{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9992357661444402,
  "eval_steps": 500,
  "global_step": 4578,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 35.743499547325705,
      "learning_rate": 7.246376811594204e-08,
      "loss": 1.6456,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 36.34008861481412,
      "learning_rate": 1.4492753623188408e-07,
      "loss": 1.6348,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 37.36187958216525,
      "learning_rate": 2.173913043478261e-07,
      "loss": 1.6449,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 36.30638096963061,
      "learning_rate": 2.8985507246376816e-07,
      "loss": 1.618,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 35.679564094117055,
      "learning_rate": 3.623188405797102e-07,
      "loss": 1.6395,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 36.13337241594072,
      "learning_rate": 4.347826086956522e-07,
      "loss": 1.623,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 35.086916250827905,
      "learning_rate": 5.072463768115942e-07,
      "loss": 1.5927,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 34.05474051231125,
      "learning_rate": 5.797101449275363e-07,
      "loss": 1.608,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 26.627208964845597,
      "learning_rate": 6.521739130434783e-07,
      "loss": 1.474,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 26.854735533859316,
      "learning_rate": 7.246376811594204e-07,
      "loss": 1.4655,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 26.041124675341802,
      "learning_rate": 7.971014492753623e-07,
      "loss": 1.4145,
      "step": 11
    },
    {
      "epoch": 0.01,
      "grad_norm": 20.523845573903163,
      "learning_rate": 8.695652173913044e-07,
      "loss": 1.1603,
      "step": 12
    },
    {
      "epoch": 0.01,
      "grad_norm": 19.21595606649398,
      "learning_rate": 9.420289855072465e-07,
      "loss": 1.1729,
      "step": 13
    },
    {
      "epoch": 0.01,
      "grad_norm": 18.710045751106716,
      "learning_rate": 1.0144927536231885e-06,
      "loss": 1.1364,
      "step": 14
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.953104267892833,
      "learning_rate": 1.0869565217391306e-06,
      "loss": 1.0985,
      "step": 15
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.384852335577659,
      "learning_rate": 1.1594202898550726e-06,
      "loss": 0.8935,
      "step": 16
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.634648458052343,
      "learning_rate": 1.2318840579710147e-06,
      "loss": 0.8067,
      "step": 17
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.3668041383007,
      "learning_rate": 1.3043478260869566e-06,
      "loss": 0.8068,
      "step": 18
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.932996794835114,
      "learning_rate": 1.3768115942028987e-06,
      "loss": 0.772,
      "step": 19
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.69108960888401,
      "learning_rate": 1.4492753623188408e-06,
      "loss": 0.7046,
      "step": 20
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.942398509821498,
      "learning_rate": 1.521739130434783e-06,
      "loss": 0.6825,
      "step": 21
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.123174632197531,
      "learning_rate": 1.5942028985507246e-06,
      "loss": 0.5726,
      "step": 22
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.2095098980048755,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 0.532,
      "step": 23
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.178920987073522,
      "learning_rate": 1.7391304347826088e-06,
      "loss": 0.5105,
      "step": 24
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.9822435981362494,
      "learning_rate": 1.8115942028985508e-06,
      "loss": 0.4827,
      "step": 25
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6888922293699338,
      "learning_rate": 1.884057971014493e-06,
      "loss": 0.4571,
      "step": 26
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.201882745382438,
      "learning_rate": 1.956521739130435e-06,
      "loss": 0.4426,
      "step": 27
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4241753035632,
      "learning_rate": 2.028985507246377e-06,
      "loss": 0.4689,
      "step": 28
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9759662130371587,
      "learning_rate": 2.101449275362319e-06,
      "loss": 0.4248,
      "step": 29
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5031113741446216,
      "learning_rate": 2.173913043478261e-06,
      "loss": 0.4312,
      "step": 30
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3120159812268084,
      "learning_rate": 2.246376811594203e-06,
      "loss": 0.4263,
      "step": 31
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.7790514330363623,
      "learning_rate": 2.3188405797101453e-06,
      "loss": 0.4284,
      "step": 32
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.098461717369773,
      "learning_rate": 2.391304347826087e-06,
      "loss": 0.4175,
      "step": 33
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.343678474328642,
      "learning_rate": 2.4637681159420295e-06,
      "loss": 0.4115,
      "step": 34
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.266475262122465,
      "learning_rate": 2.5362318840579714e-06,
      "loss": 0.4116,
      "step": 35
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8685471210762972,
      "learning_rate": 2.6086956521739132e-06,
      "loss": 0.4088,
      "step": 36
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.125836615634899,
      "learning_rate": 2.6811594202898555e-06,
      "loss": 0.3961,
      "step": 37
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7234915647692612,
      "learning_rate": 2.7536231884057974e-06,
      "loss": 0.3905,
      "step": 38
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.628765802172586,
      "learning_rate": 2.8260869565217393e-06,
      "loss": 0.3766,
      "step": 39
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9206344186016662,
      "learning_rate": 2.8985507246376816e-06,
      "loss": 0.3874,
      "step": 40
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8177874254692634,
      "learning_rate": 2.9710144927536235e-06,
      "loss": 0.3997,
      "step": 41
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.636462643052999,
      "learning_rate": 3.043478260869566e-06,
      "loss": 0.3652,
      "step": 42
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7184756073082947,
      "learning_rate": 3.1159420289855073e-06,
      "loss": 0.3696,
      "step": 43
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8146189941480078,
      "learning_rate": 3.188405797101449e-06,
      "loss": 0.3965,
      "step": 44
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5803819007784765,
      "learning_rate": 3.2608695652173914e-06,
      "loss": 0.3378,
      "step": 45
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5411267643563353,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.3931,
      "step": 46
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.643105409923438,
      "learning_rate": 3.4057971014492756e-06,
      "loss": 0.3675,
      "step": 47
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5590020300513736,
      "learning_rate": 3.4782608695652175e-06,
      "loss": 0.3801,
      "step": 48
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5316820477372075,
      "learning_rate": 3.55072463768116e-06,
      "loss": 0.3639,
      "step": 49
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2596857541635127,
      "learning_rate": 3.6231884057971017e-06,
      "loss": 0.3577,
      "step": 50
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.453460096326509,
      "learning_rate": 3.6956521739130436e-06,
      "loss": 0.3531,
      "step": 51
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.565099356750303,
      "learning_rate": 3.768115942028986e-06,
      "loss": 0.3697,
      "step": 52
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.47593642945337,
      "learning_rate": 3.840579710144928e-06,
      "loss": 0.3651,
      "step": 53
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4509800552416452,
      "learning_rate": 3.91304347826087e-06,
      "loss": 0.3596,
      "step": 54
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4305717589930722,
      "learning_rate": 3.9855072463768115e-06,
      "loss": 0.3462,
      "step": 55
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.502978505547465,
      "learning_rate": 4.057971014492754e-06,
      "loss": 0.3601,
      "step": 56
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3830605623837062,
      "learning_rate": 4.130434782608696e-06,
      "loss": 0.3363,
      "step": 57
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8133388547802471,
      "learning_rate": 4.202898550724638e-06,
      "loss": 0.3721,
      "step": 58
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5069437897848317,
      "learning_rate": 4.27536231884058e-06,
      "loss": 0.3769,
      "step": 59
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3265285955723023,
      "learning_rate": 4.347826086956522e-06,
      "loss": 0.3458,
      "step": 60
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3260408163079735,
      "learning_rate": 4.4202898550724645e-06,
      "loss": 0.3642,
      "step": 61
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3171098466777544,
      "learning_rate": 4.492753623188406e-06,
      "loss": 0.3505,
      "step": 62
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.518081100397385,
      "learning_rate": 4.565217391304348e-06,
      "loss": 0.3479,
      "step": 63
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4204385582474985,
      "learning_rate": 4.637681159420291e-06,
      "loss": 0.3612,
      "step": 64
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3823191120868878,
      "learning_rate": 4.710144927536232e-06,
      "loss": 0.3569,
      "step": 65
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5273511644380464,
      "learning_rate": 4.782608695652174e-06,
      "loss": 0.3541,
      "step": 66
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2441124764791391,
      "learning_rate": 4.855072463768117e-06,
      "loss": 0.3322,
      "step": 67
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2949549789380388,
      "learning_rate": 4.927536231884059e-06,
      "loss": 0.3282,
      "step": 68
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.290287218020303,
      "learning_rate": 5e-06,
      "loss": 0.3415,
      "step": 69
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.290812046048033,
      "learning_rate": 5.072463768115943e-06,
      "loss": 0.3485,
      "step": 70
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4514824503794477,
      "learning_rate": 5.144927536231884e-06,
      "loss": 0.3463,
      "step": 71
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2910864287797188,
      "learning_rate": 5.2173913043478265e-06,
      "loss": 0.3506,
      "step": 72
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4536920664502713,
      "learning_rate": 5.289855072463769e-06,
      "loss": 0.3557,
      "step": 73
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3072250506359264,
      "learning_rate": 5.362318840579711e-06,
      "loss": 0.335,
      "step": 74
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3544861070346121,
      "learning_rate": 5.4347826086956525e-06,
      "loss": 0.3373,
      "step": 75
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4373789708827887,
      "learning_rate": 5.507246376811595e-06,
      "loss": 0.3712,
      "step": 76
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3464418523437418,
      "learning_rate": 5.579710144927537e-06,
      "loss": 0.3308,
      "step": 77
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4524331424598784,
      "learning_rate": 5.652173913043479e-06,
      "loss": 0.3645,
      "step": 78
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.639252607043935,
      "learning_rate": 5.724637681159421e-06,
      "loss": 0.3369,
      "step": 79
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.456006229277705,
      "learning_rate": 5.797101449275363e-06,
      "loss": 0.3598,
      "step": 80
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4090244928946682,
      "learning_rate": 5.8695652173913055e-06,
      "loss": 0.3396,
      "step": 81
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6623788305366574,
      "learning_rate": 5.942028985507247e-06,
      "loss": 0.3082,
      "step": 82
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6481771573550903,
      "learning_rate": 6.014492753623189e-06,
      "loss": 0.3311,
      "step": 83
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3970845563649328,
      "learning_rate": 6.086956521739132e-06,
      "loss": 0.3525,
      "step": 84
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4306602228206378,
      "learning_rate": 6.159420289855072e-06,
      "loss": 0.3491,
      "step": 85
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4665269227203954,
      "learning_rate": 6.2318840579710145e-06,
      "loss": 0.3467,
      "step": 86
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2443139496069866,
      "learning_rate": 6.304347826086958e-06,
      "loss": 0.3055,
      "step": 87
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5145407494610197,
      "learning_rate": 6.376811594202898e-06,
      "loss": 0.3197,
      "step": 88
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.789731222536862,
      "learning_rate": 6.449275362318841e-06,
      "loss": 0.3575,
      "step": 89
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6114710348619425,
      "learning_rate": 6.521739130434783e-06,
      "loss": 0.3373,
      "step": 90
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5533153170068343,
      "learning_rate": 6.594202898550725e-06,
      "loss": 0.3544,
      "step": 91
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5025664875111162,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.3367,
      "step": 92
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.331996073270413,
      "learning_rate": 6.739130434782609e-06,
      "loss": 0.3558,
      "step": 93
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.608671480618854,
      "learning_rate": 6.811594202898551e-06,
      "loss": 0.3276,
      "step": 94
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4613321461474105,
      "learning_rate": 6.884057971014493e-06,
      "loss": 0.3535,
      "step": 95
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5305756296188198,
      "learning_rate": 6.956521739130435e-06,
      "loss": 0.3682,
      "step": 96
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4094744561212325,
      "learning_rate": 7.028985507246377e-06,
      "loss": 0.3628,
      "step": 97
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3272100598714962,
      "learning_rate": 7.10144927536232e-06,
      "loss": 0.3348,
      "step": 98
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.376530405302754,
      "learning_rate": 7.173913043478261e-06,
      "loss": 0.3464,
      "step": 99
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2447713314104958,
      "learning_rate": 7.246376811594203e-06,
      "loss": 0.3302,
      "step": 100
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1615319006759874,
      "learning_rate": 7.318840579710146e-06,
      "loss": 0.3355,
      "step": 101
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1963628589162758,
      "learning_rate": 7.391304347826087e-06,
      "loss": 0.3254,
      "step": 102
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3711701213537446,
      "learning_rate": 7.4637681159420295e-06,
      "loss": 0.3377,
      "step": 103
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3627327250437926,
      "learning_rate": 7.536231884057972e-06,
      "loss": 0.3359,
      "step": 104
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.290172193156435,
      "learning_rate": 7.608695652173914e-06,
      "loss": 0.3399,
      "step": 105
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2416463811490523,
      "learning_rate": 7.681159420289856e-06,
      "loss": 0.3284,
      "step": 106
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2857864320860644,
      "learning_rate": 7.753623188405797e-06,
      "loss": 0.3234,
      "step": 107
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2742153148078255,
      "learning_rate": 7.82608695652174e-06,
      "loss": 0.3273,
      "step": 108
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2995890311690974,
      "learning_rate": 7.898550724637682e-06,
      "loss": 0.3268,
      "step": 109
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2301364751998172,
      "learning_rate": 7.971014492753623e-06,
      "loss": 0.3536,
      "step": 110
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2533189254648132,
      "learning_rate": 8.043478260869566e-06,
      "loss": 0.3035,
      "step": 111
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2980337881233197,
      "learning_rate": 8.115942028985508e-06,
      "loss": 0.3543,
      "step": 112
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.312075117816281,
      "learning_rate": 8.188405797101449e-06,
      "loss": 0.3294,
      "step": 113
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3116852471642093,
      "learning_rate": 8.260869565217392e-06,
      "loss": 0.3409,
      "step": 114
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1187694105533332,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.3203,
      "step": 115
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4944652089205355,
      "learning_rate": 8.405797101449275e-06,
      "loss": 0.3184,
      "step": 116
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2810277871413815,
      "learning_rate": 8.478260869565218e-06,
      "loss": 0.3285,
      "step": 117
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2888696631196663,
      "learning_rate": 8.55072463768116e-06,
      "loss": 0.3553,
      "step": 118
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2428639558993837,
      "learning_rate": 8.623188405797103e-06,
      "loss": 0.3279,
      "step": 119
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2685069749410183,
      "learning_rate": 8.695652173913044e-06,
      "loss": 0.3506,
      "step": 120
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3472732746728293,
      "learning_rate": 8.768115942028986e-06,
      "loss": 0.3607,
      "step": 121
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2960177949577205,
      "learning_rate": 8.840579710144929e-06,
      "loss": 0.3041,
      "step": 122
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4507360613087965,
      "learning_rate": 8.91304347826087e-06,
      "loss": 0.3419,
      "step": 123
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4509700648609702,
      "learning_rate": 8.985507246376812e-06,
      "loss": 0.3574,
      "step": 124
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3440253000152425,
      "learning_rate": 9.057971014492755e-06,
      "loss": 0.316,
      "step": 125
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3475608477250094,
      "learning_rate": 9.130434782608697e-06,
      "loss": 0.3518,
      "step": 126
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.381578497955178,
      "learning_rate": 9.202898550724638e-06,
      "loss": 0.3373,
      "step": 127
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2859606244189608,
      "learning_rate": 9.275362318840581e-06,
      "loss": 0.3195,
      "step": 128
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.37555122603588,
      "learning_rate": 9.347826086956523e-06,
      "loss": 0.3083,
      "step": 129
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5649247671871276,
      "learning_rate": 9.420289855072464e-06,
      "loss": 0.3469,
      "step": 130
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4618512979646894,
      "learning_rate": 9.492753623188407e-06,
      "loss": 0.3337,
      "step": 131
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5430561886820622,
      "learning_rate": 9.565217391304349e-06,
      "loss": 0.3548,
      "step": 132
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.540941170490441,
      "learning_rate": 9.63768115942029e-06,
      "loss": 0.3262,
      "step": 133
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.384886618104957,
      "learning_rate": 9.710144927536233e-06,
      "loss": 0.3402,
      "step": 134
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1425615396586657,
      "learning_rate": 9.782608695652175e-06,
      "loss": 0.3507,
      "step": 135
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1906103638753842,
      "learning_rate": 9.855072463768118e-06,
      "loss": 0.3186,
      "step": 136
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1750337005154639,
      "learning_rate": 9.927536231884058e-06,
      "loss": 0.3021,
      "step": 137
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2073246217580407,
      "learning_rate": 1e-05,
      "loss": 0.3411,
      "step": 138
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0988798777006437,
      "learning_rate": 9.999998748376249e-06,
      "loss": 0.335,
      "step": 139
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2633781647607132,
      "learning_rate": 9.999994993505624e-06,
      "loss": 0.3414,
      "step": 140
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.267078273852445,
      "learning_rate": 9.999988735390003e-06,
      "loss": 0.3346,
      "step": 141
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2155779920582388,
      "learning_rate": 9.99997997403252e-06,
      "loss": 0.3266,
      "step": 142
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.167741019706484,
      "learning_rate": 9.999968709437563e-06,
      "loss": 0.2943,
      "step": 143
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1599848618753443,
      "learning_rate": 9.999954941610767e-06,
      "loss": 0.3447,
      "step": 144
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1568552345748158,
      "learning_rate": 9.999938670559031e-06,
      "loss": 0.3351,
      "step": 145
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2361621267053484,
      "learning_rate": 9.999919896290497e-06,
      "loss": 0.3734,
      "step": 146
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.145361364938489,
      "learning_rate": 9.999898618814565e-06,
      "loss": 0.3159,
      "step": 147
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2616066822905272,
      "learning_rate": 9.999874838141888e-06,
      "loss": 0.3512,
      "step": 148
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1719870090616111,
      "learning_rate": 9.999848554284371e-06,
      "loss": 0.3567,
      "step": 149
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.21148779015622,
      "learning_rate": 9.999819767255175e-06,
      "loss": 0.3198,
      "step": 150
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.111667553154608,
      "learning_rate": 9.99978847706871e-06,
      "loss": 0.3344,
      "step": 151
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1236293283141912,
      "learning_rate": 9.999754683740643e-06,
      "loss": 0.3493,
      "step": 152
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1109766857603933,
      "learning_rate": 9.999718387287891e-06,
      "loss": 0.313,
      "step": 153
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1863145155026278,
      "learning_rate": 9.999679587728626e-06,
      "loss": 0.3379,
      "step": 154
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1756223776481258,
      "learning_rate": 9.999638285082276e-06,
      "loss": 0.3378,
      "step": 155
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1564756575773238,
      "learning_rate": 9.999594479369515e-06,
      "loss": 0.3065,
      "step": 156
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1059922713261994,
      "learning_rate": 9.999548170612276e-06,
      "loss": 0.3375,
      "step": 157
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.015390046153437,
      "learning_rate": 9.999499358833745e-06,
      "loss": 0.2995,
      "step": 158
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2322933284015154,
      "learning_rate": 9.999448044058357e-06,
      "loss": 0.3509,
      "step": 159
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.134312379138924,
      "learning_rate": 9.999394226311804e-06,
      "loss": 0.3415,
      "step": 160
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1751250854014545,
      "learning_rate": 9.999337905621031e-06,
      "loss": 0.3323,
      "step": 161
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1755867319723134,
      "learning_rate": 9.999279082014233e-06,
      "loss": 0.3204,
      "step": 162
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1428103373697378,
      "learning_rate": 9.99921775552086e-06,
      "loss": 0.3195,
      "step": 163
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.065018042516367,
      "learning_rate": 9.999153926171617e-06,
      "loss": 0.3205,
      "step": 164
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0715461555196553,
      "learning_rate": 9.999087593998458e-06,
      "loss": 0.3245,
      "step": 165
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0444706483983466,
      "learning_rate": 9.999018759034594e-06,
      "loss": 0.323,
      "step": 166
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1182854350492766,
      "learning_rate": 9.998947421314485e-06,
      "loss": 0.3435,
      "step": 167
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1404485320061553,
      "learning_rate": 9.998873580873848e-06,
      "loss": 0.3124,
      "step": 168
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1276032035326833,
      "learning_rate": 9.998797237749651e-06,
      "loss": 0.358,
      "step": 169
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0154195210735386,
      "learning_rate": 9.998718391980113e-06,
      "loss": 0.3146,
      "step": 170
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.102522240431025,
      "learning_rate": 9.998637043604711e-06,
      "loss": 0.3434,
      "step": 171
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0747308374798556,
      "learning_rate": 9.998553192664172e-06,
      "loss": 0.3262,
      "step": 172
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0333844706198283,
      "learning_rate": 9.998466839200474e-06,
      "loss": 0.3067,
      "step": 173
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.217057046580774,
      "learning_rate": 9.998377983256851e-06,
      "loss": 0.34,
      "step": 174
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1344913629131501,
      "learning_rate": 9.998286624877786e-06,
      "loss": 0.3451,
      "step": 175
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1463815622338729,
      "learning_rate": 9.998192764109023e-06,
      "loss": 0.3157,
      "step": 176
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.104712965849977,
      "learning_rate": 9.99809640099755e-06,
      "loss": 0.3204,
      "step": 177
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.085433813137142,
      "learning_rate": 9.99799753559161e-06,
      "loss": 0.334,
      "step": 178
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.3490736882982115,
      "learning_rate": 9.997896167940701e-06,
      "loss": 0.3473,
      "step": 179
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3191830059626324,
      "learning_rate": 9.997792298095572e-06,
      "loss": 0.3347,
      "step": 180
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1587984280751413,
      "learning_rate": 9.99768592610823e-06,
      "loss": 0.3406,
      "step": 181
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1792906985931957,
      "learning_rate": 9.997577052031922e-06,
      "loss": 0.3277,
      "step": 182
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2016253532528236,
      "learning_rate": 9.997465675921163e-06,
      "loss": 0.3448,
      "step": 183
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.081569826710216,
      "learning_rate": 9.99735179783171e-06,
      "loss": 0.3123,
      "step": 184
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1277139571667543,
      "learning_rate": 9.997235417820577e-06,
      "loss": 0.3265,
      "step": 185
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2658608583526312,
      "learning_rate": 9.997116535946028e-06,
      "loss": 0.303,
      "step": 186
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0726458908591978,
      "learning_rate": 9.996995152267581e-06,
      "loss": 0.3389,
      "step": 187
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2334304960757767,
      "learning_rate": 9.99687126684601e-06,
      "loss": 0.3356,
      "step": 188
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2441195609066136,
      "learning_rate": 9.996744879743336e-06,
      "loss": 0.3109,
      "step": 189
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0708496952216062,
      "learning_rate": 9.996615991022835e-06,
      "loss": 0.3381,
      "step": 190
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1083070295239057,
      "learning_rate": 9.996484600749035e-06,
      "loss": 0.3454,
      "step": 191
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1175404351232427,
      "learning_rate": 9.996350708987715e-06,
      "loss": 0.3431,
      "step": 192
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3742347707107079,
      "learning_rate": 9.99621431580591e-06,
      "loss": 0.3311,
      "step": 193
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2802775697460793,
      "learning_rate": 9.996075421271905e-06,
      "loss": 0.3373,
      "step": 194
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.220201419616746,
      "learning_rate": 9.995934025455236e-06,
      "loss": 0.3415,
      "step": 195
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2078810042185277,
      "learning_rate": 9.995790128426693e-06,
      "loss": 0.322,
      "step": 196
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.219682483554145,
      "learning_rate": 9.99564373025832e-06,
      "loss": 0.2908,
      "step": 197
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0852533050290962,
      "learning_rate": 9.99549483102341e-06,
      "loss": 0.3186,
      "step": 198
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9728195306225799,
      "learning_rate": 9.995343430796507e-06,
      "loss": 0.3021,
      "step": 199
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9469257921819585,
      "learning_rate": 9.995189529653412e-06,
      "loss": 0.3041,
      "step": 200
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0144504287701925,
      "learning_rate": 9.995033127671174e-06,
      "loss": 0.3179,
      "step": 201
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0717323014922537,
      "learning_rate": 9.9948742249281e-06,
      "loss": 0.3374,
      "step": 202
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.150164803030314,
      "learning_rate": 9.994712821503737e-06,
      "loss": 0.3152,
      "step": 203
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1778435057711354,
      "learning_rate": 9.994548917478899e-06,
      "loss": 0.3105,
      "step": 204
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9947886182264449,
      "learning_rate": 9.99438251293564e-06,
      "loss": 0.313,
      "step": 205
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1383824907314235,
      "learning_rate": 9.994213607957273e-06,
      "loss": 0.3229,
      "step": 206
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.141780280038623,
      "learning_rate": 9.994042202628356e-06,
      "loss": 0.3202,
      "step": 207
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1133707299044613,
      "learning_rate": 9.993868297034709e-06,
      "loss": 0.3388,
      "step": 208
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2024293731159965,
      "learning_rate": 9.993691891263393e-06,
      "loss": 0.3637,
      "step": 209
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0754402701975092,
      "learning_rate": 9.993512985402724e-06,
      "loss": 0.3437,
      "step": 210
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9920502632609169,
      "learning_rate": 9.993331579542278e-06,
      "loss": 0.3243,
      "step": 211
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1414150360002602,
      "learning_rate": 9.993147673772869e-06,
      "loss": 0.3213,
      "step": 212
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0486823581154605,
      "learning_rate": 9.992961268186575e-06,
      "loss": 0.2991,
      "step": 213
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1175090000269068,
      "learning_rate": 9.992772362876714e-06,
      "loss": 0.3131,
      "step": 214
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0321635817160209,
      "learning_rate": 9.992580957937868e-06,
      "loss": 0.3209,
      "step": 215
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0848922827332514,
      "learning_rate": 9.992387053465857e-06,
      "loss": 0.3206,
      "step": 216
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9976574776723222,
      "learning_rate": 9.992190649557765e-06,
      "loss": 0.3355,
      "step": 217
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9883603175552789,
      "learning_rate": 9.991991746311916e-06,
      "loss": 0.3043,
      "step": 218
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0087945287799105,
      "learning_rate": 9.991790343827896e-06,
      "loss": 0.331,
      "step": 219
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0197352362980432,
      "learning_rate": 9.991586442206533e-06,
      "loss": 0.3148,
      "step": 220
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0652645566166723,
      "learning_rate": 9.991380041549911e-06,
      "loss": 0.3263,
      "step": 221
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0590027308111558,
      "learning_rate": 9.99117114196137e-06,
      "loss": 0.3358,
      "step": 222
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9763506062420368,
      "learning_rate": 9.990959743545487e-06,
      "loss": 0.3034,
      "step": 223
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.202825364824694,
      "learning_rate": 9.990745846408102e-06,
      "loss": 0.321,
      "step": 224
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0986187762975559,
      "learning_rate": 9.990529450656303e-06,
      "loss": 0.3254,
      "step": 225
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2424400499128183,
      "learning_rate": 9.99031055639843e-06,
      "loss": 0.3122,
      "step": 226
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.217483054533846,
      "learning_rate": 9.99008916374407e-06,
      "loss": 0.3573,
      "step": 227
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2205901834162767,
      "learning_rate": 9.989865272804064e-06,
      "loss": 0.3058,
      "step": 228
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0380083837175575,
      "learning_rate": 9.989638883690501e-06,
      "loss": 0.3293,
      "step": 229
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3261492857534325,
      "learning_rate": 9.989409996516726e-06,
      "loss": 0.315,
      "step": 230
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.243277026835093,
      "learning_rate": 9.989178611397328e-06,
      "loss": 0.3414,
      "step": 231
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0437312108132286,
      "learning_rate": 9.98894472844815e-06,
      "loss": 0.3533,
      "step": 232
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.15396061115585,
      "learning_rate": 9.988708347786288e-06,
      "loss": 0.3243,
      "step": 233
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1419664891962085,
      "learning_rate": 9.988469469530086e-06,
      "loss": 0.3209,
      "step": 234
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1202122010099798,
      "learning_rate": 9.988228093799136e-06,
      "loss": 0.3156,
      "step": 235
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9384963505896191,
      "learning_rate": 9.987984220714283e-06,
      "loss": 0.3056,
      "step": 236
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1443540825504839,
      "learning_rate": 9.987737850397623e-06,
      "loss": 0.3248,
      "step": 237
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.911081940046987,
      "learning_rate": 9.9874889829725e-06,
      "loss": 0.2945,
      "step": 238
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0599729208419308,
      "learning_rate": 9.987237618563509e-06,
      "loss": 0.3039,
      "step": 239
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1295570270989106,
      "learning_rate": 9.986983757296498e-06,
      "loss": 0.3296,
      "step": 240
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9446157050103396,
      "learning_rate": 9.98672739929856e-06,
      "loss": 0.3226,
      "step": 241
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9747730955791443,
      "learning_rate": 9.986468544698041e-06,
      "loss": 0.3298,
      "step": 242
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0499827182501116,
      "learning_rate": 9.986207193624537e-06,
      "loss": 0.338,
      "step": 243
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0508860107407663,
      "learning_rate": 9.985943346208893e-06,
      "loss": 0.3269,
      "step": 244
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0587063604629994,
      "learning_rate": 9.985677002583204e-06,
      "loss": 0.3407,
      "step": 245
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0621606264458223,
      "learning_rate": 9.985408162880813e-06,
      "loss": 0.3285,
      "step": 246
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0309675236218658,
      "learning_rate": 9.98513682723632e-06,
      "loss": 0.3009,
      "step": 247
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9045194175325502,
      "learning_rate": 9.984862995785564e-06,
      "loss": 0.3025,
      "step": 248
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9940678351806522,
      "learning_rate": 9.984586668665641e-06,
      "loss": 0.3183,
      "step": 249
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0291794881616612,
      "learning_rate": 9.984307846014891e-06,
      "loss": 0.3307,
      "step": 250
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0906825222573526,
      "learning_rate": 9.98402652797291e-06,
      "loss": 0.3208,
      "step": 251
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9989460572880086,
      "learning_rate": 9.983742714680538e-06,
      "loss": 0.3434,
      "step": 252
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9360845004843308,
      "learning_rate": 9.983456406279866e-06,
      "loss": 0.3146,
      "step": 253
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.056118352774009,
      "learning_rate": 9.983167602914236e-06,
      "loss": 0.2937,
      "step": 254
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0498895566512552,
      "learning_rate": 9.982876304728233e-06,
      "loss": 0.2993,
      "step": 255
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.176591419449869,
      "learning_rate": 9.9825825118677e-06,
      "loss": 0.3417,
      "step": 256
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9921048229780709,
      "learning_rate": 9.98228622447972e-06,
      "loss": 0.3266,
      "step": 257
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1100742197293856,
      "learning_rate": 9.981987442712634e-06,
      "loss": 0.3338,
      "step": 258
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.170961631431727,
      "learning_rate": 9.981686166716023e-06,
      "loss": 0.3245,
      "step": 259
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1725616433186266,
      "learning_rate": 9.981382396640721e-06,
      "loss": 0.3519,
      "step": 260
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0043266092796233,
      "learning_rate": 9.981076132638811e-06,
      "loss": 0.3304,
      "step": 261
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9746850626564973,
      "learning_rate": 9.980767374863627e-06,
      "loss": 0.3515,
      "step": 262
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.056164642059128,
      "learning_rate": 9.980456123469743e-06,
      "loss": 0.3406,
      "step": 263
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.060794716737408,
      "learning_rate": 9.98014237861299e-06,
      "loss": 0.3211,
      "step": 264
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0547422544468457,
      "learning_rate": 9.979826140450443e-06,
      "loss": 0.3315,
      "step": 265
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9987854635267067,
      "learning_rate": 9.97950740914043e-06,
      "loss": 0.3233,
      "step": 266
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1034315795980063,
      "learning_rate": 9.979186184842517e-06,
      "loss": 0.3428,
      "step": 267
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9802496217624141,
      "learning_rate": 9.978862467717532e-06,
      "loss": 0.3183,
      "step": 268
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0261217689364168,
      "learning_rate": 9.97853625792754e-06,
      "loss": 0.3035,
      "step": 269
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9175499819868345,
      "learning_rate": 9.978207555635856e-06,
      "loss": 0.3161,
      "step": 270
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9845216501453553,
      "learning_rate": 9.977876361007049e-06,
      "loss": 0.3179,
      "step": 271
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9451150754651583,
      "learning_rate": 9.977542674206927e-06,
      "loss": 0.312,
      "step": 272
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0509120720090523,
      "learning_rate": 9.977206495402554e-06,
      "loss": 0.3463,
      "step": 273
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0280805637145571,
      "learning_rate": 9.976867824762236e-06,
      "loss": 0.3324,
      "step": 274
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9583523285824739,
      "learning_rate": 9.976526662455528e-06,
      "loss": 0.3213,
      "step": 275
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9552427241149758,
      "learning_rate": 9.976183008653232e-06,
      "loss": 0.3131,
      "step": 276
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.003576471081345,
      "learning_rate": 9.975836863527402e-06,
      "loss": 0.3335,
      "step": 277
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9539922212087029,
      "learning_rate": 9.97548822725133e-06,
      "loss": 0.3176,
      "step": 278
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9846322013387402,
      "learning_rate": 9.975137099999566e-06,
      "loss": 0.3368,
      "step": 279
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0072665372505596,
      "learning_rate": 9.974783481947896e-06,
      "loss": 0.3163,
      "step": 280
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0376362443788094,
      "learning_rate": 9.974427373273364e-06,
      "loss": 0.317,
      "step": 281
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9799721143375604,
      "learning_rate": 9.974068774154252e-06,
      "loss": 0.3254,
      "step": 282
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9152161989515294,
      "learning_rate": 9.973707684770095e-06,
      "loss": 0.3301,
      "step": 283
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.88812838124082,
      "learning_rate": 9.97334410530167e-06,
      "loss": 0.3067,
      "step": 284
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.034054883536243,
      "learning_rate": 9.972978035931003e-06,
      "loss": 0.3615,
      "step": 285
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8979377820540138,
      "learning_rate": 9.972609476841368e-06,
      "loss": 0.3088,
      "step": 286
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9930995263069547,
      "learning_rate": 9.972238428217283e-06,
      "loss": 0.3426,
      "step": 287
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1101055672222442,
      "learning_rate": 9.971864890244514e-06,
      "loss": 0.3201,
      "step": 288
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9749840830540123,
      "learning_rate": 9.971488863110071e-06,
      "loss": 0.3397,
      "step": 289
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8844402440518165,
      "learning_rate": 9.971110347002213e-06,
      "loss": 0.3001,
      "step": 290
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8785914206965239,
      "learning_rate": 9.970729342110446e-06,
      "loss": 0.2886,
      "step": 291
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9341778237596841,
      "learning_rate": 9.970345848625516e-06,
      "loss": 0.3233,
      "step": 292
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9685580474096556,
      "learning_rate": 9.96995986673942e-06,
      "loss": 0.3253,
      "step": 293
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0389661356556137,
      "learning_rate": 9.9695713966454e-06,
      "loss": 0.3159,
      "step": 294
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9375482686692821,
      "learning_rate": 9.969180438537944e-06,
      "loss": 0.3337,
      "step": 295
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9385198035089628,
      "learning_rate": 9.968786992612785e-06,
      "loss": 0.3232,
      "step": 296
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9289012501315866,
      "learning_rate": 9.9683910590669e-06,
      "loss": 0.3148,
      "step": 297
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.042140218385059,
      "learning_rate": 9.967992638098517e-06,
      "loss": 0.3062,
      "step": 298
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9510289874195472,
      "learning_rate": 9.967591729907098e-06,
      "loss": 0.3374,
      "step": 299
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9595366321106535,
      "learning_rate": 9.967188334693364e-06,
      "loss": 0.3368,
      "step": 300
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0391457297637343,
      "learning_rate": 9.966782452659272e-06,
      "loss": 0.298,
      "step": 301
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.024493100354939,
      "learning_rate": 9.966374084008025e-06,
      "loss": 0.3014,
      "step": 302
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.938837070470008,
      "learning_rate": 9.965963228944077e-06,
      "loss": 0.3262,
      "step": 303
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0822481742478234,
      "learning_rate": 9.965549887673121e-06,
      "loss": 0.3372,
      "step": 304
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1051513889027265,
      "learning_rate": 9.965134060402092e-06,
      "loss": 0.3235,
      "step": 305
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8978151500656456,
      "learning_rate": 9.964715747339178e-06,
      "loss": 0.3028,
      "step": 306
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9791718255721538,
      "learning_rate": 9.964294948693805e-06,
      "loss": 0.3373,
      "step": 307
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9212990093241396,
      "learning_rate": 9.963871664676647e-06,
      "loss": 0.3221,
      "step": 308
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.95092709001035,
      "learning_rate": 9.963445895499622e-06,
      "loss": 0.2977,
      "step": 309
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9685630509752409,
      "learning_rate": 9.963017641375888e-06,
      "loss": 0.2991,
      "step": 310
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9467929983752023,
      "learning_rate": 9.962586902519851e-06,
      "loss": 0.3394,
      "step": 311
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9821260520395496,
      "learning_rate": 9.962153679147162e-06,
      "loss": 0.318,
      "step": 312
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9111719555757573,
      "learning_rate": 9.961717971474714e-06,
      "loss": 0.3033,
      "step": 313
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.051629525245098,
      "learning_rate": 9.961279779720641e-06,
      "loss": 0.3116,
      "step": 314
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.973704142434735,
      "learning_rate": 9.960839104104326e-06,
      "loss": 0.3253,
      "step": 315
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0202466585963772,
      "learning_rate": 9.960395944846394e-06,
      "loss": 0.3432,
      "step": 316
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9392372429398523,
      "learning_rate": 9.95995030216871e-06,
      "loss": 0.308,
      "step": 317
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8315223580265233,
      "learning_rate": 9.959502176294384e-06,
      "loss": 0.3048,
      "step": 318
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0281115789239117,
      "learning_rate": 9.959051567447775e-06,
      "loss": 0.3371,
      "step": 319
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9274973110332339,
      "learning_rate": 9.958598475854474e-06,
      "loss": 0.3152,
      "step": 320
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9937655796384761,
      "learning_rate": 9.958142901741325e-06,
      "loss": 0.3142,
      "step": 321
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9569741737306057,
      "learning_rate": 9.957684845336409e-06,
      "loss": 0.3249,
      "step": 322
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0083798308902425,
      "learning_rate": 9.957224306869053e-06,
      "loss": 0.2904,
      "step": 323
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.953730643146498,
      "learning_rate": 9.956761286569824e-06,
      "loss": 0.2887,
      "step": 324
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9238729974150489,
      "learning_rate": 9.956295784670536e-06,
      "loss": 0.3118,
      "step": 325
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9108850097674973,
      "learning_rate": 9.955827801404237e-06,
      "loss": 0.3367,
      "step": 326
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0909125328051226,
      "learning_rate": 9.955357337005226e-06,
      "loss": 0.3237,
      "step": 327
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0551273867960516,
      "learning_rate": 9.954884391709043e-06,
      "loss": 0.3112,
      "step": 328
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9009841578289884,
      "learning_rate": 9.954408965752463e-06,
      "loss": 0.3556,
      "step": 329
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9187872253974906,
      "learning_rate": 9.95393105937351e-06,
      "loss": 0.3079,
      "step": 330
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9072166278160997,
      "learning_rate": 9.953450672811449e-06,
      "loss": 0.2924,
      "step": 331
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9680006568132059,
      "learning_rate": 9.952967806306784e-06,
      "loss": 0.3054,
      "step": 332
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0671519669297564,
      "learning_rate": 9.95248246010126e-06,
      "loss": 0.3203,
      "step": 333
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0431228132702526,
      "learning_rate": 9.951994634437866e-06,
      "loss": 0.3109,
      "step": 334
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0593701873645796,
      "learning_rate": 9.951504329560835e-06,
      "loss": 0.3094,
      "step": 335
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.955015886264525,
      "learning_rate": 9.951011545715637e-06,
      "loss": 0.3309,
      "step": 336
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8762597784878597,
      "learning_rate": 9.95051628314898e-06,
      "loss": 0.3161,
      "step": 337
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9674533807833485,
      "learning_rate": 9.950018542108818e-06,
      "loss": 0.3051,
      "step": 338
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9946387780121904,
      "learning_rate": 9.94951832284435e-06,
      "loss": 0.3393,
      "step": 339
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9626726878231822,
      "learning_rate": 9.949015625606007e-06,
      "loss": 0.325,
      "step": 340
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9955464958133412,
      "learning_rate": 9.94851045064546e-06,
      "loss": 0.3176,
      "step": 341
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8643819034011909,
      "learning_rate": 9.948002798215632e-06,
      "loss": 0.312,
      "step": 342
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9338835079332041,
      "learning_rate": 9.947492668570675e-06,
      "loss": 0.3331,
      "step": 343
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.932507005123103,
      "learning_rate": 9.946980061965985e-06,
      "loss": 0.2986,
      "step": 344
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9390680463973057,
      "learning_rate": 9.9464649786582e-06,
      "loss": 0.3051,
      "step": 345
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0866570343095472,
      "learning_rate": 9.945947418905193e-06,
      "loss": 0.3214,
      "step": 346
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.026145208144877,
      "learning_rate": 9.945427382966084e-06,
      "loss": 0.3194,
      "step": 347
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.078326300867755,
      "learning_rate": 9.944904871101227e-06,
      "loss": 0.32,
      "step": 348
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.98569377504982,
      "learning_rate": 9.944379883572217e-06,
      "loss": 0.2993,
      "step": 349
