518323,
      "learning_rate": 1.9421781882055447e-06,
      "loss": 0.1834,
      "step": 1644
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1874159758219804,
      "learning_rate": 1.936583027966279e-06,
      "loss": 0.2238,
      "step": 1645
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.175577383916632,
      "learning_rate": 1.93099400251895e-06,
      "loss": 0.2002,
      "step": 1646
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2291923595408072,
      "learning_rate": 1.9254111230561282e-06,
      "loss": 0.2408,
      "step": 1647
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.260548344765347,
      "learning_rate": 1.9198344007580776e-06,
      "loss": 0.1847,
      "step": 1648
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3703990008436915,
      "learning_rate": 1.9142638467927254e-06,
      "loss": 0.2191,
      "step": 1649
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3514236359046263,
      "learning_rate": 1.908699472315651e-06,
      "loss": 0.2006,
      "step": 1650
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.361510666612423,
      "learning_rate": 1.9031412884700612e-06,
      "loss": 0.1901,
      "step": 1651
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.170262793311354,
      "learning_rate": 1.8975893063867635e-06,
      "loss": 0.1419,
      "step": 1652
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1773994346012162,
      "learning_rate": 1.8920435371841394e-06,
      "loss": 0.2068,
      "step": 1653
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.178200649652803,
      "learning_rate": 1.8865039919681377e-06,
      "loss": 0.1786,
      "step": 1654
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8049210312835253,
      "learning_rate": 1.8809706818322375e-06,
      "loss": 0.2083,
      "step": 1655
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0302158428206758,
      "learning_rate": 1.8754436178574358e-06,
      "loss": 0.1193,
      "step": 1656
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.270108587770916,
      "learning_rate": 1.8699228111122146e-06,
      "loss": 0.2129,
      "step": 1657
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1103055388390088,
      "learning_rate": 1.8644082726525313e-06,
      "loss": 0.1834,
      "step": 1658
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2345507286271047,
      "learning_rate": 1.8589000135217882e-06,
      "loss": 0.1982,
      "step": 1659
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3630012285057358,
      "learning_rate": 1.8533980447508138e-06,
      "loss": 0.2381,
      "step": 1660
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3404516924298537,
      "learning_rate": 1.8479023773578354e-06,
      "loss": 0.1957,
      "step": 1661
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2747527608956302,
      "learning_rate": 1.8424130223484655e-06,
      "loss": 0.1782,
      "step": 1662
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3601906990918577,
      "learning_rate": 1.8369299907156745e-06,
      "loss": 0.2031,
      "step": 1663
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2305738708299787,
      "learning_rate": 1.831453293439771e-06,
      "loss": 0.1839,
      "step": 1664
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2424489905655156,
      "learning_rate": 1.8259829414883728e-06,
      "loss": 0.1885,
      "step": 1665
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3181509300346776,
      "learning_rate": 1.8205189458163964e-06,
      "loss": 0.1936,
      "step": 1666
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.446796954967567,
      "learning_rate": 1.815061317366027e-06,
      "loss": 0.2166,
      "step": 1667
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1998631416453354,
      "learning_rate": 1.809610067066701e-06,
      "loss": 0.1982,
      "step": 1668
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4121970228033887,
      "learning_rate": 1.8041652058350768e-06,
      "loss": 0.1803,
      "step": 1669
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2502316602310533,
      "learning_rate": 1.7987267445750224e-06,
      "loss": 0.1734,
      "step": 1670
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1809447704161513,
      "learning_rate": 1.7932946941775882e-06,
      "loss": 0.1859,
      "step": 1671
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.261945854566483,
      "learning_rate": 1.7878690655209862e-06,
      "loss": 0.1934,
      "step": 1672
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.491368169039514,
      "learning_rate": 1.78244986947057e-06,
      "loss": 0.1948,
      "step": 1673
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3766334458299292,
      "learning_rate": 1.7770371168788042e-06,
      "loss": 0.2203,
      "step": 1674
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3853553171468593,
      "learning_rate": 1.7716308185852605e-06,
      "loss": 0.1698,
      "step": 1675
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2533002980669838,
      "learning_rate": 1.7662309854165788e-06,
      "loss": 0.196,
      "step": 1676
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3072641228959843,
      "learning_rate": 1.7608376281864502e-06,
      "loss": 0.221,
      "step": 1677
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.321575713472952,
      "learning_rate": 1.755450757695601e-06,
      "loss": 0.2489,
      "step": 1678
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3165782817023004,
      "learning_rate": 1.7500703847317663e-06,
      "loss": 0.1177,
      "step": 1679
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1418308565907473,
      "learning_rate": 1.7446965200696702e-06,
      "loss": 0.1926,
      "step": 1680
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4993423759978417,
      "learning_rate": 1.739329174470999e-06,
      "loss": 0.2274,
      "step": 1681
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2727322691360112,
      "learning_rate": 1.733968358684388e-06,
      "loss": 0.2318,
      "step": 1682
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.288315095029625,
      "learning_rate": 1.7286140834453958e-06,
      "loss": 0.2098,
      "step": 1683
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2571520427224463,
      "learning_rate": 1.723266359476483e-06,
      "loss": 0.1798,
      "step": 1684
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0738563787851196,
      "learning_rate": 1.7179251974869858e-06,
      "loss": 0.1541,
      "step": 1685
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.170948621320043,
      "learning_rate": 1.7125906081731064e-06,
      "loss": 0.1435,
      "step": 1686
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3998340811222758,
      "learning_rate": 1.7072626022178796e-06,
      "loss": 0.1545,
      "step": 1687
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1679562109749448,
      "learning_rate": 1.7019411902911609e-06,
      "loss": 0.163,
      "step": 1688
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.274256924318675,
      "learning_rate": 1.6966263830495939e-06,
      "loss": 0.1685,
      "step": 1689
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.292145509096811,
      "learning_rate": 1.6913181911366017e-06,
      "loss": 0.2588,
      "step": 1690
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4087536266416896,
      "learning_rate": 1.6860166251823574e-06,
      "loss": 0.2601,
      "step": 1691
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3939298578686656,
      "learning_rate": 1.6807216958037664e-06,
      "loss": 0.224,
      "step": 1692
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.037105512093906,
      "learning_rate": 1.6754334136044398e-06,
      "loss": 0.1684,
      "step": 1693
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2628618032715853,
      "learning_rate": 1.6701517891746805e-06,
      "loss": 0.2117,
      "step": 1694
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.289636307709651,
      "learning_rate": 1.6648768330914578e-06,
      "loss": 0.179,
      "step": 1695
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2292330077971607,
      "learning_rate": 1.6596085559183872e-06,
      "loss": 0.1725,
      "step": 1696
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3490585889640219,
      "learning_rate": 1.6543469682057105e-06,
      "loss": 0.2343,
      "step": 1697
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9688902088643485,
      "learning_rate": 1.649092080490266e-06,
      "loss": 0.1154,
      "step": 1698
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3262554691186115,
      "learning_rate": 1.6438439032954857e-06,
      "loss": 0.2268,
      "step": 1699
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.242127053049811,
      "learning_rate": 1.6386024471313561e-06,
      "loss": 0.1926,
      "step": 1700
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1526715025158047,
      "learning_rate": 1.633367722494404e-06,
      "loss": 0.2137,
      "step": 1701
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4965766974619594,
      "learning_rate": 1.6281397398676774e-06,
      "loss": 0.2527,
      "step": 1702
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3458642620496255,
      "learning_rate": 1.6229185097207228e-06,
      "loss": 0.2278,
      "step": 1703
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3254272959483266,
      "learning_rate": 1.6177040425095664e-06,
      "loss": 0.2265,
      "step": 1704
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2688453609642667,
      "learning_rate": 1.6124963486766844e-06,
      "loss": 0.2511,
      "step": 1705
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3820038071244962,
      "learning_rate": 1.6072954386509948e-06,
      "loss": 0.2456,
      "step": 1706
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3638555905954544,
      "learning_rate": 1.6021013228478277e-06,
      "loss": 0.1777,
      "step": 1707
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3513557635816023,
      "learning_rate": 1.5969140116689102e-06,
      "loss": 0.244,
      "step": 1708
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.254274264924666,
      "learning_rate": 1.5917335155023368e-06,
      "loss": 0.1343,
      "step": 1709
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0750962696929593,
      "learning_rate": 1.5865598447225576e-06,
      "loss": 0.2035,
      "step": 1710
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0762779360147328,
      "learning_rate": 1.5813930096903562e-06,
      "loss": 0.1133,
      "step": 1711
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3995278515913345,
      "learning_rate": 1.5762330207528249e-06,
      "loss": 0.23,
      "step": 1712
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2089777440947118,
      "learning_rate": 1.5710798882433432e-06,
      "loss": 0.2153,
      "step": 1713
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2996653746185673,
      "learning_rate": 1.5659336224815642e-06,
      "loss": 0.1965,
      "step": 1714
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1230001257427074,
      "learning_rate": 1.5607942337733884e-06,
      "loss": 0.1808,
      "step": 1715
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3376016552854764,
      "learning_rate": 1.555661732410944e-06,
      "loss": 0.1283,
      "step": 1716
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1275560472274624,
      "learning_rate": 1.5505361286725679e-06,
      "loss": 0.1517,
      "step": 1717
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3719229530246535,
      "learning_rate": 1.5454174328227794e-06,
      "loss": 0.2138,
      "step": 1718
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3603312904547626,
      "learning_rate": 1.5403056551122697e-06,
      "loss": 0.1863,
      "step": 1719
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1011272808392907,
      "learning_rate": 1.5352008057778723e-06,
      "loss": 0.1719,
      "step": 1720
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4027478917702458,
      "learning_rate": 1.5301028950425494e-06,
      "loss": 0.1669,
      "step": 1721
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3930858559707602,
      "learning_rate": 1.5250119331153606e-06,
      "loss": 0.2647,
      "step": 1722
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2705957951892055,
      "learning_rate": 1.5199279301914592e-06,
      "loss": 0.1904,
      "step": 1723
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1615898592768472,
      "learning_rate": 1.5148508964520586e-06,
      "loss": 0.1874,
      "step": 1724
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1099511848408585,
      "learning_rate": 1.5097808420644117e-06,
      "loss": 0.1678,
      "step": 1725
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.283234719363446,
      "learning_rate": 1.5047177771817989e-06,
      "loss": 0.1417,
      "step": 1726
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2850072718723042,
      "learning_rate": 1.4996617119435036e-06,
      "loss": 0.1837,
      "step": 1727
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0950083722855832,
      "learning_rate": 1.494612656474792e-06,
      "loss": 0.1402,
      "step": 1728
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2430556335890295,
      "learning_rate": 1.4895706208868876e-06,
      "loss": 0.1935,
      "step": 1729
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.419316176917452,
      "learning_rate": 1.4845356152769618e-06,
      "loss": 0.2731,
      "step": 1730
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3261387187236975,
      "learning_rate": 1.4795076497281052e-06,
      "loss": 0.2012,
      "step": 1731
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2237271193424537,
      "learning_rate": 1.4744867343093112e-06,
      "loss": 0.1848,
      "step": 1732
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4074192955234428,
      "learning_rate": 1.4694728790754514e-06,
      "loss": 0.22,
      "step": 1733
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1217970889366191,
      "learning_rate": 1.4644660940672628e-06,
      "loss": 0.159,
      "step": 1734
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1201716755909223,
      "learning_rate": 1.459466389311321e-06,
      "loss": 0.1949,
      "step": 1735
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2347307638759668,
      "learning_rate": 1.4544737748200238e-06,
      "loss": 0.1562,
      "step": 1736
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1618863607241832,
      "learning_rate": 1.4494882605915717e-06,
      "loss": 0.1896,
      "step": 1737
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3085477800870828,
      "learning_rate": 1.4445098566099409e-06,
      "loss": 0.1928,
      "step": 1738
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3198056385761283,
      "learning_rate": 1.439538572844873e-06,
      "loss": 0.1632,
      "step": 1739
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1414852451938027,
      "learning_rate": 1.4345744192518507e-06,
      "loss": 0.1786,
      "step": 1740
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.303600514116103,
      "learning_rate": 1.4296174057720775e-06,
      "loss": 0.2062,
      "step": 1741
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.249653791894533,
      "learning_rate": 1.4246675423324546e-06,
      "loss": 0.1208,
      "step": 1742
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2701855084260536,
      "learning_rate": 1.4197248388455693e-06,
      "loss": 0.2039,
      "step": 1743
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0505020763632942,
      "learning_rate": 1.4147893052096684e-06,
      "loss": 0.1238,
      "step": 1744
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3970390148217988,
      "learning_rate": 1.4098609513086414e-06,
      "loss": 0.235,
      "step": 1745
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1143254945281351,
      "learning_rate": 1.4049397870119962e-06,
      "loss": 0.1889,
      "step": 1746
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3291592673380654,
      "learning_rate": 1.4000258221748452e-06,
      "loss": 0.155,
      "step": 1747
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2221355733859423,
      "learning_rate": 1.3951190666378884e-06,
      "loss": 0.2071,
      "step": 1748
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3643223035613377,
      "learning_rate": 1.390219530227378e-06,
      "loss": 0.217,
      "step": 1749
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1824090754733534,
      "learning_rate": 1.3853272227551167e-06,
      "loss": 0.1697,
      "step": 1750
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3894138031586178,
      "learning_rate": 1.3804421540184282e-06,
      "loss": 0.1871,
      "step": 1751
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0745755907783694,
      "learning_rate": 1.3755643338001411e-06,
      "loss": 0.1658,
      "step": 1752
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3455316523465097,
      "learning_rate": 1.3706937718685654e-06,
      "loss": 0.2383,
      "step": 1753
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3428879720147906,
      "learning_rate": 1.3658304779774784e-06,
      "loss": 0.239,
      "step": 1754
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2587739129649136,
      "learning_rate": 1.3609744618661013e-06,
      "loss": 0.1744,
      "step": 1755
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2500301161424654,
      "learning_rate": 1.3561257332590816e-06,
      "loss": 0.1876,
      "step": 1756
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2900191701007295,
      "learning_rate": 1.3512843018664728e-06,
      "loss": 0.2115,
      "step": 1757
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1936073158095306,
      "learning_rate": 1.3464501773837124e-06,
      "loss": 0.1777,
      "step": 1758
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2232208400437021,
      "learning_rate": 1.3416233694916086e-06,
      "loss": 0.1778,
      "step": 1759
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2230166015810229,
      "learning_rate": 1.3368038878563154e-06,
      "loss": 0.1522,
      "step": 1760
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3430526731275074,
      "learning_rate": 1.3319917421293182e-06,
      "loss": 0.1789,
      "step": 1761
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2552017280293333,
      "learning_rate": 1.3271869419474059e-06,
      "loss": 0.2179,
      "step": 1762
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.128058212989182,
      "learning_rate": 1.3223894969326623e-06,
      "loss": 0.1462,
      "step": 1763
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4136471605496954,
      "learning_rate": 1.3175994166924394e-06,
      "loss": 0.2264,
      "step": 1764
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3767774981670622,
      "learning_rate": 1.312816710819343e-06,
      "loss": 0.1301,
      "step": 1765
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.478722316735654,
      "learning_rate": 1.3080413888912057e-06,
      "loss": 0.2162,
      "step": 1766
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3504115469546392,
      "learning_rate": 1.3032734604710784e-06,
      "loss": 0.1804,
      "step": 1767
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4118799789055272,
      "learning_rate": 1.298512935107204e-06,
      "loss": 0.1781,
      "step": 1768
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3062807250574344,
      "learning_rate": 1.2937598223330006e-06,
      "loss": 0.2307,
      "step": 1769
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2952821042088734,
      "learning_rate": 1.289014131667039e-06,
      "loss": 0.1818,
      "step": 1770
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.509314454286657,
      "learning_rate": 1.2842758726130283e-06,
      "loss": 0.2311,
      "step": 1771
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2301652221593593,
      "learning_rate": 1.2795450546597999e-06,
      "loss": 0.1825,
      "step": 1772
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5097625172616274,
      "learning_rate": 1.2748216872812747e-06,
      "loss": 0.1755,
      "step": 1773
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2883982553370696,
      "learning_rate": 1.2701057799364591e-06,
      "loss": 0.1698,
      "step": 1774
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5333581907196376,
      "learning_rate": 1.2653973420694182e-06,
      "loss": 0.1813,
      "step": 1775
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0977079632570401,
      "learning_rate": 1.26069638310926e-06,
      "loss": 0.1611,
      "step": 1776
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2831364079556953,
      "learning_rate": 1.256002912470115e-06,
      "loss": 0.1996,
      "step": 1777
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3871232764432957,
      "learning_rate": 1.2513169395511143e-06,
      "loss": 0.2224,
      "step": 1778
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3097709211052897,
      "learning_rate": 1.246638473736378e-06,
      "loss": 0.2035,
      "step": 1779
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2585891580877602,
      "learning_rate": 1.2419675243949914e-06,
      "loss": 0.1731,
      "step": 1780
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4280778527160025,
      "learning_rate": 1.2373041008809888e-06,
      "loss": 0.2566,
      "step": 1781
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1671393507935142,
      "learning_rate": 1.2326482125333284e-06,
      "loss": 0.1741,
      "step": 1782
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3887562079953006,
      "learning_rate": 1.2279998686758833e-06,
      "loss": 0.1576,
      "step": 1783
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1899081533867129,
      "learning_rate": 1.223359078617416e-06,
      "loss": 0.1732,
      "step": 1784
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4106502881963106,
      "learning_rate": 1.2187258516515644e-06,
      "loss": 0.2688,
      "step": 1785
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.243709524825908,
      "learning_rate": 1.214100197056815e-06,
      "loss": 0.2213,
      "step": 1786
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2122333455683616,
      "learning_rate": 1.2094821240964955e-06,
      "loss": 0.1624,
      "step": 1787
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.229906164409389,
      "learning_rate": 1.2048716420187473e-06,
      "loss": 0.1703,
      "step": 1788
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.303587634515647,
      "learning_rate": 1.2002687600565138e-06,
      "loss": 0.1686,
      "step": 1789
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0645101293038606,
      "learning_rate": 1.1956734874275127e-06,
      "loss": 0.1541,
      "step": 1790
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1719673454806228,
      "learning_rate": 1.1910858333342279e-06,
      "loss": 0.1836,
      "step": 1791
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5399407627071362,
      "learning_rate": 1.1865058069638857e-06,
      "loss": 0.257,
      "step": 1792
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3907923455559228,
      "learning_rate": 1.1819334174884378e-06,
      "loss": 0.2037,
      "step": 1793
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1774271096407545,
      "learning_rate": 1.1773686740645384e-06,
      "loss": 0.1784,
      "step": 1794
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2779374935818797,
      "learning_rate": 1.172811585833532e-06,
      "loss": 0.1567,
      "step": 1795
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5222868158369,
      "learning_rate": 1.168262161921438e-06,
      "loss": 0.2256,
      "step": 1796
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2133438047788971,
      "learning_rate": 1.1637204114389179e-06,
      "loss": 0.133,
      "step": 1797
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4239177083940993,
      "learning_rate": 1.1591863434812722e-06,
      "loss": 0.1916,
      "step": 1798
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3135023315512546,
      "learning_rate": 1.1546599671284158e-06,
      "loss": 0.2081,
      "step": 1799
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2333884839665845,
      "learning_rate": 1.1501412914448596e-06,
      "loss": 0.203,
      "step": 1800
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9927154626465802,
      "learning_rate": 1.1456303254796941e-06,
      "loss": 0.1397,
      "step": 1801
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2941995207863888,
      "learning_rate": 1.141127078266568e-06,
      "loss": 0.1998,
      "step": 1802
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1780324422929405,
      "learning_rate": 1.1366315588236743e-06,
      "loss": 0.1402,
      "step": 1803
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2201697538182048,
      "learning_rate": 1.1321437761537307e-06,
      "loss": 0.1676,
      "step": 1804
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3018060855473053,
      "learning_rate": 1.127663739243962e-06,
      "loss": 0.178,
      "step": 1805