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8825334016612482,
      "learning_rate": 9.943852420641888e-06,
      "loss": 0.311,
      "step": 350
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9584947454668095,
      "learning_rate": 9.943322482574316e-06,
      "loss": 0.3227,
      "step": 351
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.87184559482003,
      "learning_rate": 9.942790069634814e-06,
      "loss": 0.3272,
      "step": 352
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9758199475409963,
      "learning_rate": 9.94225518208993e-06,
      "loss": 0.3025,
      "step": 353
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8934170471514837,
      "learning_rate": 9.941717820207461e-06,
      "loss": 0.3008,
      "step": 354
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8604851796936462,
      "learning_rate": 9.941177984256436e-06,
      "loss": 0.289,
      "step": 355
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8999164204346828,
      "learning_rate": 9.94063567450712e-06,
      "loss": 0.3323,
      "step": 356
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.964145736382101,
      "learning_rate": 9.940090891231024e-06,
      "loss": 0.3112,
      "step": 357
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8571866987245382,
      "learning_rate": 9.939543634700891e-06,
      "loss": 0.3006,
      "step": 358
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8718564516367789,
      "learning_rate": 9.938993905190706e-06,
      "loss": 0.3039,
      "step": 359
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9187587371509705,
      "learning_rate": 9.938441702975689e-06,
      "loss": 0.3059,
      "step": 360
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9326065252882119,
      "learning_rate": 9.937887028332303e-06,
      "loss": 0.3326,
      "step": 361
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8318296642901971,
      "learning_rate": 9.937329881538242e-06,
      "loss": 0.3015,
      "step": 362
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9407580449117474,
      "learning_rate": 9.936770262872444e-06,
      "loss": 0.2989,
      "step": 363
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9105914541153703,
      "learning_rate": 9.936208172615081e-06,
      "loss": 0.2932,
      "step": 364
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9042531678173799,
      "learning_rate": 9.93564361104756e-06,
      "loss": 0.2959,
      "step": 365
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9382026852359685,
      "learning_rate": 9.935076578452535e-06,
      "loss": 0.2878,
      "step": 366
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8920857863294961,
      "learning_rate": 9.934507075113886e-06,
      "loss": 0.3172,
      "step": 367
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9560992811321057,
      "learning_rate": 9.933935101316735e-06,
      "loss": 0.2906,
      "step": 368
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9017416968626356,
      "learning_rate": 9.933360657347441e-06,
      "loss": 0.3077,
      "step": 369
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9885690702439038,
      "learning_rate": 9.932783743493599e-06,
      "loss": 0.3266,
      "step": 370
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9858772839680632,
      "learning_rate": 9.932204360044042e-06,
      "loss": 0.3148,
      "step": 371
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9469562525482141,
      "learning_rate": 9.931622507288835e-06,
      "loss": 0.3136,
      "step": 372
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8433063935833467,
      "learning_rate": 9.931038185519285e-06,
      "loss": 0.3156,
      "step": 373
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8981494637270862,
      "learning_rate": 9.930451395027932e-06,
      "loss": 0.3058,
      "step": 374
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0001385083401668,
      "learning_rate": 9.92986213610855e-06,
      "loss": 0.3266,
      "step": 375
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.907086341572844,
      "learning_rate": 9.929270409056154e-06,
      "loss": 0.3135,
      "step": 376
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9150027644183985,
      "learning_rate": 9.92867621416699e-06,
      "loss": 0.3201,
      "step": 377
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9014109421686484,
      "learning_rate": 9.928079551738542e-06,
      "loss": 0.338,
      "step": 378
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9411423417908064,
      "learning_rate": 9.927480422069531e-06,
      "loss": 0.3166,
      "step": 379
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8767047203841716,
      "learning_rate": 9.926878825459907e-06,
      "loss": 0.3137,
      "step": 380
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.01743857672353,
      "learning_rate": 9.926274762210862e-06,
      "loss": 0.3381,
      "step": 381
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8366348601678808,
      "learning_rate": 9.925668232624819e-06,
      "loss": 0.2827,
      "step": 382
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8531399662299582,
      "learning_rate": 9.925059237005437e-06,
      "loss": 0.3318,
      "step": 383
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9446954417295104,
      "learning_rate": 9.924447775657607e-06,
      "loss": 0.3067,
      "step": 384
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9742526114533888,
      "learning_rate": 9.92383384888746e-06,
      "loss": 0.3194,
      "step": 385
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9786314131876327,
      "learning_rate": 9.923217457002356e-06,
      "loss": 0.3312,
      "step": 386
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8955752707071684,
      "learning_rate": 9.922598600310893e-06,
      "loss": 0.3215,
      "step": 387
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9255855121906975,
      "learning_rate": 9.9219772791229e-06,
      "loss": 0.2983,
      "step": 388
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8671493852118374,
      "learning_rate": 9.921353493749443e-06,
      "loss": 0.3297,
      "step": 389
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8596494269077187,
      "learning_rate": 9.92072724450282e-06,
      "loss": 0.3074,
      "step": 390
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9459935523065338,
      "learning_rate": 9.920098531696558e-06,
      "loss": 0.3464,
      "step": 391
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9316426423763245,
      "learning_rate": 9.919467355645425e-06,
      "loss": 0.3207,
      "step": 392
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9043336182544588,
      "learning_rate": 9.91883371666542e-06,
      "loss": 0.2933,
      "step": 393
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8923675419249251,
      "learning_rate": 9.91819761507377e-06,
      "loss": 0.3375,
      "step": 394
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8761509979470802,
      "learning_rate": 9.917559051188943e-06,
      "loss": 0.3184,
      "step": 395
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9567253691523504,
      "learning_rate": 9.916918025330635e-06,
      "loss": 0.2974,
      "step": 396
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9042933170220128,
      "learning_rate": 9.916274537819774e-06,
      "loss": 0.3213,
      "step": 397
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8309914634610538,
      "learning_rate": 9.915628588978522e-06,
      "loss": 0.3072,
      "step": 398
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8910499617941405,
      "learning_rate": 9.914980179130273e-06,
      "loss": 0.3154,
      "step": 399
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9381710511783101,
      "learning_rate": 9.914329308599655e-06,
      "loss": 0.3289,
      "step": 400
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9392321466139855,
      "learning_rate": 9.913675977712522e-06,
      "loss": 0.3276,
      "step": 401
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8559259895097117,
      "learning_rate": 9.913020186795967e-06,
      "loss": 0.3014,
      "step": 402
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8885127454151469,
      "learning_rate": 9.912361936178312e-06,
      "loss": 0.3166,
      "step": 403
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8998851368292131,
      "learning_rate": 9.911701226189105e-06,
      "loss": 0.3296,
      "step": 404
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8004306030984101,
      "learning_rate": 9.911038057159136e-06,
      "loss": 0.2886,
      "step": 405
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.975131175309984,
      "learning_rate": 9.910372429420416e-06,
      "loss": 0.3296,
      "step": 406
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.79899133962689,
      "learning_rate": 9.909704343306194e-06,
      "loss": 0.2859,
      "step": 407
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9013410146014118,
      "learning_rate": 9.909033799150947e-06,
      "loss": 0.2977,
      "step": 408
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9995945826730133,
      "learning_rate": 9.908360797290379e-06,
      "loss": 0.311,
      "step": 409
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9225850188201667,
      "learning_rate": 9.907685338061433e-06,
      "loss": 0.307,
      "step": 410
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8853733911578154,
      "learning_rate": 9.907007421802273e-06,
      "loss": 0.2974,
      "step": 411
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8969890674625314,
      "learning_rate": 9.906327048852301e-06,
      "loss": 0.3218,
      "step": 412
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8867733249502174,
      "learning_rate": 9.90564421955214e-06,
      "loss": 0.2913,
      "step": 413
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.876654611249085,
      "learning_rate": 9.904958934243655e-06,
      "loss": 0.3171,
      "step": 414
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8554359622542629,
      "learning_rate": 9.90427119326993e-06,
      "loss": 0.3182,
      "step": 415
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9021786017308604,
      "learning_rate": 9.903580996975282e-06,
      "loss": 0.3056,
      "step": 416
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9393719075864526,
      "learning_rate": 9.902888345705257e-06,
      "loss": 0.325,
      "step": 417
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9450210504720629,
      "learning_rate": 9.902193239806634e-06,
      "loss": 0.3481,
      "step": 418
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9476851784295617,
      "learning_rate": 9.901495679627415e-06,
      "loss": 0.3222,
      "step": 419
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9305561418534978,
      "learning_rate": 9.90079566551683e-06,
      "loss": 0.2718,
      "step": 420
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9238427903708284,
      "learning_rate": 9.900093197825349e-06,
      "loss": 0.3311,
      "step": 421
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9309238671979215,
      "learning_rate": 9.899388276904653e-06,
      "loss": 0.3385,
      "step": 422
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9686938788560661,
      "learning_rate": 9.898680903107668e-06,
      "loss": 0.3284,
      "step": 423
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8869047227837544,
      "learning_rate": 9.897971076788532e-06,
      "loss": 0.3254,
      "step": 424
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8609700513122369,
      "learning_rate": 9.897258798302626e-06,
      "loss": 0.3151,
      "step": 425
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9625173189694815,
      "learning_rate": 9.89654406800655e-06,
      "loss": 0.3457,
      "step": 426
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.867327394716504,
      "learning_rate": 9.895826886258134e-06,
      "loss": 0.2956,
      "step": 427
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9002350737478294,
      "learning_rate": 9.895107253416434e-06,
      "loss": 0.3254,
      "step": 428
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9071837260689016,
      "learning_rate": 9.894385169841731e-06,
      "loss": 0.3321,
      "step": 429
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.891301246358432,
      "learning_rate": 9.893660635895542e-06,
      "loss": 0.3152,
      "step": 430
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.859193199077908,
      "learning_rate": 9.892933651940597e-06,
      "loss": 0.2805,
      "step": 431
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9629055056587158,
      "learning_rate": 9.892204218340866e-06,
      "loss": 0.3318,
      "step": 432
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8685055682797603,
      "learning_rate": 9.891472335461537e-06,
      "loss": 0.2902,
      "step": 433
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9850143672462021,
      "learning_rate": 9.890738003669029e-06,
      "loss": 0.3162,
      "step": 434
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8353386418665399,
      "learning_rate": 9.890001223330983e-06,
      "loss": 0.2857,
      "step": 435
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8613788406341032,
      "learning_rate": 9.889261994816266e-06,
      "loss": 0.3314,
      "step": 436
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0155239422104487,
      "learning_rate": 9.888520318494976e-06,
      "loss": 0.3252,
      "step": 437
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9277339740882299,
      "learning_rate": 9.887776194738433e-06,
      "loss": 0.3168,
      "step": 438
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8258251003444398,
      "learning_rate": 9.887029623919178e-06,
      "loss": 0.308,
      "step": 439
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8540798901656225,
      "learning_rate": 9.886280606410984e-06,
      "loss": 0.3362,
      "step": 440
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8847906894839436,
      "learning_rate": 9.885529142588845e-06,
      "loss": 0.3407,
      "step": 441
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9822732857979704,
      "learning_rate": 9.884775232828985e-06,
      "loss": 0.3065,
      "step": 442
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8039615941346897,
      "learning_rate": 9.884018877508844e-06,
      "loss": 0.2926,
      "step": 443
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7988415134816012,
      "learning_rate": 9.883260077007092e-06,
      "loss": 0.2994,
      "step": 444
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.903325071048877,
      "learning_rate": 9.882498831703623e-06,
      "loss": 0.3191,
      "step": 445
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9286071225216349,
      "learning_rate": 9.881735141979554e-06,
      "loss": 0.3101,
      "step": 446
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.845018672035287,
      "learning_rate": 9.880969008217225e-06,
      "loss": 0.3068,
      "step": 447
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9494684194455351,
      "learning_rate": 9.8802004308002e-06,
      "loss": 0.3088,
      "step": 448
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9491045328105907,
      "learning_rate": 9.87942941011327e-06,
      "loss": 0.3087,
      "step": 449
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9323861359756581,
      "learning_rate": 9.878655946542443e-06,
      "loss": 0.3105,
      "step": 450
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.836479597192937,
      "learning_rate": 9.877880040474954e-06,
      "loss": 0.2937,
      "step": 451
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9800072424495623,
      "learning_rate": 9.87710169229926e-06,
      "loss": 0.3018,
      "step": 452
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8881390852163128,
      "learning_rate": 9.876320902405041e-06,
      "loss": 0.317,
      "step": 453
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8756649502338383,
      "learning_rate": 9.8755376711832e-06,
      "loss": 0.3269,
      "step": 454
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9526897194012715,
      "learning_rate": 9.87475199902586e-06,
      "loss": 0.3213,
      "step": 455
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9443202897557392,
      "learning_rate": 9.873963886326365e-06,
      "loss": 0.3142,
      "step": 456
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9419498679641796,
      "learning_rate": 9.873173333479289e-06,
      "loss": 0.3339,
      "step": 457
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8479377620359323,
      "learning_rate": 9.872380340880416e-06,
      "loss": 0.3089,
      "step": 458
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0126850701572503,
      "learning_rate": 9.871584908926763e-06,
      "loss": 0.3062,
      "step": 459
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8835967160098732,
      "learning_rate": 9.870787038016556e-06,
      "loss": 0.3089,
      "step": 460
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8807928521099033,
      "learning_rate": 9.869986728549254e-06,
      "loss": 0.2945,
      "step": 461
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9204339784161943,
      "learning_rate": 9.869183980925531e-06,
      "loss": 0.2901,
      "step": 462
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9608702300286008,
      "learning_rate": 9.86837879554728e-06,
      "loss": 0.3282,
      "step": 463
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8525096968683844,
      "learning_rate": 9.867571172817617e-06,
      "loss": 0.3124,
      "step": 464
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.026891309310467,
      "learning_rate": 9.866761113140881e-06,
      "loss": 0.2953,
      "step": 465
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9761465095132643,
      "learning_rate": 9.865948616922623e-06,
      "loss": 0.3064,
      "step": 466
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9357969023846976,
      "learning_rate": 9.865133684569622e-06,
      "loss": 0.3148,
      "step": 467
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9046289043660181,
      "learning_rate": 9.864316316489873e-06,
      "loss": 0.3224,
      "step": 468
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0267923312629916,
      "learning_rate": 9.863496513092592e-06,
      "loss": 0.2972,
      "step": 469
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8367992173339784,
      "learning_rate": 9.862674274788212e-06,
      "loss": 0.2954,
      "step": 470
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9394207287413068,
      "learning_rate": 9.861849601988384e-06,
      "loss": 0.322,
      "step": 471
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9622632792200031,
      "learning_rate": 9.861022495105983e-06,
      "loss": 0.3029,
      "step": 472
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8859155113373961,
      "learning_rate": 9.860192954555099e-06,
      "loss": 0.3122,
      "step": 473
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9311785464055797,
      "learning_rate": 9.85936098075104e-06,
      "loss": 0.3208,
      "step": 474
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9045927763127819,
      "learning_rate": 9.858526574110336e-06,
      "loss": 0.2952,
      "step": 475
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8363892930517255,
      "learning_rate": 9.85768973505073e-06,
      "loss": 0.3023,
      "step": 476
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9232188244867702,
      "learning_rate": 9.856850463991186e-06,
      "loss": 0.2942,
      "step": 477
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9118556408215819,
      "learning_rate": 9.856008761351882e-06,
      "loss": 0.2844,
      "step": 478
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8220516806329685,
      "learning_rate": 9.855164627554219e-06,
      "loss": 0.306,
      "step": 479
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8272057862805671,
      "learning_rate": 9.85431806302081e-06,
      "loss": 0.3079,
      "step": 480
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8366389446873932,
      "learning_rate": 9.853469068175491e-06,
      "loss": 0.3089,
      "step": 481
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8445115223523502,
      "learning_rate": 9.85261764344331e-06,
      "loss": 0.3075,
      "step": 482
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7963940052947625,
      "learning_rate": 9.851763789250526e-06,
      "loss": 0.3108,
      "step": 483
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8546236208385868,
      "learning_rate": 9.850907506024627e-06,
      "loss": 0.328,
      "step": 484
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8909637085789561,
      "learning_rate": 9.850048794194309e-06,
      "loss": 0.3123,
      "step": 485
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8626655390717736,
      "learning_rate": 9.849187654189486e-06,
      "loss": 0.2938,
      "step": 486
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8380424142998599,
      "learning_rate": 9.848324086441287e-06,
      "loss": 0.3153,
      "step": 487
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9307254031905737,
      "learning_rate": 9.847458091382057e-06,
      "loss": 0.3195,
      "step": 488
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8599025145206229,
      "learning_rate": 9.846589669445356e-06,
      "loss": 0.3141,
      "step": 489
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8594073969680509,
      "learning_rate": 9.845718821065957e-06,
      "loss": 0.3054,
      "step": 490
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8725081357271073,
      "learning_rate": 9.844845546679852e-06,
      "loss": 0.306,
      "step": 491
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.811198394388389,
      "learning_rate": 9.843969846724247e-06,
      "loss": 0.2969,
      "step": 492
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9391436639230167,
      "learning_rate": 9.843091721637559e-06,
      "loss": 0.3085,
      "step": 493
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8044040440534261,
      "learning_rate": 9.842211171859418e-06,
      "loss": 0.2828,
      "step": 494
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8866882886668681,
      "learning_rate": 9.841328197830676e-06,
      "loss": 0.3371,
      "step": 495
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8466808869641862,
      "learning_rate": 9.840442799993389e-06,
      "loss": 0.2958,
      "step": 496
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8890554075080945,
      "learning_rate": 9.839554978790832e-06,
      "loss": 0.3154,
      "step": 497
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8678972271729204,
      "learning_rate": 9.838664734667496e-06,
      "loss": 0.3073,
      "step": 498
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8906339956357324,
      "learning_rate": 9.837772068069076e-06,
      "loss": 0.3235,
      "step": 499
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8762968749553193,
      "learning_rate": 9.83687697944249e-06,
      "loss": 0.298,
      "step": 500
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8800412939729314,
      "learning_rate": 9.835979469235857e-06,
      "loss": 0.3093,
      "step": 501
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9250650829835099,
      "learning_rate": 9.835079537898523e-06,
      "loss": 0.3254,
      "step": 502
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8363787714590211,
      "learning_rate": 9.834177185881033e-06,
      "loss": 0.3129,
      "step": 503
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8528261148222819,
      "learning_rate": 9.833272413635151e-06,
      "loss": 0.3207,
      "step": 504
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8091442677932296,
      "learning_rate": 9.83236522161385e-06,
      "loss": 0.2783,
      "step": 505
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9148869830225208,
      "learning_rate": 9.831455610271315e-06,
      "loss": 0.2896,
      "step": 506
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8796599072491815,
      "learning_rate": 9.830543580062944e-06,
      "loss": 0.2875,
      "step": 507
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8141572454569102,
      "learning_rate": 9.829629131445342e-06,
      "loss": 0.32,
      "step": 508
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8042033587032246,
      "learning_rate": 9.82871226487633e-06,
      "loss": 0.2916,
      "step": 509
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8653821890021641,
      "learning_rate": 9.827792980814934e-06,
      "loss": 0.3139,
      "step": 510
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9233752011230871,
      "learning_rate": 9.826871279721396e-06,
      "loss": 0.3146,
      "step": 511
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.85286027250392,
      "learning_rate": 9.825947162057163e-06,
      "loss": 0.3057,
      "step": 512
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9482280065600975,
      "learning_rate": 9.825020628284896e-06,
      "loss": 0.3174,
      "step": 513
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8207247678794052,
      "learning_rate": 9.82409167886846e-06,
      "loss": 0.295,
      "step": 514
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.935026510571052,
      "learning_rate": 9.823160314272937e-06,
      "loss": 0.3121,
      "step": 515
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8631808386809859,
      "learning_rate": 9.822226534964613e-06,
      "loss": 0.2977,
      "step": 516
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8291711041760845,
      "learning_rate": 9.821290341410985e-06,
      "loss": 0.279,
      "step": 517
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8782610389989516,
      "learning_rate": 9.820351734080754e-06,
      "loss": 0.3262,
      "step": 518
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9141515537044687,
      "learning_rate": 9.819410713443837e-06,
      "loss": 0.3083,
      "step": 519
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8434403540257611,
      "learning_rate": 9.818467279971356e-06,
      "loss": 0.2991,
      "step": 520
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8254408599356445,
      "learning_rate": 9.817521434135637e-06,
      "loss": 0.3201,
      "step": 521
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8667796673416904,
      "learning_rate": 9.81657317641022e-06,
      "loss": 0.2745,
      "step": 522
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8927865696447629,
      "learning_rate": 9.81562250726985e-06,
      "loss": 0.31,
      "step": 523
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8347203763916002,
      "learning_rate": 9.814669427190477e-06,
      "loss": 0.3092,
      "step": 524
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9175131095734524,
      "learning_rate": 9.813713936649261e-06,
      "loss": 0.3261,
      "step": 525
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8983544046300977,
      "learning_rate": 9.81275603612457e-06,
      "loss": 0.3234,
      "step": 526
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8551675767667206,
      "learning_rate": 9.811795726095971e-06,
      "loss": 0.3205,
      "step": 527
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9059615454619293,
      "learning_rate": 9.810833007044247e-06,
      "loss": 0.3176,
      "step": 528
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9395562299299104,
      "learning_rate": 9.809867879451384e-06,
      "loss": 0.3088,
      "step": 529
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7930995305134051,
      "learning_rate": 9.808900343800568e-06,
      "loss": 0.2841,
      "step": 530
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.831544056682001,
      "learning_rate": 9.807930400576199e-06,
      "loss": 0.2995,
      "step": 531
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9123723054776914,
      "learning_rate": 9.80695805026388e-06,
      "loss": 0.3053,
      "step": 532
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8534145500116296,
      "learning_rate": 9.805983293350413e-06,
      "loss": 0.2979,
      "step": 533
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8836977348027187,
      "learning_rate": 9.80500613032381e-06,
      "loss": 0.2803,
      "step": 534
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8083822756118431,
      "learning_rate": 9.80402656167329e-06,
      "loss": 0.3081,
      "step": 535
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8530062894515141,
      "learning_rate": 9.803044587889273e-06,
      "loss": 0.3024,
      "step": 536
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8710877893636996,
      "learning_rate": 9.802060209463382e-06,
      "loss": 0.2837,
      "step": 537
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8311955451325409,
      "learning_rate": 9.801073426888447e-06,
      "loss": 0.3038,
      "step": 538
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8483596607742161,
      "learning_rate": 9.800084240658502e-06,
      "loss": 0.3089,
      "step": 539
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8342162763325259,
      "learning_rate": 9.799092651268778e-06,
      "loss": 0.314,
      "step": 540
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9422811959925851,
      "learning_rate": 9.798098659215719e-06,
      "loss": 0.3273,
      "step": 541
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9499853026250528,
      "learning_rate": 9.79710226499696e-06,
      "loss": 0.3294,
      "step": 542
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8771397413381753,
      "learning_rate": 9.796103469111352e-06,
      "loss": 0.274,
      "step": 543
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8223503352281302,
      "learning_rate": 9.795102272058937e-06,
      "loss": 0.3226,
      "step": 544
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9168426187216421,
      "learning_rate": 9.794098674340966e-06,
      "loss": 0.3071,
      "step": 545
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8812945348880485,
      "learning_rate": 9.79309267645989e-06,
      "loss": 0.296,
      "step": 546
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8562482793782277,
      "learning_rate": 9.79208427891936e-06,
      "loss": 0.2971,
      "step": 547
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.802258562365172,
      "learning_rate": 9.791073482224229e-06,
      "loss": 0.3187,
      "step": 548
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8654662604104817,
      "learning_rate": 9.790060286880556e-06,
      "loss": 0.3026,
      "step": 549
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8891913706882685,
      "learning_rate": 9.789044693395593e-06,
      "loss": 0.3261,
      "step": 550
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.828482668221415,
      "learning_rate": 9.788026702277796e-06,
      "loss": 0.3085,
      "step": 551
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8505749104751756,
      "learning_rate": 9.787006314036824e-06,
      "loss": 0.2764,
      "step": 552
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8832485957086176,
      "learning_rate": 9.785983529183533e-06,
      "loss": 0.2982,
      "step": 553
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.838962271357617,
      "learning_rate": 9.784958348229978e-06,
      "loss": 0.3143,
      "step": 554
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8591865486157055,
      "learning_rate": 9.783930771689419e-06,
      "loss": 0.3166,
      "step": 555
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8526978809744375,
      "learning_rate": 9.78290080007631e-06,
      "loss": 0.2977,
      "step": 556
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8037103442597819,
      "learning_rate": 9.781868433906303e-06,
      "loss": 0.3227,
      "step": 557
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9313550536691406,
      "learning_rate": 9.780833673696255e-06,
      "loss": 0.3077,
      "step": 558
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8801548098766419,
      "learning_rate": 9.779796519964217e-06,
      "loss": 0.3188,
      "step": 559
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8540692360309182,
      "learning_rate": 9.778756973229442e-06,
      "loss": 0.3259,
      "step": 560
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8980445109742562,
      "learning_rate": 9.777715034012373e-06,
      "loss": 0.3082,
      "step": 561
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8614661946672458,
      "learning_rate": 9.776670702834662e-06,
      "loss": 0.3121,
      "step": 562
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9156857740663258,
      "learning_rate": 9.775623980219149e-06,
      "loss": 0.3419,
      "step": 563
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8198463363050524,
      "learning_rate": 9.774574866689878e-06,
      "loss": 0.3214,
      "step": 564
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9485678411514338,
      "learning_rate": 9.773523362772084e-06,
      "loss": 0.3192,
      "step": 565
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8429233938358672,
      "learning_rate": 9.772469468992206e-06,
      "loss": 0.3208,
      "step": 566
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8456917281524947,
      "learning_rate": 9.771413185877872e-06,
      "loss": 0.345,
      "step": 567
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8587422187096292,
      "learning_rate": 9.77035451395791e-06,
      "loss": 0.302,
      "step": 568
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9142647641681578,
      "learning_rate": 9.769293453762344e-06,
      "loss": 0.3204,
      "step": 569
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9020119357986989,
      "learning_rate": 9.768230005822394e-06,
      "loss": 0.3002,
      "step": 570
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8046264349273933,
      "learning_rate": 9.767164170670476e-06,
      "loss": 0.2972,
      "step": 571
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8262308472253456,
      "learning_rate": 9.766095948840195e-06,
      "loss": 0.3079,
      "step": 572
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8828556967724037,
      "learning_rate": 9.76502534086636e-06,
      "loss": 0.3528,
      "step": 573
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7645971591760368,
      "learning_rate": 9.76395234728497e-06,
      "loss": 0.288,
      "step": 574
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8338060398145799,
      "learning_rate": 9.762876968633218e-06,
      "loss": 0.3181,
      "step": 575
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8490633069425626,
      "learning_rate": 9.761799205449492e-06,
      "loss": 0.3244,
      "step": 576
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8659877472457195,
      "learning_rate": 9.760719058273371e-06,
      "loss": 0.3115,
      "step": 577
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8735481917307428,
      "learning_rate": 9.759636527645633e-06,
      "loss": 0.3129,
      "step": 578
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8605468363189038,
      "learning_rate": 9.758551614108247e-06,
      "loss": 0.3112,
      "step": 579
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7918563262565805,
      "learning_rate": 9.757464318204374e-06,
      "loss": 0.2858,
      "step": 580
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7806457261288615,
      "learning_rate": 9.756374640478363e-06,
      "loss": 0.2875,
      "step": 581
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.803943270251532,
      "learning_rate": 9.755282581475769e-06,
      "loss": 0.3155,
      "step": 582
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.829358747335961,
      "learning_rate": 9.754188141743326e-06,
      "loss": 0.3158,
      "step": 583
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8555882051458769,
      "learning_rate": 9.753091321828964e-06,
      "loss": 0.314,
      "step": 584
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9059662915528343,
      "learning_rate": 9.751992122281809e-06,
      "loss": 0.3392,
      "step": 585
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8357800162392397,
      "learning_rate": 9.750890543652169e-06,
      "loss": 0.2983,
      "step": 586
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9125315990366426,
      "learning_rate": 9.749786586491557e-06,
      "loss": 0.3198,
      "step": 587
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8376922659394419,
      "learning_rate": 9.74868025135266e-06,
      "loss": 0.2972,
      "step": 588
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.871109250801388,
      "learning_rate": 9.74757153878937e-06,
      "loss": 0.2965,
      "step": 589
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9581192613061694,
      "learning_rate": 9.74646044935676e-06,
      "loss": 0.3154,
      "step": 590
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9063475073647052,
      "learning_rate": 9.7453469836111e-06,
      "loss": 0.3068,
      "step": 591
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8266369173135708,
      "learning_rate": 9.744231142109842e-06,
      "loss": 0.2965,
      "step": 592
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7825926477043852,
      "learning_rate": 9.743112925411633e-06,
      "loss": 0.2934,
      "step": 593
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8236338211431135,
      "learning_rate": 9.741992334076309e-06,
      "loss": 0.2928,
      "step": 594
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8178654040433289,
      "learning_rate": 9.740869368664894e-06,
      "loss": 0.2969,
      "step": 595
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9281531100784395,
      "learning_rate": 9.739744029739596e-06,
      "loss": 0.3301,
      "step": 596
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7951363732868675,
      "learning_rate": 9.738616317863819e-06,
      "loss": 0.2822,
      "step": 597
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8725134740556669,
      "learning_rate": 9.737486233602149e-06,
      "loss": 0.3154,
      "step": 598
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9955093794855342,
      "learning_rate": 9.736353777520365e-06,
      "loss": 0.3444,
      "step": 599
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8073441627922676,
      "learning_rate": 9.735218950185428e-06,
      "loss": 0.2831,
      "step": 600
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8481350389823527,
      "learning_rate": 9.734081752165491e-06,
      "loss": 0.2858,
      "step": 601
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8274907956395389,
      "learning_rate": 9.73294218402989e-06,
      "loss": 0.2905,
      "step": 602
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8602188034500966,
      "learning_rate": 9.731800246349149e-06,
      "loss": 0.313,
      "step": 603
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8700185600035144,
      "learning_rate": 9.730655939694979e-06,
      "loss": 0.306,
      "step": 604
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7913774539503676,
      "learning_rate": 9.729509264640276e-06,
      "loss": 0.2747,
      "step": 605
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8835373579018234,
      "learning_rate": 9.728360221759125e-06,
      "loss": 0.3274,
      "step": 606
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9264367931609369,
      "learning_rate": 9.72720881162679e-06,
      "loss": 0.3195,
      "step": 607
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8128263300153058,
      "learning_rate": 9.726055034819726e-06,
      "loss": 0.3144,
      "step": 608
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8085666000686689,
      "learning_rate": 9.72489889191557e-06,
      "loss": 0.2989,
      "step": 609
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8844918825677812,
      "learning_rate": 9.723740383493146e-06,
      "loss": 0.3298,
      "step": 610
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8766993303197957,
      "learning_rate": 9.722579510132458e-06,
      "loss": 0.2878,
      "step": 611
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8035261507159634,
      "learning_rate": 9.721416272414698e-06,
      "loss": 0.285,
      "step": 612
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8360838303904843,
      "learning_rate": 9.720250670922242e-06,
      "loss": 0.3023,
      "step": 613
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7800936287454288,
      "learning_rate": 9.719082706238645e-06,
      "loss": 0.2836,
      "step": 614
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7823835394460914,
      "learning_rate": 9.717912378948651e-06,
      "loss": 0.2918,
      "step": 615
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8126368826768864,
      "learning_rate": 9.716739689638179e-06,
      "loss": 0.2852,
      "step": 616
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8025546026194916,
      "learning_rate": 9.715564638894342e-06,
      "loss": 0.288,
      "step": 617
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7882541143888225,
      "learning_rate": 9.714387227305422e-06,
      "loss": 0.2613,
      "step": 618
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8608805887176507,
      "learning_rate": 9.713207455460893e-06,
      "loss": 0.2921,
      "step": 619
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.784994812248838,
      "learning_rate": 9.712025323951406e-06,
      "loss": 0.2802,
      "step": 620
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9126176597642491,
      "learning_rate": 9.710840833368796e-06,
      "loss": 0.3134,
      "step": 621
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9084183711116995,
      "learning_rate": 9.709653984306077e-06,
      "loss": 0.3348,
      "step": 622
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8138550855694298,
      "learning_rate": 9.708464777357444e-06,
      "loss": 0.2934,
      "step": 623
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8280322768366619,
      "learning_rate": 9.70727321311827e-06,
      "loss": 0.3185,
      "step": 624
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8299154276413793,
      "learning_rate": 9.706079292185119e-06,
      "loss": 0.3085,
      "step": 625
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9024474762550658,
      "learning_rate": 9.70488301515572e-06,
      "loss": 0.3049,
      "step": 626
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8261585489187697,
      "learning_rate": 9.70368438262899e-06,
      "loss": 0.3316,
      "step": 627
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.865025379678509,
      "learning_rate": 9.702483395205023e-06,
      "loss": 0.2948,
      "step": 628
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7852853849545657,
      "learning_rate": 9.701280053485096e-06,
      "loss": 0.3034,
      "step": 629
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8106515669989987,
      "learning_rate": 9.700074358071658e-06,
      "loss": 0.2915,
      "step": 630
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.789864876252586,
      "learning_rate": 9.698866309568341e-06,
      "loss": 0.316,
      "step": 631
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8260838858522519,
      "learning_rate": 9.697655908579958e-06,
      "loss": 0.3117,
      "step": 632
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8323290453896688,
      "learning_rate": 9.696443155712488e-06,
      "loss": 0.3236,
      "step": 633
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.780537861555731,
      "learning_rate": 9.695228051573098e-06,
      "loss": 0.318,
      "step": 634
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.811626786760629,
      "learning_rate": 9.694010596770132e-06,
      "loss": 0.2906,
      "step": 635
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8455696385273999,
      "learning_rate": 9.692790791913107e-06,
      "loss": 0.3167,
      "step": 636
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8052964227338938,
      "learning_rate": 9.691568637612714e-06,
      "loss": 0.3022,
      "step": 637
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8672567291134518,
      "learning_rate": 9.69034413448083e-06,
      "loss": 0.331,
      "step": 638
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8719530480237986,
      "learning_rate": 9.689117283130498e-06,
      "loss": 0.331,
      "step": 639
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8834071580114524,
      "learning_rate": 9.68788808417594e-06,
      "loss": 0.3231,
      "step": 640
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8100938500160219,
      "learning_rate": 9.686656538232558e-06,
      "loss": 0.3031,
      "step": 641
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.810517388510763,
      "learning_rate": 9.68542264591692e-06,
      "loss": 0.2922,
      "step": 642
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8060123500161532,
      "learning_rate": 9.684186407846774e-06,
      "loss": 0.2967,
      "step": 643
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8803655852805512,
      "learning_rate": 9.682947824641045e-06,
      "loss": 0.3087,
      "step": 644
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8119361214128186,
      "learning_rate": 9.681706896919829e-06,
      "loss": 0.3021,
      "step": 645
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8981388422153652,
      "learning_rate": 9.680463625304393e-06,
      "loss": 0.2877,
      "step": 646
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8652520574486162,
      "learning_rate": 9.679218010417182e-06,
      "loss": 0.3054,
      "step": 647
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7563843575223439,
      "learning_rate": 9.677970052881811e-06,
      "loss": 0.2977,
      "step": 648
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8436782591487094,
      "learning_rate": 9.676719753323074e-06,
      "loss": 0.2988,
      "step": 649
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9123457779371446,
      "learning_rate": 9.675467112366925e-06,
      "loss": 0.294,
      "step": 650
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8163498088577138,
      "learning_rate": 9.674212130640506e-06,
      "loss": 0.3001,
      "step": 651
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7619345515130359,
      "learning_rate": 9.672954808772118e-06,
      "loss": 0.2865,
      "step": 652
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8270041790158729,
      "learning_rate": 9.67169514739124e-06,
      "loss": 0.3236,
      "step": 653
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8040635199746899,
      "learning_rate": 9.670433147128522e-06,
      "loss": 0.2766,
      "step": 654
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8168285004127489,
      "learning_rate": 9.669168808615781e-06,
      "loss": 0.3122,
      "step": 655
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8434994670899189,
      "learning_rate": 9.667902132486009e-06,
      "loss": 0.325,
      "step": 656
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8684229345439404,
      "learning_rate": 9.666633119373368e-06,
      "loss": 0.3304,
      "step": 657
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0086778319501606,
      "learning_rate": 9.665361769913187e-06,
      "loss": 0.3193,
      "step": 658
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8184393697727449,
      "learning_rate": 9.664088084741969e-06,
      "loss": 0.305,
      "step": 659
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8216883779236501,
      "learning_rate": 9.66281206449738e-06,
      "loss": 0.2919,
      "step": 660
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8249417427106482,
      "learning_rate": 9.661533709818261e-06,
      "loss": 0.2878,
      "step": 661
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7683224430080128,
      "learning_rate": 9.660253021344618e-06,
      "loss": 0.2873,
      "step": 662
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8439795410035752,
      "learning_rate": 9.658969999717631e-06,
      "loss": 0.3025,
      "step": 663
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9243889925522719,
      "learning_rate": 9.65768464557964e-06,
      "loss": 0.3287,
      "step": 664
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8608808619223792,
      "learning_rate": 9.656396959574161e-06,
      "loss": 0.3069,
      "step": 665
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8105806764833788,
      "learning_rate": 9.65510694234587e-06,
      "loss": 0.2865,
      "step": 666
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.802151954280416,
      "learning_rate": 9.653814594540614e-06,
      "loss": 0.2979,
      "step": 667
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.813828149469213,
      "learning_rate": 9.652519916805406e-06,
      "loss": 0.3125,
      "step": 668
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8198285990081854,
      "learning_rate": 9.651222909788428e-06,
      "loss": 0.2718,
      "step": 669
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9076996404684354,
      "learning_rate": 9.649923574139023e-06,
      "loss": 0.3403,
      "step": 670
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.848846002086066,
      "learning_rate": 9.648621910507704e-06,
      "loss": 0.33,
      "step": 671
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8550609863044285,
      "learning_rate": 9.64731791954615e-06,
      "loss": 0.2968,
      "step": 672
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8171664341664818,
      "learning_rate": 9.6460116019072e-06,
      "loss": 0.2798,
      "step": 673
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.870850060737812,
      "learning_rate": 9.644702958244865e-06,
      "loss": 0.3139,
      "step": 674
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7697584025081888,
      "learning_rate": 9.643391989214312e-06,
      "loss": 0.2837,
      "step": 675
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8393453909488604,
      "learning_rate": 9.642078695471882e-06,
      "loss": 0.304,
      "step": 676
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7808790363606153,
      "learning_rate": 9.640763077675072e-06,
      "loss": 0.2911,
      "step": 677
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8520556385301543,
      "learning_rate": 9.639445136482549e-06,
      "loss": 0.292,
      "step": 678
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9256725390982844,
      "learning_rate": 9.638124872554134e-06,
      "loss": 0.3391,
      "step": 679
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.808141370340325,
      "learning_rate": 9.636802286550817e-06,
      "loss": 0.3031,
      "step": 680
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.901205954193458,
      "learning_rate": 9.635477379134756e-06,
      "loss": 0.293,
      "step": 681
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9412083197734361,
      "learning_rate": 9.634150150969259e-06,
      "loss": 0.3106,
      "step": 682
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.797614593887533,
      "learning_rate": 9.632820602718806e-06,
      "loss": 0.2815,
      "step": 683
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8352159676570906,
      "learning_rate": 9.631488735049033e-06,
      "loss": 0.2908,
      "step": 684
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8855573768890841,
      "learning_rate": 9.630154548626737e-06,
      "loss": 0.3433,
      "step": 685
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8042250661815635,
      "learning_rate": 9.628818044119884e-06,
      "loss": 0.2808,
      "step": 686
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7632232524213021,
      "learning_rate": 9.627479222197587e-06,
      "loss": 0.3051,
      "step": 687
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8171019446877292,
      "learning_rate": 9.62613808353013e-06,
      "loss": 0.3251,
      "step": 688
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8868909907862387,
      "learning_rate": 9.624794628788954e-06,
      "loss": 0.3061,
      "step": 689
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8272844221647714,
      "learning_rate": 9.623448858646658e-06,
      "loss": 0.3122,
      "step": 690
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8659969349750338,
      "learning_rate": 9.622100773777002e-06,
      "loss": 0.3194,
      "step": 691
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7543652858670681,
      "learning_rate": 9.620750374854901e-06,
      "loss": 0.2907,
      "step": 692
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8188232511362676,
      "learning_rate": 9.619397662556434e-06,
      "loss": 0.3513,
      "step": 693
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7956860643304635,
      "learning_rate": 9.618042637558838e-06,
      "loss": 0.2858,
      "step": 694
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8115295108142085,
      "learning_rate": 9.6166853005405e-06,
      "loss": 0.3122,
      "step": 695
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8611591543126575,
      "learning_rate": 9.615325652180976e-06,
      "loss": 0.3191,
      "step": 696
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.823279774154387,
      "learning_rate": 9.613963693160968e-06,
      "loss": 0.3001,
      "step": 697
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8435109251498641,
      "learning_rate": 9.612599424162344e-06,
      "loss": 0.3006,
      "step": 698
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8361758183666801,
      "learning_rate": 9.611232845868124e-06,
      "loss": 0.2941,
      "step": 699
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8656493806097808,
      "learning_rate": 9.609863958962483e-06,
      "loss": 0.3293,
      "step": 700
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7886140028397411,
      "learning_rate": 9.608492764130752e-06,
      "loss": 0.3026,
      "step": 701
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8716097816128152,
      "learning_rate": 9.607119262059426e-06,
      "loss": 0.3026,
      "step": 702
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7482595777036986,
      "learning_rate": 9.60574345343614e-06,
      "loss": 0.2686,
      "step": 703
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8301368821282225,
      "learning_rate": 9.604365338949697e-06,
      "loss": 0.3091,
      "step": 704
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7776251325574932,
      "learning_rate": 9.602984919290049e-06,
      "loss": 0.2983,
      "step": 705
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8481678128854991,
      "learning_rate": 9.601602195148299e-06,
      "loss": 0.2692,
      "step": 706
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8883164617678649,
      "learning_rate": 9.60021716721671e-06,
      "loss": 0.3202,
      "step": 707
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7856869710682669,
      "learning_rate": 9.598829836188694e-06,
      "loss": 0.2986,
      "step": 708
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7597128679975649,
      "learning_rate": 9.59744020275882e-06,
      "loss": 0.2916,
      "step": 709
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7803457300014387,
      "learning_rate": 9.596048267622806e-06,
      "loss": 0.3061,
      "step": 710
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8921925793858808,
      "learning_rate": 9.594654031477522e-06,
      "loss": 0.3036,
      "step": 711
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8601805530127352,
      "learning_rate": 9.593257495020993e-06,
      "loss": 0.3014,
      "step": 712
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.799017303774288,
      "learning_rate": 9.591858658952396e-06,
      "loss": 0.2811,
      "step": 713
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8952309066333284,
      "learning_rate": 9.590457523972055e-06,
      "loss": 0.2914,
      "step": 714
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8560352223745608,
      "learning_rate": 9.58905409078145e-06,
      "loss": 0.2889,
      "step": 715
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8169443364430691,
      "learning_rate": 9.587648360083205e-06,
      "loss": 0.3107,
      "step": 716
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.751671602217226,
      "learning_rate": 9.586240332581101e-06,
      "loss": 0.2847,
      "step": 717
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9043051167541025,
      "learning_rate": 9.584830008980068e-06,
      "loss": 0.3123,
      "step": 718
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8254904450612416,
      "learning_rate": 9.58341738998618e-06,
      "loss": 0.314,
      "step": 719
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8134472323573736,
      "learning_rate": 9.582002476306668e-06,
      "loss": 0.2929,
      "step": 720
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8040519359121954,
      "learning_rate": 9.580585268649904e-06,
      "loss": 0.3021,
      "step": 721
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7475873786194283,
      "learning_rate": 9.579165767725416e-06,
      "loss": 0.2685,
      "step": 722
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8652557051708345,
      "learning_rate": 9.577743974243875e-06,
      "loss": 0.2906,
      "step": 723
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8899841623764232,
      "learning_rate": 9.576319888917099e-06,
      "loss": 0.2963,
      "step": 724
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7891547478999151,
      "learning_rate": 9.574893512458057e-06,
      "loss": 0.2775,
      "step": 725
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8372668415260228,
      "learning_rate": 9.573464845580864e-06,
      "loss": 0.3029,
      "step": 726
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7785904509036484,
      "learning_rate": 9.572033889000783e-06,
      "loss": 0.2939,
      "step": 727
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7447504744895234,
      "learning_rate": 9.570600643434217e-06,
      "loss": 0.3102,
      "step": 728
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7834308827894126,
      "learning_rate": 9.569165109598725e-06,
      "loss": 0.3117,
      "step": 729
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7620849648799713,
      "learning_rate": 9.567727288213005e-06,
      "loss": 0.2812,
      "step": 730
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.822747848997475,
      "learning_rate": 9.5662871799969e-06,
      "loss": 0.29,
      "step": 731
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8260357113714262,
      "learning_rate": 9.5648447856714e-06,
      "loss": 0.3098,
      "step": 732
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7991547165233592,
      "learning_rate": 9.563400105958638e-06,
      "loss": 0.299,
      "step": 733
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8126439850949115,
      "learning_rate": 9.561953141581895e-06,
      "loss": 0.3061,
      "step": 734
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8263272773822083,
      "learning_rate": 9.560503893265589e-06,
      "loss": 0.293,
      "step": 735
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7572113991853195,
      "learning_rate": 9.559052361735288e-06,
      "loss": 0.253,
      "step": 736
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8032682702626579,
      "learning_rate": 9.557598547717702e-06,
      "loss": 0.2977,
      "step": 737
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7798549697507038,
      "learning_rate": 9.55614245194068e-06,
      "loss": 0.2839,
      "step": 738
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8071062767922684,
      "learning_rate": 9.554684075133215e-06,
      "loss": 0.3171,
      "step": 739
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.841512718563835,
      "learning_rate": 9.553223418025443e-06,
      "loss": 0.328,
      "step": 740
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.812478578893319,
      "learning_rate": 9.551760481348644e-06,
      "loss": 0.2854,
      "step": 741
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8195809412992848,
      "learning_rate": 9.550295265835235e-06,
      "loss": 0.3246,
      "step": 742
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7431057105951361,
      "learning_rate": 9.548827772218772e-06,
      "loss": 0.2767,
      "step": 743
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7732871494184458,
      "learning_rate": 9.547358001233959e-06,
      "loss": 0.3075,
      "step": 744
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7975585273336443,
      "learning_rate": 9.545885953616636e-06,
      "loss": 0.2982,
      "step": 745
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7528030722776308,
      "learning_rate": 9.544411630103782e-06,
      "loss": 0.2877,
      "step": 746
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7773930191461702,
      "learning_rate": 9.542935031433515e-06,
      "loss": 0.2908,
      "step": 747
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8226263314818045,
      "learning_rate": 9.541456158345094e-06,
      "loss": 0.292,
      "step": 748
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8615497428558148,
      "learning_rate": 9.539975011578918e-06,
      "loss": 0.3156,
      "step": 749
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7540926287364709,
      "learning_rate": 9.538491591876522e-06,
      "loss": 0.3092,
      "step": 750
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7970933790808782,
      "learning_rate": 9.537005899980576e-06,
      "loss": 0.2748,
      "step": 751
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8648695764426515,
      "learning_rate": 9.535517936634895e-06,
      "loss": 0.2987,
      "step": 752
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7780262383130099,
      "learning_rate": 9.534027702584425e-06,
      "loss": 0.2881,
      "step": 753
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8006503073185438,
      "learning_rate": 9.53253519857525e-06,
      "loss": 0.3047,
      "step": 754
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9001589234491454,
      "learning_rate": 9.531040425354595e-06,
      "loss": 0.2941,
      "step": 755
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8682791512103175,
      "learning_rate": 9.529543383670814e-06,
      "loss": 0.3075,
      "step": 756
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8564328791187629,
      "learning_rate": 9.5280440742734e-06,
      "loss": 0.327,
      "step": 757
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8947097019058883,
      "learning_rate": 9.526542497912984e-06,
      "loss": 0.3007,
      "step": 758
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8485344819921078,
      "learning_rate": 9.52503865534133e-06,
      "loss": 0.2976,
      "step": 759
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8169501392652272,
      "learning_rate": 9.523532547311332e-06,
      "loss": 0.3122,
      "step": 760
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.819871994724038,
      "learning_rate": 9.522024174577025e-06,
      "loss": 0.3119,
      "step": 761
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8149377276236295,
      "learning_rate": 9.520513537893574e-06,
      "loss": 0.2924,
      "step": 762
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8879847586058934,
      "learning_rate": 9.51900063801728e-06,
      "loss": 0.3433,
      "step": 763
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9012331124890404,
      "learning_rate": 9.517485475705574e-06,
      "loss": 0.3319,
      "step": 764
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7541937370053963,
      "learning_rate": 9.515968051717022e-06,
      "loss": 0.2853,
      "step": 765
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8390855365644445,
      "learning_rate": 9.514448366811322e-06,
      "loss": 0.2894,
      "step": 766
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8754239927358629,
      "learning_rate": 9.512926421749305e-06,
      "loss": 0.2949,
      "step": 767
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7835030089891242,
      "learning_rate": 9.511402217292927e-06,
      "loss": 0.3072,
      "step": 768
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8440657409892679,
      "learning_rate": 9.509875754205284e-06,
      "loss": 0.3009,
      "step": 769
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7924212322032882,
      "learning_rate": 9.508347033250597e-06,
      "loss": 0.2933,
      "step": 770
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7910035108147596,
      "learning_rate": 9.506816055194223e-06,
      "loss": 0.2952,
      "step": 771
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8032288503079156,
      "learning_rate": 9.505282820802642e-06,
      "loss": 0.3239,
      "step": 772
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8184046159781967,
      "learning_rate": 9.503747330843468e-06,
      "loss": 0.2887,
      "step": 773
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8360062699307895,
      "learning_rate": 9.502209586085444e-06,
      "loss": 0.3202,
      "step": 774
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8140277023647602,
      "learning_rate": 9.50066958729844e-06,
      "loss": 0.3163,
      "step": 775
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.812845901012854,
      "learning_rate": 9.499127335253456e-06,
      "loss": 0.2965,
      "step": 776
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8225465676606755,
      "learning_rate": 9.497582830722617e-06,
      "loss": 0.2954,
      "step": 777
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.750593252791941,
      "learning_rate": 9.496036074479184e-06,
      "loss": 0.314,
      "step": 778
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8238367194340097,
      "learning_rate": 9.494487067297538e-06,
      "loss": 0.2928,
      "step": 779
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7847472294144454,
      "learning_rate": 9.492935809953185e-06,
      "loss": 0.2839,
      "step": 780
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8908414368963439,
      "learning_rate": 9.491382303222765e-06,
      "loss": 0.3349,
      "step": 781
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.824158127190454,
      "learning_rate": 9.48982654788404e-06,
      "loss": 0.2807,
      "step": 782
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8410510352484827,
      "learning_rate": 9.488268544715897e-06,
      "loss": 0.3053,
      "step": 783
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8091492647666575,
      "learning_rate": 9.486708294498348e-06,
      "loss": 0.3018,
      "step": 784
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7643924662168611,
      "learning_rate": 9.485145798012536e-06,
      "loss": 0.2952,
      "step": 785
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8486317669254456,
      "learning_rate": 9.48358105604072e-06,
      "loss": 0.2808,
      "step": 786
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9140127340062632,
      "learning_rate": 9.482014069366288e-06,
      "loss": 0.2763,
      "step": 787
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7960284952428938,
      "learning_rate": 9.480444838773753e-06,
      "loss": 0.2953,
      "step": 788
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8249342177513971,
      "learning_rate": 9.478873365048748e-06,
      "loss": 0.2984,
      "step": 789
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8026873194313119,
      "learning_rate": 9.47729964897803e-06,
      "loss": 0.3124,
      "step": 790
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7391566658844789,
      "learning_rate": 9.475723691349482e-06,
      "loss": 0.2871,
      "step": 791
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7423890541239292,
      "learning_rate": 9.474145492952102e-06,
      "loss": 0.3005,
      "step": 792
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8032394747341511,
      "learning_rate": 9.472565054576017e-06,
      "loss": 0.3149,
      "step": 793
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8630003577402069,
      "learning_rate": 9.470982377012475e-06,
      "loss": 0.3019,
      "step": 794
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8675593784333347,
      "learning_rate": 9.469397461053838e-06,
      "loss": 0.2928,
      "step": 795
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7507585918414789,
      "learning_rate": 9.467810307493595e-06,
      "loss": 0.2986,
      "step": 796
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7879760623115833,
      "learning_rate": 9.466220917126354e-06,
      "loss": 0.2983,
      "step": 797
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8040848336359661,
      "learning_rate": 9.464629290747844e-06,
      "loss": 0.3291,
      "step": 798
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8569984302549308,
      "learning_rate": 9.463035429154909e-06,
      "loss": 0.3255,
      "step": 799
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8082790290493358,
      "learning_rate": 9.461439333145516e-06,
      "loss": 0.3067,
      "step": 800
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7911231389366751,
      "learning_rate": 9.459841003518752e-06,
      "loss": 0.3006,
      "step": 801
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8067439015366548,
      "learning_rate": 9.458240441074818e-06,
      "loss": 0.3143,
      "step": 802
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7872708961100191,
      "learning_rate": 9.456637646615035e-06,
      "loss": 0.3005,
      "step": 803
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7806318642140346,
      "learning_rate": 9.45503262094184e-06,
      "loss": 0.2963,
      "step": 804
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7991048719547609,
      "learning_rate": 9.45342536485879e-06,
      "loss": 0.3151,
      "step": 805
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8414188109372203,
      "learning_rate": 9.451815879170556e-06,
      "loss": 0.3294,
      "step": 806
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7533833074299724,
      "learning_rate": 9.450204164682928e-06,
      "loss": 0.2959,
      "step": 807
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8055056761899321,
      "learning_rate": 9.448590222202808e-06,
      "loss": 0.3193,
      "step": 808
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8195633743330886,
      "learning_rate": 9.446974052538216e-06,
      "loss": 0.3122,
      "step": 809
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8592491635970398,
      "learning_rate": 9.445355656498284e-06,
      "loss": 0.3102,
      "step": 810
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9003442568970289,
      "learning_rate": 9.44373503489327e-06,
      "loss": 0.317,
      "step": 811
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7162480687143082,
      "learning_rate": 9.442112188534526e-06,
      "loss": 0.2396,
      "step": 812
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7649002889367367,
      "learning_rate": 9.440487118234536e-06,
      "loss": 0.2913,
      "step": 813
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8397943581457566,
      "learning_rate": 9.438859824806889e-06,
      "loss": 0.2821,
      "step": 814
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8508499610305502,
      "learning_rate": 9.43723030906629e-06,
      "loss": 0.3271,
      "step": 815
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8215990680536386,
      "learning_rate": 9.435598571828552e-06,
      "loss": 0.3074,
      "step": 816
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8937952172327064,
      "learning_rate": 9.433964613910608e-06,
      "loss": 0.32,
      "step": 817
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7546855577600474,
      "learning_rate": 9.432328436130493e-06,
      "loss": 0.2929,
      "step": 818
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.838541897372278,
      "learning_rate": 9.430690039307364e-06,
      "loss": 0.2789,
      "step": 819
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7934991321373599,
      "learning_rate": 9.429049424261478e-06,
      "loss": 0.308,
      "step": 820
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7910209494837097,
      "learning_rate": 9.427406591814213e-06,
      "loss": 0.3025,
      "step": 821
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.877111389265839,
      "learning_rate": 9.425761542788049e-06,
      "loss": 0.2996,
      "step": 822
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8137851908102863,
      "learning_rate": 9.42411427800658e-06,
      "loss": 0.2967,
      "step": 823
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.820120953321379,
      "learning_rate": 9.422464798294509e-06,
      "loss": 0.3158,
      "step": 824
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7645307240516589,
      "learning_rate": 9.420813104477647e-06,
      "loss": 0.2954,
      "step": 825
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7768059004061552,
      "learning_rate": 9.419159197382913e-06,
      "loss": 0.2954,
      "step": 826
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8940971803648271,
      "learning_rate": 9.417503077838334e-06,
      "loss": 0.293,
      "step": 827
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.831757375408671,
      "learning_rate": 9.415844746673047e-06,
      "loss": 0.3153,
      "step": 828
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7705481359802209,
      "learning_rate": 9.414184204717294e-06,
      "loss": 0.3132,
      "step": 829
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7792053043685653,
      "learning_rate": 9.412521452802425e-06,
      "loss": 0.3244,
      "step": 830
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8191129400730877,
      "learning_rate": 9.410856491760896e-06,
      "loss": 0.2826,
      "step": 831
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9689029535935347,
      "learning_rate": 9.409189322426267e-06,
      "loss": 0.3145,
      "step": 832
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8305804020628365,
      "learning_rate": 9.40751994563321e-06,
      "loss": 0.3166,
      "step": 833
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7376517188733894,
      "learning_rate": 9.40584836221749e-06,
      "loss": 0.2965,
      "step": 834
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8192911149618435,
      "learning_rate": 9.404174573015992e-06,
      "loss": 0.3161,
      "step": 835
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8322825841116768,
      "learning_rate": 9.402498578866695e-06,
      "loss": 0.3062,
      "step": 836
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8173436607098817,
      "learning_rate": 9.400820380608684e-06,
      "loss": 0.2878,
      "step": 837
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8079788411029656,
      "learning_rate": 9.399139979082148e-06,
      "loss": 0.2826,
      "step": 838
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7740760846747937,
      "learning_rate": 9.39745737512838e-06,
      "loss": 0.2933,
      "step": 839
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8249674271348998,
      "learning_rate": 9.395772569589774e-06,
      "loss": 0.3038,
      "step": 840
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8563732667123785,
      "learning_rate": 9.394085563309827e-06,
      "loss": 0.3023,
      "step": 841
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7948160864554199,
      "learning_rate": 9.39239635713314e-06,
      "loss": 0.301,
      "step": 842
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8228514954418829,
      "learning_rate": 9.390704951905412e-06,
      "loss": 0.3009,
      "step": 843
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7890645213151456,
      "learning_rate": 9.389011348473442e-06,
      "loss": 0.2961,
      "step": 844
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7909544231025358,
      "learning_rate": 9.387315547685134e-06,
      "loss": 0.3075,
      "step": 845
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7290172074628286,
      "learning_rate": 9.38561755038949e-06,
      "loss": 0.2833,
      "step": 846
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7968146044372687,
      "learning_rate": 9.38391735743661e-06,
      "loss": 0.301,
      "step": 847
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7410567591858324,
      "learning_rate": 9.382214969677697e-06,
      "loss": 0.2911,
      "step": 848
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7572626371519674,
      "learning_rate": 9.380510387965047e-06,
      "loss": 0.3018,
      "step": 849
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7832703119725869,
      "learning_rate": 9.37880361315206e-06,
      "loss": 0.2783,
      "step": 850
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7636798444390851,
      "learning_rate": 9.377094646093234e-06,
      "loss": 0.2812,
      "step": 851
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7990375043196898,
      "learning_rate": 9.375383487644161e-06,
      "loss": 0.3059,
      "step": 852
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.789985947424577,
      "learning_rate": 9.37367013866153e-06,
      "loss": 0.303,
      "step": 853
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7803353136321165,
      "learning_rate": 9.37195460000313e-06,
      "loss": 0.3012,
      "step": 854
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7591229876293307,
      "learning_rate": 9.370236872527845e-06,
      "loss": 0.2822,
      "step": 855
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7999271207281221,
      "learning_rate": 9.368516957095652e-06,
      "loss": 0.3208,
      "step": 856
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7459012137967113,
      "learning_rate": 9.366794854567627e-06,
      "loss": 0.292,
      "step": 857
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7900876522723778,
      "learning_rate": 9.365070565805941e-06,
      "loss": 0.3141,
      "step": 858
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8421131207579161,
      "learning_rate": 9.363344091673858e-06,
      "loss": 0.2726,
      "step": 859
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8090028842478079,
      "learning_rate": 9.361615433035736e-06,
      "loss": 0.2877,
      "step": 860
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7237225056929641,
      "learning_rate": 9.359884590757027e-06,
      "loss": 0.2679,
      "step": 861
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7611245350705664,
      "learning_rate": 9.358151565704275e-06,
      "loss": 0.2865,
      "step": 862
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7991156382018092,
      "learning_rate": 9.356416358745119e-06,
      "loss": 0.3193,
      "step": 863
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.829623902251678,
      "learning_rate": 9.35467897074829e-06,
      "loss": 0.2848,
      "step": 864
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8067849475228096,
      "learning_rate": 9.352939402583611e-06,
      "loss": 0.3001,
      "step": 865
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7451000910091446,
      "learning_rate": 9.351197655121995e-06,
      "loss": 0.2812,
      "step": 866
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8403069378873598,
      "learning_rate": 9.349453729235447e-06,
      "loss": 0.2921,
      "step": 867
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7889312361078833,
      "learning_rate": 9.347707625797062e-06,
      "loss": 0.2924,
      "step": 868
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9168213228964419,
      "learning_rate": 9.345959345681028e-06,
      "loss": 0.3104,
      "step": 869
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7715333603734136,
      "learning_rate": 9.34420888976262e-06,
      "loss": 0.2928,
      "step": 870
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8209652956291212,
      "learning_rate": 9.342456258918201e-06,
      "loss": 0.3072,
      "step": 871
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7684000556042018,
      "learning_rate": 9.340701454025224e-06,
      "loss": 0.2959,
      "step": 872
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7284979161964286,
      "learning_rate": 9.338944475962236e-06,
      "loss": 0.2858,
      "step": 873
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8684075118781358,
      "learning_rate": 9.337185325608864e-06,
      "loss": 0.3085,
      "step": 874
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.732199507679947,
      "learning_rate": 9.335424003845825e-06,
      "loss": 0.2832,
      "step": 875
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8060486656830288,
      "learning_rate": 9.333660511554926e-06,
      "loss": 0.2922,
      "step": 876
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.81229975742817,
      "learning_rate": 9.331894849619056e-06,
      "loss": 0.3189,
      "step": 877
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8868055773844985,
      "learning_rate": 9.330127018922195e-06,
      "loss": 0.3282,
      "step": 878
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7490929774023006,
      "learning_rate": 9.328357020349405e-06,
      "loss": 0.2912,
      "step": 879
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.74378040939903,
      "learning_rate": 9.326584854786835e-06,
      "loss": 0.2859,
      "step": 880
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8079750251420251,
      "learning_rate": 9.32481052312172e-06,
      "loss": 0.2959,
      "step": 881
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7622364076324933,
      "learning_rate": 9.323034026242378e-06,
      "loss": 0.2819,
      "step": 882
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7792176687747565,
      "learning_rate": 9.32125536503821e-06,
      "loss": 0.3044,
      "step": 883
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7876368741665177,
      "learning_rate": 9.319474540399704e-06,
      "loss": 0.2858,
      "step": 884
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8672801639554721,
      "learning_rate": 9.317691553218427e-06,
      "loss": 0.2996,
      "step": 885
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7968374841260473,
      "learning_rate": 9.315906404387032e-06,
      "loss": 0.2944,
      "step": 886
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7946008218875935,
      "learning_rate": 9.314119094799251e-06,
      "loss": 0.3056,
      "step": 887
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7648899443961412,
      "learning_rate": 9.312329625349903e-06,
      "loss": 0.2689,
      "step": 888
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7602215230126487,
      "learning_rate": 9.310537996934882e-06,
      "loss": 0.2581,
      "step": 889
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8304389865539289,
      "learning_rate": 9.308744210451167e-06,
      "loss": 0.3071,
      "step": 890
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7800714593868289,
      "learning_rate": 9.306948266796815e-06,
      "loss": 0.2962,
      "step": 891
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8000533093650183,
      "learning_rate": 9.305150166870966e-06,
      "loss": 0.3024,
      "step": 892
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7583598834293337,
      "learning_rate": 9.303349911573838e-06,
      "loss": 0.2886,
      "step": 893
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8649075171275785,
      "learning_rate": 9.301547501806725e-06,
      "loss": 0.3013,
      "step": 894
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8018643341552107,
      "learning_rate": 9.299742938472007e-06,
      "loss": 0.2937,
      "step": 895
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7935991591856959,
      "learning_rate": 9.297936222473134e-06,
      "loss": 0.2763,
      "step": 896
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8489301391155455,
      "learning_rate": 9.29612735471464e-06,
      "loss": 0.3026,
      "step": 897
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8303744880842548,
      "learning_rate": 9.294316336102132e-06,
      "loss": 0.3041,
      "step": 898
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.766548827959563,
      "learning_rate": 9.292503167542295e-06,
      "loss": 0.29,
      "step": 899
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.812919118031871,
      "learning_rate": 9.290687849942893e-06,
      "loss": 0.2831,
      "step": 900
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7671043520557768,
      "learning_rate": 9.288870384212762e-06,
      "loss": 0.3183,
      "step": 901
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8143329446136075,
      "learning_rate": 9.287050771261818e-06,
      "loss": 0.3188,
      "step": 902
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7832022429411961,
      "learning_rate": 9.285229012001047e-06,
      "loss": 0.328,
      "step": 903
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7762259861820453,
      "learning_rate": 9.283405107342512e-06,
      "loss": 0.2737,
      "step": 904
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8212055691298046,
      "learning_rate": 9.281579058199351e-06,
      "loss": 0.2901,
      "step": 905
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7610564066658664,
      "learning_rate": 9.279750865485772e-06,
      "loss": 0.2944,
      "step": 906
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7772102921526046,
      "learning_rate": 9.277920530117064e-06,
      "loss": 0.2976,
      "step": 907
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7651426679076685,
      "learning_rate": 9.276088053009578e-06,
      "loss": 0.2596,
      "step": 908
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7957305718404895,
      "learning_rate": 9.274253435080745e-06,
      "loss": 0.2632,
      "step": 909
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.794532068904075,
      "learning_rate": 9.272416677249067e-06,
      "loss": 0.3025,
      "step": 910
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7788098750213445,
      "learning_rate": 9.270577780434114e-06,
      "loss": 0.2996,
      "step": 911
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.767945598649621,
      "learning_rate": 9.268736745556528e-06,
      "loss": 0.303,
      "step": 912
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.752112570218429,
      "learning_rate": 9.266893573538023e-06,
      "loss": 0.2918,
      "step": 913
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8451521518625214,
      "learning_rate": 9.265048265301385e-06,
      "loss": 0.3079,
      "step": 914
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7796194736871458,
      "learning_rate": 9.263200821770462e-06,
      "loss": 0.286,
      "step": 915
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8201720983206979,
      "learning_rate": 9.261351243870178e-06,
      "loss": 0.3117,
      "step": 916
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8148254937097175,
      "learning_rate": 9.259499532526523e-06,
      "loss": 0.3055,
      "step": 917
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7533681678346078,
      "learning_rate": 9.257645688666557e-06,
      "loss": 0.2931,
      "step": 918
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8031253944606287,
      "learning_rate": 9.255789713218403e-06,
      "loss": 0.3015,
      "step": 919
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.736058436061908,
      "learning_rate": 9.253931607111257e-06,
      "loss": 0.3101,
      "step": 920
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7661488584428052,
      "learning_rate": 9.252071371275378e-06,
      "loss": 0.2912,
      "step": 921
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7595062432740963,
      "learning_rate": 9.25020900664209e-06,
      "loss": 0.2782,
      "step": 922
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7394450237395668,
      "learning_rate": 9.248344514143786e-06,
      "loss": 0.2784,
      "step": 923
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7363465633254676,
      "learning_rate": 9.246477894713925e-06,
      "loss": 0.2823,
      "step": 924
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7598091794327279,
      "learning_rate": 9.244609149287029e-06,
      "loss": 0.2963,
      "step": 925
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8087535300139654,
      "learning_rate": 9.242738278798682e-06,
      "loss": 0.2933,
      "step": 926
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7393690896608293,
      "learning_rate": 9.240865284185536e-06,
      "loss": 0.2805,
      "step": 927
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8485750092063818,
      "learning_rate": 9.238990166385304e-06,
      "loss": 0.3304,
      "step": 928
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8424040996413581,
      "learning_rate": 9.237112926336764e-06,
      "loss": 0.2891,
      "step": 929
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8327632090783766,
      "learning_rate": 9.235233564979756e-06,
      "loss": 0.3226,
      "step": 930
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.780710831093105,
      "learning_rate": 9.233352083255177e-06,
      "loss": 0.2947,
      "step": 931
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8162983451385465,
      "learning_rate": 9.231468482104994e-06,
      "loss": 0.2982,
      "step": 932
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.844031331898971,
      "learning_rate": 9.229582762472232e-06,
      "loss": 0.2948,
      "step": 933
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8462000817563199,
      "learning_rate": 9.227694925300972e-06,
      "loss": 0.2994,
      "step": 934
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7608879508687887,
      "learning_rate": 9.22580497153636e-06,
      "loss": 0.2825,
      "step": 935
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8157866329023847,
      "learning_rate": 9.223912902124601e-06,
      "loss": 0.294,
      "step": 936
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7871485436214074,
      "learning_rate": 9.222018718012957e-06,
      "loss": 0.2994,
      "step": 937
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8110872233130274,
      "learning_rate": 9.220122420149753e-06,
      "loss": 0.2749,
      "step": 938
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8505689754113855,
      "learning_rate": 9.218224009484367e-06,
      "loss": 0.2926,
      "step": 939
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8156406261262299,
      "learning_rate": 9.216323486967239e-06,
      "loss": 0.2928,
      "step": 940
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8615798914597762,
      "learning_rate": 9.214420853549864e-06,
      "loss": 0.3126,
      "step": 941
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7436878099325009,
      "learning_rate": 9.212516110184794e-06,
      "loss": 0.2841,
      "step": 942
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8794073142288895,
      "learning_rate": 9.21060925782564e-06,
      "loss": 0.286,
      "step": 943
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7772225063146398,
      "learning_rate": 9.208700297427063e-06,
      "loss": 0.2844,
      "step": 944
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7197413533658618,
      "learning_rate": 9.206789229944786e-06,
      "loss": 0.2709,
      "step": 945
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.756855899179061,
      "learning_rate": 9.204876056335584e-06,
      "loss": 0.2648,
      "step": 946
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.74192197001665,
      "learning_rate": 9.202960777557284e-06,
      "loss": 0.2864,
      "step": 947
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7838203577584194,
      "learning_rate": 9.201043394568773e-06,
      "loss": 0.2842,
      "step": 948
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7762947527742773,
      "learning_rate": 9.199123908329983e-06,
      "loss": 0.3021,
      "step": 949
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8473024993108265,
      "learning_rate": 9.197202319801908e-06,
      "loss": 0.2955,
      "step": 950
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7867054383410023,
      "learning_rate": 9.195278629946588e-06,
      "loss": 0.3077,
      "step": 951
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7451705137883774,
      "learning_rate": 9.193352839727122e-06,
      "loss": 0.2912,
      "step": 952
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7312447103512066,
      "learning_rate": 9.191424950107648e-06,
      "loss": 0.3026,
      "step": 953
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7942172248950031,
      "learning_rate": 9.18949496205337e-06,
      "loss": 0.2755,
      "step": 954
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.765296631237537,
      "learning_rate": 9.187562876530528e-06,
      "loss": 0.3026,
      "step": 955
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8002153150476999,
      "learning_rate": 9.18562869450643e-06,
      "loss": 0.3148,
      "step": 956
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7239565461777733,
      "learning_rate": 9.183692416949415e-06,
      "loss": 0.2781,
      "step": 957
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8321618745643802,
      "learning_rate": 9.181754044828882e-06,
      "loss": 0.2986,
      "step": 958
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7840697778099418,
      "learning_rate": 9.179813579115276e-06,
      "loss": 0.301,
      "step": 959
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.857398392048463,
      "learning_rate": 9.17787102078009e-06,
      "loss": 0.3011,
      "step": 960
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7700595104500547,
      "learning_rate": 9.175926370795865e-06,
      "loss": 0.2735,
      "step": 961
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8033342414576251,
      "learning_rate": 9.17397963013619e-06,
      "loss": 0.303,
      "step": 962
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8066032140279419,
      "learning_rate": 9.172030799775698e-06,
      "loss": 0.3118,
      "step": 963
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.785470834416065,
      "learning_rate": 9.170079880690072e-06,
      "loss": 0.2991,
      "step": 964
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8203729288713701,
      "learning_rate": 9.168126873856035e-06,
      "loss": 0.2869,
      "step": 965
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7202385851539883,
      "learning_rate": 9.166171780251365e-06,
      "loss": 0.28,
      "step": 966
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.855613294581656,
      "learning_rate": 9.164214600854872e-06,
      "loss": 0.2934,
      "step": 967
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8038381024714932,
      "learning_rate": 9.162255336646422e-06,
      "loss": 0.3149,
      "step": 968
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7655922115881096,
      "learning_rate": 9.160293988606917e-06,
      "loss": 0.3033,
      "step": 969
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8186707988984755,
      "learning_rate": 9.158330557718304e-06,
      "loss": 0.2822,
      "step": 970
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8372332133423401,
      "learning_rate": 9.156365044963575e-06,
      "loss": 0.3074,
      "step": 971
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8334423875474783,
      "learning_rate": 9.154397451326766e-06,
      "loss": 0.2945,
      "step": 972
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7806476263445227,
      "learning_rate": 9.152427777792947e-06,
      "loss": 0.2955,
      "step": 973
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7444421047752595,
      "learning_rate": 9.150456025348236e-06,
      "loss": 0.2999,
      "step": 974
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7914148598534815,
      "learning_rate": 9.14848219497979e-06,
      "loss": 0.2997,
      "step": 975
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8480103558838515,
      "learning_rate": 9.146506287675805e-06,
      "loss": 0.2853,
      "step": 976
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.912549394646951,
      "learning_rate": 9.14452830442552e-06,
      "loss": 0.2865,
      "step": 977
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7627263706325143,
      "learning_rate": 9.142548246219212e-06,
      "loss": 0.3003,
      "step": 978
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8111974475252385,
      "learning_rate": 9.140566114048193e-06,
      "loss": 0.2873,
      "step": 979
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8350227856792054,
      "learning_rate": 9.138581908904818e-06,
      "loss": 0.2971,
      "step": 980
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8667092025766343,
      "learning_rate": 9.136595631782478e-06,
      "loss": 0.3017,
      "step": 981
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8180452000812636,
      "learning_rate": 9.134607283675601e-06,
      "loss": 0.3008,
      "step": 982
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7828358580335144,
      "learning_rate": 9.132616865579655e-06,
      "loss": 0.3078,
      "step": 983
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7741028343645261,
      "learning_rate": 9.13062437849114e-06,
      "loss": 0.2762,
      "step": 984
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8061022982099546,
      "learning_rate": 9.128629823407594e-06,
      "loss": 0.3097,
      "step": 985
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8380846666332286,
      "learning_rate": 9.12663320132759e-06,
      "loss": 0.2895,
      "step": 986
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8132499360139261,
      "learning_rate": 9.124634513250736e-06,
      "loss": 0.2958,
      "step": 987
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7599886375883173,
      "learning_rate": 9.122633760177674e-06,
      "loss": 0.305,
      "step": 988
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8805149217344087,
      "learning_rate": 9.120630943110078e-06,
      "loss": 0.3175,
      "step": 989
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8862074915032321,
      "learning_rate": 9.118626063050661e-06,
      "loss": 0.3028,
      "step": 990
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8198180404558864,
      "learning_rate": 9.116619121003165e-06,
      "loss": 0.3001,
      "step": 991
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8347765243068982,
      "learning_rate": 9.114610117972362e-06,
      "loss": 0.317,
      "step": 992
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7582195001535587,
      "learning_rate": 9.112599054964058e-06,
      "loss": 0.2885,
      "step": 993
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8889926793044187,
      "learning_rate": 9.110585932985095e-06,
      "loss": 0.2992,
      "step": 994
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9014517036183125,
      "learning_rate": 9.108570753043335e-06,
      "loss": 0.317,
      "step": 995
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8168195530211014,
      "learning_rate": 9.106553516147683e-06,
      "loss": 0.3054,
      "step": 996
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8028404752784611,
      "learning_rate": 9.104534223308065e-06,
      "loss": 0.2951,
      "step": 997
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8754361419975237,
      "learning_rate": 9.102512875535439e-06,
      "loss": 0.2606,
      "step": 998
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0283010577597889,
      "learning_rate": 9.100489473841791e-06,
      "loss": 0.2962,
      "step": 999
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.747671711860443,
      "learning_rate": 9.098464019240137e-06,
      "loss": 0.2856,
      "step": 1000
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7939961911283938,
      "learning_rate": 9.09643651274452e-06,
      "loss": 0.2975,
      "step": 1001
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8374615735993098,
      "learning_rate": 9.09440695537001e-06,
      "loss": 0.2841,
      "step": 1002
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8039858097682666,
      "learning_rate": 9.092375348132704e-06,
      "loss": 0.3441,
      "step": 1003
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7348867268124865,
      "learning_rate": 9.090341692049725e-06,
      "loss": 0.2721,
      "step": 1004
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8728037213530544,
      "learning_rate": 9.088305988139222e-06,
      "loss": 0.2953,
      "step": 1005
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.84403118501091,
      "learning_rate": 9.086268237420368e-06,
      "loss": 0.3242,
      "step": 1006
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8042705351932206,
      "learning_rate": 9.084228440913363e-06,
      "loss": 0.3091,
      "step": 1007
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.85917361225115,
      "learning_rate": 9.082186599639429e-06,
      "loss": 0.3132,
      "step": 1008
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7479738084558707,
      "learning_rate": 9.080142714620814e-06,
      "loss": 0.2827,
      "step": 1009
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.800382689861599,
      "learning_rate": 9.07809678688079e-06,
      "loss": 0.2787,
      "step": 1010
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7966754982225652,
      "learning_rate": 9.076048817443646e-06,
      "loss": 0.3032,
      "step": 1011
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7475737323295671,
      "learning_rate": 9.073998807334698e-06,
      "loss": 0.2911,
      "step": 1012
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8445770270157533,
      "learning_rate": 9.071946757580282e-06,
      "loss": 0.2903,
      "step": 1013
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7035271944529282,
      "learning_rate": 9.069892669207757e-06,
      "loss": 0.2689,
      "step": 1014
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7606016129666269,
      "learning_rate": 9.067836543245502e-06,
      "loss": 0.3092,
      "step": 1015
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8738850072495035,
      "learning_rate": 9.065778380722914e-06,
      "loss": 0.2859,
      "step": 1016
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.778992340272719,
      "learning_rate": 9.06371818267041e-06,
      "loss": 0.3044,
      "step": 1017
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.769133524285956,
      "learning_rate": 9.06165595011943e-06,
      "loss": 0.2943,
      "step": 1018
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7534957300883848,
      "learning_rate": 9.059591684102429e-06,
      "loss": 0.2891,
      "step": 1019
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8388393247551896,
      "learning_rate": 9.057525385652877e-06,
      "loss": 0.3135,
      "step": 1020
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7767258967626035,
      "learning_rate": 9.05545705580527e-06,
      "loss": 0.3007,
      "step": 1021
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7534069173050336,
      "learning_rate": 9.053386695595114e-06,
      "loss": 0.2847,
      "step": 1022
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7550646812259267,
      "learning_rate": 9.051314306058934e-06,
      "loss": 0.2997,
      "step": 1023
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8337388890209925,
      "learning_rate": 9.049239888234271e-06,
      "loss": 0.2943,
      "step": 1024
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8045047651550278,
      "learning_rate": 9.047163443159682e-06,
      "loss": 0.314,
      "step": 1025
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7688942082002721,
      "learning_rate": 9.045084971874738e-06,
      "loss": 0.2885,
      "step": 1026
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7878558974418302,
      "learning_rate": 9.043004475420022e-06,
      "loss": 0.3,
      "step": 1027
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8331290751592837,
      "learning_rate": 9.040921954837139e-06,
      "loss": 0.3177,
      "step": 1028
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7740373276050339,
      "learning_rate": 9.038837411168697e-06,
      "loss": 0.289,
      "step": 1029
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7770428718516068,
      "learning_rate": 9.036750845458323e-06,
      "loss": 0.2939,
      "step": 1030
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8352442588458575,
      "learning_rate": 9.034662258750653e-06,
      "loss": 0.3057,
      "step": 1031
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7911992808374942,
      "learning_rate": 9.032571652091341e-06,
      "loss": 0.3121,
      "step": 1032
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7905847859454433,
      "learning_rate": 9.030479026527048e-06,
      "loss": 0.3118,
      "step": 1033
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.77257641601518,
      "learning_rate": 9.028384383105444e-06,
      "loss": 0.2974,
      "step": 1034
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7159883170969386,
      "learning_rate": 9.02628772287521e-06,
      "loss": 0.2855,
      "step": 1035
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7120497989808979,
      "learning_rate": 9.02418904688604e-06,
      "loss": 0.2789,
      "step": 1036
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8076719783641084,
      "learning_rate": 9.022088356188633e-06,
      "loss": 0.2882,
      "step": 1037
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7806523713290034,
      "learning_rate": 9.019985651834703e-06,
      "loss": 0.2894,
      "step": 1038
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7809270341360246,
      "learning_rate": 9.017880934876965e-06,
      "loss": 0.2866,
      "step": 1039
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7801111269173077,
      "learning_rate": 9.015774206369144e-06,
      "loss": 0.283,
      "step": 1040
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7847266942337844,
      "learning_rate": 9.013665467365972e-06,
      "loss": 0.2843,
      "step": 1041
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.719882786243959,
      "learning_rate": 9.01155471892319e-06,
      "loss": 0.2987,
      "step": 1042
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7591198544607971,
      "learning_rate": 9.009441962097543e-06,
      "loss": 0.3039,
      "step": 1043
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.79468228124708,
      "learning_rate": 9.007327197946782e-06,
      "loss": 0.3277,
      "step": 1044
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7902735981169708,
      "learning_rate": 9.00521042752966e-06,
      "loss": 0.2578,