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.416845150383203,
      "learning_rate": 1.1231914570660774e-06,
      "loss": 0.1992,
      "step": 1806
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3077483386624218,
      "learning_rate": 1.1187269385762605e-06,
      "loss": 0.2074,
      "step": 1807
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2026240136254998,
      "learning_rate": 1.1142701927151456e-06,
      "loss": 0.1258,
      "step": 1808
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.089111474818945,
      "learning_rate": 1.1098212284078037e-06,
      "loss": 0.1714,
      "step": 1809
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3626592912801458,
      "learning_rate": 1.1053800545637173e-06,
      "loss": 0.2009,
      "step": 1810
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1230396156416167,
      "learning_rate": 1.1009466800767738e-06,
      "loss": 0.1566,
      "step": 1811
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0858125587348708,
      "learning_rate": 1.0965211138252373e-06,
      "loss": 0.1135,
      "step": 1812
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2730488627364718,
      "learning_rate": 1.09210336467174e-06,
      "loss": 0.1842,
      "step": 1813
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2625009560288838,
      "learning_rate": 1.0876934414632523e-06,
      "loss": 0.1728,
      "step": 1814
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3434178610203948,
      "learning_rate": 1.0832913530310784e-06,
      "loss": 0.1894,
      "step": 1815
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1291776056774816,
      "learning_rate": 1.0788971081908306e-06,
      "loss": 0.2159,
      "step": 1816
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.241963580818774,
      "learning_rate": 1.0745107157424155e-06,
      "loss": 0.1819,
      "step": 1817
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2323936940691738,
      "learning_rate": 1.0701321844700097e-06,
      "loss": 0.21,
      "step": 1818
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2228465402205304,
      "learning_rate": 1.0657615231420492e-06,
      "loss": 0.1641,
      "step": 1819
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2411590578337208,
      "learning_rate": 1.0613987405112153e-06,
      "loss": 0.2129,
      "step": 1820
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2412657792827688,
      "learning_rate": 1.0570438453144044e-06,
      "loss": 0.1656,
      "step": 1821
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.292210142189328,
      "learning_rate": 1.0526968462727189e-06,
      "loss": 0.2268,
      "step": 1822
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3303160358905615,
      "learning_rate": 1.0483577520914495e-06,
      "loss": 0.185,
      "step": 1823
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5957577026848928,
      "learning_rate": 1.0440265714600573e-06,
      "loss": 0.2127,
      "step": 1824
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.058640688695683,
      "learning_rate": 1.0397033130521554e-06,
      "loss": 0.1592,
      "step": 1825
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.304238700863844,
      "learning_rate": 1.0353879855254895e-06,
      "loss": 0.1898,
      "step": 1826
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.485945213127029,
      "learning_rate": 1.0310805975219256e-06,
      "loss": 0.242,
      "step": 1827
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.258828437675275,
      "learning_rate": 1.0267811576674298e-06,
      "loss": 0.1554,
      "step": 1828
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4281343270872078,
      "learning_rate": 1.0224896745720513e-06,
      "loss": 0.2031,
      "step": 1829
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1305972993584896,
      "learning_rate": 1.0182061568299019e-06,
      "loss": 0.163,
      "step": 1830
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1665688448309384,
      "learning_rate": 1.0139306130191463e-06,
      "loss": 0.1489,
      "step": 1831
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3397657460468895,
      "learning_rate": 1.009663051701979e-06,
      "loss": 0.2364,
      "step": 1832
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3388498066341785,
      "learning_rate": 1.0054034814246093e-06,
      "loss": 0.2501,
      "step": 1833
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3230069244297658,
      "learning_rate": 1.0011519107172413e-06,
      "loss": 0.177,
      "step": 1834
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3091812112412538,
      "learning_rate": 9.969083480940617e-07,
      "loss": 0.1983,
      "step": 1835
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2407828477176117,
      "learning_rate": 9.926728020532196e-07,
      "loss": 0.134,
      "step": 1836
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1654803638579565,
      "learning_rate": 9.884452810768115e-07,
      "loss": 0.1319,
      "step": 1837
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.31383026460989,
      "learning_rate": 9.842257936308586e-07,
      "loss": 0.1424,
      "step": 1838
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3261326205381938,
      "learning_rate": 9.80014348165298e-07,
      "loss": 0.2114,
      "step": 1839
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3570551272393518,
      "learning_rate": 9.758109531139614e-07,
      "loss": 0.1443,
      "step": 1840
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2632009889187288,
      "learning_rate": 9.71615616894559e-07,
      "loss": 0.2083,
      "step": 1841
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3781868409936022,
      "learning_rate": 9.6742834790866e-07,
      "loss": 0.1813,
      "step": 1842
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.357159183554714,
      "learning_rate": 9.632491545416788e-07,
      "loss": 0.2136,
      "step": 1843
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2390748417415824,
      "learning_rate": 9.590780451628617e-07,
      "loss": 0.1759,
      "step": 1844
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0887085174910929,
      "learning_rate": 9.549150281252633e-07,
      "loss": 0.1535,
      "step": 1845
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2151270275885326,
      "learning_rate": 9.507601117657289e-07,
      "loss": 0.1706,
      "step": 1846
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.403121519824073,
      "learning_rate": 9.466133044048865e-07,
      "loss": 0.1601,
      "step": 1847
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1896777597860735,
      "learning_rate": 9.424746143471231e-07,
      "loss": 0.183,
      "step": 1848
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.282295429829092,
      "learning_rate": 9.383440498805712e-07,
      "loss": 0.2203,
      "step": 1849
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.259323411380295,
      "learning_rate": 9.34221619277087e-07,
      "loss": 0.2051,
      "step": 1850
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.167843818494609,
      "learning_rate": 9.301073307922432e-07,
      "loss": 0.1713,
      "step": 1851
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1761750824900112,
      "learning_rate": 9.260011926653034e-07,
      "loss": 0.2081,
      "step": 1852
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2426936309309375,
      "learning_rate": 9.219032131192124e-07,
      "loss": 0.1835,
      "step": 1853
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1083454900486123,
      "learning_rate": 9.178134003605721e-07,
      "loss": 0.1877,
      "step": 1854
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2375076006702388,
      "learning_rate": 9.137317625796338e-07,
      "loss": 0.1731,
      "step": 1855
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1967042407169066,
      "learning_rate": 9.096583079502763e-07,
      "loss": 0.1935,
      "step": 1856
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3443387358483094,
      "learning_rate": 9.055930446299916e-07,
      "loss": 0.194,
      "step": 1857
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2815288618127072,
      "learning_rate": 9.015359807598645e-07,
      "loss": 0.2259,
      "step": 1858
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2226681812596751,
      "learning_rate": 8.974871244645628e-07,
      "loss": 0.1693,
      "step": 1859
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.292985129629486,
      "learning_rate": 8.934464838523183e-07,
      "loss": 0.1923,
      "step": 1860
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2107517638991738,
      "learning_rate": 8.894140670149082e-07,
      "loss": 0.236,
      "step": 1861
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.261777088784674,
      "learning_rate": 8.853898820276402e-07,
      "loss": 0.2333,
      "step": 1862
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4117682744618694,
      "learning_rate": 8.813739369493395e-07,
      "loss": 0.2228,
      "step": 1863
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2148284714585194,
      "learning_rate": 8.773662398223276e-07,
      "loss": 0.2086,
      "step": 1864
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3220173207901618,
      "learning_rate": 8.733667986724109e-07,
      "loss": 0.2008,
      "step": 1865
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3123532830120657,
      "learning_rate": 8.693756215088617e-07,
      "loss": 0.2199,
      "step": 1866
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3488838845911664,
      "learning_rate": 8.653927163243986e-07,
      "loss": 0.2113,
      "step": 1867
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1874317454231904,
      "learning_rate": 8.61418091095183e-07,
      "loss": 0.2103,
      "step": 1868
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1581049025399375,
      "learning_rate": 8.574517537807897e-07,
      "loss": 0.1216,
      "step": 1869
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.218166783042652,
      "learning_rate": 8.534937123241949e-07,
      "loss": 0.2055,
      "step": 1870
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3469250175488354,
      "learning_rate": 8.495439746517648e-07,
      "loss": 0.1927,
      "step": 1871
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2228236314284617,
      "learning_rate": 8.45602548673235e-07,
      "loss": 0.2339,
      "step": 1872
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4240511594961212,
      "learning_rate": 8.416694422816974e-07,
      "loss": 0.2358,
      "step": 1873
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1980206183818527,
      "learning_rate": 8.377446633535797e-07,
      "loss": 0.194,
      "step": 1874
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2544128345834233,
      "learning_rate": 8.338282197486364e-07,
      "loss": 0.1884,
      "step": 1875
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1627977380329644,
      "learning_rate": 8.299201193099287e-07,
      "loss": 0.1514,
      "step": 1876
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2738494498512514,
      "learning_rate": 8.260203698638114e-07,
      "loss": 0.1959,
      "step": 1877
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3556597874720973,
      "learning_rate": 8.221289792199117e-07,
      "loss": 0.2513,
      "step": 1878
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2313018783550131,
      "learning_rate": 8.182459551711197e-07,
      "loss": 0.2231,
      "step": 1879
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1501533259590175,
      "learning_rate": 8.143713054935715e-07,
      "loss": 0.166,
      "step": 1880
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4060456998228874,
      "learning_rate": 8.105050379466334e-07,
      "loss": 0.2094,
      "step": 1881
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.435912103676395,
      "learning_rate": 8.066471602728804e-07,
      "loss": 0.1914,
      "step": 1882
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3062398628409833,
      "learning_rate": 8.027976801980924e-07,
      "loss": 0.2019,
      "step": 1883
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2504305730554923,
      "learning_rate": 7.989566054312286e-07,
      "loss": 0.2125,
      "step": 1884
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2026508962252367,
      "learning_rate": 7.95123943664417e-07,
      "loss": 0.1441,
      "step": 1885
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.286048527405349,
      "learning_rate": 7.91299702572938e-07,
      "loss": 0.1782,
      "step": 1886
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.181762670369151,