      "step": 1045
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7438011281533328,
      "learning_rate": 9.00309165190594e-06,
      "loss": 0.2882,
      "step": 1046
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7485614819674792,
      "learning_rate": 9.000970872136383e-06,
      "loss": 0.2909,
      "step": 1047
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7956834197340957,
      "learning_rate": 8.99884808928276e-06,
      "loss": 0.3164,
      "step": 1048
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7286899655369156,
      "learning_rate": 8.99672330440784e-06,
      "loss": 0.2982,
      "step": 1049
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7371753085182782,
      "learning_rate": 8.994596518575393e-06,
      "loss": 0.2748,
      "step": 1050
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7800730360004422,
      "learning_rate": 8.992467732850194e-06,
      "loss": 0.3012,
      "step": 1051
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7657360165151705,
      "learning_rate": 8.990336948298023e-06,
      "loss": 0.2841,
      "step": 1052
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7826668228652792,
      "learning_rate": 8.98820416598565e-06,
      "loss": 0.304,
      "step": 1053
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7667643878930182,
      "learning_rate": 8.986069386980855e-06,
      "loss": 0.2884,
      "step": 1054
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7319932001808329,
      "learning_rate": 8.983932612352413e-06,
      "loss": 0.2666,
      "step": 1055
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7740686808444169,
      "learning_rate": 8.9817938431701e-06,
      "loss": 0.3039,
      "step": 1056
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.84260192758322,
      "learning_rate": 8.979653080504689e-06,
      "loss": 0.2864,
      "step": 1057
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7997686020709208,
      "learning_rate": 8.97751032542795e-06,
      "loss": 0.3037,
      "step": 1058
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7457271261884206,
      "learning_rate": 8.975365579012656e-06,
      "loss": 0.2816,
      "step": 1059
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8446485277616422,
      "learning_rate": 8.97321884233257e-06,
      "loss": 0.2954,
      "step": 1060
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7113877511642576,
      "learning_rate": 8.971070116462457e-06,
      "loss": 0.2682,
      "step": 1061
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8155144317368913,
      "learning_rate": 8.968919402478076e-06,
      "loss": 0.2907,
      "step": 1062
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8333943328970775,
      "learning_rate": 8.966766701456177e-06,
      "loss": 0.3094,
      "step": 1063
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7949476531757625,
      "learning_rate": 8.964612014474512e-06,
      "loss": 0.3155,
      "step": 1064
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7699908762382,
      "learning_rate": 8.962455342611821e-06,
      "loss": 0.311,
      "step": 1065
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7410398277019268,
      "learning_rate": 8.960296686947846e-06,
      "loss": 0.2819,
      "step": 1066
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8231315919397555,
      "learning_rate": 8.958136048563311e-06,
      "loss": 0.284,
      "step": 1067
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8234723602977069,
      "learning_rate": 8.955973428539943e-06,
      "loss": 0.2981,
      "step": 1068
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7765854779165084,
      "learning_rate": 8.953808827960455e-06,
      "loss": 0.2979,
      "step": 1069
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8072883305374614,
      "learning_rate": 8.95164224790855e-06,
      "loss": 0.3163,
      "step": 1070
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7350252096155895,
      "learning_rate": 8.949473689468931e-06,
      "loss": 0.2791,
      "step": 1071
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8341662449448992,
      "learning_rate": 8.947303153727283e-06,
      "loss": 0.2872,
      "step": 1072
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7504273382489525,
      "learning_rate": 8.945130641770281e-06,
      "loss": 0.2834,
      "step": 1073
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8196627935330668,
      "learning_rate": 8.942956154685596e-06,
      "loss": 0.3066,
      "step": 1074
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7689543097584358,
      "learning_rate": 8.940779693561883e-06,
      "loss": 0.2727,
      "step": 1075
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.764742324691488,
      "learning_rate": 8.938601259488784e-06,
      "loss": 0.3097,
      "step": 1076
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7249761945705442,
      "learning_rate": 8.936420853556935e-06,
      "loss": 0.2946,
      "step": 1077
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7666710332635401,
      "learning_rate": 8.93423847685795e-06,
      "loss": 0.297,
      "step": 1078
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8131314454309119,
      "learning_rate": 8.932054130484439e-06,
      "loss": 0.2885,
      "step": 1079
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7615398299441541,
      "learning_rate": 8.929867815529992e-06,
      "loss": 0.3016,
      "step": 1080
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7881318088455816,
      "learning_rate": 8.927679533089188e-06,
      "loss": 0.3223,
      "step": 1081
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8170196392875678,
      "learning_rate": 8.925489284257587e-06,
      "loss": 0.2952,
      "step": 1082
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7540619682704766,
      "learning_rate": 8.923297070131738e-06,
      "loss": 0.2935,
      "step": 1083
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7960614942294164,
      "learning_rate": 8.921102891809171e-06,
      "loss": 0.2748,
      "step": 1084
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7845517471656036,
      "learning_rate": 8.9189067503884e-06,
      "loss": 0.3142,
      "step": 1085
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.816481966208708,
      "learning_rate": 8.916708646968924e-06,
      "loss": 0.2991,
      "step": 1086
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8725196356361503,
      "learning_rate": 8.91450858265122e-06,
      "loss": 0.2983,
      "step": 1087
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8937269363757484,
      "learning_rate": 8.91230655853675e-06,
      "loss": 0.2869,
      "step": 1088
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7358373033232506,
      "learning_rate": 8.910102575727957e-06,
      "loss": 0.2849,
      "step": 1089
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8508345698370803,
      "learning_rate": 8.907896635328262e-06,
      "loss": 0.3202,
      "step": 1090
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8383205301322761,
      "learning_rate": 8.90568873844207e-06,
      "loss": 0.3055,
      "step": 1091
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8469577186841311,
      "learning_rate": 8.903478886174763e-06,
      "loss": 0.3075,
      "step": 1092
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8303219445005313,
      "learning_rate": 8.901267079632703e-06,
      "loss": 0.2829,
      "step": 1093
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6910068917879951,
      "learning_rate": 8.899053319923228e-06,
      "loss": 0.2852,
      "step": 1094
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7742193845663957,
      "learning_rate": 8.896837608154656e-06,
      "loss": 0.2932,
      "step": 1095
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8061781656061993,
      "learning_rate": 8.894619945436285e-06,
      "loss": 0.2983,
      "step": 1096
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.967508459741498,
      "learning_rate": 8.892400332878381e-06,
      "loss": 0.2722,
      "step": 1097
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.767452367324214,
      "learning_rate": 8.890178771592198e-06,
      "loss": 0.2776,
      "step": 1098
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8048173918683627,
      "learning_rate": 8.887955262689956e-06,
      "loss": 0.2748,
      "step": 1099
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.671864100336097,
      "learning_rate": 8.885729807284855e-06,
      "loss": 0.3147,
      "step": 1100
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7880112706421164,
      "learning_rate": 8.883502406491067e-06,
      "loss": 0.2879,
      "step": 1101
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.773477614232053,
      "learning_rate": 8.881273061423741e-06,
      "loss": 0.3102,
      "step": 1102
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8174684031914559,
      "learning_rate": 8.879041773198996e-06,
      "loss": 0.291,
      "step": 1103
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8178252116646675,
      "learning_rate": 8.876808542933924e-06,
      "loss": 0.3179,
      "step": 1104
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7961555176048901,
      "learning_rate": 8.874573371746593e-06,
      "loss": 0.2956,
      "step": 1105
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7590906093788874,
      "learning_rate": 8.87233626075604e-06,
      "loss": 0.2992,
      "step": 1106
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.747264543752027,
      "learning_rate": 8.87009721108227e-06,
      "loss": 0.2885,
      "step": 1107
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8089862456978382,
      "learning_rate": 8.86785622384627e-06,
      "loss": 0.2989,
      "step": 1108
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7628391704479965,
      "learning_rate": 8.865613300169982e-06,
      "loss": 0.2842,
      "step": 1109
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7157140969458778,
      "learning_rate": 8.863368441176326e-06,
      "loss": 0.2984,
      "step": 1110
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7154660921821866,
      "learning_rate": 8.861121647989191e-06,
      "loss": 0.2579,
      "step": 1111
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7646214168525789,
      "learning_rate": 8.858872921733433e-06,
      "loss": 0.3081,
      "step": 1112
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7134564380666014,
      "learning_rate": 8.856622263534875e-06,
      "loss": 0.2896,
      "step": 1113
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7320606061498681,
      "learning_rate": 8.854369674520306e-06,
      "loss": 0.2805,
      "step": 1114
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7417464229824602,
      "learning_rate": 8.852115155817488e-06,
      "loss": 0.2785,
      "step": 1115
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7862547581817724,
      "learning_rate": 8.84985870855514e-06,
      "loss": 0.316,
      "step": 1116
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8558338966238921,
      "learning_rate": 8.847600333862956e-06,
      "loss": 0.304,
      "step": 1117
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7994690319330271,
      "learning_rate": 8.845340032871584e-06,
      "loss": 0.3027,
      "step": 1118
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7669871044581422,
      "learning_rate": 8.843077806712648e-06,
      "loss": 0.3065,
      "step": 1119
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8020683647945936,
      "learning_rate": 8.840813656518728e-06,
      "loss": 0.3122,
      "step": 1120
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7641418580476265,
      "learning_rate": 8.838547583423372e-06,
      "loss": 0.2891,
      "step": 1121
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7904379010650131,
      "learning_rate": 8.836279588561084e-06,
      "loss": 0.2857,
      "step": 1122
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8291853946654066,
      "learning_rate": 8.834009673067337e-06,
      "loss": 0.2966,
      "step": 1123
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.825924229279286,
      "learning_rate": 8.831737838078562e-06,
      "loss": 0.2824,
      "step": 1124
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8047860751689978,
      "learning_rate": 8.829464084732155e-06,
      "loss": 0.2969,
      "step": 1125
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7944171148473961,
      "learning_rate": 8.827188414166468e-06,
      "loss": 0.2864,
      "step": 1126
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7234294044728933,
      "learning_rate": 8.824910827520813e-06,
      "loss": 0.2686,
      "step": 1127
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7192835079445673,
      "learning_rate": 8.822631325935463e-06,
      "loss": 0.2731,
      "step": 1128
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8103107997889915,
      "learning_rate": 8.820349910551651e-06,
      "loss": 0.2817,
      "step": 1129
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.792158944955739,
      "learning_rate": 8.818066582511564e-06,
      "loss": 0.2749,
      "step": 1130
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8647208757724733,
      "learning_rate": 8.815781342958352e-06,
      "loss": 0.3014,
      "step": 1131
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7365742520729354,
      "learning_rate": 8.813494193036115e-06,
      "loss": 0.2891,
      "step": 1132
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8404013951020166,
      "learning_rate": 8.811205133889917e-06,
      "loss": 0.3068,
      "step": 1133
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8356683191965253,
      "learning_rate": 8.808914166665773e-06,
      "loss": 0.2996,
      "step": 1134
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8181147075558484,
      "learning_rate": 8.806621292510655e-06,
      "loss": 0.3182,
      "step": 1135
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7990128719810651,
      "learning_rate": 8.80432651257249e-06,
      "loss": 0.3011,
      "step": 1136
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8069591368065677,
      "learning_rate": 8.802029828000157e-06,
      "loss": 0.3098,
      "step": 1137
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7639387407002959,
      "learning_rate": 8.799731239943488e-06,
      "loss": 0.2951,
      "step": 1138
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7205740179323434,
      "learning_rate": 8.797430749553275e-06,
      "loss": 0.2659,
      "step": 1139
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7811956116503154,
      "learning_rate": 8.795128357981253e-06,
      "loss": 0.3103,
      "step": 1140
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8287860577606632,
      "learning_rate": 8.792824066380116e-06,
      "loss": 0.2913,
      "step": 1141
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7798356398151342,
      "learning_rate": 8.790517875903506e-06,
      "loss": 0.2989,
      "step": 1142
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7376137033761311,
      "learning_rate": 8.788209787706014e-06,
      "loss": 0.2779,
      "step": 1143
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7509149618178763,
      "learning_rate": 8.785899802943186e-06,
      "loss": 0.2773,
      "step": 1144
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6963581579295547,
      "learning_rate": 8.783587922771514e-06,
      "loss": 0.282,
      "step": 1145
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7850390021139333,
      "learning_rate": 8.781274148348438e-06,
      "loss": 0.2876,
      "step": 1146
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7630072479590001,
      "learning_rate": 8.778958480832349e-06,
      "loss": 0.2709,
      "step": 1147
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8486171923653862,
      "learning_rate": 8.776640921382585e-06,
      "loss": 0.3161,
      "step": 1148
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7888473226161546,
      "learning_rate": 8.77432147115943e-06,
      "loss": 0.2825,
      "step": 1149
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7392849437103308,
      "learning_rate": 8.772000131324117e-06,
      "loss": 0.2735,
      "step": 1150
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7121837061325459,
      "learning_rate": 8.769676903038824e-06,
      "loss": 0.2854,
      "step": 1151
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7465281900263809,
      "learning_rate": 8.767351787466674e-06,
      "loss": 0.3064,
      "step": 1152
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7725467157795138,
      "learning_rate": 8.765024785771732e-06,
      "loss": 0.2791,
      "step": 1153
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8212495508429666,
      "learning_rate": 8.762695899119012e-06,
      "loss": 0.3188,
      "step": 1154
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7954017450416061,
      "learning_rate": 8.760365128674473e-06,
      "loss": 0.2682,
      "step": 1155
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7443231013437491,
      "learning_rate": 8.75803247560501e-06,
      "loss": 0.2719,
      "step": 1156
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7609478593082782,
      "learning_rate": 8.755697941078465e-06,
      "loss": 0.2837,
      "step": 1157
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7379679136337791,
      "learning_rate": 8.753361526263622e-06,
      "loss": 0.2895,
      "step": 1158
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8365870967444168,
      "learning_rate": 8.751023232330208e-06,
      "loss": 0.2742,
      "step": 1159
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7886536617375809,
      "learning_rate": 8.748683060448886e-06,
      "loss": 0.3105,
      "step": 1160
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7623985066954413,
      "learning_rate": 8.746341011791265e-06,
      "loss": 0.2986,
      "step": 1161
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7524167233151875,
      "learning_rate": 8.743997087529886e-06,
      "loss": 0.2909,
      "step": 1162
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7870985560388182,
      "learning_rate": 8.741651288838237e-06,
      "loss": 0.3053,
      "step": 1163
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.749882647315831,
      "learning_rate": 8.739303616890739e-06,
      "loss": 0.2996,
      "step": 1164
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7457503250502238,
      "learning_rate": 8.736954072862755e-06,
      "loss": 0.2965,
      "step": 1165
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.771634242257192,
      "learning_rate": 8.734602657930582e-06,
      "loss": 0.2923,
      "step": 1166
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7783818624826354,
      "learning_rate": 8.732249373271455e-06,
      "loss": 0.3162,
      "step": 1167
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.740969907781502,
      "learning_rate": 8.729894220063542e-06,
      "loss": 0.2968,
      "step": 1168
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7710579732077328,
      "learning_rate": 8.727537199485953e-06,
      "loss": 0.2829,
      "step": 1169
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7823070432137538,
      "learning_rate": 8.725178312718727e-06,
      "loss": 0.2913,
      "step": 1170
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8097901152797413,
      "learning_rate": 8.72281756094284e-06,
      "loss": 0.3093,
      "step": 1171
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.811465928363141,
      "learning_rate": 8.720454945340202e-06,
      "loss": 0.2764,
      "step": 1172
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7917664699175073,
      "learning_rate": 8.718090467093654e-06,
      "loss": 0.3013,
      "step": 1173
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6913066164046763,
      "learning_rate": 8.715724127386971e-06,
      "loss": 0.271,
      "step": 1174
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7764607868050696,
      "learning_rate": 8.713355927404863e-06,
      "loss": 0.2828,
      "step": 1175
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8880941056631532,
      "learning_rate": 8.710985868332964e-06,
      "loss": 0.3055,
      "step": 1176
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.809276083862809,
      "learning_rate": 8.708613951357842e-06,
      "loss": 0.2888,
      "step": 1177
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7236214141178897,
      "learning_rate": 8.706240177667003e-06,
      "loss": 0.2896,
      "step": 1178
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7620164566867083,
      "learning_rate": 8.703864548448868e-06,
      "loss": 0.2992,
      "step": 1179
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.847635453221742,
      "learning_rate": 8.701487064892797e-06,
      "loss": 0.3149,
      "step": 1180
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7673598752317049,
      "learning_rate": 8.699107728189078e-06,
      "loss": 0.2793,
      "step": 1181
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7777601738429618,
      "learning_rate": 8.696726539528924e-06,
      "loss": 0.306,
      "step": 1182
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7334099827184369,
      "learning_rate": 8.694343500104474e-06,
      "loss": 0.3064,
      "step": 1183
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7720865862679176,
      "learning_rate": 8.691958611108795e-06,
      "loss": 0.3038,
      "step": 1184
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7699869716116633,
      "learning_rate": 8.689571873735884e-06,
      "loss": 0.2884,
      "step": 1185
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7679091943939459,
      "learning_rate": 8.687183289180659e-06,
      "loss": 0.2844,
      "step": 1186
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7209259744690923,
      "learning_rate": 8.684792858638962e-06,
      "loss": 0.2937,
      "step": 1187
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8141989209849668,
      "learning_rate": 8.682400583307562e-06,
      "loss": 0.2916,
      "step": 1188
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.791505738187637,
      "learning_rate": 8.68000646438415e-06,
      "loss": 0.3101,
      "step": 1189
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8219115630698021,
      "learning_rate": 8.67761050306734e-06,
      "loss": 0.2992,
      "step": 1190
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7474286268935181,
      "learning_rate": 8.675212700556667e-06,
      "loss": 0.2918,
      "step": 1191
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7479624374002041,
      "learning_rate": 8.672813058052596e-06,
      "loss": 0.2797,
      "step": 1192
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8367863214612048,
      "learning_rate": 8.670411576756502e-06,
      "loss": 0.3022,
      "step": 1193
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8103024135215984,
      "learning_rate": 8.668008257870684e-06,
      "loss": 0.2983,
      "step": 1194
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.74524061562655,
      "learning_rate": 8.665603102598365e-06,
      "loss": 0.2921,
      "step": 1195
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7270317388195752,
      "learning_rate": 8.663196112143685e-06,
      "loss": 0.2737,
      "step": 1196
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7809824533527292,
      "learning_rate": 8.660787287711703e-06,
      "loss": 0.301,
      "step": 1197
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7769465761689845,
      "learning_rate": 8.658376630508391e-06,
      "loss": 0.2921,
      "step": 1198
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6933340710527789,
      "learning_rate": 8.655964141740652e-06,
      "loss": 0.2972,
      "step": 1199
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7270918587977633,
      "learning_rate": 8.65354982261629e-06,
      "loss": 0.2998,
      "step": 1200
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7202024623405247,
      "learning_rate": 8.651133674344035e-06,
      "loss": 0.3078,
      "step": 1201
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7086939886657582,
      "learning_rate": 8.648715698133529e-06,
      "loss": 0.2693,
      "step": 1202
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7445464783703207,
      "learning_rate": 8.646295895195334e-06,
      "loss": 0.2737,
      "step": 1203
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7284163769860235,
      "learning_rate": 8.64387426674092e-06,
      "loss": 0.2667,
      "step": 1204
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8357634721721298,
      "learning_rate": 8.641450813982675e-06,
      "loss": 0.2818,
      "step": 1205
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.752950482943812,
      "learning_rate": 8.639025538133899e-06,
      "loss": 0.28,
      "step": 1206
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7515853712583751,
      "learning_rate": 8.636598440408808e-06,
      "loss": 0.283,
      "step": 1207
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7759728866647259,
      "learning_rate": 8.634169522022522e-06,
      "loss": 0.3066,
      "step": 1208
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7111311221378603,
      "learning_rate": 8.631738784191083e-06,
      "loss": 0.2752,
      "step": 1209
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7842120162051587,
      "learning_rate": 8.629306228131435e-06,
      "loss": 0.3002,
      "step": 1210
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8220183355922953,
      "learning_rate": 8.626871855061438e-06,
      "loss": 0.3009,
      "step": 1211
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7424707676176341,
      "learning_rate": 8.62443566619986e-06,
      "loss": 0.2748,
      "step": 1212
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7493773696114902,
      "learning_rate": 8.621997662766378e-06,
      "loss": 0.2907,
      "step": 1213
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7708797606694638,
      "learning_rate": 8.619557845981571e-06,
      "loss": 0.2963,
      "step": 1214
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7782874558273118,
      "learning_rate": 8.617116217066943e-06,
      "loss": 0.2705,
      "step": 1215
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.712290139065947,
      "learning_rate": 8.614672777244885e-06,
      "loss": 0.2829,
      "step": 1216
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7873661010908337,
      "learning_rate": 8.612227527738707e-06,
      "loss": 0.2826,
      "step": 1217
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7910595350871498,
      "learning_rate": 8.609780469772623e-06,
      "loss": 0.3302,
      "step": 1218
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8606268514253068,
      "learning_rate": 8.607331604571751e-06,
      "loss": 0.3413,
      "step": 1219
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7915993621939872,
      "learning_rate": 8.604880933362113e-06,
      "loss": 0.3296,
      "step": 1220
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7704201080701503,
      "learning_rate": 8.602428457370638e-06,
      "loss": 0.2902,
      "step": 1221
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.788902615504492,
      "learning_rate": 8.599974177825155e-06,
      "loss": 0.316,
      "step": 1222
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7415764713158516,
      "learning_rate": 8.597518095954399e-06,
      "loss": 0.3063,
      "step": 1223
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8247977941848798,
      "learning_rate": 8.595060212988006e-06,
      "loss": 0.3114,
      "step": 1224
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7139926048435267,
      "learning_rate": 8.592600530156514e-06,
      "loss": 0.2641,
      "step": 1225
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7634283300242657,
      "learning_rate": 8.590139048691361e-06,
      "loss": 0.3272,
      "step": 1226
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8085266861613166,
      "learning_rate": 8.587675769824889e-06,
      "loss": 0.3415,
      "step": 1227
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6955940076760438,
      "learning_rate": 8.585210694790333e-06,
      "loss": 0.2625,
      "step": 1228
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7430993923217873,
      "learning_rate": 8.582743824821835e-06,
      "loss": 0.2889,
      "step": 1229
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7266829967966106,
      "learning_rate": 8.580275161154432e-06,
      "loss": 0.2695,
      "step": 1230
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7375497243797405,
      "learning_rate": 8.577804705024059e-06,
      "loss": 0.2751,
      "step": 1231
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.731081048446917,
      "learning_rate": 8.575332457667546e-06,
      "loss": 0.2845,
      "step": 1232
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7149219128937959,
      "learning_rate": 8.572858420322626e-06,
      "loss": 0.2979,
      "step": 1233
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7840261832579003,
      "learning_rate": 8.570382594227923e-06,
      "loss": 0.2886,
      "step": 1234
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.757715388983263,
      "learning_rate": 8.567904980622961e-06,
      "loss": 0.2763,
      "step": 1235
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7590469670970169,
      "learning_rate": 8.56542558074815e-06,
      "loss": 0.2893,
      "step": 1236
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.769160001301231,
      "learning_rate": 8.562944395844806e-06,
      "loss": 0.2848,
      "step": 1237
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7859741759447499,
      "learning_rate": 8.56046142715513e-06,
      "loss": 0.3185,
      "step": 1238
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7348920856613551,
      "learning_rate": 8.557976675922217e-06,
      "loss": 0.3016,
      "step": 1239
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7233794450286786,
      "learning_rate": 8.555490143390061e-06,
      "loss": 0.2666,
      "step": 1240
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8225140227032416,
      "learning_rate": 8.55300183080354e-06,
      "loss": 0.3071,
      "step": 1241
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7416582243098098,
      "learning_rate": 8.550511739408428e-06,
      "loss": 0.301,
      "step": 1242
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7218480002244899,
      "learning_rate": 8.548019870451391e-06,
      "loss": 0.2572,
      "step": 1243
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.759131136095604,
      "learning_rate": 8.545526225179976e-06,
      "loss": 0.2922,
      "step": 1244
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6841513670560221,
      "learning_rate": 8.543030804842629e-06,
      "loss": 0.2575,
      "step": 1245
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7121061193115176,
      "learning_rate": 8.540533610688679e-06,
      "loss": 0.2929,
      "step": 1246
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7969497401509449,
      "learning_rate": 8.538034643968347e-06,
      "loss": 0.3112,
      "step": 1247
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7947209032876851,
      "learning_rate": 8.535533905932739e-06,
      "loss": 0.2941,
      "step": 1248
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7564317326429182,
      "learning_rate": 8.533031397833846e-06,
      "loss": 0.2831,
      "step": 1249
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7954880195822315,
      "learning_rate": 8.53052712092455e-06,
      "loss": 0.3028,
      "step": 1250
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8717325205740224,
      "learning_rate": 8.528021076458614e-06,
      "loss": 0.3084,
      "step": 1251
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.746081380487922,
      "learning_rate": 8.52551326569069e-06,
      "loss": 0.2931,
      "step": 1252
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7951902524165705,
      "learning_rate": 8.523003689876312e-06,
      "loss": 0.2799,
      "step": 1253
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7282459021169742,
      "learning_rate": 8.520492350271895e-06,
      "loss": 0.2972,
      "step": 1254
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7898380899620301,
      "learning_rate": 8.517979248134744e-06,
      "loss": 0.279,
      "step": 1255
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7713541672335588,
      "learning_rate": 8.51546438472304e-06,
      "loss": 0.3055,
      "step": 1256
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.745296607985663,
      "learning_rate": 8.512947761295847e-06,
      "loss": 0.2601,
      "step": 1257
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7671792410396591,
      "learning_rate": 8.510429379113114e-06,
      "loss": 0.3155,
      "step": 1258
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7657474656717499,
      "learning_rate": 8.507909239435665e-06,
      "loss": 0.2904,
      "step": 1259
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8201423304712395,
      "learning_rate": 8.50538734352521e-06,
      "loss": 0.32,
      "step": 1260
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7260308709413577,
      "learning_rate": 8.502863692644332e-06,
      "loss": 0.2892,
      "step": 1261
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8071043032172873,
      "learning_rate": 8.500338288056496e-06,
      "loss": 0.3055,
      "step": 1262
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7864538427410644,
      "learning_rate": 8.497811131026046e-06,
      "loss": 0.2785,
      "step": 1263
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7501507140180073,
      "learning_rate": 8.495282222818202e-06,
      "loss": 0.2734,
      "step": 1264
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7839979879597031,
      "learning_rate": 8.492751564699058e-06,
      "loss": 0.2873,
      "step": 1265
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8143059781931694,
      "learning_rate": 8.490219157935589e-06,
      "loss": 0.3043,
      "step": 1266
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7624616159128208,
      "learning_rate": 8.487685003795643e-06,
      "loss": 0.2963,
      "step": 1267
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7278041159976553,
      "learning_rate": 8.485149103547943e-06,
      "loss": 0.2816,
      "step": 1268
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7404278685848755,
      "learning_rate": 8.482611458462084e-06,
      "loss": 0.2908,
      "step": 1269
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7162130465273232,
      "learning_rate": 8.480072069808541e-06,
      "loss": 0.2669,
      "step": 1270
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7438068668343463,
      "learning_rate": 8.477530938858654e-06,
      "loss": 0.2562,
      "step": 1271
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7323371388530332,
      "learning_rate": 8.47498806688464e-06,
      "loss": 0.2774,
      "step": 1272
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.776998044507676,
      "learning_rate": 8.472443455159586e-06,
      "loss": 0.2891,
      "step": 1273
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7909055476404823,
      "learning_rate": 8.469897104957454e-06,
      "loss": 0.3105,
      "step": 1274
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7773805488946723,
      "learning_rate": 8.467349017553067e-06,
      "loss": 0.2993,
      "step": 1275
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7537334839142156,
      "learning_rate": 8.464799194222129e-06,
      "loss": 0.2876,
      "step": 1276
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7335153773198086,
      "learning_rate": 8.462247636241206e-06,
      "loss": 0.2922,
      "step": 1277
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7403341064116997,
      "learning_rate": 8.459694344887732e-06,
      "loss": 0.2879,
      "step": 1278
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8036826615443928,
      "learning_rate": 8.457139321440014e-06,
      "loss": 0.2901,
      "step": 1279
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6957806505616353,
      "learning_rate": 8.454582567177223e-06,
      "loss": 0.281,
      "step": 1280
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8300382274076942,
      "learning_rate": 8.452024083379396e-06,
      "loss": 0.2835,
      "step": 1281
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7557578029660051,
      "learning_rate": 8.449463871327434e-06,
      "loss": 0.2771,
      "step": 1282
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7823252834110997,
      "learning_rate": 8.446901932303112e-06,
      "loss": 0.2805,
      "step": 1283
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.780226524270819,
      "learning_rate": 8.444338267589057e-06,
      "loss": 0.3076,
      "step": 1284
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8948693644922551,
      "learning_rate": 8.44177287846877e-06,
      "loss": 0.2905,
      "step": 1285
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7692049177827204,
      "learning_rate": 8.439205766226613e-06,
      "loss": 0.2736,
      "step": 1286
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7099821288616858,
      "learning_rate": 8.436636932147805e-06,
      "loss": 0.2808,
      "step": 1287
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8050003973598048,
      "learning_rate": 8.434066377518437e-06,
      "loss": 0.311,
      "step": 1288
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7790485043337708,
      "learning_rate": 8.431494103625452e-06,
      "loss": 0.2917,
      "step": 1289
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8219248728627466,
      "learning_rate": 8.428920111756658e-06,
      "loss": 0.3133,
      "step": 1290
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7533549462994611,
      "learning_rate": 8.426344403200723e-06,
      "loss": 0.2956,
      "step": 1291
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7548542186445102,
      "learning_rate": 8.423766979247176e-06,
      "loss": 0.3042,
      "step": 1292
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7544297315404401,
      "learning_rate": 8.421187841186402e-06,
      "loss": 0.2842,
      "step": 1293
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8361291651493266,
      "learning_rate": 8.418606990309645e-06,
      "loss": 0.2857,
      "step": 1294
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7178418742894652,
      "learning_rate": 8.416024427909005e-06,
      "loss": 0.2871,
      "step": 1295
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7520138953978427,
      "learning_rate": 8.413440155277444e-06,
      "loss": 0.2833,
      "step": 1296
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7261421880720567,
      "learning_rate": 8.410854173708773e-06,
      "loss": 0.2802,
      "step": 1297
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7078100551710003,
      "learning_rate": 8.408266484497664e-06,
      "loss": 0.2735,
      "step": 1298
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7205566918141185,
      "learning_rate": 8.405677088939644e-06,
      "loss": 0.2794,
      "step": 1299
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.78035579057961,
      "learning_rate": 8.403085988331092e-06,
      "loss": 0.2856,
      "step": 1300
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6939634853648301,
      "learning_rate": 8.400493183969238e-06,
      "loss": 0.2854,
      "step": 1301
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7084613093888837,
      "learning_rate": 8.397898677152173e-06,
      "loss": 0.2849,
      "step": 1302
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7178701226238656,
      "learning_rate": 8.395302469178832e-06,
      "loss": 0.2704,
      "step": 1303
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.798462481642841,
      "learning_rate": 8.392704561349006e-06,
      "loss": 0.3084,
      "step": 1304
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.772425587842253,
      "learning_rate": 8.390104954963338e-06,
      "loss": 0.2871,
      "step": 1305
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7500539991172688,
      "learning_rate": 8.387503651323317e-06,
      "loss": 0.2737,
      "step": 1306
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7359831910921674,
      "learning_rate": 8.384900651731287e-06,
      "loss": 0.2976,
      "step": 1307
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7640165113808717,
      "learning_rate": 8.382295957490435e-06,
      "loss": 0.2858,
      "step": 1308
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.767934638027254,
      "learning_rate": 8.379689569904802e-06,
      "loss": 0.2846,
      "step": 1309
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7611963935827673,
      "learning_rate": 8.377081490279278e-06,
      "loss": 0.3237,
      "step": 1310
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7024912590244442,
      "learning_rate": 8.37447171991959e-06,
      "loss": 0.2834,
      "step": 1311
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7323522059299582,
      "learning_rate": 8.371860260132323e-06,
      "loss": 0.2957,
      "step": 1312
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6818449463851284,
      "learning_rate": 8.369247112224901e-06,
      "loss": 0.2787,
      "step": 1313
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7362541978655156,
      "learning_rate": 8.366632277505598e-06,
      "loss": 0.2656,
      "step": 1314
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7354371129182355,
      "learning_rate": 8.364015757283524e-06,
      "loss": 0.3123,
      "step": 1315
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7821356045204925,
      "learning_rate": 8.361397552868645e-06,
      "loss": 0.3099,
      "step": 1316
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.735733571209374,
      "learning_rate": 8.35877766557176e-06,
      "loss": 0.2969,
      "step": 1317
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7462325297764701,
      "learning_rate": 8.356156096704516e-06,
      "loss": 0.2887,
      "step": 1318
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7520159299966603,
      "learning_rate": 8.353532847579397e-06,
      "loss": 0.2931,
      "step": 1319
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7075112518699778,
      "learning_rate": 8.350907919509734e-06,
      "loss": 0.2831,
      "step": 1320
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7803144421561764,
      "learning_rate": 8.348281313809697e-06,
      "loss": 0.2805,
      "step": 1321
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7903454793542005,
      "learning_rate": 8.345653031794292e-06,
      "loss": 0.3058,
      "step": 1322
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8002194533373697,
      "learning_rate": 8.343023074779368e-06,
      "loss": 0.2922,
      "step": 1323
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7604908886164446,
      "learning_rate": 8.340391444081615e-06,
      "loss": 0.2714,
      "step": 1324
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7491544093119393,
      "learning_rate": 8.337758141018551e-06,
      "loss": 0.2842,
      "step": 1325
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7454005063938735,
      "learning_rate": 8.335123166908544e-06,
      "loss": 0.2919,
      "step": 1326
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7862560388380391,
      "learning_rate": 8.332486523070789e-06,
      "loss": 0.2839,
      "step": 1327
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7835603899591577,
      "learning_rate": 8.329848210825322e-06,
      "loss": 0.3122,
      "step": 1328
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7449549501691531,
      "learning_rate": 8.32720823149301e-06,
      "loss": 0.2884,
      "step": 1329
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7319720755108379,
      "learning_rate": 8.324566586395563e-06,
      "loss": 0.2907,
      "step": 1330
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7878648344882258,
      "learning_rate": 8.321923276855514e-06,
      "loss": 0.2965,
      "step": 1331
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7855838785761003,
      "learning_rate": 8.319278304196237e-06,
      "loss": 0.3155,
      "step": 1332
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.774784739346344,
      "learning_rate": 8.316631669741934e-06,
      "loss": 0.2866,
      "step": 1333
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8524048649926468,
      "learning_rate": 8.313983374817643e-06,
      "loss": 0.2865,
      "step": 1334
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7188076761364358,
      "learning_rate": 8.311333420749232e-06,
      "loss": 0.303,
      "step": 1335
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.801717732054723,
      "learning_rate": 8.3086818088634e-06,
      "loss": 0.2949,
      "step": 1336
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7658546946410473,
      "learning_rate": 8.306028540487671e-06,
      "loss": 0.2986,
      "step": 1337
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7542168506164785,
      "learning_rate": 8.303373616950408e-06,
      "loss": 0.295,
      "step": 1338
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7208538518098174,
      "learning_rate": 8.300717039580792e-06,
      "loss": 0.3038,
      "step": 1339
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7997688205933359,
      "learning_rate": 8.298058809708841e-06,
      "loss": 0.285,
      "step": 1340
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.780114045377116,
      "learning_rate": 8.295398928665395e-06,
      "loss": 0.2793,
      "step": 1341
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7773874423641671,
      "learning_rate": 8.29273739778212e-06,
      "loss": 0.2884,
      "step": 1342
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7937231036441694,
      "learning_rate": 8.290074218391515e-06,
      "loss": 0.2784,
      "step": 1343
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7945685464731257,
      "learning_rate": 8.287409391826895e-06,
      "loss": 0.3147,
      "step": 1344
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7391599652914709,
      "learning_rate": 8.284742919422406e-06,
      "loss": 0.2956,
      "step": 1345
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7556502227698848,
      "learning_rate": 8.282074802513016e-06,
      "loss": 0.2811,
      "step": 1346
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7291009613699867,
      "learning_rate": 8.279405042434515e-06,
      "loss": 0.3014,
      "step": 1347
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8614652943470504,
      "learning_rate": 8.27673364052352e-06,
      "loss": 0.291,
      "step": 1348
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7068877935500261,
      "learning_rate": 8.274060598117464e-06,
      "loss": 0.2643,
      "step": 1349
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7575830084062646,
      "learning_rate": 8.271385916554605e-06,
      "loss": 0.2802,
      "step": 1350
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7451999583562087,
      "learning_rate": 8.268709597174022e-06,
      "loss": 0.3105,
      "step": 1351
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7583332847347342,
      "learning_rate": 8.266031641315612e-06,
      "loss": 0.2881,
      "step": 1352
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.72663870091445,
      "learning_rate": 8.263352050320094e-06,
      "loss": 0.2812,
      "step": 1353
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6944811365523024,
      "learning_rate": 8.260670825529002e-06,
      "loss": 0.276,
      "step": 1354
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7503090074098998,
      "learning_rate": 8.257987968284691e-06,
      "loss": 0.2888,
      "step": 1355
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7590387986371336,
      "learning_rate": 8.255303479930332e-06,
      "loss": 0.2818,
      "step": 1356
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.764791786245814,
      "learning_rate": 8.252617361809911e-06,
      "loss": 0.2954,
      "step": 1357
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8226582922323359,
      "learning_rate": 8.249929615268234e-06,
      "loss": 0.2718,
      "step": 1358
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7276746028464334,
      "learning_rate": 8.247240241650918e-06,
      "loss": 0.2905,
      "step": 1359
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7145156138392166,
      "learning_rate": 8.2445492423044e-06,
      "loss": 0.2649,
      "step": 1360
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7469495922965518,
      "learning_rate": 8.241856618575923e-06,
      "loss": 0.2785,
      "step": 1361
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9414348071326497,
      "learning_rate": 8.239162371813552e-06,
      "loss": 0.3095,
      "step": 1362
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7427572726769913,
      "learning_rate": 8.236466503366155e-06,
      "loss": 0.2743,
      "step": 1363
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7498650762630708,
      "learning_rate": 8.233769014583423e-06,
      "loss": 0.2873,
      "step": 1364
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7723576036499131,
      "learning_rate": 8.231069906815847e-06,
      "loss": 0.3028,
      "step": 1365
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7902489494686233,
      "learning_rate": 8.22836918141474e-06,
      "loss": 0.2914,
      "step": 1366
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7527115398817333,
      "learning_rate": 8.225666839732213e-06,
      "loss": 0.3175,
      "step": 1367
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.717567132461287,
      "learning_rate": 8.222962883121196e-06,
      "loss": 0.3074,
      "step": 1368
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7677379535180257,
      "learning_rate": 8.220257312935422e-06,
      "loss": 0.2967,
      "step": 1369
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.74593352632947,
      "learning_rate": 8.217550130529433e-06,
      "loss": 0.2974,
      "step": 1370
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7456006949055594,
      "learning_rate": 8.214841337258579e-06,
      "loss": 0.2755,
      "step": 1371
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7550083092129399,
      "learning_rate": 8.212130934479015e-06,
      "loss": 0.281,
      "step": 1372
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7207901643133077,
      "learning_rate": 8.209418923547706e-06,
      "loss": 0.2835,
      "step": 1373
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7692984625068721,
      "learning_rate": 8.206705305822414e-06,
      "loss": 0.2821,
      "step": 1374
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7414994662321883,
      "learning_rate": 8.203990082661714e-06,
      "loss": 0.2953,
      "step": 1375
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7697919923255098,
      "learning_rate": 8.201273255424979e-06,
      "loss": 0.2834,
      "step": 1376
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7702376993784855,
      "learning_rate": 8.19855482547239e-06,
      "loss": 0.2714,
      "step": 1377
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8139100473042642,
      "learning_rate": 8.195834794164925e-06,
      "loss": 0.2823,
      "step": 1378
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.733058520614886,
      "learning_rate": 8.193113162864368e-06,
      "loss": 0.2965,
      "step": 1379
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8148467671286884,
      "learning_rate": 8.190389932933301e-06,
      "loss": 0.306,
      "step": 1380
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8241336266773468,
      "learning_rate": 8.187665105735108e-06,
      "loss": 0.3086,
      "step": 1381
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7548684922985603,
      "learning_rate": 8.184938682633973e-06,
      "loss": 0.2917,
      "step": 1382
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.738940438561531,
      "learning_rate": 8.182210664994879e-06,
      "loss": 0.2839,
      "step": 1383
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7703483987966112,
      "learning_rate": 8.179481054183605e-06,
      "loss": 0.2828,
      "step": 1384
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7203210002403654,
      "learning_rate": 8.17674985156673e-06,
      "loss": 0.2736,
      "step": 1385
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8043273548686652,
      "learning_rate": 8.17401705851163e-06,
      "loss": 0.2834,
      "step": 1386
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7209183010488207,
      "learning_rate": 8.171282676386474e-06,
      "loss": 0.2668,
      "step": 1387
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7381835521868071,
      "learning_rate": 8.168546706560231e-06,
      "loss": 0.2866,
      "step": 1388
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7651266801251816,
      "learning_rate": 8.165809150402664e-06,
      "loss": 0.3192,
      "step": 1389
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7546629474538247,
      "learning_rate": 8.163070009284326e-06,
      "loss": 0.2885,
      "step": 1390
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7986799663919103,
      "learning_rate": 8.16032928457657e-06,
      "loss": 0.2654,
      "step": 1391
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7387859839110954,
      "learning_rate": 8.157586977651534e-06,
      "loss": 0.2795,
      "step": 1392
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7452495349232148,
      "learning_rate": 8.154843089882159e-06,
      "loss": 0.2884,
      "step": 1393
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7501775343230713,
      "learning_rate": 8.152097622642167e-06,
      "loss": 0.2969,
      "step": 1394
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6733412672453386,
      "learning_rate": 8.149350577306075e-06,
      "loss": 0.2709,
      "step": 1395
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7970788287834962,
      "learning_rate": 8.146601955249187e-06,
      "loss": 0.3063,
      "step": 1396
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7439892928257333,
      "learning_rate": 8.143851757847606e-06,
      "loss": 0.2822,
      "step": 1397
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7232086848039125,
      "learning_rate": 8.141099986478212e-06,
      "loss": 0.2793,
      "step": 1398
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7332000023525236,
      "learning_rate": 8.138346642518679e-06,
      "loss": 0.288,
      "step": 1399
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7428183342220739,
      "learning_rate": 8.135591727347469e-06,
      "loss": 0.2896,
      "step": 1400
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7197737917049155,
      "learning_rate": 8.132835242343828e-06,
      "loss": 0.2831,
      "step": 1401
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6998331123716713,
      "learning_rate": 8.130077188887786e-06,
      "loss": 0.2901,
      "step": 1402
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7301625187477289,
      "learning_rate": 8.127317568360164e-06,
      "loss": 0.281,
      "step": 1403
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7924739206350182,
      "learning_rate": 8.124556382142566e-06,
      "loss": 0.323,
      "step": 1404
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7885214086179252,
      "learning_rate": 8.121793631617375e-06,
      "loss": 0.3004,
      "step": 1405
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7788424581378318,
      "learning_rate": 8.119029318167763e-06,
      "loss": 0.3057,
      "step": 1406
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8241156859557702,
      "learning_rate": 8.116263443177681e-06,
      "loss": 0.2842,
      "step": 1407
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7342352013854734,
      "learning_rate": 8.113496008031863e-06,
      "loss": 0.2912,
      "step": 1408
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7455111302616158,
      "learning_rate": 8.110727014115825e-06,
      "loss": 0.2991,
      "step": 1409
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7457519931838082,
      "learning_rate": 8.107956462815862e-06,
      "loss": 0.2827,
      "step": 1410
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7554037517933672,
      "learning_rate": 8.105184355519047e-06,
      "loss": 0.2902,
      "step": 1411
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7304227772833274,
      "learning_rate": 8.102410693613239e-06,
      "loss": 0.3027,
      "step": 1412
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8041569853112263,
      "learning_rate": 8.099635478487064e-06,
      "loss": 0.287,
      "step": 1413
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7590561079726307,
      "learning_rate": 8.096858711529938e-06,
      "loss": 0.2733,
      "step": 1414
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7312106655440528,
      "learning_rate": 8.094080394132045e-06,
      "loss": 0.2873,
      "step": 1415
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7528587815721207,
      "learning_rate": 8.09130052768435e-06,
      "loss": 0.2769,
      "step": 1416
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7886380123432277,
      "learning_rate": 8.088519113578588e-06,
      "loss": 0.29,
      "step": 1417
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7437865824034989,
      "learning_rate": 8.085736153207277e-06,
      "loss": 0.2974,
      "step": 1418
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7287065490763478,
      "learning_rate": 8.082951647963702e-06,
      "loss": 0.2714,
      "step": 1419
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7788611792662266,
      "learning_rate": 8.080165599241924e-06,
      "loss": 0.2886,
      "step": 1420
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7629671555162344,
      "learning_rate": 8.07737800843678e-06,
      "loss": 0.2972,
      "step": 1421
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7747075240826687,
      "learning_rate": 8.074588876943872e-06,
      "loss": 0.3016,
      "step": 1422
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7702982904428332,
      "learning_rate": 8.07179820615958e-06,
      "loss": 0.292,
      "step": 1423
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7936407765610451,
      "learning_rate": 8.069005997481053e-06,
      "loss": 0.3135,
      "step": 1424
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7429645517296968,
      "learning_rate": 8.066212252306204e-06,
      "loss": 0.2785,
      "step": 1425
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7122866515935513,
      "learning_rate": 8.063416972033725e-06,
      "loss": 0.2754,
      "step": 1426
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7446594126370105,
      "learning_rate": 8.060620158063067e-06,
      "loss": 0.3026,
      "step": 1427
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7534575396384771,
      "learning_rate": 8.057821811794457e-06,
      "loss": 0.2676,
      "step": 1428
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7875360364040008,
      "learning_rate": 8.055021934628886e-06,
      "loss": 0.2944,
      "step": 1429
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7602768666502419,
      "learning_rate": 8.05222052796811e-06,
      "loss": 0.2851,
      "step": 1430
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7190098773183358,
      "learning_rate": 8.049417593214652e-06,
      "loss": 0.2705,
      "step": 1431
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7429609473381991,
      "learning_rate": 8.046613131771798e-06,
      "loss": 0.2759,
      "step": 1432
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7563430786262899,
      "learning_rate": 8.043807145043604e-06,
      "loss": 0.2711,
      "step": 1433
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7712320112603126,
      "learning_rate": 8.040999634434883e-06,
      "loss": 0.3076,
      "step": 1434
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7696742969839653,
      "learning_rate": 8.038190601351215e-06,
      "loss": 0.2944,
      "step": 1435
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7465855216662858,
      "learning_rate": 8.03538004719894e-06,
      "loss": 0.2574,
      "step": 1436
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8007070442789042,
      "learning_rate": 8.032567973385162e-06,
      "loss": 0.3328,
      "step": 1437
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7098602234495919,
      "learning_rate": 8.029754381317741e-06,
      "loss": 0.2703,
      "step": 1438
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7880394852937116,
      "learning_rate": 8.026939272405306e-06,
      "loss": 0.2963,
      "step": 1439
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7099626171261811,
      "learning_rate": 8.024122648057234e-06,
      "loss": 0.2788,
      "step": 1440
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8053408785186219,
      "learning_rate": 8.021304509683672e-06,
      "loss": 0.295,
      "step": 1441
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.742584496775178,
      "learning_rate": 8.018484858695514e-06,
      "loss": 0.2859,
      "step": 1442
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7317819621584302,
      "learning_rate": 8.015663696504424e-06,
      "loss": 0.2771,
      "step": 1443
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7706170264232871,
      "learning_rate": 8.012841024522808e-06,
      "loss": 0.3289,
      "step": 1444
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7184238679676791,
      "learning_rate": 8.01001684416384e-06,
      "loss": 0.3038,
      "step": 1445
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7501361908271242,
      "learning_rate": 8.00719115684144e-06,
      "loss": 0.302,
      "step": 1446
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7020868212026468,
      "learning_rate": 8.004363963970294e-06,
      "loss": 0.278,
      "step": 1447
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6845960292639687,
      "learning_rate": 8.001535266965829e-06,
      "loss": 0.2828,
      "step": 1448
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7433721533360018,
      "learning_rate": 7.998705067244232e-06,
      "loss": 0.313,
      "step": 1449
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7432115348856927,
      "learning_rate": 7.995873366222442e-06,
      "loss": 0.2667,
      "step": 1450
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7283932570390398,
      "learning_rate": 7.993040165318148e-06,
      "loss": 0.283,
      "step": 1451
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7603083047474957,
      "learning_rate": 7.990205465949792e-06,
      "loss": 0.2921,
      "step": 1452
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7247068032455323,
      "learning_rate": 7.987369269536563e-06,
      "loss": 0.2636,
      "step": 1453
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7396356379233954,
      "learning_rate": 7.9845315774984e-06,
      "loss": 0.3071,
      "step": 1454
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7276302593382149,
      "learning_rate": 7.981692391255996e-06,
      "loss": 0.2606,
      "step": 1455
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7336721538100802,
      "learning_rate": 7.978851712230786e-06,
      "loss": 0.3047,
      "step": 1456
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8125328508227085,
      "learning_rate": 7.976009541844956e-06,
      "loss": 0.2799,
      "step": 1457
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.750471889888569,
      "learning_rate": 7.973165881521435e-06,
      "loss": 0.2733,
      "step": 1458
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7294594068599813,
      "learning_rate": 7.9703207326839e-06,
      "loss": 0.2873,
      "step": 1459
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6979831033198769,
      "learning_rate": 7.967474096756775e-06,
      "loss": 0.2606,
      "step": 1460
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7496144564132814,
      "learning_rate": 7.964625975165225e-06,
      "loss": 0.2943,
      "step": 1461
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7184143913475328,
      "learning_rate": 7.961776369335164e-06,
      "loss": 0.2767,
      "step": 1462
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7364900982656133,
      "learning_rate": 7.958925280693243e-06,
      "loss": 0.2646,
      "step": 1463
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7415402167140391,
      "learning_rate": 7.95607271066686e-06,
      "loss": 0.3049,
      "step": 1464
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7131932565661578,
      "learning_rate": 7.95321866068415e-06,
      "loss": 0.241,
      "step": 1465
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7672738847524183,
      "learning_rate": 7.950363132173992e-06,
      "loss": 0.3023,
      "step": 1466
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7507416152737614,
      "learning_rate": 7.947506126566009e-06,
      "loss": 0.2766,
      "step": 1467
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7749070277937928,
      "learning_rate": 7.944647645290555e-06,
      "loss": 0.3042,
      "step": 1468
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7811655209844243,
      "learning_rate": 7.94178768977873e-06,
      "loss": 0.3224,
      "step": 1469
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7216903467621463,
      "learning_rate": 7.938926261462366e-06,
      "loss": 0.2791,
      "step": 1470
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7873214973245648,
      "learning_rate": 7.93606336177404e-06,
      "loss": 0.3082,
      "step": 1471
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7339913960601033,
      "learning_rate": 7.933198992147057e-06,
      "loss": 0.2904,
      "step": 1472
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7994580403778336,
      "learning_rate": 7.930333154015467e-06,
      "loss": 0.279,
      "step": 1473
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7306141020951299,
      "learning_rate": 7.927465848814045e-06,
      "loss": 0.2928,
      "step": 1474
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8153875292482803,
      "learning_rate": 7.92459707797831e-06,
      "loss": 0.2822,
      "step": 1475
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7298911863261782,
      "learning_rate": 7.921726842944507e-06,
      "loss": 0.2871,
      "step": 1476
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7688270973555564,
      "learning_rate": 7.918855145149622e-06,
      "loss": 0.3173,
      "step": 1477
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7836786349357487,
      "learning_rate": 7.915981986031367e-06,
      "loss": 0.2747,
      "step": 1478
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.805261831184872,
      "learning_rate": 7.913107367028188e-06,
      "loss": 0.3033,
      "step": 1479
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7412962703181099,
      "learning_rate": 7.910231289579263e-06,
      "loss": 0.2697,
      "step": 1480
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.763945598506891,
      "learning_rate": 7.907353755124493e-06,
      "loss": 0.2753,
      "step": 1481
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.809846693497088,
      "learning_rate": 7.90447476510452e-06,
      "loss": 0.2902,
      "step": 1482
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7400592980950179,
      "learning_rate": 7.901594320960709e-06,
      "loss": 0.2808,
      "step": 1483
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8151243752278778,
      "learning_rate": 7.898712424135149e-06,
      "loss": 0.2837,
      "step": 1484
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7294262852258118,
      "learning_rate": 7.895829076070663e-06,
      "loss": 0.3047,
      "step": 1485
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8023733808179765,
      "learning_rate": 7.892944278210796e-06,
      "loss": 0.2757,
      "step": 1486
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9007914578799444,
      "learning_rate": 7.890058031999825e-06,
      "loss": 0.304,
      "step": 1487
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7635137553884721,
      "learning_rate": 7.887170338882742e-06,
      "loss": 0.2746,
      "step": 1488
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7455530910372065,
      "learning_rate": 7.884281200305272e-06,
      "loss": 0.296,
      "step": 1489
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7643866162733286,
      "learning_rate": 7.88139061771386e-06,
      "loss": 0.2686,
      "step": 1490
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7181667396744196,
      "learning_rate": 7.878498592555675e-06,
      "loss": 0.2786,
      "step": 1491
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.890352587498433,
      "learning_rate": 7.875605126278608e-06,
      "loss": 0.2994,
      "step": 1492
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8428130076017047,
      "learning_rate": 7.872710220331271e-06,
      "loss": 0.3161,
      "step": 1493
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7366937949801327,
      "learning_rate": 7.869813876162997e-06,
      "loss": 0.2828,
      "step": 1494
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.676880920244091,
      "learning_rate": 7.866916095223842e-06,
      "loss": 0.2527,
      "step": 1495
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8137587737294969,
      "learning_rate": 7.864016878964575e-06,
      "loss": 0.3083,
      "step": 1496
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7583016511069617,
      "learning_rate": 7.86111622883669e-06,
      "loss": 0.2719,
      "step": 1497
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7292255578040546,
      "learning_rate": 7.858214146292394e-06,
      "loss": 0.3065,
      "step": 1498
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7276848870795923,
      "learning_rate": 7.855310632784615e-06,
      "loss": 0.2656,
      "step": 1499
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7099965049771809,
      "learning_rate": 7.852405689766993e-06,
      "loss": 0.2846,
      "step": 1500
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7023165769322881,
      "learning_rate": 7.849499318693887e-06,
      "loss": 0.2933,
      "step": 1501
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7380842541184434,
      "learning_rate": 7.846591521020373e-06,
      "loss": 0.2735,
      "step": 1502
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.724088210864372,
      "learning_rate": 7.843682298202235e-06,
      "loss": 0.295,
      "step": 1503
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7208917841914452,
      "learning_rate": 7.840771651695976e-06,
      "loss": 0.2778,
      "step": 1504
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7204684445429352,
      "learning_rate": 7.83785958295881e-06,
      "loss": 0.2973,
      "step": 1505
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7753084470357962,
      "learning_rate": 7.834946093448658e-06,
      "loss": 0.2775,
      "step": 1506
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7010744364163912,
      "learning_rate": 7.832031184624165e-06,
      "loss": 0.2665,
      "step": 1507
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7347107280032372,
      "learning_rate": 7.829114857944672e-06,
      "loss": 0.2891,
      "step": 1508
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7563768969741221,
      "learning_rate": 7.826197114870243e-06,
      "loss": 0.3011,
      "step": 1509
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7335109379477206,
      "learning_rate": 7.823277956861635e-06,
      "loss": 0.2825,
      "step": 1510
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7511757105558683,
      "learning_rate": 7.820357385380333e-06,
      "loss": 0.273,
      "step": 1511
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7552883213662275,
      "learning_rate": 7.817435401888514e-06,
      "loss": 0.2804,
      "step": 1512
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8455965463236557,
      "learning_rate": 7.814512007849069e-06,
      "loss": 0.3081,
      "step": 1513
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8047520274236665,
      "learning_rate": 7.811587204725593e-06,
      "loss": 0.3048,
      "step": 1514
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7556808576319619,
      "learning_rate": 7.808660993982389e-06,
      "loss": 0.2809,
      "step": 1515
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7866279767366957,
      "learning_rate": 7.80573337708446e-06,
      "loss": 0.293,
      "step": 1516
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6766703384069377,
      "learning_rate": 7.802804355497519e-06,
      "loss": 0.2559,
      "step": 1517
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7918484830306554,
      "learning_rate": 7.799873930687979e-06,
      "loss": 0.297,
      "step": 1518
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7928007241141743,
      "learning_rate": 7.796942104122953e-06,
      "loss": 0.2854,
      "step": 1519
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7724397952858703,
      "learning_rate": 7.79400887727026e-06,
      "loss": 0.3049,
      "step": 1520
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7989689433618362,
      "learning_rate": 7.79107425159842e-06,
      "loss": 0.2812,
      "step": 1521
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8426075271715305,
      "learning_rate": 7.78813822857665e-06,
      "loss": 0.3013,
      "step": 1522
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7244990509808025,
      "learning_rate": 7.785200809674869e-06,
      "loss": 0.2653,
      "step": 1523
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.741836237150017,
      "learning_rate": 7.782261996363694e-06,
      "loss": 0.294,
      "step": 1524
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7227176809815988,
      "learning_rate": 7.779321790114442e-06,
      "loss": 0.2741,
      "step": 1525
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6954608803556982,
      "learning_rate": 7.776380192399122e-06,
      "loss": 0.2812,
      "step": 1526
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7305467324447517,
      "learning_rate": 7.773437204690449e-06,
      "loss": 0.2855,
      "step": 1527
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7784995206240991,
      "learning_rate": 7.770492828461824e-06,
      "loss": 0.3266,
      "step": 1528
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7407342397284418,
      "learning_rate": 7.767547065187349e-06,
      "loss": 0.2819,
      "step": 1529
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7183460984786887,
      "learning_rate": 7.764599916341817e-06,
      "loss": 0.2765,
      "step": 1530
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7080598190987735,
      "learning_rate": 7.761651383400722e-06,
      "loss": 0.2707,
      "step": 1531
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7274877030370703,
      "learning_rate": 7.75870146784024e-06,
      "loss": 0.2803,
      "step": 1532
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.721955957306433,
      "learning_rate": 7.755750171137245e-06,
      "loss": 0.2904,
      "step": 1533
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7922684473767118,
      "learning_rate": 7.752797494769307e-06,
      "loss": 0.3015,
      "step": 1534
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7492318776553276,
      "learning_rate": 7.749843440214675e-06,
      "loss": 0.3015,
      "step": 1535
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7329948889270185,
      "learning_rate": 7.7468880089523e-06,
      "loss": 0.2681,
      "step": 1536
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7548062842364311,
      "learning_rate": 7.743931202461818e-06,
      "loss": 0.2792,
      "step": 1537
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.752621614310855,
      "learning_rate": 7.74097302222355e-06,
      "loss": 0.2746,
      "step": 1538
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7194751964517884,
      "learning_rate": 7.738013469718507e-06,
      "loss": 0.3015,
      "step": 1539
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7578421324501132,
      "learning_rate": 7.73505254642839e-06,
      "loss": 0.2836,
      "step": 1540
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7434183807253829,
      "learning_rate": 7.73209025383558e-06,
      "loss": 0.2738,
      "step": 1541
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7713380119289839,
      "learning_rate": 7.729126593423151e-06,
      "loss": 0.315,
      "step": 1542
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7356147769598325,
      "learning_rate": 7.726161566674856e-06,
      "loss": 0.2846,
      "step": 1543
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6875634330782326,
      "learning_rate": 7.723195175075136e-06,
      "loss": 0.2725,
      "step": 1544
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7289174144206966,
      "learning_rate": 7.720227420109112e-06,
      "loss": 0.288,
      "step": 1545
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8173339700916057,
      "learning_rate": 7.71725830326259e-06,
      "loss": 0.2964,
      "step": 1546
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8150622254388097,
      "learning_rate": 7.714287826022056e-06,
      "loss": 0.2967,
      "step": 1547
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7052522597922545,
      "learning_rate": 7.711315989874677e-06,
      "loss": 0.2692,
      "step": 1548
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6744847035177614,
      "learning_rate": 7.708342796308303e-06,
      "loss": 0.2541,
      "step": 1549
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7320941472194075,
      "learning_rate": 7.70536824681146e-06,
      "loss": 0.2806,
      "step": 1550
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7590199039500177,
      "learning_rate": 7.702392342873358e-06,
      "loss": 0.3061,
      "step": 1551
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.754345653987312,
      "learning_rate": 7.69941508598388e-06,
      "loss": 0.2872,
      "step": 1552
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7611982238611191,
      "learning_rate": 7.696436477633588e-06,
      "loss": 0.2898,
      "step": 1553
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7095741950151112,
      "learning_rate": 7.69345651931372e-06,
      "loss": 0.2621,
      "step": 1554
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7178074141973462,
      "learning_rate": 7.690475212516191e-06,
      "loss": 0.2899,
      "step": 1555
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7392640711910687,
      "learning_rate": 7.687492558733593e-06,
      "loss": 0.2799,
      "step": 1556
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7748651298577264,
      "learning_rate": 7.684508559459188e-06,
      "loss": 0.3197,
      "step": 1557
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7542639563994519,
      "learning_rate": 7.681523216186912e-06,
      "loss": 0.2869,
      "step": 1558
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.763789087106946,
      "learning_rate": 7.67853653041138e-06,
      "loss": 0.2858,
      "step": 1559
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7816471536730466,
      "learning_rate": 7.675548503627872e-06,
      "loss": 0.286,
      "step": 1560
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7904533867123814,
      "learning_rate": 7.672559137332343e-06,
      "loss": 0.3116,
      "step": 1561
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7498406263119081,
      "learning_rate": 7.669568433021418e-06,
      "loss": 0.2876,
      "step": 1562
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7003685557798964,
      "learning_rate": 7.666576392192389e-06,
      "loss": 0.2854,
      "step": 1563
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7634250617956412,
      "learning_rate": 7.663583016343224e-06,
      "loss": 0.2951,
      "step": 1564
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7794573035276008,
      "learning_rate": 7.66058830697255e-06,
      "loss": 0.3047,
      "step": 1565
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7556930360711428,
      "learning_rate": 7.65759226557967e-06,
      "loss": 0.3003,
      "step": 1566
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7525429704111921,
      "learning_rate": 7.65459489366455e-06,
      "loss": 0.284,
      "step": 1567
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7543106325589166,
      "learning_rate": 7.651596192727826e-06,
      "loss": 0.2951,
      "step": 1568
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7952143849659247,
      "learning_rate": 7.648596164270792e-06,
      "loss": 0.3064,
      "step": 1569
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.747902109313066,
      "learning_rate": 7.64559480979541e-06,
      "loss": 0.2962,
      "step": 1570
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7760035051592336,
      "learning_rate": 7.64259213080431e-06,
      "loss": 0.3173,
      "step": 1571
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8564546452448831,
      "learning_rate": 7.63958812880078e-06,
      "loss": 0.2551,
      "step": 1572
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7699120115014312,
      "learning_rate": 7.636582805288771e-06,
      "loss": 0.2909,
      "step": 1573
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6985172696816996,
      "learning_rate": 7.633576161772898e-06,
      "loss": 0.2638,
      "step": 1574
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7633404543616388,
      "learning_rate": 7.630568199758436e-06,
      "loss": 0.2799,
      "step": 1575
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7243893949130152,
      "learning_rate": 7.627558920751321e-06,
      "loss": 0.2859,
      "step": 1576
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7668370120644918,
      "learning_rate": 7.624548326258144e-06,
      "loss": 0.2858,
      "step": 1577
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7365316368613518,
      "learning_rate": 7.621536417786159e-06,
      "loss": 0.2856,
      "step": 1578
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7509842706855119,
      "learning_rate": 7.618523196843276e-06,
      "loss": 0.304,
      "step": 1579
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7620773694600913,
      "learning_rate": 7.615508664938065e-06,
      "loss": 0.3076,
      "step": 1580
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7122668254442069,
      "learning_rate": 7.612492823579744e-06,
      "loss": 0.2588,
      "step": 1581
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7159585981857514,
      "learning_rate": 7.6094756742782e-06,
      "loss": 0.2681,
      "step": 1582
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.784099472908839,
      "learning_rate": 7.606457218543961e-06,
      "loss": 0.2733,
      "step": 1583
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7929282085792096,
      "learning_rate": 7.6034374578882205e-06,
      "loss": 0.2817,
      "step": 1584
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.788850958462567,
      "learning_rate": 7.600416393822816e-06,
      "loss": 0.2915,
      "step": 1585
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7059482724935049,
      "learning_rate": 7.597394027860243e-06,
      "loss": 0.2669,
      "step": 1586
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8105429252261422,
      "learning_rate": 7.594370361513648e-06,
      "loss": 0.2844,
      "step": 1587
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7671418089117007,
      "learning_rate": 7.5913453962968296e-06,
      "loss": 0.3181,
      "step": 1588
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7559070951038315,
      "learning_rate": 7.5883191337242315e-06,
      "loss": 0.2886,
      "step": 1589
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.727023773099087,
      "learning_rate": 7.585291575310952e-06,
      "loss": 0.2752,
      "step": 1590
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7484733859413457,
      "learning_rate": 7.582262722572739e-06,
      "loss": 0.2841,
      "step": 1591
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.784398695667266,
      "learning_rate": 7.579232577025984e-06,
      "loss": 0.2977,
      "step": 1592
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7491976623849015,
      "learning_rate": 7.576201140187727e-06,
      "loss": 0.2797,
      "step": 1593
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6967027221990336,
      "learning_rate": 7.5731684135756566e-06,
      "loss": 0.2618,
      "step": 1594
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7662008008115204,
      "learning_rate": 7.570134398708106e-06,
      "loss": 0.2742,
      "step": 1595
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7601941879813825,
      "learning_rate": 7.567099097104053e-06,
      "loss": 0.265,
      "step": 1596
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7558374813870496,
      "learning_rate": 7.5640625102831215e-06,
      "loss": 0.2939,
      "step": 1597
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6944610115700677,
      "learning_rate": 7.5610246397655715e-06,
      "loss": 0.2762,
      "step": 1598
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7154487221447293,
      "learning_rate": 7.557985487072318e-06,
      "loss": 0.2713,
      "step": 1599
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7365284337531287,
      "learning_rate": 7.5549450537249096e-06,
      "loss": 0.2689,
      "step": 1600
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7171992622543646,
      "learning_rate": 7.551903341245534e-06,
      "loss": 0.2794,
      "step": 1601
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7008117150387158,
      "learning_rate": 7.548860351157028e-06,
      "loss": 0.2845,
      "step": 1602
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6551272359386591,
      "learning_rate": 7.54581608498286e-06,
      "loss": 0.2334,
      "step": 1603
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.760134220254608,
      "learning_rate": 7.542770544247141e-06,
      "loss": 0.2915,
      "step": 1604
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7404761746098899,
      "learning_rate": 7.53972373047462e-06,
      "loss": 0.2897,
      "step": 1605
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7062880649209463,
      "learning_rate": 7.536675645190682e-06,
      "loss": 0.2796,
      "step": 1606
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7090246529624624,
      "learning_rate": 7.53362628992135e-06,
      "loss": 0.2856,
      "step": 1607
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7281717436620446,
      "learning_rate": 7.530575666193283e-06,
      "loss": 0.2727,
      "step": 1608
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6833530758572677,
      "learning_rate": 7.527523775533771e-06,
      "loss": 0.2684,
      "step": 1609
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7643302326004493,
      "learning_rate": 7.524470619470745e-06,
      "loss": 0.3037,
      "step": 1610
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7211107495749812,
      "learning_rate": 7.521416199532765e-06,
      "loss": 0.2799,
      "step": 1611
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7408549996031762,
      "learning_rate": 7.518360517249024e-06,
      "loss": 0.2932,
      "step": 1612
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7691205165281905,
      "learning_rate": 7.515303574149348e-06,
      "loss": 0.2725,
      "step": 1613
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7219359659352503,
      "learning_rate": 7.512245371764197e-06,
      "loss": 0.3,
      "step": 1614
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7537107186474019,
      "learning_rate": 7.509185911624653e-06,
      "loss": 0.2861,
      "step": 1615
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7029989531456574,
      "learning_rate": 7.506125195262437e-06,
      "loss": 0.2878,
      "step": 1616
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.693036544447862,
      "learning_rate": 7.503063224209896e-06,
      "loss": 0.2587,
      "step": 1617
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7662842693077363,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.2874,
      "step": 1618
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7602852571940877,
      "learning_rate": 7.4969355241663554e-06,
      "loss": 0.296,
      "step": 1619
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7805575991264153,
      "learning_rate": 7.493869798243188e-06,
      "loss": 0.2748,
      "step": 1620
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.727537732927315,
      "learning_rate": 7.490802823765352e-06,
      "loss": 0.2836,
      "step": 1621
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7721836370030607,
      "learning_rate": 7.487734602268326e-06,
      "loss": 0.2883,
      "step": 1622
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7028611372861732,
      "learning_rate": 7.484665135288214e-06,
      "loss": 0.2773,
      "step": 1623
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7150590281986862,
      "learning_rate": 7.4815944243617455e-06,
      "loss": 0.2895,
      "step": 1624
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7400411329227471,
      "learning_rate": 7.478522471026267e-06,
      "loss": 0.2967,
      "step": 1625
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8413759890426574,
      "learning_rate": 7.475449276819753e-06,
      "loss": 0.299,
      "step": 1626
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7634492482459837,
      "learning_rate": 7.472374843280794e-06,
      "loss": 0.2795,
      "step": 1627
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7729430795494269,
      "learning_rate": 7.469299171948608e-06,
      "loss": 0.3041,
      "step": 1628
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6719164635774132,
      "learning_rate": 7.466222264363022e-06,
      "loss": 0.2892,
      "step": 1629
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7113422523861179,
      "learning_rate": 7.463144122064494e-06,
      "loss": 0.2797,
      "step": 1630
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7770699428277016,
      "learning_rate": 7.460064746594091e-06,
      "loss": 0.3042,
      "step": 1631
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7189297616193195,
      "learning_rate": 7.456984139493502e-06,
      "loss": 0.2852,
      "step": 1632
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6845229159450801,
      "learning_rate": 7.453902302305032e-06,
      "loss": 0.2756,
      "step": 1633
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7367925709568842,
      "learning_rate": 7.4508192365716005e-06,
      "loss": 0.2958,
      "step": 1634
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7309699747332289,
      "learning_rate": 7.447734943836742e-06,
      "loss": 0.3009,
      "step": 1635
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8097880237554279,
      "learning_rate": 7.444649425644606e-06,
      "loss": 0.2918,
      "step": 1636
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7536850508206001,
      "learning_rate": 7.441562683539959e-06,
      "loss": 0.2887,
      "step": 1637
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7230659894358514,
      "learning_rate": 7.438474719068174e-06,
      "loss": 0.3025,
      "step": 1638
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6936632551794452,
      "learning_rate": 7.435385533775239e-06,
      "loss": 0.2508,
      "step": 1639
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7508320811372227,
      "learning_rate": 7.432295129207753e-06,
      "loss": 0.2771,
      "step": 1640
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7159677287157274,
      "learning_rate": 7.429203506912927e-06,
      "loss": 0.2672,
      "step": 1641
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.745298054833925,
      "learning_rate": 7.426110668438578e-06,
      "loss": 0.2946,
      "step": 1642
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7767202426737069,
      "learning_rate": 7.423016615333135e-06,
      "loss": 0.2975,
      "step": 1643
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7028097710463417,
      "learning_rate": 7.419921349145635e-06,
      "loss": 0.2762,
      "step": 1644
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8241215632635451,
      "learning_rate": 7.4168248714257204e-06,
      "loss": 0.2764,
      "step": 1645
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7513413939008636,
      "learning_rate": 7.413727183723642e-06,
      "loss": 0.2835,
      "step": 1646
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7428203179126651,
      "learning_rate": 7.410628287590254e-06,
      "loss": 0.2943,
      "step": 1647
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8042045186549074,
      "learning_rate": 7.4075281845770196e-06,
      "loss": 0.2668,
      "step": 1648
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7739140770784639,
      "learning_rate": 7.404426876236003e-06,
      "loss": 0.3077,
      "step": 1649
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7474931454894871,
      "learning_rate": 7.401324364119872e-06,
      "loss": 0.2764,
      "step": 1650
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7700193109151653,
      "learning_rate": 7.3982206497818965e-06,
      "loss": 0.2857,
      "step": 1651
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6987617377170268,
      "learning_rate": 7.395115734775953e-06,
      "loss": 0.2803,
      "step": 1652
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6610345019721695,
      "learning_rate": 7.392009620656513e-06,
      "loss": 0.2578,
      "step": 1653
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7738089743795716,
      "learning_rate": 7.388902308978652e-06,
      "loss": 0.2965,
      "step": 1654
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8330387965571734,
      "learning_rate": 7.3857938012980425e-06,
      "loss": 0.2833,
      "step": 1655
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7463488359420019,
      "learning_rate": 7.382684099170959e-06,
      "loss": 0.2708,
      "step": 1656
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7307305567199199,
      "learning_rate": 7.379573204154274e-06,
      "loss": 0.2903,
      "step": 1657
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.71687693085662,
      "learning_rate": 7.37646111780545e-06,
      "loss": 0.276,
      "step": 1658
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7915179388027526,
      "learning_rate": 7.3733478416825565e-06,
      "loss": 0.3013,
      "step": 1659
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7120494832613936,
      "learning_rate": 7.370233377344252e-06,
      "loss": 0.2828,
      "step": 1660
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.754984184198239,
      "learning_rate": 7.367117726349793e-06,
      "loss": 0.2818,
      "step": 1661
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7355901805103071,
      "learning_rate": 7.364000890259024e-06,
      "loss": 0.2834,
      "step": 1662
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7840418059734934,
      "learning_rate": 7.360882870632393e-06,
      "loss": 0.2893,
      "step": 1663
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.723959638161678,
      "learning_rate": 7.357763669030931e-06,
      "loss": 0.2886,
      "step": 1664
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7679121399594483,
      "learning_rate": 7.354643287016268e-06,
      "loss": 0.3094,
      "step": 1665
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7269119392274389,
      "learning_rate": 7.351521726150619e-06,
      "loss": 0.2894,
      "step": 1666
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7488054664522558,
      "learning_rate": 7.348398987996792e-06,
      "loss": 0.2791,
      "step": 1667
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7039730816574329,
      "learning_rate": 7.3452750741181855e-06,
      "loss": 0.2821,
      "step": 1668
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8492409327488027,
      "learning_rate": 7.342149986078786e-06,
      "loss": 0.2977,
      "step": 1669
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8063292669237198,
      "learning_rate": 7.339023725443166e-06,
      "loss": 0.2999,
      "step": 1670
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7082028998384131,
      "learning_rate": 7.335896293776486e-06,
      "loss": 0.2837,
      "step": 1671
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7106121274840315,
      "learning_rate": 7.332767692644493e-06,
      "loss": 0.2869,
      "step": 1672
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.736671422375547,
      "learning_rate": 7.329637923613522e-06,
      "loss": 0.286,
      "step": 1673
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7308148139155677,
      "learning_rate": 7.326506988250488e-06,
      "loss": 0.2941,
      "step": 1674
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7695656029693637,
      "learning_rate": 7.323374888122893e-06,
      "loss": 0.2799,
      "step": 1675
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7059002690446718,
      "learning_rate": 7.320241624798821e-06,
      "loss": 0.2703,
      "step": 1676
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.809849046097867,
      "learning_rate": 7.31710719984694e-06,
      "loss": 0.316,
      "step": 1677
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7256096662437354,
      "learning_rate": 7.313971614836496e-06,
      "loss": 0.2869,
      "step": 1678
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7121018382014945,
      "learning_rate": 7.310834871337319e-06,
      "loss": 0.2577,
      "step": 1679
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7582693055480179,
      "learning_rate": 7.307696970919819e-06,
      "loss": 0.2893,
      "step": 1680
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7460825937277334,
      "learning_rate": 7.3045579151549836e-06,
      "loss": 0.2947,
      "step": 1681
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7681186688124774,
      "learning_rate": 7.3014177056143795e-06,
      "loss": 0.2871,
      "step": 1682
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6890651840277093,
      "learning_rate": 7.298276343870152e-06,
      "loss": 0.2902,
      "step": 1683
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7272284318217722,
      "learning_rate": 7.295133831495021e-06,
      "loss": 0.2989,
      "step": 1684
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7185193591989671,
      "learning_rate": 7.291990170062283e-06,
      "loss": 0.2821,
      "step": 1685
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7408240642597359,
      "learning_rate": 7.288845361145812e-06,
      "loss": 0.2777,
      "step": 1686
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7814933778586025,
      "learning_rate": 7.285699406320055e-06,
      "loss": 0.2754,
      "step": 1687
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7415625791932362,
      "learning_rate": 7.282552307160033e-06,
      "loss": 0.286,
      "step": 1688
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7251885187615795,
      "learning_rate": 7.2794040652413374e-06,
      "loss": 0.2805,
      "step": 1689
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7625631964088719,
      "learning_rate": 7.276254682140137e-06,
      "loss": 0.2895,
      "step": 1690
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7612552314716403,
      "learning_rate": 7.273104159433166e-06,
      "loss": 0.2653,
      "step": 1691
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7925748914659962,
      "learning_rate": 7.269952498697734e-06,
      "loss": 0.2949,
      "step": 1692
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8232305876000426,
      "learning_rate": 7.26679970151172e-06,
      "loss": 0.297,
      "step": 1693
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8195688745895094,
      "learning_rate": 7.2636457694535655e-06,
      "loss": 0.312,
      "step": 1694
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7286069677268305,
      "learning_rate": 7.260490704102287e-06,
      "loss": 0.2711,
      "step": 1695
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8090765255892995,
      "learning_rate": 7.2573345070374675e-06,
      "loss": 0.2739,
      "step": 1696
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7061827667959241,
      "learning_rate": 7.254177179839257e-06,
      "loss": 0.2896,
      "step": 1697
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7116392751479221,
      "learning_rate": 7.251018724088367e-06,
      "loss": 0.2638,
      "step": 1698
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7586518107637933,
      "learning_rate": 7.247859141366077e-06,
      "loss": 0.2978,
      "step": 1699
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7346136920994816,
      "learning_rate": 7.244698433254231e-06,
      "loss": 0.2737,
      "step": 1700
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8362305383311675,
      "learning_rate": 7.241536601335237e-06,
      "loss": 0.2884,
      "step": 1701
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6974933762465726,
      "learning_rate": 7.238373647192063e-06,
      "loss": 0.2865,
      "step": 1702
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7269665067353086,
      "learning_rate": 7.235209572408241e-06,
      "loss": 0.2849,
      "step": 1703
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7640844568758955,
      "learning_rate": 7.232044378567864e-06,
      "loss": 0.2979,
      "step": 1704
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7735987082373242,
      "learning_rate": 7.228878067255585e-06,
      "loss": 0.2985,
      "step": 1705
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7212616661501471,
      "learning_rate": 7.225710640056614e-06,
      "loss": 0.2891,
      "step": 1706
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7353531561381292,
      "learning_rate": 7.222542098556722e-06,
      "loss": 0.2722,
      "step": 1707
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8518282825869693,
      "learning_rate": 7.2193724443422405e-06,
      "loss": 0.2939,
      "step": 1708
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7685726217958676,
      "learning_rate": 7.2162016790000545e-06,
      "loss": 0.2744,
      "step": 1709
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.734026254259307,
      "learning_rate": 7.213029804117604e-06,
      "loss": 0.3021,
      "step": 1710
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7384511975779009,
      "learning_rate": 7.209856821282888e-06,
      "loss": 0.2676,
      "step": 1711
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.700036630340313,
      "learning_rate": 7.206682732084458e-06,
      "loss": 0.2844,
      "step": 1712
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7442138585660607,
      "learning_rate": 7.203507538111423e-06,
      "loss": 0.2794,
      "step": 1713
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7331261185233161,
      "learning_rate": 7.200331240953439e-06,
      "loss": 0.2796,
      "step": 1714
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8032576114470661,
      "learning_rate": 7.197153842200718e-06,
      "loss": 0.2844,
      "step": 1715
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7327471469504304,
      "learning_rate": 7.193975343444024e-06,
      "loss": 0.2642,
      "step": 1716
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6973397291699059,
      "learning_rate": 7.190795746274673e-06,
      "loss": 0.2844,
      "step": 1717
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.826015876954625,
      "learning_rate": 7.187615052284522e-06,
      "loss": 0.2987,
      "step": 1718
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7365335454499613,
      "learning_rate": 7.184433263065991e-06,
      "loss": 0.2628,
      "step": 1719
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7214239399862475,
      "learning_rate": 7.181250380212037e-06,
      "loss": 0.3032,
      "step": 1720
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7072576204371066,
      "learning_rate": 7.178066405316172e-06,
      "loss": 0.2602,
      "step": 1721
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.739647779800438,
      "learning_rate": 7.174881339972448e-06,
      "loss": 0.2776,
      "step": 1722
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7639618978256241,
      "learning_rate": 7.171695185775468e-06,
      "loss": 0.2652,
      "step": 1723
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7742314929778945,
      "learning_rate": 7.1685079443203784e-06,
      "loss": 0.2993,
      "step": 1724
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7246603025426065,
      "learning_rate": 7.165319617202871e-06,
      "loss": 0.2855,
      "step": 1725
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7577004453478776,
      "learning_rate": 7.16213020601918e-06,
      "loss": 0.2831,
      "step": 1726
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7410955395700459,
      "learning_rate": 7.158939712366079e-06,
      "loss": 0.2912,
      "step": 1727
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6943434207052043,
      "learning_rate": 7.155748137840892e-06,
      "loss": 0.2596,
      "step": 1728
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7831264997023479,
      "learning_rate": 7.1525554840414765e-06,
      "loss": 0.2829,
      "step": 1729
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7993603073806127,
      "learning_rate": 7.149361752566233e-06,
      "loss": 0.2842,
      "step": 1730
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7221128291863135,
      "learning_rate": 7.146166945014102e-06,
      "loss": 0.2694,
      "step": 1731
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7691867260370905,
      "learning_rate": 7.142971062984563e-06,
      "loss": 0.276,
      "step": 1732
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7709245250737005,
      "learning_rate": 7.139774108077633e-06,
      "loss": 0.3023,
      "step": 1733
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7399215631632148,
      "learning_rate": 7.136576081893863e-06,
      "loss": 0.2682,
      "step": 1734
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6999385308918328,
      "learning_rate": 7.133376986034346e-06,
      "loss": 0.2747,
      "step": 1735
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7742099022031292,
      "learning_rate": 7.130176822100707e-06,
      "loss": 0.2723,
      "step": 1736
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7828516966323655,