      "learning_rate": 7.87483889815207e-07,
      "loss": 0.1893,
      "step": 1887
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.265289719180237,
      "learning_rate": 7.836765130327622e-07,
      "loss": 0.1713,
      "step": 1888
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2421572851516827,
      "learning_rate": 7.798775798502484e-07,
      "loss": 0.2061,
      "step": 1889
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4349770634594494,
      "learning_rate": 7.760870978754015e-07,
      "loss": 0.2223,
      "step": 1890
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.346583116800994,
      "learning_rate": 7.723050746990291e-07,
      "loss": 0.2215,
      "step": 1891
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.149419435307701,
      "learning_rate": 7.685315178950048e-07,
      "loss": 0.1576,
      "step": 1892
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4069700175355921,
      "learning_rate": 7.647664350202461e-07,
      "loss": 0.2224,
      "step": 1893
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1616807715828747,
      "learning_rate": 7.610098336146965e-07,
      "loss": 0.1644,
      "step": 1894
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2422494318063484,
      "learning_rate": 7.572617212013189e-07,
      "loss": 0.1534,
      "step": 1895
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.254934448974616,
      "learning_rate": 7.535221052860747e-07,
      "loss": 0.2073,
      "step": 1896
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0902745336531452,
      "learning_rate": 7.497909933579117e-07,
      "loss": 0.1304,
      "step": 1897
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4393151949154304,
      "learning_rate": 7.460683928887441e-07,
      "loss": 0.1394,
      "step": 1898
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2436638381436822,
      "learning_rate": 7.423543113334436e-07,
      "loss": 0.1683,
      "step": 1899
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2125877408533787,
      "learning_rate": 7.386487561298223e-07,
      "loss": 0.1657,
      "step": 1900
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3419906684959158,
      "learning_rate": 7.349517346986173e-07,
      "loss": 0.2142,
      "step": 1901
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.315367478711032,
      "learning_rate": 7.312632544434739e-07,
      "loss": 0.1459,
      "step": 1902
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2218850903009488,
      "learning_rate": 7.275833227509344e-07,
      "loss": 0.1926,
      "step": 1903
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3615005296215414,
      "learning_rate": 7.239119469904227e-07,
      "loss": 0.2051,
      "step": 1904
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0064233728407215,
      "learning_rate": 7.202491345142288e-07,
      "loss": 0.1125,
      "step": 1905
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2978861271997346,
      "learning_rate": 7.165948926574895e-07,
      "loss": 0.1803,
      "step": 1906
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1886365746235097,
      "learning_rate": 7.129492287381828e-07,
      "loss": 0.1403,
      "step": 1907
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1291098594919629,
      "learning_rate": 7.093121500571082e-07,
      "loss": 0.1768,
      "step": 1908
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.347150852983031,
      "learning_rate": 7.056836638978698e-07,
      "loss": 0.1429,
      "step": 1909
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0804353250345222,
      "learning_rate": 7.020637775268674e-07,
      "loss": 0.1507,
      "step": 1910
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.137434790784438,
      "learning_rate": 6.984524981932756e-07,
      "loss": 0.1573,
      "step": 1911
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5590346656478187,
      "learning_rate": 6.948498331290349e-07,
      "loss": 0.275,
      "step": 1912
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4190669317281916,
      "learning_rate": 6.912557895488342e-07,
      "loss": 0.1736,
      "step": 1913
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3003425525313195,
      "learning_rate": 6.876703746500984e-07,
      "loss": 0.1901,
      "step": 1914
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3228905803335385,
      "learning_rate": 6.84093595612968e-07,
      "loss": 0.181,
      "step": 1915
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2827454002382066,
      "learning_rate": 6.805254596002959e-07,
      "loss": 0.167,
      "step": 1916
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1184453790831377,
      "learning_rate": 6.769659737576229e-07,
      "loss": 0.2028,
      "step": 1917
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2830105253482098,
      "learning_rate": 6.73415145213166e-07,
      "loss": 0.1692,
      "step": 1918
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3087914346724872,
      "learning_rate": 6.698729810778065e-07,
      "loss": 0.1925,
      "step": 1919
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3818333188632634,
      "learning_rate": 6.663394884450752e-07,
      "loss": 0.1736,
      "step": 1920
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3749191016119608,
      "learning_rate": 6.628146743911374e-07,
      "loss": 0.1912,
      "step": 1921
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3032064373349173,
      "learning_rate": 6.592985459747753e-07,
      "loss": 0.2077,
      "step": 1922
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.313815915207631,
      "learning_rate": 6.55791110237381e-07,
      "loss": 0.1845,
      "step": 1923
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2909048367281875,
      "learning_rate": 6.522923742029374e-07,
      "loss": 0.177,
      "step": 1924
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3297607371863842,
      "learning_rate": 6.488023448780068e-07,
      "loss": 0.1981,
      "step": 1925
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4653384724089213,
      "learning_rate": 6.453210292517114e-07,
      "loss": 0.2041,
      "step": 1926
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2126491103673396,
      "learning_rate": 6.418484342957277e-07,
      "loss": 0.1857,
      "step": 1927
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1847248739366616,
      "learning_rate": 6.383845669642657e-07,
      "loss": 0.1866,
      "step": 1928
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2752935398411473,
      "learning_rate": 6.349294341940593e-07,
      "loss": 0.2089,
      "step": 1929
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1140175695142172,
      "learning_rate": 6.3148304290435e-07,
      "loss": 0.2074,
      "step": 1930
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1171130115667371,
      "learning_rate": 6.280453999968711e-07,
      "loss": 0.1536,
      "step": 1931
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1346056892193983,
      "learning_rate": 6.246165123558401e-07,
      "loss": 0.1687,
      "step": 1932
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1797051963489857,
      "learning_rate": 6.211963868479393e-07,
      "loss": 0.1858,
      "step": 1933
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3673216664142904,
      "learning_rate": 6.177850303223059e-07,
      "loss": 0.2033,
      "step": 1934
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3876440667897372,
      "learning_rate": 6.143824496105122e-07,
      "loss": 0.2488,
      "step": 1935
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3726093834821678,
      "learning_rate": 6.109886515265595e-07,
      "loss": 0.246,
      "step": 1936
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.319549023374216,
      "learning_rate": 6.076036428668613e-07,
      "loss": 0.1535,
      "step": 1937
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1994564108506884,
      "learning_rate": 6.042274304102281e-07,
      "loss": 0.2717,
      "step": 1938
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3099751998160054,
      "learning_rate": 6.008600209178539e-07,
      "loss": 0.1896,
      "step": 1939
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4156301832933946,
      "learning_rate": 5.975014211333053e-07,
      "loss": 0.1962,
      "step": 1940
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.379235557395966,
      "learning_rate": 5.941516377825102e-07,
      "loss": 0.1929,
      "step": 1941
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1948606465675655,
      "learning_rate": 5.908106775737332e-07,
      "loss": 0.1917,
      "step": 1942
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3125569479615604,
      "learning_rate": 5.874785471975753e-07,
      "loss": 0.1497,
      "step": 1943
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3531211981123952,
      "learning_rate": 5.841552533269534e-07,
      "loss": 0.2085,
      "step": 1944
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0932633334641242,
      "learning_rate": 5.808408026170892e-07,
      "loss": 0.1834,
      "step": 1945
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5142494122660635,
      "learning_rate": 5.775352017054919e-07,
      "loss": 0.1868,
      "step": 1946
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.059804146801211,
      "learning_rate": 5.742384572119519e-07,
      "loss": 0.1607,
      "step": 1947
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1135029041439424,
      "learning_rate": 5.70950575738522e-07,
      "loss": 0.1747,
      "step": 1948
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.199006602806134,
      "learning_rate": 5.676715638695063e-07,
      "loss": 0.2165,
      "step": 1949
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.231534541927178,
      "learning_rate": 5.644014281714483e-07,
      "loss": 0.1399,
      "step": 1950
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2517469147924858,
      "learning_rate": 5.611401751931112e-07,
      "loss": 0.1837,
      "step": 1951
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.102110248319947,
      "learning_rate": 5.578878114654746e-07,
      "loss": 0.1606,
      "step": 1952
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2124819735729768,
      "learning_rate": 5.546443435017146e-07,
      "loss": 0.2241,
      "step": 1953
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3968296738213606,
      "learning_rate": 5.514097777971939e-07,
      "loss": 0.1831,
      "step": 1954
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4616723183434965,
      "learning_rate": 5.481841208294447e-07,
      "loss": 0.2279,
      "step": 1955
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2916089872930405,
      "learning_rate": 5.449673790581611e-07,
      "loss": 0.1825,
      "step": 1956
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.318282976618061,
      "learning_rate": 5.417595589251828e-07,
      "loss": 0.1845,
      "step": 1957
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3488100114917823,
      "learning_rate": 5.385606668544846e-07,
      "loss": 0.1928,
      "step": 1958
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2547335540201183,
      "learning_rate": 5.353707092521581e-07,
      "loss": 0.1955,
      "step": 1959
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1701814041131633,
      "learning_rate": 5.321896925064069e-07,
      "loss": 0.2096,
      "step": 1960
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3259695057410739,
      "learning_rate": 5.290176229875271e-07,
      "loss": 0.1747,
      "step": 1961
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.16400928612747,
      "learning_rate": 5.258545070478999e-07,
      "loss": 0.1353,
      "step": 1962
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.446616319904846,
      "learning_rate": 5.227003510219714e-07,
      "loss": 0.2127,
      "step": 1963
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.286185862704154,
      "learning_rate": 5.195551612262478e-07,
      "loss": 0.2422,
      "step": 1964
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4743599915363579,
      "learning_rate": 5.16418943959282e-07,
      "loss": 0.2173,
      "step": 1965
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3966270262741722,
      "learning_rate": 5.132917055016523e-07,
      "loss": 0.2564,
      "step": 1966
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0410315522478588,
      "learning_rate": 5.101734521159613e-07,
      "loss": 0.1528,
      "step": 1967
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.37521194412789,
      "learning_rate": 5.070641900468149e-07,
      "loss": 0.2037,
      "step": 1968
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2886098350655402,
      "learning_rate": 5.039639255208156e-07,
      "loss": 0.1982,
      "step": 1969
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.182912628162159,
      "learning_rate": 5.008726647465468e-07,
      "loss": 0.2002,
      "step": 1970
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2263063041602589,
      "learning_rate": 4.977904139145579e-07,
      "loss": 0.208,
      "step": 1971
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5139200237079866,
      "learning_rate": 4.947171791973587e-07,
      "loss": 0.2171,
      "step": 1972
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2648065594204205,
      "learning_rate": 4.916529667494024e-07,
      "loss": 0.1397,
      "step": 1973
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1288137023604738,
      "learning_rate": 4.885977827070748e-07,
      "loss": 0.1304,
      "step": 1974
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2718568874288427,
      "learning_rate": 4.855516331886783e-07,
      "loss": 0.1951,
      "step": 1975
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3296415135484243,
      "learning_rate": 4.825145242944262e-07,
      "loss": 0.2328,
      "step": 1976
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3888862720299298,
      "learning_rate": 4.794864621064266e-07,
      "loss": 0.1753,
      "step": 1977
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3213319436199353,
      "learning_rate": 4.7646745268867034e-07,
      "loss": 0.2399,
      "step": 1978
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2333503447001917,
      "learning_rate": 4.734575020870169e-07,
      "loss": 0.1874,
      "step": 1979
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.209390686764583,
      "learning_rate": 4.704566163291879e-07,
      "loss": 0.1435,
      "step": 1980
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3091664319877228,
      "learning_rate": 4.6746480142475015e-07,
      "loss": 0.1939,
      "step": 1981
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0464161679088697,
      "learning_rate": 4.6448206336510646e-07,
      "loss": 0.1381,
      "step": 1982
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3710687693363783,
      "learning_rate": 4.6150840812348e-07,
      "loss": 0.2438,
      "step": 1983
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1509485517177542,
      "learning_rate": 4.5854384165490596e-07,
      "loss": 0.1538,
      "step": 1984
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2432655602925171,
      "learning_rate": 4.5558836989621933e-07,
      "loss": 0.1483,
      "step": 1985
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1909678514556068,
      "learning_rate": 4.526419987660419e-07,
      "loss": 0.1932,
      "step": 1986
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2685267376405467,
      "learning_rate": 4.497047341647676e-07,
      "loss": 0.1984,
      "step": 1987
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2241316267185736,
      "learning_rate": 4.467765819745562e-07,
      "loss": 0.1761,
      "step": 1988
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3830124887444974,
      "learning_rate": 4.43857548059321e-07,
      "loss": 0.2439,
      "step": 1989
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3422229874878324,
      "learning_rate": 4.409476382647121e-07,
      "loss": 0.1922,
      "step": 1990
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2780503551242646,
      "learning_rate": 4.380468584181069e-07,
      "loss": 0.1697,
      "step": 1991
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4188385377083876,
      "learning_rate": 4.351552143286014e-07,
      "loss": 0.1771,
      "step": 1992
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4165626863317407,
      "learning_rate": 4.322727117869951e-07,
      "loss": 0.2459,
      "step": 1993
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2479007358579757,
      "learning_rate": 4.293993565657828e-07,
      "loss": 0.2114,
      "step": 1994
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2702705421522464,
      "learning_rate": 4.265351544191365e-07,
      "loss": 0.166,
      "step": 1995
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.279450383108828,
      "learning_rate": 4.2368011108290306e-07,
      "loss": 0.1499,
      "step": 1996
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.314233139356742,
      "learning_rate": 4.208342322745845e-07,
      "loss": 0.1968,
      "step": 1997
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3003940335968336,
      "learning_rate": 4.1799752369333313e-07,
      "loss": 0.1754,
      "step": 1998
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3259737382452104,
      "learning_rate": 4.151699910199336e-07,
      "loss": 0.2223,
      "step": 1999
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4832440058381506,
      "learning_rate": 4.123516399167965e-07,
      "loss": 0.2192,
      "step": 2000
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3450842246525638,
      "learning_rate": 4.095424760279454e-07,
      "loss": 0.2131,
      "step": 2001
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4208047899871914,
      "learning_rate": 4.067425049790075e-07,
      "loss": 0.1962,
      "step": 2002
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1945485461550016,
      "learning_rate": 4.039517323771958e-07,
      "loss": 0.1468,
      "step": 2003
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.173695489482996,
      "learning_rate": 4.0117016381130636e-07,
      "loss": 0.1688,
      "step": 2004
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2140543576624758,
      "learning_rate": 3.983978048517029e-07,
      "loss": 0.1884,
      "step": 2005
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2318507778891028,
      "learning_rate": 3.956346610503048e-07,
      "loss": 0.226,
      "step": 2006
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1851185171832812,
      "learning_rate": 3.9288073794057634e-07,
      "loss": 0.1579,
      "step": 2007
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1825710351829433,
      "learning_rate": 3.9013604103751933e-07,
      "loss": 0.1484,
      "step": 2008
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.331756812111433,
      "learning_rate": 3.87400575837657e-07,
      "loss": 0.1784,
      "step": 2009
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3749818791617356,
      "learning_rate": 3.8467434781902624e-07,
      "loss": 0.2253,
      "step": 2010
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.218211547712307,
      "learning_rate": 3.819573624411643e-07,
      "loss": 0.1863,
      "step": 2011
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.116587794734974,
      "learning_rate": 3.792496251451e-07,
      "loss": 0.2057,
      "step": 2012
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3693277972400748,
      "learning_rate": 3.765511413533429e-07,
      "loss": 0.2145,
      "step": 2013
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3392854767531168,
      "learning_rate": 3.7386191646987094e-07,
      "loss": 0.2507,
      "step": 2014
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1685837645401649,
      "learning_rate": 3.711819558801183e-07,
      "loss": 0.1949,
      "step": 2015
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2550272734732957,
      "learning_rate": 3.685112649509681e-07,
      "loss": 0.2083,
      "step": 2016
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0840986894727525,
      "learning_rate": 3.6584984903074104e-07,
      "loss": 0.1383,
      "step": 2017
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2549413123516158,
      "learning_rate": 3.631977134491832e-07,
      "loss": 0.1991,
      "step": 2018
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0883281206522657,
      "learning_rate": 3.6055486351745327e-07,
      "loss": 0.1527,
      "step": 2019
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2424368827617824,
      "learning_rate": 3.5792130452811825e-07,
      "loss": 0.1343,
      "step": 2020
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2591685299300674,
      "learning_rate": 3.5529704175513604e-07,
      "loss": 0.2001,
      "step": 2021
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2520001203204056,
      "learning_rate": 3.5268208045385167e-07,
      "loss": 0.2325,
      "step": 2022
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.352354085681633,
      "learning_rate": 3.500764258609779e-07,
      "loss": 0.2007,
      "step": 2023
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1325920673407366,
      "learning_rate": 3.4748008319459457e-07,
      "loss": 0.2105,
      "step": 2024
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2751121576599556,
      "learning_rate": 3.4489305765413097e-07,
      "loss": 0.1948,
      "step": 2025
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2197346339791628,
      "learning_rate": 3.4231535442035947e-07,
      "loss": 0.1408,
      "step": 2026
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6160860610662047,
      "learning_rate": 3.3974697865538195e-07,
      "loss": 0.1695,
      "step": 2027
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3836305043620842,
      "learning_rate": 3.37187935502622e-07,
      "loss": 0.2147,
      "step": 2028
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2208176572414642,
      "learning_rate": 3.346382300868134e-07,
      "loss": 0.1754,
      "step": 2029
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2301202472858639,
      "learning_rate": 3.320978675139919e-07,
      "loss": 0.1854,
      "step": 2030
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.263004595929867,
      "learning_rate": 3.2956685287148014e-07,
      "loss": 0.1616,
      "step": 2031
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.411099629677307,
      "learning_rate": 3.2704519122788324e-07,
      "loss": 0.1873,
      "step": 2032
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3764933024324102,
      "learning_rate": 3.24532887633075e-07,
      "loss": 0.2314,
      "step": 2033
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2138616353003004,
      "learning_rate": 3.220299471181898e-07,
      "loss": 0.1776,
      "step": 2034
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2333780931974916,
      "learning_rate": 3.195363746956087e-07,
      "loss": 0.1886,
      "step": 2035
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4053546200223321,
      "learning_rate": 3.1705217535895504e-07,
      "loss": 0.2476,
      "step": 2036
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.404414412555447,
      "learning_rate": 3.1457735408308153e-07,
      "loss": 0.1763,
      "step": 2037
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2518004942378818,
      "learning_rate": 3.121119158240604e-07,
      "loss": 0.165,
      "step": 2038
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3190697293916485,
      "learning_rate": 3.096558655191706e-07,
      "loss": 0.1992,
      "step": 2039
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4673219476705341,
      "learning_rate": 3.0720920808689435e-07,
      "loss": 0.1979,
      "step": 2040
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1091518372400866,
      "learning_rate": 3.047719484269018e-07,
      "loss": 0.133,
      "step": 2041
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1855074423224201,
      "learning_rate": 3.023440914200448e-07,
      "loss": 0.1458,
      "step": 2042
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1851932736361306,
      "learning_rate": 2.9992564192834253e-07,
      "loss": 0.1599,
      "step": 2043
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2664632408181773,
      "learning_rate": 2.9751660479497737e-07,
      "loss": 0.1585,
      "step": 2044
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3857931363870346,
      "learning_rate": 2.9511698484428144e-07,
      "loss": 0.2653,
      "step": 2045
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.322073725507458,
      "learning_rate": 2.9272678688172905e-07,
      "loss": 0.1961,
      "step": 2046
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4141188695895708,
      "learning_rate": 2.9034601569392417e-07,
      "loss": 0.2176,
      "step": 2047
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3570332611001996,
      "learning_rate": 2.879746760485946e-07,
      "loss": 0.2204,
      "step": 2048
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.191614024284905,
      "learning_rate": 2.85612772694579e-07,