      "learning_rate": 7.126975591695108e-06,
      "loss": 0.2824,
      "step": 1737
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.772614644374149,
      "learning_rate": 7.12377329642024e-06,
      "loss": 0.2814,
      "step": 1738
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7321831959037887,
      "learning_rate": 7.120569937879333e-06,
      "loss": 0.2887,
      "step": 1739
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7272674098429573,
      "learning_rate": 7.117365517676146e-06,
      "loss": 0.2814,
      "step": 1740
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7485951546959866,
      "learning_rate": 7.11416003741497e-06,
      "loss": 0.2723,
      "step": 1741
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6854265212338791,
      "learning_rate": 7.110953498700629e-06,
      "loss": 0.2608,
      "step": 1742
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7884692498077537,
      "learning_rate": 7.107745903138472e-06,
      "loss": 0.3037,
      "step": 1743
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7444031897860472,
      "learning_rate": 7.104537252334383e-06,
      "loss": 0.2864,
      "step": 1744
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7607257739793158,
      "learning_rate": 7.101327547894771e-06,
      "loss": 0.2774,
      "step": 1745
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7414661531763037,
      "learning_rate": 7.09811679142657e-06,
      "loss": 0.2623,
      "step": 1746
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7104003326212947,
      "learning_rate": 7.0949049845372475e-06,
      "loss": 0.2753,
      "step": 1747
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7244026137442655,
      "learning_rate": 7.09169212883479e-06,
      "loss": 0.2834,
      "step": 1748
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6913675763708204,
      "learning_rate": 7.0884782259277155e-06,
      "loss": 0.2618,
      "step": 1749
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.714410003636145,
      "learning_rate": 7.085263277425058e-06,
      "loss": 0.2715,
      "step": 1750
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7230840360701067,
      "learning_rate": 7.082047284936385e-06,
      "loss": 0.2692,
      "step": 1751
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7284809144259901,
      "learning_rate": 7.078830250071778e-06,
      "loss": 0.2731,
      "step": 1752
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6909535111682021,
      "learning_rate": 7.075612174441846e-06,
      "loss": 0.2637,
      "step": 1753
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7338783491639093,
      "learning_rate": 7.072393059657716e-06,
      "loss": 0.3082,
      "step": 1754
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7080055338207052,
      "learning_rate": 7.069172907331034e-06,
      "loss": 0.2636,
      "step": 1755
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7337336283599293,
      "learning_rate": 7.065951719073973e-06,
      "loss": 0.2825,
      "step": 1756
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7196313692470484,
      "learning_rate": 7.0627294964992145e-06,
      "loss": 0.2775,
      "step": 1757
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7731317778952145,
      "learning_rate": 7.059506241219964e-06,
      "loss": 0.2921,
      "step": 1758
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6933275627762446,
      "learning_rate": 7.056281954849944e-06,
      "loss": 0.2793,
      "step": 1759
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7189155573307195,
      "learning_rate": 7.053056639003389e-06,
      "loss": 0.2893,
      "step": 1760
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7217565074905102,
      "learning_rate": 7.049830295295057e-06,
      "loss": 0.2513,
      "step": 1761
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7013772893271781,
      "learning_rate": 7.046602925340209e-06,
      "loss": 0.2655,
      "step": 1762
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7039109079079224,
      "learning_rate": 7.04337453075463e-06,
      "loss": 0.2755,
      "step": 1763
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7057273968797994,
      "learning_rate": 7.0401451131546125e-06,
      "loss": 0.2724,
      "step": 1764
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6992008323423498,
      "learning_rate": 7.036914674156964e-06,
      "loss": 0.2766,
      "step": 1765
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7144359482320997,
      "learning_rate": 7.033683215379002e-06,
      "loss": 0.2737,
      "step": 1766
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7146174001497837,
      "learning_rate": 7.0304507384385535e-06,
      "loss": 0.2572,
      "step": 1767
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.716444633997443,
      "learning_rate": 7.027217244953958e-06,
      "loss": 0.2839,
      "step": 1768
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.702232758040956,
      "learning_rate": 7.023982736544063e-06,
      "loss": 0.2825,
      "step": 1769
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6988764741463727,
      "learning_rate": 7.020747214828221e-06,
      "loss": 0.2926,
      "step": 1770
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7739548850767979,
      "learning_rate": 7.017510681426296e-06,
      "loss": 0.3018,
      "step": 1771
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.698250707025481,
      "learning_rate": 7.014273137958656e-06,
      "loss": 0.2854,
      "step": 1772
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7400352670818225,
      "learning_rate": 7.011034586046177e-06,
      "loss": 0.2676,
      "step": 1773
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6953520845095524,
      "learning_rate": 7.007795027310236e-06,
      "loss": 0.2841,
      "step": 1774
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7550472311001029,
      "learning_rate": 7.004554463372719e-06,
      "loss": 0.2912,
      "step": 1775
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7118847806420704,
      "learning_rate": 7.001312895856011e-06,
      "loss": 0.2785,
      "step": 1776
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7924071516174388,
      "learning_rate": 6.998070326383002e-06,
      "loss": 0.2974,
      "step": 1777
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.712259808841367,
      "learning_rate": 6.994826756577082e-06,
      "loss": 0.2744,
      "step": 1778
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7187777705086817,
      "learning_rate": 6.991582188062144e-06,
      "loss": 0.2731,
      "step": 1779
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7683771170222067,
      "learning_rate": 6.988336622462578e-06,
      "loss": 0.2818,
      "step": 1780
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7733858555496339,
      "learning_rate": 6.985090061403277e-06,
      "loss": 0.296,
      "step": 1781
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6670991353645158,
      "learning_rate": 6.981842506509626e-06,
      "loss": 0.2651,
      "step": 1782
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7783556626593441,
      "learning_rate": 6.978593959407516e-06,
      "loss": 0.2993,
      "step": 1783
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6910240393977983,
      "learning_rate": 6.975344421723329e-06,
      "loss": 0.2734,
      "step": 1784
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7357763846398442,
      "learning_rate": 6.9720938950839445e-06,
      "loss": 0.2694,
      "step": 1785
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7037652288062016,
      "learning_rate": 6.968842381116736e-06,
      "loss": 0.2799,
      "step": 1786
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.754108237450581,
      "learning_rate": 6.965589881449574e-06,
      "loss": 0.2689,
      "step": 1787
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7937726598275043,
      "learning_rate": 6.962336397710819e-06,
      "loss": 0.2939,
      "step": 1788
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6807082557668076,
      "learning_rate": 6.959081931529328e-06,
      "loss": 0.2728,
      "step": 1789
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7246728093879773,
      "learning_rate": 6.955826484534447e-06,
      "loss": 0.2815,
      "step": 1790
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7272075139697921,
      "learning_rate": 6.952570058356013e-06,
      "loss": 0.2694,
      "step": 1791
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8309995985973556,
      "learning_rate": 6.949312654624355e-06,
      "loss": 0.3233,
      "step": 1792
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7679308168174795,
      "learning_rate": 6.946054274970292e-06,
      "loss": 0.3139,
      "step": 1793
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7310030082480796,
      "learning_rate": 6.942794921025127e-06,
      "loss": 0.286,
      "step": 1794
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7614878203686738,
      "learning_rate": 6.939534594420656e-06,
      "loss": 0.273,
      "step": 1795
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7478335521126207,
      "learning_rate": 6.93627329678916e-06,
      "loss": 0.3044,
      "step": 1796
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7469399271757192,
      "learning_rate": 6.933011029763406e-06,
      "loss": 0.2876,
      "step": 1797
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6562143354826936,
      "learning_rate": 6.9297477949766445e-06,
      "loss": 0.2552,
      "step": 1798
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7233801160250665,
      "learning_rate": 6.9264835940626145e-06,
      "loss": 0.2972,
      "step": 1799
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6895916260577517,
      "learning_rate": 6.9232184286555345e-06,
      "loss": 0.2566,
      "step": 1800
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6942665694915807,
      "learning_rate": 6.919952300390111e-06,
      "loss": 0.2569,
      "step": 1801
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7313070693372549,
      "learning_rate": 6.916685210901527e-06,
      "loss": 0.3034,
      "step": 1802
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7392235731602189,
      "learning_rate": 6.913417161825449e-06,
      "loss": 0.2755,
      "step": 1803
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7028923459637796,
      "learning_rate": 6.910148154798026e-06,
      "loss": 0.2729,
      "step": 1804
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7317224385164145,
      "learning_rate": 6.906878191455883e-06,
      "loss": 0.2828,
      "step": 1805
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.765612665429997,
      "learning_rate": 6.903607273436128e-06,
      "loss": 0.2819,
      "step": 1806
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8031190226170362,
      "learning_rate": 6.900335402376341e-06,
      "loss": 0.2957,
      "step": 1807
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7128903653009858,
      "learning_rate": 6.897062579914587e-06,
      "loss": 0.2807,
      "step": 1808
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7389933700167703,
      "learning_rate": 6.8937888076893964e-06,
      "loss": 0.2746,
      "step": 1809
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7340607238717314,
      "learning_rate": 6.890514087339789e-06,
      "loss": 0.3004,
      "step": 1810
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6945561537052288,
      "learning_rate": 6.887238420505247e-06,
      "loss": 0.291,
      "step": 1811
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6922943448894608,
      "learning_rate": 6.883961808825732e-06,
      "loss": 0.2791,
      "step": 1812
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7104107962393728,
      "learning_rate": 6.88068425394168e-06,
      "loss": 0.28,
      "step": 1813
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6872784778895678,
      "learning_rate": 6.8774057574939956e-06,
      "loss": 0.267,
      "step": 1814
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.773570244240207,
      "learning_rate": 6.874126321124057e-06,
      "loss": 0.2677,
      "step": 1815
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7251962299691589,
      "learning_rate": 6.870845946473711e-06,
      "loss": 0.282,
      "step": 1816
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7227300957731865,
      "learning_rate": 6.867564635185279e-06,
      "loss": 0.2864,
      "step": 1817
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7328454249959027,
      "learning_rate": 6.864282388901544e-06,
      "loss": 0.3022,
      "step": 1818
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6720193037757877,
      "learning_rate": 6.860999209265762e-06,
      "loss": 0.2695,
      "step": 1819
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.787068724419666,
      "learning_rate": 6.857715097921657e-06,
      "loss": 0.2865,
      "step": 1820
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.693337891819729,
      "learning_rate": 6.854430056513417e-06,
      "loss": 0.2878,
      "step": 1821
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6844607768431952,
      "learning_rate": 6.851144086685695e-06,
      "loss": 0.2587,
      "step": 1822
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.737441477172111,
      "learning_rate": 6.847857190083611e-06,
      "loss": 0.2775,
      "step": 1823
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7663873680739703,
      "learning_rate": 6.844569368352748e-06,
      "loss": 0.2858,
      "step": 1824
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7030054270632352,
      "learning_rate": 6.841280623139153e-06,
      "loss": 0.2788,
      "step": 1825
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7304440455013389,
      "learning_rate": 6.837990956089334e-06,
      "loss": 0.2994,
      "step": 1826
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7255711293522648,
      "learning_rate": 6.8347003688502614e-06,
      "loss": 0.2798,
      "step": 1827
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.722578673154131,
      "learning_rate": 6.831408863069364e-06,
      "loss": 0.2707,
      "step": 1828
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.703284873156879,
      "learning_rate": 6.828116440394536e-06,
      "loss": 0.2658,
      "step": 1829
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7146847129946112,
      "learning_rate": 6.824823102474127e-06,
      "loss": 0.2788,
      "step": 1830
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7612513203468594,
      "learning_rate": 6.821528850956942e-06,
      "loss": 0.2625,
      "step": 1831
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7258109690212726,
      "learning_rate": 6.818233687492248e-06,
      "loss": 0.2672,
      "step": 1832
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7351835310991254,
      "learning_rate": 6.814937613729766e-06,
      "loss": 0.3108,
      "step": 1833
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7231492007501611,
      "learning_rate": 6.811640631319677e-06,
      "loss": 0.2785,
      "step": 1834
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7164697978731931,
      "learning_rate": 6.80834274191261e-06,
      "loss": 0.2828,
      "step": 1835
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7576451927360557,
      "learning_rate": 6.805043947159652e-06,
      "loss": 0.2708,
      "step": 1836
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.709433116990722,
      "learning_rate": 6.801744248712345e-06,
      "loss": 0.2761,
      "step": 1837
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7225747225211355,
      "learning_rate": 6.79844364822268e-06,
      "loss": 0.2807,
      "step": 1838
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7692222121263681,
      "learning_rate": 6.795142147343101e-06,
      "loss": 0.2742,
      "step": 1839
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8079729585601091,
      "learning_rate": 6.7918397477265e-06,
      "loss": 0.2858,
      "step": 1840
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.757406258523277,
      "learning_rate": 6.78853645102623e-06,
      "loss": 0.2828,
      "step": 1841
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6805046587465851,
      "learning_rate": 6.785232258896078e-06,
      "loss": 0.2656,
      "step": 1842
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7611497976385118,
      "learning_rate": 6.781927172990285e-06,
      "loss": 0.2917,
      "step": 1843
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7337840011651761,
      "learning_rate": 6.778621194963545e-06,
      "loss": 0.2793,
      "step": 1844
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7129347122202683,
      "learning_rate": 6.775314326470992e-06,
      "loss": 0.3042,
      "step": 1845
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.72244979834723,
      "learning_rate": 6.77200656916821e-06,
      "loss": 0.2699,
      "step": 1846
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7466622397890397,
      "learning_rate": 6.768697924711223e-06,
      "loss": 0.2813,
      "step": 1847
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7456655984978305,
      "learning_rate": 6.765388394756504e-06,
      "loss": 0.2737,
      "step": 1848
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7670987815559354,
      "learning_rate": 6.762077980960968e-06,
      "loss": 0.2994,
      "step": 1849
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7975017336761592,
      "learning_rate": 6.758766684981972e-06,
      "loss": 0.2946,
      "step": 1850
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.715610976630328,
      "learning_rate": 6.755454508477313e-06,
      "loss": 0.2587,
      "step": 1851
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7594491425675967,
      "learning_rate": 6.752141453105231e-06,
      "loss": 0.2884,
      "step": 1852
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7703317131552213,
      "learning_rate": 6.748827520524406e-06,
      "loss": 0.2753,
      "step": 1853
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7805959690171534,
      "learning_rate": 6.745512712393958e-06,
      "loss": 0.2752,
      "step": 1854
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7522737789111458,
      "learning_rate": 6.742197030373442e-06,
      "loss": 0.2857,
      "step": 1855
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7423496694280001,
      "learning_rate": 6.7388804761228535e-06,
      "loss": 0.2943,
      "step": 1856
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6947920666199925,
      "learning_rate": 6.735563051302622e-06,
      "loss": 0.2825,
      "step": 1857
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7264128171037199,
      "learning_rate": 6.732244757573619e-06,
      "loss": 0.2511,
      "step": 1858
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8025757094051992,
      "learning_rate": 6.728925596597142e-06,
      "loss": 0.2772,
      "step": 1859
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7220656522636424,
      "learning_rate": 6.725605570034928e-06,
      "loss": 0.2732,
      "step": 1860
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7691670892171477,
      "learning_rate": 6.722284679549149e-06,
      "loss": 0.2877,
      "step": 1861
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6739897479567528,
      "learning_rate": 6.718962926802407e-06,
      "loss": 0.269,
      "step": 1862
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7766511869962528,
      "learning_rate": 6.715640313457733e-06,
      "loss": 0.2924,
      "step": 1863
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.782919908702191,
      "learning_rate": 6.712316841178592e-06,
      "loss": 0.3044,
      "step": 1864
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7567324680821782,
      "learning_rate": 6.708992511628882e-06,
      "loss": 0.2856,
      "step": 1865
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7531972971371086,
      "learning_rate": 6.705667326472926e-06,
      "loss": 0.285,
      "step": 1866
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7673400854710007,
      "learning_rate": 6.702341287375473e-06,
      "loss": 0.2699,
      "step": 1867
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7621457301133813,
      "learning_rate": 6.699014396001707e-06,
      "loss": 0.2712,
      "step": 1868
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7104863647858227,
      "learning_rate": 6.6956866540172305e-06,
      "loss": 0.2576,
      "step": 1869
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7099038525771183,
      "learning_rate": 6.69235806308808e-06,
      "loss": 0.2928,
      "step": 1870
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7222098143884347,
      "learning_rate": 6.6890286248807104e-06,
      "loss": 0.2818,
      "step": 1871
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7438677462603746,
      "learning_rate": 6.685698341062003e-06,
      "loss": 0.3019,
      "step": 1872
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8142918422287644,
      "learning_rate": 6.682367213299264e-06,
      "loss": 0.2864,
      "step": 1873
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7488593205534378,
      "learning_rate": 6.67903524326022e-06,
      "loss": 0.2808,
      "step": 1874
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7308252400763589,
      "learning_rate": 6.6757024326130204e-06,
      "loss": 0.2762,
      "step": 1875
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.722395768431837,
      "learning_rate": 6.672368783026235e-06,
      "loss": 0.2703,
      "step": 1876
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.766572371597163,
      "learning_rate": 6.669034296168855e-06,
      "loss": 0.2748,
      "step": 1877
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7387973217083316,
      "learning_rate": 6.665698973710289e-06,
      "loss": 0.3054,
      "step": 1878
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7028372013273649,
      "learning_rate": 6.662362817320364e-06,
      "loss": 0.2728,
      "step": 1879
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7569034634029541,
      "learning_rate": 6.659025828669323e-06,
      "loss": 0.2706,
      "step": 1880
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7888189216182079,
      "learning_rate": 6.655688009427832e-06,
      "loss": 0.2856,
      "step": 1881
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7337321151773692,
      "learning_rate": 6.652349361266967e-06,
      "loss": 0.2928,
      "step": 1882
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.744117810536084,
      "learning_rate": 6.6490098858582176e-06,
      "loss": 0.2808,
      "step": 1883
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7623648005908948,
      "learning_rate": 6.645669584873495e-06,
      "loss": 0.2906,
      "step": 1884
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6972254502550064,
      "learning_rate": 6.6423284599851145e-06,
      "loss": 0.2731,
      "step": 1885
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6447361229806764,
      "learning_rate": 6.638986512865814e-06,
      "loss": 0.257,
      "step": 1886
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7544951125015439,
      "learning_rate": 6.6356437451887336e-06,
      "loss": 0.3068,
      "step": 1887
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7488555658006846,
      "learning_rate": 6.632300158627427e-06,
      "loss": 0.2948,
      "step": 1888
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6916503567439573,
      "learning_rate": 6.628955754855865e-06,
      "loss": 0.3186,
      "step": 1889
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6925711559700464,
      "learning_rate": 6.625610535548418e-06,
      "loss": 0.2616,
      "step": 1890
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7178290946178079,
      "learning_rate": 6.6222645023798675e-06,
      "loss": 0.2767,
      "step": 1891
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.65149917122455,
      "learning_rate": 6.6189176570254045e-06,
      "loss": 0.2575,
      "step": 1892
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6929584703150365,
      "learning_rate": 6.615570001160626e-06,
      "loss": 0.2828,
      "step": 1893
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7757814971468718,
      "learning_rate": 6.612221536461534e-06,
      "loss": 0.3145,
      "step": 1894
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7234787441004236,
      "learning_rate": 6.608872264604536e-06,
      "loss": 0.2838,
      "step": 1895
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7467444434018466,
      "learning_rate": 6.605522187266441e-06,
      "loss": 0.2926,
      "step": 1896
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7407189080198748,
      "learning_rate": 6.6021713061244655e-06,
      "loss": 0.2966,
      "step": 1897
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6999494411887094,
      "learning_rate": 6.598819622856227e-06,
      "loss": 0.2697,
      "step": 1898
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7010614195592602,
      "learning_rate": 6.595467139139743e-06,
      "loss": 0.2607,
      "step": 1899
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7099805642476235,
      "learning_rate": 6.592113856653433e-06,
      "loss": 0.2687,
      "step": 1900
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7537616015392117,
      "learning_rate": 6.588759777076116e-06,
      "loss": 0.2578,
      "step": 1901
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6933827750929232,
      "learning_rate": 6.585404902087011e-06,
      "loss": 0.2559,
      "step": 1902
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7156369362165526,
      "learning_rate": 6.582049233365734e-06,
      "loss": 0.2773,
      "step": 1903
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.693364929531636,
      "learning_rate": 6.578692772592298e-06,
      "loss": 0.2792,
      "step": 1904
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.722080397925299,
      "learning_rate": 6.575335521447114e-06,
      "loss": 0.2901,
      "step": 1905
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7208474726801739,
      "learning_rate": 6.571977481610989e-06,
      "loss": 0.2787,
      "step": 1906
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7754158382618878,
      "learning_rate": 6.568618654765124e-06,
      "loss": 0.2785,
      "step": 1907
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7309309647460863,
      "learning_rate": 6.565259042591112e-06,
      "loss": 0.2779,
      "step": 1908
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7370019326823413,
      "learning_rate": 6.561898646770943e-06,
      "loss": 0.2753,
      "step": 1909
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6888455501241898,
      "learning_rate": 6.558537468986998e-06,
      "loss": 0.2537,
      "step": 1910
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6855716702129053,
      "learning_rate": 6.555175510922047e-06,
      "loss": 0.2776,
      "step": 1911
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7818237132147583,
      "learning_rate": 6.551812774259253e-06,
      "loss": 0.3056,
      "step": 1912
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.71312674005686,
      "learning_rate": 6.548449260682169e-06,
      "loss": 0.2707,
      "step": 1913
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7202360147124,
      "learning_rate": 6.545084971874738e-06,
      "loss": 0.2793,
      "step": 1914
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7122592475571917,
      "learning_rate": 6.541719909521287e-06,
      "loss": 0.2937,
      "step": 1915
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6680412993332466,
      "learning_rate": 6.538354075306532e-06,
      "loss": 0.2623,
      "step": 1916
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7342090841291561,
      "learning_rate": 6.5349874709155805e-06,
      "loss": 0.274,
      "step": 1917
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6914009737846821,
      "learning_rate": 6.531620098033919e-06,
      "loss": 0.2623,
      "step": 1918
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6870490083654114,
      "learning_rate": 6.528251958347418e-06,
      "loss": 0.2757,
      "step": 1919
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7293446309671537,
      "learning_rate": 6.524883053542339e-06,
      "loss": 0.2724,
      "step": 1920
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7373582813199379,
      "learning_rate": 6.521513385305321e-06,
      "loss": 0.2583,
      "step": 1921
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6600784170730802,
      "learning_rate": 6.518142955323389e-06,
      "loss": 0.2608,
      "step": 1922
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7398022346614388,
      "learning_rate": 6.514771765283942e-06,
      "loss": 0.2625,
      "step": 1923
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7380257970586428,
      "learning_rate": 6.511399816874769e-06,
      "loss": 0.2839,
      "step": 1924
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7556019914360496,
      "learning_rate": 6.508027111784033e-06,
      "loss": 0.2873,
      "step": 1925
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.707965036542678,
      "learning_rate": 6.504653651700278e-06,
      "loss": 0.2794,
      "step": 1926
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7482309434788189,
      "learning_rate": 6.501279438312422e-06,
      "loss": 0.2725,
      "step": 1927
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7642320599651012,
      "learning_rate": 6.497904473309766e-06,
      "loss": 0.281,
      "step": 1928
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.812749486462967,
      "learning_rate": 6.494528758381985e-06,
      "loss": 0.2715,
      "step": 1929
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6886614806254254,
      "learning_rate": 6.4911522952191275e-06,
      "loss": 0.2676,
      "step": 1930
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7339177801512771,
      "learning_rate": 6.487775085511618e-06,
      "loss": 0.2824,
      "step": 1931
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6995363228913959,
      "learning_rate": 6.4843971309502554e-06,
      "loss": 0.2852,
      "step": 1932
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7562714692009729,
      "learning_rate": 6.481018433226212e-06,
      "loss": 0.2736,
      "step": 1933
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8027112982630921,
      "learning_rate": 6.477638994031028e-06,
      "loss": 0.2974,
      "step": 1934
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7298423540627572,
      "learning_rate": 6.474258815056622e-06,
      "loss": 0.2989,
      "step": 1935
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7922466661464435,
      "learning_rate": 6.470877897995275e-06,
      "loss": 0.2986,
      "step": 1936
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7212364188505765,
      "learning_rate": 6.467496244539644e-06,
      "loss": 0.2683,
      "step": 1937
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6836809527923363,
      "learning_rate": 6.464113856382752e-06,
      "loss": 0.2997,
      "step": 1938
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7201674086906172,
      "learning_rate": 6.460730735217991e-06,
      "loss": 0.2693,
      "step": 1939
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.707918746945487,
      "learning_rate": 6.457346882739114e-06,
      "loss": 0.2687,
      "step": 1940
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6639348496158802,
      "learning_rate": 6.45396230064025e-06,
      "loss": 0.2548,
      "step": 1941
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7939904172522406,
      "learning_rate": 6.450576990615888e-06,
      "loss": 0.2984,
      "step": 1942
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.713503200012862,
      "learning_rate": 6.447190954360878e-06,
      "loss": 0.2826,
      "step": 1943
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7103303333039123,
      "learning_rate": 6.443804193570441e-06,
      "loss": 0.2795,
      "step": 1944
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7181394228496566,
      "learning_rate": 6.440416709940156e-06,
      "loss": 0.2898,
      "step": 1945
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7365614025816377,
      "learning_rate": 6.437028505165965e-06,
      "loss": 0.2852,
      "step": 1946
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6780757281629586,
      "learning_rate": 6.433639580944171e-06,
      "loss": 0.2734,
      "step": 1947
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7169533037944557,
      "learning_rate": 6.430249938971438e-06,
      "loss": 0.2668,
      "step": 1948
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7414353839004023,
      "learning_rate": 6.426859580944786e-06,
      "loss": 0.2752,
      "step": 1949
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7586495364658113,
      "learning_rate": 6.423468508561599e-06,
      "loss": 0.276,
      "step": 1950
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7389511947591709,
      "learning_rate": 6.420076723519615e-06,
      "loss": 0.2827,
      "step": 1951
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.688161221196996,
      "learning_rate": 6.416684227516927e-06,
      "loss": 0.2453,
      "step": 1952
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.706740270794062,
      "learning_rate": 6.41329102225199e-06,
      "loss": 0.2558,
      "step": 1953
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7150555217524055,
      "learning_rate": 6.409897109423607e-06,
      "loss": 0.2719,
      "step": 1954
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.79727600007681,
      "learning_rate": 6.406502490730944e-06,
      "loss": 0.3223,
      "step": 1955
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.808583091548564,
      "learning_rate": 6.403107167873509e-06,
      "loss": 0.2747,
      "step": 1956
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7051180488883962,
      "learning_rate": 6.399711142551172e-06,
      "loss": 0.2645,
      "step": 1957
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7166299386729075,
      "learning_rate": 6.396314416464151e-06,
      "loss": 0.2721,
      "step": 1958
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7372491054822993,
      "learning_rate": 6.3929169913130165e-06,
      "loss": 0.2856,
      "step": 1959
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.734089496223724,
      "learning_rate": 6.389518868798683e-06,
      "loss": 0.2908,
      "step": 1960
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7051276915042275,
      "learning_rate": 6.3861200506224255e-06,
      "loss": 0.291,
      "step": 1961
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7409988946113253,
      "learning_rate": 6.382720538485856e-06,
      "loss": 0.2964,
      "step": 1962
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7562087415988665,
      "learning_rate": 6.37932033409094e-06,
      "loss": 0.2718,
      "step": 1963
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7411996973251302,
      "learning_rate": 6.375919439139988e-06,
      "loss": 0.2853,
      "step": 1964
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7424783915594334,
      "learning_rate": 6.372517855335656e-06,
      "loss": 0.2843,
      "step": 1965
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6910629502559478,
      "learning_rate": 6.3691155843809475e-06,
      "loss": 0.288,
      "step": 1966
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7937060627190675,
      "learning_rate": 6.365712627979204e-06,
      "loss": 0.2952,
      "step": 1967
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.742232203254155,
      "learning_rate": 6.3623089878341146e-06,
      "loss": 0.2797,
      "step": 1968
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7193517768302171,
      "learning_rate": 6.358904665649712e-06,
      "loss": 0.2754,
      "step": 1969
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7499271102765878,
      "learning_rate": 6.355499663130369e-06,
      "loss": 0.2908,
      "step": 1970
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6751733273464428,
      "learning_rate": 6.352093981980796e-06,
      "loss": 0.2683,
      "step": 1971
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6923650809599412,
      "learning_rate": 6.348687623906047e-06,
      "loss": 0.2885,
      "step": 1972
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7352838691053389,
      "learning_rate": 6.345280590611512e-06,
      "loss": 0.2491,
      "step": 1973
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.70482369015348,
      "learning_rate": 6.341872883802923e-06,
      "loss": 0.2672,
      "step": 1974
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7363069686076072,
      "learning_rate": 6.338464505186344e-06,
      "loss": 0.2924,
      "step": 1975
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7355479725269142,
      "learning_rate": 6.33505545646818e-06,
      "loss": 0.3061,
      "step": 1976
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7060344565051093,
      "learning_rate": 6.331645739355168e-06,
      "loss": 0.2593,
      "step": 1977
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7354135351034425,
      "learning_rate": 6.328235355554382e-06,
      "loss": 0.3096,
      "step": 1978
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7023089145506008,
      "learning_rate": 6.324824306773231e-06,
      "loss": 0.2628,
      "step": 1979
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7166905730351398,
      "learning_rate": 6.32141259471945e-06,
      "loss": 0.2701,
      "step": 1980
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7591281801108658,
      "learning_rate": 6.318000221101115e-06,
      "loss": 0.2871,
      "step": 1981
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.702183097183241,
      "learning_rate": 6.314587187626628e-06,
      "loss": 0.2662,
      "step": 1982
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7350290987301642,
      "learning_rate": 6.311173496004723e-06,
      "loss": 0.3037,
      "step": 1983
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7621549859257039,
      "learning_rate": 6.307759147944461e-06,
      "loss": 0.2817,
      "step": 1984
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7168702136508078,
      "learning_rate": 6.304344145155234e-06,
      "loss": 0.2796,
      "step": 1985
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6969607404063658,
      "learning_rate": 6.3009284893467655e-06,
      "loss": 0.2885,
      "step": 1986
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6698061572481825,
      "learning_rate": 6.297512182229098e-06,
      "loss": 0.2495,
      "step": 1987
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7836938398515462,
      "learning_rate": 6.294095225512604e-06,
      "loss": 0.2801,
      "step": 1988
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7271813611572596,
      "learning_rate": 6.290677620907982e-06,
      "loss": 0.2879,
      "step": 1989
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7053032470515543,
      "learning_rate": 6.287259370126255e-06,
      "loss": 0.2942,
      "step": 1990
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6660821425166477,
      "learning_rate": 6.283840474878767e-06,
      "loss": 0.2646,
      "step": 1991
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7102429784608782,
      "learning_rate": 6.280420936877187e-06,
      "loss": 0.3017,
      "step": 1992
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7488093546460344,
      "learning_rate": 6.2770007578335044e-06,
      "loss": 0.2712,
      "step": 1993
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7254586693146127,
      "learning_rate": 6.273579939460032e-06,
      "loss": 0.2954,
      "step": 1994
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6843345804400118,
      "learning_rate": 6.270158483469398e-06,
      "loss": 0.2628,
      "step": 1995
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7302936035090254,
      "learning_rate": 6.266736391574553e-06,
      "loss": 0.282,
      "step": 1996
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.669534953103525,
      "learning_rate": 6.263313665488767e-06,
      "loss": 0.2581,
      "step": 1997
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7258881029137093,
      "learning_rate": 6.259890306925627e-06,
      "loss": 0.2604,
      "step": 1998
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7318863955951938,
      "learning_rate": 6.256466317599033e-06,
      "loss": 0.2663,
      "step": 1999
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6753482494921613,
      "learning_rate": 6.253041699223206e-06,
      "loss": 0.2696,
      "step": 2000
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7340689908891561,
      "learning_rate": 6.2496164535126766e-06,
      "loss": 0.2729,
      "step": 2001
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.717343380795936,
      "learning_rate": 6.246190582182296e-06,
      "loss": 0.2682,
      "step": 2002
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7458005612895554,
      "learning_rate": 6.2427640869472235e-06,
      "loss": 0.2832,
      "step": 2003
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.657578201379116,
      "learning_rate": 6.239336969522931e-06,
      "loss": 0.2555,
      "step": 2004
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.722061861043424,
      "learning_rate": 6.2359092316252035e-06,
      "loss": 0.2971,
      "step": 2005
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.763836300662379,
      "learning_rate": 6.232480874970137e-06,
      "loss": 0.2783,
      "step": 2006
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7258511399983563,
      "learning_rate": 6.229051901274137e-06,
      "loss": 0.2951,
      "step": 2007
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7284206748933852,
      "learning_rate": 6.225622312253916e-06,
      "loss": 0.2819,
      "step": 2008
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7134304499658854,
      "learning_rate": 6.222192109626496e-06,
      "loss": 0.2936,
      "step": 2009
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7588313930309825,
      "learning_rate": 6.218761295109209e-06,
      "loss": 0.2748,
      "step": 2010
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.796488874836055,
      "learning_rate": 6.215329870419688e-06,
      "loss": 0.2922,
      "step": 2011
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7328445826541862,
      "learning_rate": 6.211897837275873e-06,
      "loss": 0.2981,
      "step": 2012
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7125058468944659,
      "learning_rate": 6.208465197396013e-06,
      "loss": 0.2769,
      "step": 2013
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7101354268547426,
      "learning_rate": 6.205031952498656e-06,
      "loss": 0.2837,
      "step": 2014
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7751504856285847,
      "learning_rate": 6.201598104302653e-06,
      "loss": 0.2982,
      "step": 2015
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7558916882216088,
      "learning_rate": 6.198163654527161e-06,
      "loss": 0.2731,
      "step": 2016
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.718691147214598,
      "learning_rate": 6.194728604891632e-06,
      "loss": 0.2489,
      "step": 2017
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7640014184933723,
      "learning_rate": 6.191292957115825e-06,
      "loss": 0.2973,
      "step": 2018
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7222490723007333,
      "learning_rate": 6.187856712919795e-06,
      "loss": 0.2873,
      "step": 2019
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7053605283665517,
      "learning_rate": 6.184419874023895e-06,
      "loss": 0.2578,
      "step": 2020
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7247329154784007,
      "learning_rate": 6.1809824421487765e-06,
      "loss": 0.2816,
      "step": 2021
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7574133363091835,
      "learning_rate": 6.177544419015388e-06,
      "loss": 0.2937,
      "step": 2022
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7765994652379172,
      "learning_rate": 6.174105806344975e-06,
      "loss": 0.3003,
      "step": 2023
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7199418072922038,
      "learning_rate": 6.170666605859077e-06,
      "loss": 0.3002,
      "step": 2024
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7758685548817282,
      "learning_rate": 6.1672268192795285e-06,
      "loss": 0.2974,
      "step": 2025
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7411788462244598,
      "learning_rate": 6.163786448328455e-06,
      "loss": 0.2652,
      "step": 2026
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7216974623764857,
      "learning_rate": 6.160345494728278e-06,
      "loss": 0.2574,
      "step": 2027
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7990767580683223,
      "learning_rate": 6.156903960201709e-06,
      "loss": 0.2794,
      "step": 2028
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6896885825024462,
      "learning_rate": 6.15346184647175e-06,
      "loss": 0.2615,
      "step": 2029
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7603901723826563,
      "learning_rate": 6.150019155261694e-06,
      "loss": 0.2619,
      "step": 2030
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6545309961506429,
      "learning_rate": 6.1465758882951236e-06,
      "loss": 0.2545,
      "step": 2031
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7413965453499308,
      "learning_rate": 6.1431320472959054e-06,
      "loss": 0.2761,
      "step": 2032
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7350137768755389,
      "learning_rate": 6.1396876339882e-06,
      "loss": 0.2714,
      "step": 2033
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7204467514707223,
      "learning_rate": 6.136242650096451e-06,
      "loss": 0.267,
      "step": 2034
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7557250051657394,
      "learning_rate": 6.132797097345387e-06,
      "loss": 0.3004,
      "step": 2035
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7355696628417663,
      "learning_rate": 6.12935097746002e-06,
      "loss": 0.3074,
      "step": 2036
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7423203626009198,
      "learning_rate": 6.125904292165652e-06,
      "loss": 0.276,
      "step": 2037
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6529137696645775,
      "learning_rate": 6.122457043187863e-06,
      "loss": 0.2638,
      "step": 2038
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7850559649044939,
      "learning_rate": 6.119009232252517e-06,
      "loss": 0.269,
      "step": 2039
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7234511429227397,
      "learning_rate": 6.115560861085757e-06,
      "loss": 0.2818,
      "step": 2040
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7406225347562903,
      "learning_rate": 6.112111931414009e-06,
      "loss": 0.2751,
      "step": 2041
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6746101812975096,
      "learning_rate": 6.108662444963979e-06,
      "loss": 0.2611,
      "step": 2042
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7439493638343363,
      "learning_rate": 6.10521240346265e-06,
      "loss": 0.2799,
      "step": 2043
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6520324949739921,
      "learning_rate": 6.101761808637283e-06,
      "loss": 0.2526,
      "step": 2044
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7181872184101334,
      "learning_rate": 6.098310662215416e-06,
      "loss": 0.2802,
      "step": 2045
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7059602084487586,
      "learning_rate": 6.094858965924866e-06,
      "loss": 0.2919,
      "step": 2046
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7079887535402768,
      "learning_rate": 6.091406721493724e-06,
      "loss": 0.2524,
      "step": 2047
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6983267031910523,
      "learning_rate": 6.087953930650349e-06,
      "loss": 0.2697,
      "step": 2048
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7886005215234524,
      "learning_rate": 6.084500595123383e-06,
      "loss": 0.3055,
      "step": 2049
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6816080971482323,
      "learning_rate": 6.081046716641735e-06,
      "loss": 0.267,
      "step": 2050
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6893339397880582,
      "learning_rate": 6.077592296934592e-06,
      "loss": 0.2732,
      "step": 2051
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6761102522071097,
      "learning_rate": 6.0741373377314005e-06,
      "loss": 0.2705,
      "step": 2052
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7562336143277476,
      "learning_rate": 6.070681840761889e-06,
      "loss": 0.277,
      "step": 2053
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7348674892747358,
      "learning_rate": 6.0672258077560475e-06,
      "loss": 0.2825,
      "step": 2054
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7018865020705293,
      "learning_rate": 6.063769240444141e-06,
      "loss": 0.2617,
      "step": 2055
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7674978887822358,
      "learning_rate": 6.060312140556694e-06,
      "loss": 0.2633,
      "step": 2056
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7476872183337455,
      "learning_rate": 6.056854509824503e-06,
      "loss": 0.2567,
      "step": 2057
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6976307874586976,
      "learning_rate": 6.053396349978632e-06,
      "loss": 0.2761,
      "step": 2058
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7318873945234923,
      "learning_rate": 6.049937662750405e-06,
      "loss": 0.2719,
      "step": 2059
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7516267571647144,
      "learning_rate": 6.046478449871409e-06,
      "loss": 0.2707,
      "step": 2060
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7114507819742608,
      "learning_rate": 6.043018713073502e-06,
      "loss": 0.2701,
      "step": 2061
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7459178227181671,
      "learning_rate": 6.039558454088796e-06,
      "loss": 0.2732,
      "step": 2062
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7710137798955121,
      "learning_rate": 6.036097674649672e-06,
      "loss": 0.3052,
      "step": 2063
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7647519576554361,
      "learning_rate": 6.032636376488763e-06,
      "loss": 0.2887,
      "step": 2064
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.67637580060714,
      "learning_rate": 6.029174561338968e-06,
      "loss": 0.2413,
      "step": 2065
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7086241729947577,
      "learning_rate": 6.025712230933443e-06,
      "loss": 0.2697,
      "step": 2066
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.801705288427062,
      "learning_rate": 6.022249387005604e-06,
      "loss": 0.2869,
      "step": 2067
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7404382007236996,
      "learning_rate": 6.018786031289119e-06,
      "loss": 0.2779,
      "step": 2068
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.729155647064458,
      "learning_rate": 6.015322165517916e-06,
      "loss": 0.283,
      "step": 2069
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6661503345772429,
      "learning_rate": 6.0118577914261786e-06,
      "loss": 0.2617,
      "step": 2070
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6826057732755006,
      "learning_rate": 6.008392910748345e-06,
      "loss": 0.2667,
      "step": 2071
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7068344837012447,
      "learning_rate": 6.004927525219103e-06,
      "loss": 0.2671,
      "step": 2072
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.701373581284137,
      "learning_rate": 6.001461636573397e-06,
      "loss": 0.2577,
      "step": 2073
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7521068710700027,
      "learning_rate": 5.997995246546424e-06,
      "loss": 0.2891,
      "step": 2074
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6760844307688502,
      "learning_rate": 5.9945283568736305e-06,
      "loss": 0.2638,
      "step": 2075
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7041367475238592,
      "learning_rate": 5.9910609692907105e-06,
      "loss": 0.2577,
      "step": 2076
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7444856135565255,
      "learning_rate": 5.9875930855336116e-06,
      "loss": 0.3243,
      "step": 2077
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7830605446409774,
      "learning_rate": 5.984124707338528e-06,
      "loss": 0.277,
      "step": 2078
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7747236163803682,
      "learning_rate": 5.980655836441902e-06,
      "loss": 0.269,
      "step": 2079
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8044743964637943,
      "learning_rate": 5.977186474580421e-06,
      "loss": 0.3195,
      "step": 2080
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7062369943836458,
      "learning_rate": 5.973716623491019e-06,
      "loss": 0.2807,
      "step": 2081
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.712950705125213,