      "loss": 0.1902,
      "step": 2049
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.415143264854317,
      "learning_rate": 2.832603103618214e-07,
      "loss": 0.1992,
      "step": 2050
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3905501936929445,
      "learning_rate": 2.809172937613552e-07,
      "loss": 0.1589,
      "step": 2051
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.241016552552454,
      "learning_rate": 2.7858372758530206e-07,
      "loss": 0.2139,
      "step": 2052
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2075487344889542,
      "learning_rate": 2.762596165068548e-07,
      "loss": 0.1971,
      "step": 2053
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1801081241647395,
      "learning_rate": 2.739449651802756e-07,
      "loss": 0.1761,
      "step": 2054
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1142844732281836,
      "learning_rate": 2.7163977824087696e-07,
      "loss": 0.1877,
      "step": 2055
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5388977276175058,
      "learning_rate": 2.6934406030502235e-07,
      "loss": 0.2002,
      "step": 2056
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3915549636561924,
      "learning_rate": 2.6705781597011147e-07,
      "loss": 0.1816,
      "step": 2057
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2109876840520886,
      "learning_rate": 2.6478104981457254e-07,
      "loss": 0.1945,
      "step": 2058
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2245570745179353,
      "learning_rate": 2.6251376639785163e-07,
      "loss": 0.1944,
      "step": 2059
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2817184610088688,
      "learning_rate": 2.6025597026040463e-07,
      "loss": 0.1904,
      "step": 2060
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4350482332135566,
      "learning_rate": 2.5800766592369077e-07,
      "loss": 0.2,
      "step": 2061
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2461221421864084,
      "learning_rate": 2.5576885789015905e-07,
      "loss": 0.2065,
      "step": 2062
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2983272041809275,
      "learning_rate": 2.535395506432403e-07,
      "loss": 0.2254,
      "step": 2063
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2813854615670706,
      "learning_rate": 2.5131974864734063e-07,
      "loss": 0.2034,
      "step": 2064
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2333329725467366,
      "learning_rate": 2.4910945634783024e-07,
      "loss": 0.1466,
      "step": 2065
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2973888022263969,
      "learning_rate": 2.469086781710367e-07,
      "loss": 0.2292,
      "step": 2066
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5549615171970879,
      "learning_rate": 2.447174185242324e-07,
      "loss": 0.2391,
      "step": 2067
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1801882369883758,
      "learning_rate": 2.425356817956287e-07,
      "loss": 0.1886,
      "step": 2068
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1887484446379255,
      "learning_rate": 2.403634723543674e-07,
      "loss": 0.2322,
      "step": 2069
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2422516998153816,
      "learning_rate": 2.3820079455051048e-07,
      "loss": 0.1915,
      "step": 2070
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2267287594551057,
      "learning_rate": 2.3604765271503073e-07,
      "loss": 0.1603,
      "step": 2071
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3650363237070269,
      "learning_rate": 2.3390405115980563e-07,
      "loss": 0.1894,
      "step": 2072
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1984864407565745,
      "learning_rate": 2.3176999417760637e-07,
      "loss": 0.1465,
      "step": 2073
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1773665320610749,
      "learning_rate": 2.2964548604209214e-07,
      "loss": 0.1704,
      "step": 2074
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2748471704259836,
      "learning_rate": 2.2753053100779632e-07,
      "loss": 0.1539,
      "step": 2075
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2814478019305457,
      "learning_rate": 2.2542513331012372e-07,
      "loss": 0.2151,
      "step": 2076
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.203603486910237,
      "learning_rate": 2.2332929716533947e-07,
      "loss": 0.2257,
      "step": 2077
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1951579615047756,
      "learning_rate": 2.2124302677055954e-07,
      "loss": 0.1631,
      "step": 2078
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2891595884350577,
      "learning_rate": 2.1916632630374579e-07,
      "loss": 0.201,
      "step": 2079
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4038504352928358,
      "learning_rate": 2.1709919992369211e-07,
      "loss": 0.2433,
      "step": 2080
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3010712619505351,
      "learning_rate": 2.1504165177002212e-07,
      "loss": 0.1588,
      "step": 2081
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3505287364251428,
      "learning_rate": 2.1299368596317705e-07,
      "loss": 0.2049,
      "step": 2082
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1868987916692693,
      "learning_rate": 2.1095530660440954e-07,
      "loss": 0.1705,
      "step": 2083
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1353087608025674,
      "learning_rate": 2.0892651777577045e-07,
      "loss": 0.184,
      "step": 2084
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1259317065162704,
      "learning_rate": 2.069073235401109e-07,
      "loss": 0.1795,
      "step": 2085
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2089263639143648,
      "learning_rate": 2.0489772794106411e-07,
      "loss": 0.1389,
      "step": 2086
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4038094399279901,
      "learning_rate": 2.028977350030409e-07,
      "loss": 0.2294,
      "step": 2087
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.133791982322205,
      "learning_rate": 2.0090734873122243e-07,
      "loss": 0.1187,
      "step": 2088
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2422443525228022,
      "learning_rate": 1.989265731115525e-07,
      "loss": 0.1722,
      "step": 2089
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2334975238286843,
      "learning_rate": 1.969554121107281e-07,
      "loss": 0.1883,
      "step": 2090
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3570167317916277,
      "learning_rate": 1.9499386967619104e-07,
      "loss": 0.2178,
      "step": 2091
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.204521357086859,
      "learning_rate": 1.9304194973612244e-07,
      "loss": 0.2119,
      "step": 2092
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.243900586700778,
      "learning_rate": 1.910996561994316e-07,
      "loss": 0.2322,
      "step": 2093
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2238434898884216,
      "learning_rate": 1.8916699295575324e-07,
      "loss": 0.1361,
      "step": 2094
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2426604674579187,
      "learning_rate": 1.872439638754331e-07,
      "loss": 0.2199,
      "step": 2095
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2610322871715227,
      "learning_rate": 1.853305728095245e-07,
      "loss": 0.1584,
      "step": 2096
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3252863265782022,
      "learning_rate": 1.834268235897807e-07,
      "loss": 0.1905,
      "step": 2097
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2190231523686506,
      "learning_rate": 1.8153272002864542e-07,
      "loss": 0.2139,
      "step": 2098
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.238179559841821,
      "learning_rate": 1.7964826591924722e-07,
      "loss": 0.212,
      "step": 2099
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2382479582414045,
      "learning_rate": 1.7777346503538796e-07,
      "loss": 0.1842,
      "step": 2100
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4060577457907,
      "learning_rate": 1.7590832113154045e-07,
      "loss": 0.2057,
      "step": 2101
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3154742536993618,
      "learning_rate": 1.7405283794283745e-07,
      "loss": 0.2141,
      "step": 2102
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2464253595827326,
      "learning_rate": 1.7220701918506666e-07,
      "loss": 0.1236,
      "step": 2103
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1393875266792188,
      "learning_rate": 1.7037086855465902e-07,
      "loss": 0.2427,
      "step": 2104
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.315553485944377,
      "learning_rate": 1.685443897286859e-07,
      "loss": 0.1918,
      "step": 2105
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2803499602983919,
      "learning_rate": 1.6672758636485043e-07,
      "loss": 0.2231,
      "step": 2106
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.210086258686365,
      "learning_rate": 1.6492046210147884e-07,
      "loss": 0.1841,
      "step": 2107
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1840340621548233,
      "learning_rate": 1.6312302055751295e-07,
      "loss": 0.1479,
      "step": 2108
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2766742370321154,
      "learning_rate": 1.6133526533250566e-07,
      "loss": 0.2022,
      "step": 2109
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3367490361980268,
      "learning_rate": 1.5955720000661312e-07,
      "loss": 0.2113,
      "step": 2110
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.178975782949385,
      "learning_rate": 1.577888281405826e-07,
      "loss": 0.1483,
      "step": 2111
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1533786064076177,
      "learning_rate": 1.5603015327575354e-07,
      "loss": 0.1285,
      "step": 2112
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2794335225791775,
      "learning_rate": 1.5428117893404315e-07,
      "loss": 0.2062,
      "step": 2113
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3376416522330679,
      "learning_rate": 1.5254190861794415e-07,
      "loss": 0.1845,
      "step": 2114
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2905090456096535,
      "learning_rate": 1.5081234581051484e-07,
      "loss": 0.1693,
      "step": 2115
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.364779033529992,
      "learning_rate": 1.490924939753735e-07,
      "loss": 0.2519,
      "step": 2116
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1891452862655765,
      "learning_rate": 1.473823565566923e-07,
      "loss": 0.1967,
      "step": 2117
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0191745694542733,
      "learning_rate": 1.4568193697918843e-07,
      "loss": 0.185,
      "step": 2118
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1492425464344098,
      "learning_rate": 1.4399123864811904e-07,
      "loss": 0.1954,
      "step": 2119
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2863846488856365,
      "learning_rate": 1.4231026494927082e-07,
      "loss": 0.1876,
      "step": 2120
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2423047460583008,
      "learning_rate": 1.4063901924895985e-07,
      "loss": 0.2302,
      "step": 2121
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1285060362424193,
      "learning_rate": 1.3897750489401785e-07,
      "loss": 0.1815,
      "step": 2122
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1715245602697253,
      "learning_rate": 1.3732572521179043e-07,
      "loss": 0.1333,
      "step": 2123
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2691171026674515,
      "learning_rate": 1.3568368351012718e-07,
      "loss": 0.1962,
      "step": 2124
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1564453190399404,
      "learning_rate": 1.3405138307737765e-07,
      "loss": 0.1774,
      "step": 2125
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4461229159253561,
      "learning_rate": 1.324288271823826e-07,
      "loss": 0.1835,
      "step": 2126
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.125141280407893,
      "learning_rate": 1.3081601907447007e-07,
      "loss": 0.1651,
      "step": 2127
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2440918821783193,
      "learning_rate": 1.2921296198344424e-07,
      "loss": 0.2059,
      "step": 2128
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2191193460163952,
      "learning_rate": 1.2761965911958385e-07,
      "loss": 0.1888,
      "step": 2129
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4053605073943205,
      "learning_rate": 1.260361136736349e-07,
      "loss": 0.2221,
      "step": 2130
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4014856242373879,
      "learning_rate": 1.244623288168012e-07,
      "loss": 0.195,
      "step": 2131
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3477977656615363,
      "learning_rate": 1.2289830770074064e-07,
      "loss": 0.1676,
      "step": 2132
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.227271178558234,
      "learning_rate": 1.2134405345755773e-07,
      "loss": 0.1906,
      "step": 2133
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3434318210918845,
      "learning_rate": 1.1979956919979996e-07,
      "loss": 0.1411,
      "step": 2134
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4950289060623951,
      "learning_rate": 1.1826485802044707e-07,
      "loss": 0.2124,
      "step": 2135
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2975032120948635,
      "learning_rate": 1.1673992299290893e-07,
      "loss": 0.1787,
      "step": 2136
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.269447026934572,
      "learning_rate": 1.1522476717101605e-07,
      "loss": 0.2174,
      "step": 2137
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2702432541717803,
      "learning_rate": 1.137193935890174e-07,
      "loss": 0.1621,
      "step": 2138
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.230129191650044,
      "learning_rate": 1.1222380526156929e-07,
      "loss": 0.1867,
      "step": 2139
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2822221602208423,
      "learning_rate": 1.1073800518373423e-07,
      "loss": 0.187,
      "step": 2140
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.213845790487509,
      "learning_rate": 1.0926199633097156e-07,
      "loss": 0.2001,
      "step": 2141
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3177922444749732,
      "learning_rate": 1.0779578165913463e-07,
      "loss": 0.2057,
      "step": 2142
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2451168178563221,
      "learning_rate": 1.0633936410446077e-07,
      "loss": 0.1699,
      "step": 2143
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2823270411599066,
      "learning_rate": 1.0489274658356808e-07,
      "loss": 0.1751,
      "step": 2144
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0237105000464255,
      "learning_rate": 1.0345593199344972e-07,
      "loss": 0.1668,
      "step": 2145
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1883665672253314,
      "learning_rate": 1.0202892321146796e-07,
      "loss": 0.1599,
      "step": 2146
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2202100965986942,
      "learning_rate": 1.0061172309534739e-07,
      "loss": 0.1705,
      "step": 2147
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1645264321408326,
      "learning_rate": 9.92043344831689e-08,
      "loss": 0.1702,
      "step": 2148
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2774428084546872,
      "learning_rate": 9.780676019336632e-08,
      "loss": 0.1531,
      "step": 2149
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3767493293621793,
      "learning_rate": 9.641900302471918e-08,
      "loss": 0.1798,
      "step": 2150
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.242986176881077,
      "learning_rate": 9.504106575634664e-08,
      "loss": 0.1557,
      "step": 2151
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5435638892595602,
      "learning_rate": 9.367295114770191e-08,
      "loss": 0.2143,
      "step": 2152
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4223499779612792,
      "learning_rate": 9.231466193856897e-08,
      "loss": 0.2395,
      "step": 2153
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2724425389942924,
      "learning_rate": 9.096620084905472e-08,
      "loss": 0.1611,
      "step": 2154
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1607024699674449,
      "learning_rate": 8.96275705795846e-08,
      "loss": 0.2144,
      "step": 2155
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0858609695733623,
      "learning_rate": 8.829877381089536e-08,
      "loss": 0.1793,
      "step": 2156
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.248966818892033,
      "learning_rate": 8.697981320403337e-08,
      "loss": 0.1476,
      "step": 2157
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3053472714599363,
      "learning_rate": 8.56706914003469e-08,
      "loss": 0.1985,
      "step": 2158
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3087178328668159,
      "learning_rate": 8.437141102147883e-08,
      "loss": 0.2314,
      "step": 2159
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3591384680921896,
      "learning_rate": 8.308197466936619e-08,
      "loss": 0.2168,
      "step": 2160
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0969583729172139,
      "learning_rate": 8.18023849262306e-08,
      "loss": 0.1486,
      "step": 2161
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2305682707580319,
      "learning_rate": 8.05326443545762e-08,
      "loss": 0.2,
      "step": 2162
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3392378983111652,
      "learning_rate": 7.927275549718228e-08,
      "loss": 0.304,
      "step": 2163
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3510061955352857,
      "learning_rate": 7.802272087709951e-08,
      "loss": 0.1985,
      "step": 2164
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.187405961894408,
      "learning_rate": 7.678254299764431e-08,
      "loss": 0.1767,
      "step": 2165
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2646623909210142,
      "learning_rate": 7.555222434239395e-08,
      "loss": 0.1778,
      "step": 2166
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2962652546743378,
      "learning_rate": 7.433176737518255e-08,
      "loss": 0.1774,
      "step": 2167
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2127745344331717,
      "learning_rate": 7.312117454009337e-08,
      "loss": 0.1779,
      "step": 2168
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.38217562004809,
      "learning_rate": 7.192044826145772e-08,
      "loss": 0.1293,
      "step": 2169
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1895482770866936,
      "learning_rate": 7.072959094384713e-08,
      "loss": 0.1813,
      "step": 2170
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.435233322776814,
      "learning_rate": 6.954860497207006e-08,
      "loss": 0.2323,
      "step": 2171
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.383240719125952,
      "learning_rate": 6.837749271116578e-08,
      "loss": 0.2234,
      "step": 2172
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.335960591207556,
      "learning_rate": 6.721625650640163e-08,
      "loss": 0.2264,
      "step": 2173
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.297403733052241,
      "learning_rate": 6.606489868326571e-08,
      "loss": 0.1651,
      "step": 2174
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1995449325745497,
      "learning_rate": 6.49234215474659e-08,
      "loss": 0.1939,
      "step": 2175
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1835483330860597,
      "learning_rate": 6.379182738492085e-08,
      "loss": 0.1888,
      "step": 2176
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3777062339507429,
      "learning_rate": 6.267011846175896e-08,
      "loss": 0.2018,
      "step": 2177
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2734706375844962,
      "learning_rate": 6.15582970243117e-08,
      "loss": 0.1676,
      "step": 2178
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2079609111497778,
      "learning_rate": 6.045636529911025e-08,
      "loss": 0.197,
      "step": 2179
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3522291066863223,
      "learning_rate": 5.9364325492880514e-08,
      "loss": 0.2358,
      "step": 2180
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2280779858300284,
      "learning_rate": 5.82821797925387e-08,
      "loss": 0.2513,
      "step": 2181
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2677222591095199,
      "learning_rate": 5.720993036518796e-08,
      "loss": 0.1844,
      "step": 2182
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2857432688954609,
      "learning_rate": 5.6147579358112324e-08,
      "loss": 0.1985,
      "step": 2183
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.254422009164354,
      "learning_rate": 5.509512889877333e-08,
      "loss": 0.209,
      "step": 2184
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1545765040289746,
      "learning_rate": 5.4052581094806713e-08,
      "loss": 0.1782,
      "step": 2185
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2207683595262873,
      "learning_rate": 5.301993803401517e-08,
      "loss": 0.1918,
      "step": 2186
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3917967806655331,
      "learning_rate": 5.19972017843684e-08,
      "loss": 0.217,
      "step": 2187
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.586118406708269,
      "learning_rate": 5.098437439399473e-08,
      "loss": 0.221,
      "step": 2188
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2668612232113428,
      "learning_rate": 4.998145789118114e-08,
      "loss": 0.1925,
      "step": 2189
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0923561868701013,
      "learning_rate": 4.898845428436494e-08,
      "loss": 0.1785,
      "step": 2190
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2512490535851164,
      "learning_rate": 4.800536556213375e-08,
      "loss": 0.1915,
      "step": 2191
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4235510655460824,
      "learning_rate": 4.7032193693217765e-08,
      "loss": 0.233,
      "step": 2192
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1516617418323452,
      "learning_rate": 4.60689406264897e-08,
      "loss": 0.1464,
      "step": 2193
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2648034925169211,
      "learning_rate": 4.511560829095818e-08,
      "loss": 0.1844,
      "step": 2194
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3145972612223862,
      "learning_rate": 4.417219859576383e-08,
      "loss": 0.2073,
      "step": 2195
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4890614692748236,
      "learning_rate": 4.3238713430176494e-08,
      "loss": 0.2051,
      "step": 2196
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3547532766040637,
      "learning_rate": 4.231515466359193e-08,
      "loss": 0.178,
      "step": 2197
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3113682768114183,
      "learning_rate": 4.140152414552678e-08,
      "loss": 0.1763,
      "step": 2198
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3263993854829428,
      "learning_rate": 4.0497823705615836e-08,
      "loss": 0.2031,
      "step": 2199
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2489182115268633,
      "learning_rate": 3.960405515360699e-08,
      "loss": 0.1605,
      "step": 2200
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2989420406947476,
      "learning_rate": 3.8720220279359065e-08,
      "loss": 0.2002,
      "step": 2201
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2326121180397507,
      "learning_rate": 3.784632085283846e-08,
      "loss": 0.1687,
      "step": 2202
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3081345582190995,
      "learning_rate": 3.698235862411359e-08,
      "loss": 0.1713,
      "step": 2203
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.199632402294394,
      "learning_rate": 3.6128335323353804e-08,
      "loss": 0.2601,
      "step": 2204
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2189195878132228,
      "learning_rate": 3.528425266082325e-08,
      "loss": 0.2101,
      "step": 2205
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3001739731282322,
      "learning_rate": 3.445011232688089e-08,
      "loss": 0.1966,
      "step": 2206
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2041198126808226,
      "learning_rate": 3.36259159919744e-08,
      "loss": 0.2028,
      "step": 2207
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.284258706201361,
      "learning_rate": 3.281166530663682e-08,
      "loss": 0.208,
      "step": 2208
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2590461340402483,
      "learning_rate": 3.2007361901485455e-08,
      "loss": 0.1991,
      "step": 2209
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.159007804247797,
      "learning_rate": 3.121300738721578e-08,
      "loss": 0.1467,
      "step": 2210
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2413395129817908,
      "learning_rate": 3.042860335460085e-08,