      "learning_rate": 5.970246284910877e-06,
      "loss": 0.2742,
      "step": 2082
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6720885660986333,
      "learning_rate": 5.966775460577418e-06,
      "loss": 0.2489,
      "step": 2083
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7108637880860625,
      "learning_rate": 5.963304152228307e-06,
      "loss": 0.2611,
      "step": 2084
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7087732773458004,
      "learning_rate": 5.959832361601453e-06,
      "loss": 0.2741,
      "step": 2085
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7046813062746912,
      "learning_rate": 5.9563600904350074e-06,
      "loss": 0.2678,
      "step": 2086
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7724387778402334,
      "learning_rate": 5.95288734046736e-06,
      "loss": 0.2921,
      "step": 2087
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7486912597711516,
      "learning_rate": 5.949414113437142e-06,
      "loss": 0.2635,
      "step": 2088
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7156118270115613,
      "learning_rate": 5.945940411083222e-06,
      "loss": 0.2681,
      "step": 2089
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7726221170771321,
      "learning_rate": 5.942466235144708e-06,
      "loss": 0.3053,
      "step": 2090
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7028071551253304,
      "learning_rate": 5.9389915873609465e-06,
      "loss": 0.2755,
      "step": 2091
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7864939478706411,
      "learning_rate": 5.935516469471514e-06,
      "loss": 0.2898,
      "step": 2092
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7178051392674606,
      "learning_rate": 5.932040883216228e-06,
      "loss": 0.2977,
      "step": 2093
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7207122592698979,
      "learning_rate": 5.9285648303351404e-06,
      "loss": 0.2591,
      "step": 2094
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7505436134651144,
      "learning_rate": 5.9250883125685355e-06,
      "loss": 0.2781,
      "step": 2095
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7074113748617105,
      "learning_rate": 5.921611331656928e-06,
      "loss": 0.2715,
      "step": 2096
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7283975579977451,
      "learning_rate": 5.918133889341067e-06,
      "loss": 0.2931,
      "step": 2097
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7632527164424444,
      "learning_rate": 5.914655987361934e-06,
      "loss": 0.2946,
      "step": 2098
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6903453756967245,
      "learning_rate": 5.911177627460739e-06,
      "loss": 0.2642,
      "step": 2099
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7318153619786458,
      "learning_rate": 5.907698811378919e-06,
      "loss": 0.2817,
      "step": 2100
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6485127916039243,
      "learning_rate": 5.904219540858143e-06,
      "loss": 0.2411,
      "step": 2101
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6439355748828809,
      "learning_rate": 5.900739817640304e-06,
      "loss": 0.2352,
      "step": 2102
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6523501582323539,
      "learning_rate": 5.897259643467528e-06,
      "loss": 0.2517,
      "step": 2103
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7998031399153838,
      "learning_rate": 5.893779020082159e-06,
      "loss": 0.3095,
      "step": 2104
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7485994385053455,
      "learning_rate": 5.890297949226768e-06,
      "loss": 0.2894,
      "step": 2105
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.703400482478022,
      "learning_rate": 5.886816432644155e-06,
      "loss": 0.271,
      "step": 2106
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6953968986583025,
      "learning_rate": 5.88333447207734e-06,
      "loss": 0.2658,
      "step": 2107
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7440071666259573,
      "learning_rate": 5.8798520692695605e-06,
      "loss": 0.2723,
      "step": 2108
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6910473724809723,
      "learning_rate": 5.876369225964283e-06,
      "loss": 0.2765,
      "step": 2109
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7243337493469587,
      "learning_rate": 5.872885943905191e-06,
      "loss": 0.2859,
      "step": 2110
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7458922464829899,
      "learning_rate": 5.869402224836187e-06,
      "loss": 0.2654,
      "step": 2111
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7563880740884433,
      "learning_rate": 5.865918070501394e-06,
      "loss": 0.2599,
      "step": 2112
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7426904943323095,
      "learning_rate": 5.862433482645151e-06,
      "loss": 0.2847,
      "step": 2113
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7268951267371104,
      "learning_rate": 5.858948463012015e-06,
      "loss": 0.2814,
      "step": 2114
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7098207882067277,
      "learning_rate": 5.855463013346763e-06,
      "loss": 0.2728,
      "step": 2115
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7165003745338722,
      "learning_rate": 5.851977135394379e-06,
      "loss": 0.2841,
      "step": 2116
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7316694395161838,
      "learning_rate": 5.848490830900067e-06,
      "loss": 0.2833,
      "step": 2117
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6953798819604824,
      "learning_rate": 5.8450041016092465e-06,
      "loss": 0.275,
      "step": 2118
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6996840088419533,
      "learning_rate": 5.841516949267544e-06,
      "loss": 0.2784,
      "step": 2119
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6939838704241327,
      "learning_rate": 5.838029375620801e-06,
      "loss": 0.2781,
      "step": 2120
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6919173623707665,
      "learning_rate": 5.83454138241507e-06,
      "loss": 0.2681,
      "step": 2121
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6948193734450919,
      "learning_rate": 5.831052971396613e-06,
      "loss": 0.2672,
      "step": 2122
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7000538887557383,
      "learning_rate": 5.8275641443119015e-06,
      "loss": 0.2784,
      "step": 2123
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6742660105540789,
      "learning_rate": 5.824074902907614e-06,
      "loss": 0.2757,
      "step": 2124
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.710999039344304,
      "learning_rate": 5.820585248930638e-06,
      "loss": 0.2675,
      "step": 2125
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7027483095579201,
      "learning_rate": 5.817095184128067e-06,
      "loss": 0.2781,
      "step": 2126
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6898644704980387,
      "learning_rate": 5.813604710247201e-06,
      "loss": 0.2538,
      "step": 2127
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7201387595503987,
      "learning_rate": 5.810113829035544e-06,
      "loss": 0.2835,
      "step": 2128
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6945639220698407,
      "learning_rate": 5.806622542240801e-06,
      "loss": 0.2692,
      "step": 2129
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6859630709683681,
      "learning_rate": 5.803130851610887e-06,
      "loss": 0.2693,
      "step": 2130
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7125857854119416,
      "learning_rate": 5.799638758893914e-06,
      "loss": 0.2728,
      "step": 2131
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7308845668456662,
      "learning_rate": 5.796146265838195e-06,
      "loss": 0.2664,
      "step": 2132
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.763422595594739,
      "learning_rate": 5.792653374192245e-06,
      "loss": 0.2989,
      "step": 2133
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7134770105161126,
      "learning_rate": 5.78916008570478e-06,
      "loss": 0.2816,
      "step": 2134
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7760260907471717,
      "learning_rate": 5.785666402124715e-06,
      "loss": 0.2759,
      "step": 2135
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7218308074271776,
      "learning_rate": 5.782172325201155e-06,
      "loss": 0.2672,
      "step": 2136
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6780017060821143,
      "learning_rate": 5.778677856683412e-06,
      "loss": 0.2834,
      "step": 2137
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6313945550876645,
      "learning_rate": 5.77518299832099e-06,
      "loss": 0.2636,
      "step": 2138
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7142917084812658,
      "learning_rate": 5.771687751863587e-06,
      "loss": 0.2759,
      "step": 2139
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7282708071345062,
      "learning_rate": 5.768192119061096e-06,
      "loss": 0.2756,
      "step": 2140
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.695437784663553,
      "learning_rate": 5.764696101663605e-06,
      "loss": 0.2853,
      "step": 2141
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.696241009607509,
      "learning_rate": 5.761199701421392e-06,
      "loss": 0.2623,
      "step": 2142
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7520159052500115,
      "learning_rate": 5.757702920084931e-06,
      "loss": 0.2685,
      "step": 2143
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7127130564132192,
      "learning_rate": 5.754205759404879e-06,
      "loss": 0.2712,
      "step": 2144
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7447871470707401,
      "learning_rate": 5.750708221132092e-06,
      "loss": 0.2856,
      "step": 2145
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6719906644036155,
      "learning_rate": 5.747210307017609e-06,
      "loss": 0.2466,
      "step": 2146
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.652867025705546,
      "learning_rate": 5.743712018812659e-06,
      "loss": 0.2737,
      "step": 2147
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.764911260072609,
      "learning_rate": 5.740213358268658e-06,
      "loss": 0.2657,
      "step": 2148
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7294374961334102,
      "learning_rate": 5.73671432713721e-06,
      "loss": 0.2664,
      "step": 2149
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7123799243972174,
      "learning_rate": 5.733214927170101e-06,
      "loss": 0.2854,
      "step": 2150
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7075869684777754,
      "learning_rate": 5.729715160119306e-06,
      "loss": 0.2574,
      "step": 2151
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7173946662727629,
      "learning_rate": 5.726215027736981e-06,
      "loss": 0.2684,
      "step": 2152
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7551276229417613,
      "learning_rate": 5.722714531775463e-06,
      "loss": 0.2824,
      "step": 2153
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7390424577832171,
      "learning_rate": 5.719213673987277e-06,
      "loss": 0.266,
      "step": 2154
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7119549337491924,
      "learning_rate": 5.715712456125127e-06,
      "loss": 0.2724,
      "step": 2155
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7447067063062862,
      "learning_rate": 5.712210879941891e-06,
      "loss": 0.2835,
      "step": 2156
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7523935256258153,
      "learning_rate": 5.7087089471906345e-06,
      "loss": 0.2924,
      "step": 2157
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6955199955988277,
      "learning_rate": 5.705206659624597e-06,
      "loss": 0.276,
      "step": 2158
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.762449459615242,
      "learning_rate": 5.701704018997198e-06,
      "loss": 0.2763,
      "step": 2159
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.753977508905572,
      "learning_rate": 5.698201027062034e-06,
      "loss": 0.2632,
      "step": 2160
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7789370107274239,
      "learning_rate": 5.694697685572873e-06,
      "loss": 0.2794,
      "step": 2161
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7358291852965582,
      "learning_rate": 5.6911939962836625e-06,
      "loss": 0.3007,
      "step": 2162
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.76020560049349,
      "learning_rate": 5.687689960948526e-06,
      "loss": 0.2921,
      "step": 2163
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7532342848726105,
      "learning_rate": 5.684185581321752e-06,
      "loss": 0.2929,
      "step": 2164
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.722066047060307,
      "learning_rate": 5.680680859157808e-06,
      "loss": 0.2754,
      "step": 2165
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.733543012128782,
      "learning_rate": 5.6771757962113325e-06,
      "loss": 0.2678,
      "step": 2166
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7171899742671882,
      "learning_rate": 5.673670394237133e-06,
      "loss": 0.2621,
      "step": 2167
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7536725311938572,
      "learning_rate": 5.670164654990189e-06,
      "loss": 0.2729,
      "step": 2168
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6965621927740518,
      "learning_rate": 5.666658580225643e-06,
      "loss": 0.2452,
      "step": 2169
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7436125545227164,
      "learning_rate": 5.663152171698813e-06,
      "loss": 0.2992,
      "step": 2170
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7326678698055892,
      "learning_rate": 5.659645431165179e-06,
      "loss": 0.2732,
      "step": 2171
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7711798626204903,
      "learning_rate": 5.656138360380391e-06,
      "loss": 0.3105,
      "step": 2172
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.734844733912963,
      "learning_rate": 5.65263096110026e-06,
      "loss": 0.2937,
      "step": 2173
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7462370081333015,
      "learning_rate": 5.649123235080762e-06,
      "loss": 0.2947,
      "step": 2174
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7743001242129894,
      "learning_rate": 5.645615184078044e-06,
      "loss": 0.2816,
      "step": 2175
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7511857455415951,
      "learning_rate": 5.642106809848405e-06,
      "loss": 0.2758,
      "step": 2176
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7138680376816384,
      "learning_rate": 5.638598114148314e-06,
      "loss": 0.2782,
      "step": 2177
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7363899578006559,
      "learning_rate": 5.635089098734394e-06,
      "loss": 0.2806,
      "step": 2178
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6938282007614922,
      "learning_rate": 5.631579765363436e-06,
      "loss": 0.2742,
      "step": 2179
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7218993877351596,
      "learning_rate": 5.628070115792384e-06,
      "loss": 0.2778,
      "step": 2180
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7416963310439845,
      "learning_rate": 5.624560151778341e-06,
      "loss": 0.2933,
      "step": 2181
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7366300663886365,
      "learning_rate": 5.621049875078571e-06,
      "loss": 0.2963,
      "step": 2182
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7026747294018718,
      "learning_rate": 5.617539287450492e-06,
      "loss": 0.2641,
      "step": 2183
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6809560906801787,
      "learning_rate": 5.614028390651676e-06,
      "loss": 0.277,
      "step": 2184
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7637688654084793,
      "learning_rate": 5.610517186439853e-06,
      "loss": 0.2723,
      "step": 2185
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7115656138086432,
      "learning_rate": 5.6070056765729065e-06,
      "loss": 0.2792,
      "step": 2186
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6801099118243467,
      "learning_rate": 5.6034938628088705e-06,
      "loss": 0.2754,
      "step": 2187
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7022701104102793,
      "learning_rate": 5.599981746905935e-06,
      "loss": 0.2705,
      "step": 2188
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6960839531798909,
      "learning_rate": 5.596469330622437e-06,
      "loss": 0.2672,
      "step": 2189
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7155444335340762,
      "learning_rate": 5.592956615716867e-06,
      "loss": 0.2866,
      "step": 2190
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6837156665133521,
      "learning_rate": 5.589443603947863e-06,
      "loss": 0.2599,
      "step": 2191
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7178367529786974,
      "learning_rate": 5.585930297074214e-06,
      "loss": 0.2669,
      "step": 2192
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7170062491106083,
      "learning_rate": 5.582416696854853e-06,
      "loss": 0.2812,
      "step": 2193
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6754863015609961,
      "learning_rate": 5.578902805048866e-06,
      "loss": 0.2648,
      "step": 2194
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.630248207133472,
      "learning_rate": 5.575388623415476e-06,
      "loss": 0.2364,
      "step": 2195
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6457557114019324,
      "learning_rate": 5.571874153714063e-06,
      "loss": 0.2555,
      "step": 2196
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.711224541204305,
      "learning_rate": 5.568359397704138e-06,
      "loss": 0.2747,
      "step": 2197
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7659741434196804,
      "learning_rate": 5.564844357145365e-06,
      "loss": 0.3048,
      "step": 2198
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7323449615024579,
      "learning_rate": 5.561329033797547e-06,
      "loss": 0.2694,
      "step": 2199
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6919922738801989,
      "learning_rate": 5.55781342942063e-06,
      "loss": 0.2631,
      "step": 2200
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.67975159011405,
      "learning_rate": 5.5542975457746966e-06,
      "loss": 0.254,
      "step": 2201
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7348642016936934,
      "learning_rate": 5.550781384619974e-06,
      "loss": 0.289,
      "step": 2202
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7348204514631128,
      "learning_rate": 5.5472649477168264e-06,
      "loss": 0.2746,
      "step": 2203
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7022250827873909,
      "learning_rate": 5.543748236825756e-06,
      "loss": 0.2845,
      "step": 2204
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7402152717556992,
      "learning_rate": 5.5402312537074035e-06,
      "loss": 0.3041,
      "step": 2205
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7366577509207204,
      "learning_rate": 5.536714000122543e-06,
      "loss": 0.2729,
      "step": 2206
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7582441821695654,
      "learning_rate": 5.5331964778320875e-06,
      "loss": 0.2971,
      "step": 2207
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7336313159638086,
      "learning_rate": 5.529678688597081e-06,
      "loss": 0.2853,
      "step": 2208
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7301499943685386,
      "learning_rate": 5.5261606341787035e-06,
      "loss": 0.2639,
      "step": 2209
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7461322317033245,
      "learning_rate": 5.522642316338268e-06,
      "loss": 0.2793,
      "step": 2210
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6644062447664929,
      "learning_rate": 5.5191237368372175e-06,
      "loss": 0.2673,
      "step": 2211
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7609907729273769,
      "learning_rate": 5.515604897437129e-06,
      "loss": 0.3009,
      "step": 2212
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7903085988017817,
      "learning_rate": 5.512085799899705e-06,
      "loss": 0.2852,
      "step": 2213
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7061781175145283,
      "learning_rate": 5.50856644598678e-06,
      "loss": 0.2711,
      "step": 2214
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7079615966170021,
      "learning_rate": 5.505046837460319e-06,
      "loss": 0.2857,
      "step": 2215
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7913423383830396,
      "learning_rate": 5.501526976082413e-06,
      "loss": 0.2749,
      "step": 2216
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7442062211278604,
      "learning_rate": 5.498006863615275e-06,
      "loss": 0.2576,
      "step": 2217
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6442651304951162,
      "learning_rate": 5.49448650182125e-06,
      "loss": 0.2554,
      "step": 2218
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7200461265765025,
      "learning_rate": 5.490965892462804e-06,
      "loss": 0.2793,
      "step": 2219
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7095581442037399,
      "learning_rate": 5.487445037302531e-06,
      "loss": 0.2824,
      "step": 2220
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6821697784626459,
      "learning_rate": 5.483923938103143e-06,
      "loss": 0.2556,
      "step": 2221
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6995473327497114,
      "learning_rate": 5.480402596627477e-06,
      "loss": 0.2523,
      "step": 2222
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6764078708196171,
      "learning_rate": 5.476881014638491e-06,
      "loss": 0.2327,
      "step": 2223
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7440140923707415,
      "learning_rate": 5.473359193899265e-06,
      "loss": 0.276,
      "step": 2224
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7101649694613822,
      "learning_rate": 5.469837136172993e-06,
      "loss": 0.2759,
      "step": 2225
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7422340778570685,
      "learning_rate": 5.466314843222993e-06,
      "loss": 0.2811,
      "step": 2226
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6733218109996224,
      "learning_rate": 5.4627923168127e-06,
      "loss": 0.2596,
      "step": 2227
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7071527931691813,
      "learning_rate": 5.459269558705667e-06,
      "loss": 0.2678,
      "step": 2228
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7175768971478804,
      "learning_rate": 5.4557465706655564e-06,
      "loss": 0.2911,
      "step": 2229
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7257927075298262,
      "learning_rate": 5.452223354456154e-06,
      "loss": 0.2619,
      "step": 2230
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6894627799704726,
      "learning_rate": 5.4486999118413555e-06,
      "loss": 0.2615,
      "step": 2231
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7416981786185953,
      "learning_rate": 5.445176244585171e-06,
      "loss": 0.2673,
      "step": 2232
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7016327945373169,
      "learning_rate": 5.441652354451721e-06,
      "loss": 0.2837,
      "step": 2233
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.714765717166,
      "learning_rate": 5.438128243205241e-06,
      "loss": 0.284,
      "step": 2234
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7052747078097773,
      "learning_rate": 5.434603912610073e-06,
      "loss": 0.2831,
      "step": 2235
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7150430483520847,
      "learning_rate": 5.431079364430677e-06,
      "loss": 0.2883,
      "step": 2236
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.751566511996753,
      "learning_rate": 5.427554600431611e-06,
      "loss": 0.2822,
      "step": 2237
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7759382389232331,
      "learning_rate": 5.4240296223775465e-06,
      "loss": 0.2975,
      "step": 2238
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7434156774226824,
      "learning_rate": 5.420504432033265e-06,
      "loss": 0.2864,
      "step": 2239
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7452663238570465,
      "learning_rate": 5.416979031163649e-06,
      "loss": 0.3054,
      "step": 2240
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6581135791164967,
      "learning_rate": 5.4134534215336895e-06,
      "loss": 0.2509,
      "step": 2241
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6734341790868424,
      "learning_rate": 5.409927604908481e-06,
      "loss": 0.2522,
      "step": 2242
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7073903575766025,
      "learning_rate": 5.406401583053222e-06,
      "loss": 0.2586,
      "step": 2243
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7016234195655463,
      "learning_rate": 5.402875357733215e-06,
      "loss": 0.2622,
      "step": 2244
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6983262708831973,
      "learning_rate": 5.399348930713859e-06,
      "loss": 0.2836,
      "step": 2245
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.758596534970875,
      "learning_rate": 5.395822303760662e-06,
      "loss": 0.2881,
      "step": 2246
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7027163774158931,
      "learning_rate": 5.392295478639226e-06,
      "loss": 0.2806,
      "step": 2247
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7213811260487687,
      "learning_rate": 5.388768457115254e-06,
      "loss": 0.2767,
      "step": 2248
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6803727030310043,
      "learning_rate": 5.38524124095455e-06,
      "loss": 0.2615,
      "step": 2249
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7792626749204622,
      "learning_rate": 5.381713831923008e-06,
      "loss": 0.2778,
      "step": 2250
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7038027533773945,
      "learning_rate": 5.378186231786629e-06,
      "loss": 0.2632,
      "step": 2251
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7472585589933701,
      "learning_rate": 5.3746584423115025e-06,
      "loss": 0.2605,
      "step": 2252
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7315423800585631,
      "learning_rate": 5.371130465263813e-06,
      "loss": 0.2841,
      "step": 2253
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6714361843221643,
      "learning_rate": 5.367602302409841e-06,
      "loss": 0.2433,
      "step": 2254
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.724955493755836,
      "learning_rate": 5.364073955515961e-06,
      "loss": 0.2638,
      "step": 2255
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6753741076803202,
      "learning_rate": 5.360545426348638e-06,
      "loss": 0.2786,
      "step": 2256
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7041669110671515,
      "learning_rate": 5.357016716674426e-06,
      "loss": 0.2683,
      "step": 2257
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7153023649007895,
      "learning_rate": 5.353487828259973e-06,
      "loss": 0.2643,
      "step": 2258
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7177549039364571,
      "learning_rate": 5.349958762872016e-06,
      "loss": 0.2575,
      "step": 2259
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7498817218134824,
      "learning_rate": 5.346429522277381e-06,
      "loss": 0.2607,
      "step": 2260
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6800050121366018,
      "learning_rate": 5.3429001082429766e-06,
      "loss": 0.2724,
      "step": 2261
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7201304003018841,
      "learning_rate": 5.339370522535805e-06,
      "loss": 0.2648,
      "step": 2262
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7115893619507485,
      "learning_rate": 5.33584076692295e-06,
      "loss": 0.2584,
      "step": 2263
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7672547829399445,
      "learning_rate": 5.3323108431715845e-06,
      "loss": 0.275,
      "step": 2264
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7107740788401015,
      "learning_rate": 5.32878075304896e-06,
      "loss": 0.2778,
      "step": 2265
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7566579514731968,
      "learning_rate": 5.325250498322417e-06,
      "loss": 0.3157,
      "step": 2266
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7263207223580583,
      "learning_rate": 5.321720080759373e-06,
      "loss": 0.2789,
      "step": 2267
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6919275223943363,
      "learning_rate": 5.318189502127332e-06,
      "loss": 0.2873,
      "step": 2268
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6944665892239648,
      "learning_rate": 5.3146587641938766e-06,
      "loss": 0.2714,
      "step": 2269
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7100507147752989,
      "learning_rate": 5.311127868726667e-06,
      "loss": 0.2623,
      "step": 2270
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7233920827898758,
      "learning_rate": 5.307596817493445e-06,
      "loss": 0.2709,
      "step": 2271
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7161097615402355,
      "learning_rate": 5.304065612262033e-06,
      "loss": 0.2617,
      "step": 2272
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6621136363327843,
      "learning_rate": 5.300534254800321e-06,
      "loss": 0.2459,
      "step": 2273
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6999756573005023,
      "learning_rate": 5.297002746876284e-06,
      "loss": 0.2876,
      "step": 2274
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.754102771871802,
      "learning_rate": 5.293471090257971e-06,
      "loss": 0.2912,
      "step": 2275
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6923303968827104,
      "learning_rate": 5.289939286713505e-06,
      "loss": 0.2713,
      "step": 2276
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7723188585722672,
      "learning_rate": 5.286407338011079e-06,
      "loss": 0.32,
      "step": 2277
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7853813250883611,
      "learning_rate": 5.282875245918963e-06,
      "loss": 0.2808,
      "step": 2278
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6752102593172056,
      "learning_rate": 5.279343012205495e-06,
      "loss": 0.2751,
      "step": 2279
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7096810066978054,
      "learning_rate": 5.275810638639088e-06,
      "loss": 0.2699,
      "step": 2280
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7797959691716102,
      "learning_rate": 5.272278126988224e-06,
      "loss": 0.3035,
      "step": 2281
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7319063097462124,
      "learning_rate": 5.26874547902145e-06,
      "loss": 0.2657,
      "step": 2282
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6746893672675572,
      "learning_rate": 5.265212696507387e-06,
      "loss": 0.2704,
      "step": 2283
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6576732213149185,
      "learning_rate": 5.2616797812147205e-06,
      "loss": 0.277,
      "step": 2284
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6900174472231235,
      "learning_rate": 5.258146734912202e-06,
      "loss": 0.263,
      "step": 2285
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7607268530587965,
      "learning_rate": 5.2546135593686484e-06,
      "loss": 0.2714,
      "step": 2286
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7112656943438306,
      "learning_rate": 5.251080256352944e-06,
      "loss": 0.2528,
      "step": 2287
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7181494694581685,
      "learning_rate": 5.247546827634035e-06,
      "loss": 0.2562,
      "step": 2288
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7420229617722425,
      "learning_rate": 5.244013274980932e-06,
      "loss": 0.253,
      "step": 2289
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.711129826844497,
      "learning_rate": 5.240479600162704e-06,
      "loss": 0.2834,
      "step": 2290
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6929769970919517,
      "learning_rate": 5.2369458049484824e-06,
      "loss": 0.2428,
      "step": 2291
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6579749794142266,
      "learning_rate": 5.233411891107464e-06,
      "loss": 0.2251,
      "step": 2292
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7227754595039692,
      "learning_rate": 5.229877860408899e-06,
      "loss": 0.2376,
      "step": 2293
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6527601354912929,
      "learning_rate": 5.226343714622096e-06,
      "loss": 0.2184,
      "step": 2294
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6368014204277112,
      "learning_rate": 5.222809455516427e-06,
      "loss": 0.2133,
      "step": 2295
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7139934477628971,
      "learning_rate": 5.219275084861315e-06,
      "loss": 0.2142,
      "step": 2296
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6550596583560256,
      "learning_rate": 5.2157406044262424e-06,
      "loss": 0.2269,
      "step": 2297
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6955516479594025,
      "learning_rate": 5.212206015980742e-06,
      "loss": 0.2234,
      "step": 2298
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6938998003128063,
      "learning_rate": 5.208671321294406e-06,
      "loss": 0.2372,
      "step": 2299
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6927658078338522,
      "learning_rate": 5.2051365221368785e-06,
      "loss": 0.2275,
      "step": 2300
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7214343101998049,
      "learning_rate": 5.201601620277854e-06,
      "loss": 0.2274,
      "step": 2301
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.751758975102136,
      "learning_rate": 5.198066617487078e-06,
      "loss": 0.2343,
      "step": 2302
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6700806646033041,
      "learning_rate": 5.194531515534349e-06,
      "loss": 0.2119,
      "step": 2303
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6776193960397937,
      "learning_rate": 5.190996316189515e-06,
      "loss": 0.2111,
      "step": 2304
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.71319950596359,
      "learning_rate": 5.18746102122247e-06,
      "loss": 0.2288,
      "step": 2305
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7269578511455247,
      "learning_rate": 5.1839256324031575e-06,
      "loss": 0.2107,
      "step": 2306
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7491915486769158,
      "learning_rate": 5.18039015150157e-06,
      "loss": 0.2352,
      "step": 2307
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6928709540420753,
      "learning_rate": 5.176854580287744e-06,
      "loss": 0.2117,
      "step": 2308
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6712305411917385,
      "learning_rate": 5.173318920531761e-06,
      "loss": 0.1917,
      "step": 2309
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6846986485905895,
      "learning_rate": 5.169783174003744e-06,
      "loss": 0.2066,
      "step": 2310
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6911629052797897,
      "learning_rate": 5.166247342473868e-06,
      "loss": 0.2137,
      "step": 2311
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7235349850510542,
      "learning_rate": 5.1627114277123435e-06,
      "loss": 0.2223,
      "step": 2312
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7395864808554553,
      "learning_rate": 5.159175431489424e-06,
      "loss": 0.2274,
      "step": 2313
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7149962586691214,
      "learning_rate": 5.155639355575402e-06,
      "loss": 0.2228,
      "step": 2314
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7096569118871925,
      "learning_rate": 5.152103201740617e-06,
      "loss": 0.2119,
      "step": 2315
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7684905130866256,
      "learning_rate": 5.14856697175544e-06,
      "loss": 0.2357,
      "step": 2316
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6825771252519807,
      "learning_rate": 5.145030667390281e-06,
      "loss": 0.195,
      "step": 2317
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7153082775794426,
      "learning_rate": 5.141494290415592e-06,
      "loss": 0.2227,
      "step": 2318
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.785298619812158,
      "learning_rate": 5.137957842601856e-06,
      "loss": 0.2227,
      "step": 2319
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.718517473943119,
      "learning_rate": 5.1344213257195965e-06,
      "loss": 0.2125,
      "step": 2320
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6798995561986827,
      "learning_rate": 5.130884741539367e-06,
      "loss": 0.2095,
      "step": 2321
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.734725042712074,
      "learning_rate": 5.127348091831755e-06,
      "loss": 0.2214,
      "step": 2322
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6989483958664038,
      "learning_rate": 5.123811378367387e-06,
      "loss": 0.2203,
      "step": 2323
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6955437107076875,
      "learning_rate": 5.120274602916915e-06,
      "loss": 0.2076,
      "step": 2324
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7646862034217813,
      "learning_rate": 5.116737767251021e-06,
      "loss": 0.2386,
      "step": 2325
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.764052524830395,
      "learning_rate": 5.1132008731404225e-06,
      "loss": 0.2272,
      "step": 2326
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7427277819053459,
      "learning_rate": 5.1096639223558645e-06,
      "loss": 0.2293,
      "step": 2327
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7062952687873117,
      "learning_rate": 5.106126916668118e-06,
      "loss": 0.2046,
      "step": 2328
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.716964851479053,
      "learning_rate": 5.102589857847984e-06,
      "loss": 0.2409,
      "step": 2329
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7008821103918589,
      "learning_rate": 5.099052747666289e-06,
      "loss": 0.2231,
      "step": 2330
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7490266423959572,
      "learning_rate": 5.0955155878938845e-06,
      "loss": 0.2291,
      "step": 2331
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7120446959748309,
      "learning_rate": 5.09197838030165e-06,
      "loss": 0.2195,
      "step": 2332
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.744831168918449,
      "learning_rate": 5.088441126660484e-06,
      "loss": 0.2391,
      "step": 2333
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6881432312887805,
      "learning_rate": 5.084903828741312e-06,
      "loss": 0.2118,
      "step": 2334
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7073126377920534,
      "learning_rate": 5.081366488315082e-06,
      "loss": 0.2154,
      "step": 2335
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7179959560539928,
      "learning_rate": 5.0778291071527596e-06,
      "loss": 0.215,
      "step": 2336
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7189701673463361,
      "learning_rate": 5.074291687025334e-06,
      "loss": 0.2264,
      "step": 2337
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7783156142931794,
      "learning_rate": 5.070754229703811e-06,
      "loss": 0.2414,
      "step": 2338
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7334504747588447,
      "learning_rate": 5.067216736959219e-06,
      "loss": 0.2179,
      "step": 2339
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.764912353220648,
      "learning_rate": 5.063679210562602e-06,
      "loss": 0.2409,
      "step": 2340
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6824417222520222,
      "learning_rate": 5.06014165228502e-06,
      "loss": 0.208,
      "step": 2341
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7845620060761387,
      "learning_rate": 5.056604063897549e-06,
      "loss": 0.2369,
      "step": 2342
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7331559702106367,
      "learning_rate": 5.053066447171282e-06,
      "loss": 0.1973,
      "step": 2343
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7070056242308872,
      "learning_rate": 5.049528803877324e-06,
      "loss": 0.2,
      "step": 2344
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7269124208361217,
      "learning_rate": 5.045991135786796e-06,
      "loss": 0.2248,
      "step": 2345
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7196854018919948,
      "learning_rate": 5.042453444670829e-06,
      "loss": 0.2215,
      "step": 2346
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7024832832694052,
      "learning_rate": 5.038915732300565e-06,
      "loss": 0.2159,
      "step": 2347
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7052193508632981,
      "learning_rate": 5.0353780004471605e-06,
      "loss": 0.2166,
      "step": 2348
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.6937767036276127,
      "learning_rate": 5.0318402508817765e-06,
      "loss": 0.2023,
      "step": 2349
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7358490928765411,
      "learning_rate": 5.028302485375588e-06,
      "loss": 0.2215,
      "step": 2350
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.6974861649886434,
      "learning_rate": 5.024764705699773e-06,
      "loss": 0.2228,
      "step": 2351
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.715820550828015,
      "learning_rate": 5.021226913625522e-06,
      "loss": 0.2272,
      "step": 2352
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.75245424042281,
      "learning_rate": 5.0176891109240265e-06,
      "loss": 0.2298,
      "step": 2353
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7343429211388731,
      "learning_rate": 5.0141512993664864e-06,
      "loss": 0.2331,
      "step": 2354
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7412516463346023,
      "learning_rate": 5.0106134807241045e-06,
      "loss": 0.2228,
      "step": 2355
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7035402177396993,
      "learning_rate": 5.007075656768091e-06,
      "loss": 0.2157,
      "step": 2356
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7329507317195972,
      "learning_rate": 5.0035378292696525e-06,
      "loss": 0.205,
      "step": 2357
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.878709813362945,
      "learning_rate": 5e-06,
      "loss": 0.2324,
      "step": 2358
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7624436747080937,
      "learning_rate": 4.9964621707303475e-06,
      "loss": 0.2315,
      "step": 2359
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7591792865782907,
      "learning_rate": 4.99292434323191e-06,
      "loss": 0.2096,
      "step": 2360
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7708219780826083,
      "learning_rate": 4.989386519275895e-06,
      "loss": 0.2215,
      "step": 2361
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.6827296646501627,
      "learning_rate": 4.985848700633515e-06,
      "loss": 0.206,
      "step": 2362
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7595058248479672,
      "learning_rate": 4.9823108890759735e-06,
      "loss": 0.2273,
      "step": 2363
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.707866425452845,
      "learning_rate": 4.97877308637448e-06,
      "loss": 0.2074,
      "step": 2364
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7109859067648852,
      "learning_rate": 4.975235294300227e-06,
      "loss": 0.2011,
      "step": 2365
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.691867160534324,
      "learning_rate": 4.971697514624414e-06,
      "loss": 0.2042,
      "step": 2366
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7189690422538112,
      "learning_rate": 4.9681597491182235e-06,
      "loss": 0.2291,
      "step": 2367
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7177669870800464,
      "learning_rate": 4.964621999552841e-06,
      "loss": 0.2087,
      "step": 2368
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7468404129525235,
      "learning_rate": 4.961084267699435e-06,
      "loss": 0.2153,
      "step": 2369
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7455319252047266,
      "learning_rate": 4.957546555329174e-06,
      "loss": 0.2237,
      "step": 2370
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7187745206339414,
      "learning_rate": 4.9540088642132044e-06,
      "loss": 0.1971,
      "step": 2371
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.749650200346472,
      "learning_rate": 4.950471196122677e-06,
      "loss": 0.2112,
      "step": 2372
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7388759568976221,
      "learning_rate": 4.94693355282872e-06,
      "loss": 0.2208,
      "step": 2373
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7678614909971043,
      "learning_rate": 4.943395936102453e-06,
      "loss": 0.2089,
      "step": 2374
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7426880572598019,
      "learning_rate": 4.939858347714981e-06,
      "loss": 0.2024,
      "step": 2375
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7098401297908847,
      "learning_rate": 4.9363207894374e-06,
      "loss": 0.2265,
      "step": 2376
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7503433716200861,
      "learning_rate": 4.9327832630407825e-06,
      "loss": 0.2092,
      "step": 2377
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.773040406014198,
      "learning_rate": 4.929245770296191e-06,
      "loss": 0.2387,
      "step": 2378
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7831415829113091,
      "learning_rate": 4.925708312974667e-06,
      "loss": 0.2156,
      "step": 2379
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.775609330551932,
      "learning_rate": 4.922170892847242e-06,
      "loss": 0.2368,
      "step": 2380
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7369213408632893,
      "learning_rate": 4.91863351168492e-06,
      "loss": 0.2248,
      "step": 2381
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.6789799624900177,
      "learning_rate": 4.91509617125869e-06,
      "loss": 0.2296,
      "step": 2382
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7107830099601917,
      "learning_rate": 4.911558873339517e-06,
      "loss": 0.2037,
      "step": 2383
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7047862295408929,
      "learning_rate": 4.908021619698352e-06,
      "loss": 0.1994,
      "step": 2384
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7958240302877663,
      "learning_rate": 4.904484412106117e-06,
      "loss": 0.2277,
      "step": 2385
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7233644775257393,
      "learning_rate": 4.900947252333712e-06,
      "loss": 0.2165,
      "step": 2386
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.731149392646508,
      "learning_rate": 4.897410142152017e-06,
      "loss": 0.2229,
      "step": 2387
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.771594336225585,
      "learning_rate": 4.8938730833318825e-06,
      "loss": 0.235,
      "step": 2388
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7384917534578418,
      "learning_rate": 4.890336077644137e-06,
      "loss": 0.2114,
      "step": 2389
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.6999577949335719,
      "learning_rate": 4.8867991268595775e-06,
      "loss": 0.2021,
      "step": 2390
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7820579943570841,
      "learning_rate": 4.88326223274898e-06,
      "loss": 0.2221,
      "step": 2391
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7027093854031,
      "learning_rate": 4.879725397083087e-06,
      "loss": 0.2134,
      "step": 2392
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7665202052831134,
      "learning_rate": 4.876188621632614e-06,
      "loss": 0.219,
      "step": 2393
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7478323244388803,
      "learning_rate": 4.872651908168244e-06,
      "loss": 0.2194,
      "step": 2394
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7552351671508765,
      "learning_rate": 4.869115258460636e-06,
      "loss": 0.2121,
      "step": 2395
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7490140732735436,
      "learning_rate": 4.865578674280406e-06,
      "loss": 0.2142,
      "step": 2396
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7057986827197743,
      "learning_rate": 4.862042157398147e-06,
      "loss": 0.2072,
      "step": 2397
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7274629799680484,
      "learning_rate": 4.85850570958441e-06,
      "loss": 0.2214,
      "step": 2398
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7457288482452573,
      "learning_rate": 4.854969332609722e-06,
      "loss": 0.2086,
      "step": 2399
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7246388144545238,
      "learning_rate": 4.851433028244562e-06,
      "loss": 0.2083,
      "step": 2400
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.6967909793693907,
      "learning_rate": 4.847896798259386e-06,
      "loss": 0.2164,
      "step": 2401
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7096938732178498,
      "learning_rate": 4.844360644424599e-06,
      "loss": 0.2076,
      "step": 2402
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.6840881107445365,
      "learning_rate": 4.840824568510579e-06,
      "loss": 0.2184,
      "step": 2403
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.6990958651957185,
      "learning_rate": 4.837288572287658e-06,
      "loss": 0.2057,
      "step": 2404
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7471311257654601,
      "learning_rate": 4.8337526575261336e-06,
      "loss": 0.213,
      "step": 2405
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7402529797725983,
      "learning_rate": 4.830216825996257e-06,
      "loss": 0.213,
      "step": 2406
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7623461328087011,
      "learning_rate": 4.82668107946824e-06,
      "loss": 0.2222,
      "step": 2407
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.6582310070540428,
      "learning_rate": 4.8231454197122575e-06,
      "loss": 0.1885,
      "step": 2408
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7238964047661943,
      "learning_rate": 4.81960984849843e-06,
      "loss": 0.2056,
      "step": 2409
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7855799270177137,
      "learning_rate": 4.816074367596843e-06,
      "loss": 0.2262,
      "step": 2410
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7520829370811484,
      "learning_rate": 4.81253897877753e-06,
      "loss": 0.2206,
      "step": 2411
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8049495365202308,
      "learning_rate": 4.809003683810487e-06,
      "loss": 0.2329,
      "step": 2412
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7332891712902533,
      "learning_rate": 4.805468484465651e-06,
      "loss": 0.216,
      "step": 2413
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.746774962538432,
      "learning_rate": 4.801933382512923e-06,
      "loss": 0.2341,
      "step": 2414
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7385456680716258,
      "learning_rate": 4.798398379722147e-06,
      "loss": 0.2106,
      "step": 2415
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7175982649543575,
      "learning_rate": 4.794863477863122e-06,
      "loss": 0.2067,
      "step": 2416
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7139691177089635,
      "learning_rate": 4.791328678705594e-06,
      "loss": 0.2074,
      "step": 2417
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7414890290559308,
      "learning_rate": 4.78779398401926e-06,
      "loss": 0.2337,
      "step": 2418
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7206939498971221,
      "learning_rate": 4.784259395573759e-06,
      "loss": 0.2134,
      "step": 2419
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7298225413978338,
      "learning_rate": 4.780724915138686e-06,
      "loss": 0.2185,
      "step": 2420
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7098769736480288,
      "learning_rate": 4.777190544483574e-06,
      "loss": 0.2003,
      "step": 2421
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.752537248760573,
      "learning_rate": 4.7736562853779066e-06,
      "loss": 0.2193,
      "step": 2422
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7514528684691797,
      "learning_rate": 4.770122139591103e-06,
      "loss": 0.2287,
      "step": 2423
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7050784575460617,
      "learning_rate": 4.766588108892538e-06,
      "loss": 0.1994,
      "step": 2424
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8006162258373204,
      "learning_rate": 4.763054195051519e-06,
      "loss": 0.2194,
      "step": 2425
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.74811145311134,
      "learning_rate": 4.7595203998373e-06,
      "loss": 0.2219,
      "step": 2426
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8374858753391072,
      "learning_rate": 4.75598672501907e-06,
      "loss": 0.2373,
      "step": 2427