      "loss": 0.2009,
      "step": 2211
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2596791831083973,
      "learning_rate": 2.9654151374485797e-08,
      "loss": 0.1597,
      "step": 2212
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.263832636307367,
      "learning_rate": 2.8889652997787233e-08,
      "loss": 0.1784,
      "step": 2213
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2513698948919239,
      "learning_rate": 2.8135109755487723e-08,
      "loss": 0.2084,
      "step": 2214
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3931115042876958,
      "learning_rate": 2.7390523158633552e-08,
      "loss": 0.2896,
      "step": 2215
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2849576656064527,
      "learning_rate": 2.6655894698331962e-08,
      "loss": 0.1888,
      "step": 2216
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1981855629515394,
      "learning_rate": 2.593122584574892e-08,
      "loss": 0.1321,
      "step": 2217
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3666010267027464,
      "learning_rate": 2.5216518052104122e-08,
      "loss": 0.1755,
      "step": 2218
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.185108510362184,
      "learning_rate": 2.4511772748669894e-08,
      "loss": 0.2148,
      "step": 2219
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1985486500887188,
      "learning_rate": 2.381699134676785e-08,
      "loss": 0.15,
      "step": 2220
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1009489672612136,
      "learning_rate": 2.3132175237765564e-08,
      "loss": 0.1365,
      "step": 2221
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0539605425336063,
      "learning_rate": 2.2457325793073803e-08,
      "loss": 0.128,
      "step": 2222
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1473219971988382,
      "learning_rate": 2.179244436414485e-08,
      "loss": 0.1948,
      "step": 2223
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.358669381972583,
      "learning_rate": 2.1137532282469176e-08,
      "loss": 0.2546,
      "step": 2224
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.193620727140209,
      "learning_rate": 2.0492590859571558e-08,
      "loss": 0.1593,
      "step": 2225
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.254880800852633,
      "learning_rate": 1.9857621387010528e-08,
      "loss": 0.2494,
      "step": 2226
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.34244442490662,
      "learning_rate": 1.9232625136374472e-08,
      "loss": 0.1629,
      "step": 2227
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.193350915624481,
      "learning_rate": 1.8617603359279977e-08,
      "loss": 0.1513,
      "step": 2228
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.333881040514611,
      "learning_rate": 1.8012557287367394e-08,
      "loss": 0.2684,
      "step": 2229
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2757803629211668,
      "learning_rate": 1.7417488132301375e-08,
      "loss": 0.1993,
      "step": 2230
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.200514810971319,
      "learning_rate": 1.6832397085765893e-08,
      "loss": 0.1734,
      "step": 2231
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3201553316556656,
      "learning_rate": 1.6257285319462578e-08,
      "loss": 0.183,
      "step": 2232
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3544238971647777,
      "learning_rate": 1.5692153985109037e-08,
      "loss": 0.2291,
      "step": 2233
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4667820113526733,
      "learning_rate": 1.513700421443609e-08,
      "loss": 0.1295,
      "step": 2234
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1689219003973155,
      "learning_rate": 1.4591837119186104e-08,
      "loss": 0.1898,
      "step": 2235
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2854415634181675,
      "learning_rate": 1.4056653791108543e-08,
      "loss": 0.2248,
      "step": 2236
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1842639673626645,
      "learning_rate": 1.3531455301960539e-08,
      "loss": 0.1785,
      "step": 2237
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1080567158644805,
      "learning_rate": 1.3016242703503545e-08,
      "loss": 0.1733,
      "step": 2238
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2190311785117172,
      "learning_rate": 1.2511017027501682e-08,
      "loss": 0.2058,
      "step": 2239
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3046774291162564,
      "learning_rate": 1.2015779285717844e-08,
      "loss": 0.2186,
      "step": 2240
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2904461349464946,
      "learning_rate": 1.1530530469914258e-08,
      "loss": 0.2171,
      "step": 2241
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.372035875112899,
      "learning_rate": 1.1055271551849156e-08,
      "loss": 0.2016,
      "step": 2242
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3340549226769352,
      "learning_rate": 1.0590003483275657e-08,
      "loss": 0.2243,
      "step": 2243
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4206942296161098,
      "learning_rate": 1.0134727195937332e-08,
      "loss": 0.1733,
      "step": 2244
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1913944226587803,
      "learning_rate": 9.689443601570425e-09,
      "loss": 0.1596,
      "step": 2245
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3243409632404697,
      "learning_rate": 9.254153591898852e-09,
      "loss": 0.2018,
      "step": 2246
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.193698324460147,
      "learning_rate": 8.828858038632538e-09,
      "loss": 0.1514,
      "step": 2247
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2945381884511842,
      "learning_rate": 8.413557793467974e-09,
      "loss": 0.1833,
      "step": 2248
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2339694938329357,
      "learning_rate": 8.008253688084888e-09,
      "loss": 0.1553,
      "step": 2249
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3092397437807914,
      "learning_rate": 7.612946534143461e-09,
      "loss": 0.184,
      "step": 2250
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0603996850598718,
      "learning_rate": 7.227637123285447e-09,
      "loss": 0.1077,
      "step": 2251
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3286167561532818,
      "learning_rate": 6.852326227130835e-09,
      "loss": 0.1752,
      "step": 2252
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2163851780043546,
      "learning_rate": 6.487014597275631e-09,
      "loss": 0.1358,
      "step": 2253
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2219834055384593,
      "learning_rate": 6.1317029652929734e-09,
      "loss": 0.1574,
      "step": 2254
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1478687194614625,
      "learning_rate": 5.7863920427286834e-09,
      "loss": 0.1542,
      "step": 2255
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1039281265054128,
      "learning_rate": 5.45108252110127e-09,
      "loss": 0.1475,
      "step": 2256
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4077657171875981,
      "learning_rate": 5.125775071901373e-09,
      "loss": 0.2434,
      "step": 2257
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.258772804652871,
      "learning_rate": 4.810470346588992e-09,
      "loss": 0.1783,
      "step": 2258
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4760337035153956,
      "learning_rate": 4.505168976592922e-09,
      "loss": 0.1647,
      "step": 2259
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2054960271714068,
      "learning_rate": 4.209871573307989e-09,
      "loss": 0.1914,
      "step": 2260
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3170021950162647,
      "learning_rate": 3.924578728096707e-09,
      "loss": 0.2131,
      "step": 2261
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1718352478509328,
      "learning_rate": 3.6492910122859493e-09,
      "loss": 0.1868,
      "step": 2262
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4215963925209592,
      "learning_rate": 3.3840089771658424e-09,
      "loss": 0.1467,
      "step": 2263
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.204970146434229,
      "learning_rate": 3.1287331539903155e-09,
      "loss": 0.1792,
      "step": 2264
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3047301523359247,
      "learning_rate": 2.8834640539737723e-09,
      "loss": 0.1925,
      "step": 2265
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.407970783923807,
      "learning_rate": 2.648202168291092e-09,
      "loss": 0.2456,
      "step": 2266
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.167676985432377,
      "learning_rate": 2.422947968078182e-09,
      "loss": 0.1486,
      "step": 2267
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5472530525512869,
      "learning_rate": 2.2077019044280946e-09,
      "loss": 0.1761,
      "step": 2268
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2175132276131597,
      "learning_rate": 2.002464408392135e-09,
      "loss": 0.1793,
      "step": 2269
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0689677187822055,
      "learning_rate": 1.8072358909781983e-09,
      "loss": 0.1466,
      "step": 2270
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2603401787729727,
      "learning_rate": 1.622016743150212e-09,
      "loss": 0.2182,
      "step": 2271
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1989919767237363,
      "learning_rate": 1.4468073358286926e-09,
      "loss": 0.1444,
      "step": 2272
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2462840691261128,
      "learning_rate": 1.2816080198868596e-09,
      "loss": 0.2006,
      "step": 2273
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0531259591638367,
      "learning_rate": 1.1264191261528557e-09,
      "loss": 0.1404,
      "step": 2274
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.369653280347741,
      "learning_rate": 9.812409654075261e-10,
      "loss": 0.2456,
      "step": 2275
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3028204273095485,
      "learning_rate": 8.460738283838643e-10,
      "loss": 0.1893,
      "step": 2276
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5729133903999457,
      "learning_rate": 7.209179857675664e-10,
      "loss": 0.2373,
      "step": 2277
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2789896264049463,
      "learning_rate": 6.057736881959209e-10,
      "loss": 0.2303,
      "step": 2278
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.338163365092118,
      "learning_rate": 5.006411662555888e-10,
      "loss": 0.2051,
      "step": 2279
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2560230669406927,
      "learning_rate": 4.0552063048593384e-10,
      "loss": 0.2466,
      "step": 2280
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3179969739134982,
      "learning_rate": 3.204122713740265e-10,
      "loss": 0.2064,
      "step": 2281
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2111829377920944,
      "learning_rate": 2.453162593585301e-10,
      "loss": 0.1696,
      "step": 2282
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1728767784604814,
      "learning_rate": 1.8023274482636965e-10,
      "loss": 0.1596,
      "step": 2283
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2659146976157374,
      "learning_rate": 1.251618581127323e-10,
      "loss": 0.1894,
      "step": 2284
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.185335508721984,
      "learning_rate": 8.01037095038426e-11,
      "loss": 0.1611,
      "step": 2285
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3538565194974914,
      "learning_rate": 4.505838923252181e-11,
      "loss": 0.2246,
      "step": 2286
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1256816644511867,
      "learning_rate": 2.0025967479853082e-11,
      "loss": 0.1814,
      "step": 2287
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2700478589176818,
      "learning_rate": 5.006494376291748e-12,
      "loss": 0.1817,
      "step": 2288
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1025430016067395,
      "learning_rate": 0.0,
      "loss": 0.126,
      "step": 2289
    },
    {
      "epoch": 1.0,
      "step": 2289,
      "total_flos": 684559975874560.0,
      "train_loss": 0.217679803739369,
      "train_runtime": 32766.6415,
      "train_samples_per_second": 8.945,
      "train_steps_per_second": 0.07
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 2289,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "total_flos": 684559975874560.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}