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7837776553915371,
      "learning_rate": 4.752453172365966e-06,
      "loss": 0.2402,
      "step": 2428
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7547394306951766,
      "learning_rate": 4.7489197436470575e-06,
      "loss": 0.2246,
      "step": 2429
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7691184734234494,
      "learning_rate": 4.745386440631354e-06,
      "loss": 0.2147,
      "step": 2430
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8126860237082211,
      "learning_rate": 4.7418532650878005e-06,
      "loss": 0.2102,
      "step": 2431
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.6944477916029542,
      "learning_rate": 4.738320218785281e-06,
      "loss": 0.2049,
      "step": 2432
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7580116609327289,
      "learning_rate": 4.734787303492615e-06,
      "loss": 0.2297,
      "step": 2433
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7617017203463653,
      "learning_rate": 4.73125452097855e-06,
      "loss": 0.2257,
      "step": 2434
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7241217011154363,
      "learning_rate": 4.727721873011778e-06,
      "loss": 0.2021,
      "step": 2435
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7116038144473354,
      "learning_rate": 4.7241893613609126e-06,
      "loss": 0.2198,
      "step": 2436
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7563361804788155,
      "learning_rate": 4.720656987794507e-06,
      "loss": 0.2279,
      "step": 2437
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7179783874414088,
      "learning_rate": 4.717124754081038e-06,
      "loss": 0.2278,
      "step": 2438
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.738598195475659,
      "learning_rate": 4.713592661988922e-06,
      "loss": 0.2139,
      "step": 2439
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7509341609920965,
      "learning_rate": 4.710060713286496e-06,
      "loss": 0.2217,
      "step": 2440
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.6757922621243581,
      "learning_rate": 4.70652890974203e-06,
      "loss": 0.2065,
      "step": 2441
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.6906698599210975,
      "learning_rate": 4.702997253123716e-06,
      "loss": 0.1992,
      "step": 2442
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7858836639058501,
      "learning_rate": 4.6994657451996815e-06,
      "loss": 0.2184,
      "step": 2443
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7641570038222286,
      "learning_rate": 4.6959343877379695e-06,
      "loss": 0.225,
      "step": 2444
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8164248579930271,
      "learning_rate": 4.692403182506557e-06,
      "loss": 0.2274,
      "step": 2445
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7268546167990062,
      "learning_rate": 4.688872131273334e-06,
      "loss": 0.2019,
      "step": 2446
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7051979632145661,
      "learning_rate": 4.685341235806126e-06,
      "loss": 0.1941,
      "step": 2447
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7119246747400949,
      "learning_rate": 4.6818104978726685e-06,
      "loss": 0.216,
      "step": 2448
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.786361667457002,
      "learning_rate": 4.678279919240629e-06,
      "loss": 0.238,
      "step": 2449
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7103921072043038,
      "learning_rate": 4.674749501677585e-06,
      "loss": 0.2078,
      "step": 2450
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7256390354988408,
      "learning_rate": 4.671219246951043e-06,
      "loss": 0.2125,
      "step": 2451
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7212663994874252,
      "learning_rate": 4.667689156828417e-06,
      "loss": 0.205,
      "step": 2452
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7896411661836725,
      "learning_rate": 4.664159233077051e-06,
      "loss": 0.2315,
      "step": 2453
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7958573914329989,
      "learning_rate": 4.660629477464197e-06,
      "loss": 0.2114,
      "step": 2454
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7189273009348249,
      "learning_rate": 4.657099891757023e-06,
      "loss": 0.2191,
      "step": 2455
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.6931473009016857,
      "learning_rate": 4.653570477722621e-06,
      "loss": 0.2176,
      "step": 2456
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.735481993193373,
      "learning_rate": 4.650041237127983e-06,
      "loss": 0.2372,
      "step": 2457
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.780866900278659,
      "learning_rate": 4.646512171740028e-06,
      "loss": 0.2461,
      "step": 2458
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7187154668369822,
      "learning_rate": 4.642983283325574e-06,
      "loss": 0.2042,
      "step": 2459
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7293195458441986,
      "learning_rate": 4.639454573651364e-06,
      "loss": 0.2228,
      "step": 2460
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7155407903347799,
      "learning_rate": 4.6359260444840395e-06,
      "loss": 0.214,
      "step": 2461
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.723566874985894,
      "learning_rate": 4.6323976975901595e-06,
      "loss": 0.2163,
      "step": 2462
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7483997621825879,
      "learning_rate": 4.628869534736187e-06,
      "loss": 0.2336,
      "step": 2463
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.6919082530727133,
      "learning_rate": 4.625341557688499e-06,
      "loss": 0.2022,
      "step": 2464
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7833426197138714,
      "learning_rate": 4.621813768213372e-06,
      "loss": 0.241,
      "step": 2465
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7427194292943368,
      "learning_rate": 4.618286168076993e-06,
      "loss": 0.2152,
      "step": 2466
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7604425933527091,
      "learning_rate": 4.614758759045452e-06,
      "loss": 0.206,
      "step": 2467
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7539317304121146,
      "learning_rate": 4.611231542884747e-06,
      "loss": 0.2218,
      "step": 2468
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.6902351527011088,
      "learning_rate": 4.6077045213607765e-06,
      "loss": 0.2039,
      "step": 2469
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7316364377905319,
      "learning_rate": 4.6041776962393405e-06,
      "loss": 0.2296,
      "step": 2470
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7185300521462094,
      "learning_rate": 4.600651069286141e-06,
      "loss": 0.2244,
      "step": 2471
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.723334399053822,
      "learning_rate": 4.597124642266788e-06,
      "loss": 0.226,
      "step": 2472
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7478522204906125,
      "learning_rate": 4.593598416946779e-06,
      "loss": 0.2258,
      "step": 2473
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7782307486505677,
      "learning_rate": 4.590072395091521e-06,
      "loss": 0.2145,
      "step": 2474
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7651031748323943,
      "learning_rate": 4.586546578466311e-06,
      "loss": 0.2356,
      "step": 2475
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7357128924409291,
      "learning_rate": 4.583020968836353e-06,
      "loss": 0.2245,
      "step": 2476
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7159248882655196,
      "learning_rate": 4.579495567966738e-06,
      "loss": 0.2043,
      "step": 2477
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7157903964814355,
      "learning_rate": 4.575970377622456e-06,
      "loss": 0.2031,
      "step": 2478
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7334090909056902,
      "learning_rate": 4.572445399568391e-06,
      "loss": 0.2094,
      "step": 2479
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7464550624778237,
      "learning_rate": 4.568920635569325e-06,
      "loss": 0.2213,
      "step": 2480
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7583074255481083,
      "learning_rate": 4.565396087389928e-06,
      "loss": 0.2367,
      "step": 2481
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7085878486837995,
      "learning_rate": 4.56187175679476e-06,
      "loss": 0.2146,
      "step": 2482
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7910039966077715,
      "learning_rate": 4.55834764554828e-06,
      "loss": 0.2555,
      "step": 2483
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7340566054707175,
      "learning_rate": 4.554823755414829e-06,
      "loss": 0.214,
      "step": 2484
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7669067241024193,
      "learning_rate": 4.551300088158645e-06,
      "loss": 0.2078,
      "step": 2485
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8854773268791286,
      "learning_rate": 4.547776645543846e-06,
      "loss": 0.2241,
      "step": 2486
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7078405384187241,
      "learning_rate": 4.544253429334444e-06,
      "loss": 0.2136,
      "step": 2487
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.6788456594677954,
      "learning_rate": 4.540730441294334e-06,
      "loss": 0.2033,
      "step": 2488
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7884763669508009,
      "learning_rate": 4.5372076831873006e-06,
      "loss": 0.2174,
      "step": 2489
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.731886458197574,
      "learning_rate": 4.533685156777008e-06,
      "loss": 0.2235,
      "step": 2490
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7398696700585058,
      "learning_rate": 4.53016286382701e-06,
      "loss": 0.199,
      "step": 2491
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7220574020444697,
      "learning_rate": 4.526640806100737e-06,
      "loss": 0.21,
      "step": 2492
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7298256054785325,
      "learning_rate": 4.523118985361511e-06,
      "loss": 0.2096,
      "step": 2493
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.726639417135202,
      "learning_rate": 4.519597403372524e-06,
      "loss": 0.2164,
      "step": 2494
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7309451828972929,
      "learning_rate": 4.516076061896859e-06,
      "loss": 0.2315,
      "step": 2495
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.722600111501639,
      "learning_rate": 4.5125549626974695e-06,
      "loss": 0.2041,
      "step": 2496
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7601430493319178,
      "learning_rate": 4.5090341075371964e-06,
      "loss": 0.2351,
      "step": 2497
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7662481350019147,
      "learning_rate": 4.505513498178752e-06,
      "loss": 0.2408,
      "step": 2498
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7046488416923309,
      "learning_rate": 4.501993136384727e-06,
      "loss": 0.2068,
      "step": 2499
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7070960454134201,
      "learning_rate": 4.4984730239175885e-06,
      "loss": 0.2098,
      "step": 2500
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7668890991265578,
      "learning_rate": 4.494953162539682e-06,
      "loss": 0.2174,
      "step": 2501
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.770113296393988,
      "learning_rate": 4.491433554013221e-06,
      "loss": 0.2325,
      "step": 2502
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.6899160973655715,
      "learning_rate": 4.487914200100296e-06,
      "loss": 0.206,
      "step": 2503
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.6927810745179973,
      "learning_rate": 4.484395102562873e-06,
      "loss": 0.2125,
      "step": 2504
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7118628081081557,
      "learning_rate": 4.4808762631627825e-06,
      "loss": 0.2297,
      "step": 2505
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7252988659506993,
      "learning_rate": 4.477357683661734e-06,
      "loss": 0.2111,
      "step": 2506
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7159232962667906,
      "learning_rate": 4.4738393658212965e-06,
      "loss": 0.2107,
      "step": 2507
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7510684689843072,
      "learning_rate": 4.47032131140292e-06,
      "loss": 0.209,
      "step": 2508
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7329497767180294,
      "learning_rate": 4.466803522167914e-06,
      "loss": 0.2254,
      "step": 2509
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7085115244850572,
      "learning_rate": 4.463285999877458e-06,
      "loss": 0.2141,
      "step": 2510
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7300378936201015,
      "learning_rate": 4.459768746292597e-06,
      "loss": 0.2138,
      "step": 2511
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7660198804201622,
      "learning_rate": 4.456251763174245e-06,
      "loss": 0.2338,
      "step": 2512
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7360771306580914,
      "learning_rate": 4.452735052283175e-06,
      "loss": 0.2272,
      "step": 2513
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7616416523177791,
      "learning_rate": 4.449218615380029e-06,
      "loss": 0.2255,
      "step": 2514
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.6996601019311994,
      "learning_rate": 4.445702454225305e-06,
      "loss": 0.1971,
      "step": 2515
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7642529387428261,
      "learning_rate": 4.442186570579373e-06,
      "loss": 0.2438,
      "step": 2516
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.711827350629488,
      "learning_rate": 4.438670966202455e-06,
      "loss": 0.2034,
      "step": 2517
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.738972992928359,
      "learning_rate": 4.435155642854637e-06,
      "loss": 0.2391,
      "step": 2518
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7160319636660881,
      "learning_rate": 4.431640602295863e-06,
      "loss": 0.2002,
      "step": 2519
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7332335902307145,
      "learning_rate": 4.42812584628594e-06,
      "loss": 0.2322,
      "step": 2520
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8332973242839641,
      "learning_rate": 4.4246113765845244e-06,
      "loss": 0.2337,
      "step": 2521
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7383860253541238,
      "learning_rate": 4.421097194951137e-06,
      "loss": 0.2115,
      "step": 2522
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7118650834544501,
      "learning_rate": 4.4175833031451475e-06,
      "loss": 0.2047,
      "step": 2523
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7383955132956579,
      "learning_rate": 4.414069702925789e-06,
      "loss": 0.2239,
      "step": 2524
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7584235441256587,
      "learning_rate": 4.4105563960521395e-06,
      "loss": 0.2266,
      "step": 2525
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7437077028187451,
      "learning_rate": 4.407043384283136e-06,
      "loss": 0.208,
      "step": 2526
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7839526880533244,
      "learning_rate": 4.4035306693775645e-06,
      "loss": 0.2313,
      "step": 2527
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7629480551086241,
      "learning_rate": 4.400018253094065e-06,
      "loss": 0.223,
      "step": 2528
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7314729543866126,
      "learning_rate": 4.396506137191131e-06,
      "loss": 0.2162,
      "step": 2529
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7515380486731745,
      "learning_rate": 4.3929943234270935e-06,
      "loss": 0.2174,
      "step": 2530
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7293338592279716,
      "learning_rate": 4.389482813560148e-06,
      "loss": 0.2162,
      "step": 2531
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.771396367499794,
      "learning_rate": 4.385971609348324e-06,
      "loss": 0.2064,
      "step": 2532
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7252961460088536,
      "learning_rate": 4.38246071254951e-06,
      "loss": 0.201,
      "step": 2533
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.740563842424557,
      "learning_rate": 4.37895012492143e-06,
      "loss": 0.2278,
      "step": 2534
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7134736908305211,
      "learning_rate": 4.375439848221661e-06,
      "loss": 0.1999,
      "step": 2535
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7988909799479279,
      "learning_rate": 4.371929884207617e-06,
      "loss": 0.221,
      "step": 2536
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7680971989164633,
      "learning_rate": 4.3684202346365655e-06,
      "loss": 0.2202,
      "step": 2537
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.718970747006994,
      "learning_rate": 4.364910901265607e-06,
      "loss": 0.2087,
      "step": 2538
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7399119868608256,
      "learning_rate": 4.3614018858516885e-06,
      "loss": 0.2217,
      "step": 2539
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7040939619442786,
      "learning_rate": 4.3578931901515955e-06,
      "loss": 0.1936,
      "step": 2540
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7986706642288883,
      "learning_rate": 4.354384815921958e-06,
      "loss": 0.2264,
      "step": 2541
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8103479912100641,
      "learning_rate": 4.350876764919238e-06,
      "loss": 0.2249,
      "step": 2542
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7629588342339396,
      "learning_rate": 4.347369038899744e-06,
      "loss": 0.211,
      "step": 2543
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8063717498781313,
      "learning_rate": 4.343861639619611e-06,
      "loss": 0.2291,
      "step": 2544
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.741489524066697,
      "learning_rate": 4.3403545688348225e-06,
      "loss": 0.2278,
      "step": 2545
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7185980970526101,
      "learning_rate": 4.336847828301189e-06,
      "loss": 0.217,
      "step": 2546
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.734000139141878,
      "learning_rate": 4.333341419774359e-06,
      "loss": 0.2048,
      "step": 2547
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7996704925300875,
      "learning_rate": 4.329835345009813e-06,
      "loss": 0.2353,
      "step": 2548
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7364004760655857,
      "learning_rate": 4.3263296057628685e-06,
      "loss": 0.2029,
      "step": 2549
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7674807735081272,
      "learning_rate": 4.322824203788669e-06,
      "loss": 0.2106,
      "step": 2550
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.686879505598947,
      "learning_rate": 4.3193191408421945e-06,
      "loss": 0.2048,
      "step": 2551
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.745736049939675,
      "learning_rate": 4.3158144186782504e-06,
      "loss": 0.2002,
      "step": 2552
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.801205368538779,
      "learning_rate": 4.312310039051476e-06,
      "loss": 0.217,
      "step": 2553
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.769212570819908,
      "learning_rate": 4.308806003716338e-06,
      "loss": 0.2173,
      "step": 2554
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.793597710469809,
      "learning_rate": 4.305302314427128e-06,
      "loss": 0.2213,
      "step": 2555
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7319368546816826,
      "learning_rate": 4.301798972937968e-06,
      "loss": 0.2023,
      "step": 2556
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.738439611489552,
      "learning_rate": 4.2982959810028025e-06,
      "loss": 0.2202,
      "step": 2557
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7308984973432795,
      "learning_rate": 4.294793340375405e-06,
      "loss": 0.2109,
      "step": 2558
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7376634980231127,
      "learning_rate": 4.291291052809366e-06,
      "loss": 0.2249,
      "step": 2559
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7449605574715518,
      "learning_rate": 4.287789120058111e-06,
      "loss": 0.226,
      "step": 2560
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7523283462364577,
      "learning_rate": 4.284287543874875e-06,
      "loss": 0.2201,
      "step": 2561
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7592022876470123,
      "learning_rate": 4.280786326012724e-06,
      "loss": 0.2184,
      "step": 2562
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7462333653425037,
      "learning_rate": 4.2772854682245365e-06,
      "loss": 0.2215,
      "step": 2563
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7730720066365665,
      "learning_rate": 4.273784972263021e-06,
      "loss": 0.2385,
      "step": 2564
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7263218787280056,
      "learning_rate": 4.270284839880696e-06,
      "loss": 0.2124,
      "step": 2565
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7273382118358755,
      "learning_rate": 4.266785072829901e-06,
      "loss": 0.2164,
      "step": 2566
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7221417839017054,
      "learning_rate": 4.263285672862792e-06,
      "loss": 0.2023,
      "step": 2567
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7182138413953524,
      "learning_rate": 4.259786641731344e-06,
      "loss": 0.1934,
      "step": 2568
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.732003958422046,
      "learning_rate": 4.256287981187343e-06,
      "loss": 0.1989,
      "step": 2569
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7590818810595491,
      "learning_rate": 4.252789692982394e-06,
      "loss": 0.2191,
      "step": 2570
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7151109881343539,
      "learning_rate": 4.2492917788679095e-06,
      "loss": 0.2058,
      "step": 2571
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7360939377226572,
      "learning_rate": 4.245794240595123e-06,
      "loss": 0.2211,
      "step": 2572
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7385222568665204,
      "learning_rate": 4.242297079915071e-06,
      "loss": 0.2123,
      "step": 2573
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7881747868102442,
      "learning_rate": 4.23880029857861e-06,
      "loss": 0.2083,
      "step": 2574
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8791867800641686,
      "learning_rate": 4.235303898336396e-06,
      "loss": 0.2365,
      "step": 2575
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7727315697617793,
      "learning_rate": 4.231807880938904e-06,
      "loss": 0.2105,
      "step": 2576
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7698216169766893,
      "learning_rate": 4.228312248136414e-06,
      "loss": 0.217,
      "step": 2577
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7674855396070213,
      "learning_rate": 4.224817001679011e-06,
      "loss": 0.2188,
      "step": 2578
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7315244401474186,
      "learning_rate": 4.221322143316589e-06,
      "loss": 0.2135,
      "step": 2579
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7445017054763987,
      "learning_rate": 4.217827674798845e-06,
      "loss": 0.2239,
      "step": 2580
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.6986398653019038,
      "learning_rate": 4.214333597875288e-06,
      "loss": 0.1952,
      "step": 2581
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7437187409621807,
      "learning_rate": 4.210839914295219e-06,
      "loss": 0.2033,
      "step": 2582
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7411811571662236,
      "learning_rate": 4.2073466258077564e-06,
      "loss": 0.2104,
      "step": 2583
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7473078504299465,
      "learning_rate": 4.203853734161806e-06,
      "loss": 0.2113,
      "step": 2584
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8128187109889882,
      "learning_rate": 4.200361241106088e-06,
      "loss": 0.2271,
      "step": 2585
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7393280637926253,
      "learning_rate": 4.196869148389114e-06,
      "loss": 0.2114,
      "step": 2586
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7421251952537092,
      "learning_rate": 4.1933774577592e-06,
      "loss": 0.2237,
      "step": 2587
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7360246248168558,
      "learning_rate": 4.189886170964458e-06,
      "loss": 0.2104,
      "step": 2588
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7673353917946846,
      "learning_rate": 4.1863952897528e-06,
      "loss": 0.2301,
      "step": 2589
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7398135798670211,
      "learning_rate": 4.1829048158719344e-06,
      "loss": 0.2147,
      "step": 2590
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7141682564669396,
      "learning_rate": 4.179414751069365e-06,
      "loss": 0.1998,
      "step": 2591
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7858871484962893,
      "learning_rate": 4.1759250970923876e-06,
      "loss": 0.2364,
      "step": 2592
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.74482481422728,
      "learning_rate": 4.172435855688101e-06,
      "loss": 0.2076,
      "step": 2593
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7974380565249126,
      "learning_rate": 4.168947028603389e-06,
      "loss": 0.2275,
      "step": 2594
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7551467378411585,
      "learning_rate": 4.165458617584933e-06,
      "loss": 0.2347,
      "step": 2595
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.730279345564403,
      "learning_rate": 4.161970624379201e-06,
      "loss": 0.2015,
      "step": 2596
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7851817292242234,
      "learning_rate": 4.158483050732458e-06,
      "loss": 0.2375,
      "step": 2597
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7233843786255273,
      "learning_rate": 4.154995898390756e-06,
      "loss": 0.1966,
      "step": 2598
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7159348876959872,
      "learning_rate": 4.151509169099935e-06,
      "loss": 0.222,
      "step": 2599
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7619570099752686,
      "learning_rate": 4.1480228646056225e-06,
      "loss": 0.226,
      "step": 2600
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7450774338999306,
      "learning_rate": 4.144536986653239e-06,
      "loss": 0.2303,
      "step": 2601
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7545921580819821,
      "learning_rate": 4.141051536987986e-06,
      "loss": 0.2191,
      "step": 2602
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7007386530575032,
      "learning_rate": 4.13756651735485e-06,
      "loss": 0.198,
      "step": 2603
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7727416920208341,
      "learning_rate": 4.1340819294986076e-06,
      "loss": 0.223,
      "step": 2604
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7948217943702968,
      "learning_rate": 4.130597775163814e-06,
      "loss": 0.2298,
      "step": 2605
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7094034651077902,
      "learning_rate": 4.127114056094811e-06,
      "loss": 0.213,
      "step": 2606
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7461287100284557,
      "learning_rate": 4.1236307740357175e-06,
      "loss": 0.2213,
      "step": 2607
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7517075455149687,
      "learning_rate": 4.12014793073044e-06,
      "loss": 0.2277,
      "step": 2608
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7111669502573054,
      "learning_rate": 4.116665527922662e-06,
      "loss": 0.2079,
      "step": 2609
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7555080628929559,
      "learning_rate": 4.113183567355846e-06,
      "loss": 0.2186,
      "step": 2610
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7606155831606336,
      "learning_rate": 4.109702050773233e-06,
      "loss": 0.2211,
      "step": 2611
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7634383385528585,
      "learning_rate": 4.106220979917845e-06,
      "loss": 0.2069,
      "step": 2612
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7490971960897952,
      "learning_rate": 4.102740356532474e-06,
      "loss": 0.2019,
      "step": 2613
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7324347135120008,
      "learning_rate": 4.099260182359698e-06,
      "loss": 0.1992,
      "step": 2614
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7017563482202213,
      "learning_rate": 4.0957804591418596e-06,
      "loss": 0.2019,
      "step": 2615
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.682628301789677,
      "learning_rate": 4.092301188621084e-06,
      "loss": 0.2027,
      "step": 2616
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.711780501026605,
      "learning_rate": 4.088822372539263e-06,
      "loss": 0.2163,
      "step": 2617
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7023812882124234,
      "learning_rate": 4.085344012638067e-06,
      "loss": 0.2019,
      "step": 2618
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7124703715760149,
      "learning_rate": 4.081866110658934e-06,
      "loss": 0.2143,
      "step": 2619
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.6919922976392654,
      "learning_rate": 4.078388668343075e-06,
      "loss": 0.2105,
      "step": 2620
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7591993670594455,
      "learning_rate": 4.074911687431466e-06,
      "loss": 0.214,
      "step": 2621
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7285779722039453,
      "learning_rate": 4.071435169664861e-06,
      "loss": 0.218,
      "step": 2622
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7006804333379847,
      "learning_rate": 4.0679591167837725e-06,
      "loss": 0.1969,
      "step": 2623
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7027012712396675,
      "learning_rate": 4.064483530528487e-06,
      "loss": 0.2129,
      "step": 2624
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7354937018485183,
      "learning_rate": 4.061008412639055e-06,
      "loss": 0.2058,
      "step": 2625
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7682082599120396,
      "learning_rate": 4.057533764855291e-06,
      "loss": 0.2379,
      "step": 2626
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7411696030068968,
      "learning_rate": 4.054059588916779e-06,
      "loss": 0.2142,
      "step": 2627
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7320744964658601,
      "learning_rate": 4.050585886562858e-06,
      "loss": 0.2074,
      "step": 2628
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7659399875396703,
      "learning_rate": 4.047112659532641e-06,
      "loss": 0.208,
      "step": 2629
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7436470949307146,
      "learning_rate": 4.043639909564993e-06,
      "loss": 0.2222,
      "step": 2630
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7379177715624754,
      "learning_rate": 4.040167638398549e-06,
      "loss": 0.2284,
      "step": 2631
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7248090412067579,
      "learning_rate": 4.036695847771694e-06,
      "loss": 0.2044,
      "step": 2632
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7250849882771937,
      "learning_rate": 4.033224539422584e-06,
      "loss": 0.2164,
      "step": 2633
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7401619283862976,
      "learning_rate": 4.029753715089124e-06,
      "loss": 0.2028,
      "step": 2634
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7407714351208877,
      "learning_rate": 4.026283376508983e-06,
      "loss": 0.2228,
      "step": 2635
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7410628649964853,
      "learning_rate": 4.02281352541958e-06,
      "loss": 0.204,
      "step": 2636
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7559770476658598,
      "learning_rate": 4.0193441635581e-06,
      "loss": 0.2285,
      "step": 2637
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7635769839690071,
      "learning_rate": 4.015875292661474e-06,
      "loss": 0.2325,
      "step": 2638
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.733215265363802,
      "learning_rate": 4.012406914466391e-06,
      "loss": 0.2159,
      "step": 2639
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7620931969781093,
      "learning_rate": 4.008939030709291e-06,
      "loss": 0.2112,
      "step": 2640
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.6865699580297544,
      "learning_rate": 4.005471643126372e-06,
      "loss": 0.1941,
      "step": 2641
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.725397187819954,
      "learning_rate": 4.002004753453577e-06,
      "loss": 0.2152,
      "step": 2642
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7300920260642259,
      "learning_rate": 3.998538363426605e-06,
      "loss": 0.2111,
      "step": 2643
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7512575002712415,
      "learning_rate": 3.995072474780899e-06,
      "loss": 0.2239,
      "step": 2644
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8124686617706962,
      "learning_rate": 3.9916070892516575e-06,
      "loss": 0.2321,
      "step": 2645
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.756459726144271,
      "learning_rate": 3.988142208573822e-06,
      "loss": 0.2057,
      "step": 2646
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7508611841350463,
      "learning_rate": 3.984677834482085e-06,
      "loss": 0.2249,
      "step": 2647
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7510073617002391,
      "learning_rate": 3.981213968710882e-06,
      "loss": 0.2051,
      "step": 2648
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.6989264591721017,
      "learning_rate": 3.977750612994397e-06,
      "loss": 0.2045,
      "step": 2649
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.6903978845348916,
      "learning_rate": 3.9742877690665575e-06,
      "loss": 0.2005,
      "step": 2650
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7248177113506664,
      "learning_rate": 3.970825438661032e-06,
      "loss": 0.2099,
      "step": 2651
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.775431196685737,
      "learning_rate": 3.967363623511238e-06,
      "loss": 0.2166,
      "step": 2652
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8008427356311503,
      "learning_rate": 3.96390232535033e-06,
      "loss": 0.2259,
      "step": 2653
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.727128637461233,
      "learning_rate": 3.960441545911205e-06,
      "loss": 0.2096,
      "step": 2654
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8236300194500459,
      "learning_rate": 3.956981286926499e-06,
      "loss": 0.2267,
      "step": 2655
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7246448406242817,
      "learning_rate": 3.953521550128591e-06,
      "loss": 0.2023,
      "step": 2656
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.6982539337183077,
      "learning_rate": 3.950062337249596e-06,
      "loss": 0.2107,
      "step": 2657
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7285883774609566,
      "learning_rate": 3.94660365002137e-06,
      "loss": 0.2136,
      "step": 2658
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7669485275022877,
      "learning_rate": 3.943145490175497e-06,
      "loss": 0.2288,
      "step": 2659
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7199473454820471,
      "learning_rate": 3.939687859443308e-06,
      "loss": 0.2167,
      "step": 2660
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7798392673722216,
      "learning_rate": 3.93623075955586e-06,
      "loss": 0.2384,
      "step": 2661
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7322338519482284,
      "learning_rate": 3.932774192243953e-06,
      "loss": 0.2233,
      "step": 2662
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7519781571610954,
      "learning_rate": 3.929318159238113e-06,
      "loss": 0.202,
      "step": 2663
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7359814381295765,
      "learning_rate": 3.925862662268602e-06,
      "loss": 0.2054,
      "step": 2664
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7378966704680454,
      "learning_rate": 3.922407703065411e-06,
      "loss": 0.2176,
      "step": 2665
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.73343593558669,
      "learning_rate": 3.918953283358266e-06,
      "loss": 0.2254,
      "step": 2666
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7244506279639855,
      "learning_rate": 3.915499404876619e-06,
      "loss": 0.2069,
      "step": 2667
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7839297640322995,
      "learning_rate": 3.912046069349654e-06,
      "loss": 0.2154,
      "step": 2668
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7411994264150708,
      "learning_rate": 3.908593278506279e-06,
      "loss": 0.1896,
      "step": 2669
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7693926575833469,
      "learning_rate": 3.905141034075135e-06,
      "loss": 0.2243,
      "step": 2670
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.786878754470664,
      "learning_rate": 3.901689337784585e-06,
      "loss": 0.2176,
      "step": 2671
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7318358745511616,
      "learning_rate": 3.898238191362718e-06,
      "loss": 0.2119,
      "step": 2672
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.743692174286769,
      "learning_rate": 3.894787596537352e-06,
      "loss": 0.2008,
      "step": 2673
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7985085355246633,
      "learning_rate": 3.891337555036022e-06,
      "loss": 0.223,
      "step": 2674
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7686450960944278,
      "learning_rate": 3.887888068585992e-06,
      "loss": 0.2072,
      "step": 2675
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7705418740876665,
      "learning_rate": 3.884439138914243e-06,
      "loss": 0.2201,
      "step": 2676
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7448335146103073,
      "learning_rate": 3.880990767747484e-06,
      "loss": 0.2185,
      "step": 2677
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7409952012634048,
      "learning_rate": 3.877542956812137e-06,
      "loss": 0.208,
      "step": 2678
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.6989857286639606,
      "learning_rate": 3.874095707834349e-06,
      "loss": 0.2034,
      "step": 2679
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7204473341759882,
      "learning_rate": 3.870649022539979e-06,
      "loss": 0.2222,
      "step": 2680
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8301449754234858,
      "learning_rate": 3.867202902654616e-06,
      "loss": 0.2337,
      "step": 2681
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7293371888379114,
      "learning_rate": 3.863757349903551e-06,
      "loss": 0.239,
      "step": 2682
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.733048248249755,
      "learning_rate": 3.860312366011802e-06,
      "loss": 0.2137,
      "step": 2683
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7111316454840059,
      "learning_rate": 3.856867952704095e-06,
      "loss": 0.2054,
      "step": 2684
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7241842390851274,
      "learning_rate": 3.853424111704879e-06,
      "loss": 0.2128,
      "step": 2685
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7593841036524549,
      "learning_rate": 3.849980844738308e-06,
      "loss": 0.2,
      "step": 2686
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7493218741959657,
      "learning_rate": 3.846538153528253e-06,
      "loss": 0.2075,
      "step": 2687
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7876824698232509,
      "learning_rate": 3.843096039798293e-06,
      "loss": 0.2227,
      "step": 2688
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.739004697931858,
      "learning_rate": 3.839654505271724e-06,
      "loss": 0.2127,
      "step": 2689
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7959799838830441,
      "learning_rate": 3.836213551671547e-06,
      "loss": 0.2128,
      "step": 2690
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7825929508176503,
      "learning_rate": 3.832773180720475e-06,
      "loss": 0.2307,
      "step": 2691
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7268358462280964,
      "learning_rate": 3.829333394140924e-06,
      "loss": 0.2156,
      "step": 2692
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.767566791455208,
      "learning_rate": 3.825894193655026e-06,
      "loss": 0.211,
      "step": 2693
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7079904549904329,
      "learning_rate": 3.822455580984613e-06,
      "loss": 0.2104,
      "step": 2694
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7836980518980754,
      "learning_rate": 3.819017557851226e-06,
      "loss": 0.2335,
      "step": 2695
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8023094768972195,
      "learning_rate": 3.8155801259761065e-06,
      "loss": 0.2225,
      "step": 2696
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7634500954557024,
      "learning_rate": 3.812143287080205e-06,
      "loss": 0.2077,
      "step": 2697
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.75025672260304,
      "learning_rate": 3.808707042884176e-06,
      "loss": 0.2215,
      "step": 2698
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.713120627904261,
      "learning_rate": 3.805271395108368e-06,
      "loss": 0.2004,
      "step": 2699
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7779801581734569,
      "learning_rate": 3.8018363454728415e-06,
      "loss": 0.2127,
      "step": 2700
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7845057330224962,
      "learning_rate": 3.7984018956973467e-06,
      "loss": 0.2183,
      "step": 2701
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7423940495954774,
      "learning_rate": 3.794968047501345e-06,
      "loss": 0.2144,
      "step": 2702
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7255269909216969,
      "learning_rate": 3.7915348026039877e-06,
      "loss": 0.2255,
      "step": 2703
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.766999167388485,
      "learning_rate": 3.7881021627241283e-06,
      "loss": 0.2354,
      "step": 2704
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7434125316031385,
      "learning_rate": 3.7846701295803136e-06,
      "loss": 0.2278,
      "step": 2705
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7630047839323075,
      "learning_rate": 3.781238704890793e-06,
      "loss": 0.2245,
      "step": 2706
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7548888253100491,
      "learning_rate": 3.7778078903735044e-06,
      "loss": 0.2141,
      "step": 2707
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7466444331533001,
      "learning_rate": 3.7743776877460864e-06,
      "loss": 0.2078,
      "step": 2708
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8103301795120881,
      "learning_rate": 3.770948098725864e-06,
      "loss": 0.229,
      "step": 2709
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7241837587353073,
      "learning_rate": 3.767519125029865e-06,
      "loss": 0.2262,
      "step": 2710
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7620035974588006,
      "learning_rate": 3.7640907683747977e-06,
      "loss": 0.2178,
      "step": 2711
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7197351516678224,
      "learning_rate": 3.760663030477072e-06,
      "loss": 0.2158,
      "step": 2712
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7863415173303022,
      "learning_rate": 3.757235913052778e-06,
      "loss": 0.2351,
      "step": 2713
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7260705789240365,
      "learning_rate": 3.7538094178177054e-06,
      "loss": 0.2099,
      "step": 2714
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7451768565374772,
      "learning_rate": 3.750383546487324e-06,
      "loss": 0.2195,
      "step": 2715
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7278759791210825,
      "learning_rate": 3.7469583007767964e-06,
      "loss": 0.2093,
      "step": 2716
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.723844406664789,
      "learning_rate": 3.743533682400968e-06,
      "loss": 0.2065,
      "step": 2717
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7497114126857488,
      "learning_rate": 3.7401096930743753e-06,
      "loss": 0.2208,
      "step": 2718
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7836238758374254,
      "learning_rate": 3.7366863345112334e-06,
      "loss": 0.2383,
      "step": 2719
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7474765038612897,
      "learning_rate": 3.733263608425447e-06,
      "loss": 0.2199,
      "step": 2720
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7322534522539459,
      "learning_rate": 3.729841516530604e-06,
      "loss": 0.2064,
      "step": 2721
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7504756842847778,
      "learning_rate": 3.72642006053997e-06,
      "loss": 0.2084,
      "step": 2722
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7193105325805006,
      "learning_rate": 3.722999242166497e-06,
      "loss": 0.2118,
      "step": 2723
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7716437814229232,
      "learning_rate": 3.7195790631228133e-06,
      "loss": 0.2299,
      "step": 2724
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7751478895756507,
      "learning_rate": 3.7161595251212334e-06,
      "loss": 0.2362,
      "step": 2725
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7893061860694346,
      "learning_rate": 3.7127406298737458e-06,
      "loss": 0.2124,
      "step": 2726
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7295461503532327,
      "learning_rate": 3.709322379092019e-06,
      "loss": 0.198,
      "step": 2727
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7563436208042505,
      "learning_rate": 3.705904774487396e-06,
      "loss": 0.2309,
      "step": 2728
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7363897699524607,
      "learning_rate": 3.702487817770903e-06,
      "loss": 0.2058,
      "step": 2729
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7332686289074092,
      "learning_rate": 3.6990715106532353e-06,
      "loss": 0.2245,
      "step": 2730
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.772817955849267,
      "learning_rate": 3.6956558548447665e-06,
      "loss": 0.2191,
      "step": 2731
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7380892717380079,
      "learning_rate": 3.69224085205554e-06,
      "loss": 0.2238,
      "step": 2732
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7380244498240087,
      "learning_rate": 3.6888265039952796e-06,
      "loss": 0.2049,
      "step": 2733
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7870407955886278,
      "learning_rate": 3.6854128123733736e-06,
      "loss": 0.2087,
      "step": 2734
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7174946441732077,
      "learning_rate": 3.681999778898887e-06,
      "loss": 0.2095,
      "step": 2735
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7444444964657252,
      "learning_rate": 3.6785874052805515e-06,
      "loss": 0.2137,
      "step": 2736
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7399968358679665,
      "learning_rate": 3.6751756932267725e-06,
      "loss": 0.2304,
      "step": 2737
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.758572592582499,
      "learning_rate": 3.6717646444456196e-06,
      "loss": 0.2168,
      "step": 2738
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7617898047258749,
      "learning_rate": 3.6683542606448347e-06,
      "loss": 0.2264,
      "step": 2739
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7148041524643864,
      "learning_rate": 3.664944543531822e-06,
      "loss": 0.2018,
      "step": 2740
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7475940413578381,
      "learning_rate": 3.6615354948136584e-06,
      "loss": 0.2273,
      "step": 2741
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7891153858035237,
      "learning_rate": 3.658127116197079e-06,
      "loss": 0.2054,
      "step": 2742
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7686298936422776,
      "learning_rate": 3.6547194093884907e-06,
      "loss": 0.2318,
      "step": 2743
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.785884546264189,
      "learning_rate": 3.651312376093955e-06,
      "loss": 0.2255,
      "step": 2744
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7418708648441357,
      "learning_rate": 3.6479060180192038e-06,
      "loss": 0.2173,
      "step": 2745
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7379252971963994,
      "learning_rate": 3.6445003368696317e-06,
      "loss": 0.2215,
      "step": 2746
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7114554663911913,
      "learning_rate": 3.6410953343502875e-06,
      "loss": 0.2004,
      "step": 2747
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7159009177030168,
      "learning_rate": 3.6376910121658867e-06,
      "loss": 0.1949,
      "step": 2748
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7363944125274376,
      "learning_rate": 3.6342873720207972e-06,
      "loss": 0.1989,
      "step": 2749
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7585429842358286,
      "learning_rate": 3.6308844156190546e-06,
      "loss": 0.2107,
      "step": 2750
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7392283663400031,
      "learning_rate": 3.6274821446643445e-06,
      "loss": 0.1957,
      "step": 2751
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7253260283445101,
      "learning_rate": 3.6240805608600137e-06,
      "loss": 0.2082,
      "step": 2752
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7185732104357594,
      "learning_rate": 3.6206796659090605e-06,
      "loss": 0.1952,
      "step": 2753
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7082110036724132,
      "learning_rate": 3.6172794615141448e-06,
      "loss": 0.203,
      "step": 2754
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7754107625655559,
      "learning_rate": 3.6138799493775758e-06,
      "loss": 0.203,
      "step": 2755
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7749105721076811,
      "learning_rate": 3.6104811312013175e-06,
      "loss": 0.2223,
      "step": 2756
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7627817062537215,
      "learning_rate": 3.607083008686985e-06,
      "loss": 0.2122,
      "step": 2757
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7375651704819707,
      "learning_rate": 3.60368558353585e-06,
      "loss": 0.2175,
      "step": 2758
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7652137954642126,
      "learning_rate": 3.6002888574488294e-06,
      "loss": 0.213,
      "step": 2759
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7649136338573297,
      "learning_rate": 3.596892832126494e-06,
      "loss": 0.2137,
      "step": 2760
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7781566392396718,
      "learning_rate": 3.5934975092690585e-06,
      "loss": 0.2114,
      "step": 2761
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7421268475534615,
      "learning_rate": 3.590102890576394e-06,
      "loss": 0.2211,
      "step": 2762
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7685428358893595,
      "learning_rate": 3.5867089777480124e-06,
      "loss": 0.2345,
      "step": 2763
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7200367728886268,
      "learning_rate": 3.5833157724830757e-06,
      "loss": 0.2289,
      "step": 2764
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7154704916991732,
      "learning_rate": 3.579923276480387e-06,
      "loss": 0.2006,
      "step": 2765
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7569267644800617,
      "learning_rate": 3.576531491438403e-06,
      "loss": 0.2086,
      "step": 2766
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7447473787553117,
      "learning_rate": 3.5731404190552154e-06,
      "loss": 0.2078,
      "step": 2767
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7451548879670098,
      "learning_rate": 3.569750061028565e-06,
      "loss": 0.2088,
      "step": 2768
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7163073018942958,
      "learning_rate": 3.56636041905583e-06,
      "loss": 0.2026,
      "step": 2769
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7135561696900274,
      "learning_rate": 3.5629714948340354e-06,
      "loss": 0.1969,
      "step": 2770
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7561453531644322,
      "learning_rate": 3.5595832900598453e-06,
      "loss": 0.2085,
      "step": 2771
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7402560068405705,
      "learning_rate": 3.5561958064295585e-06,
      "loss": 0.2167,
      "step": 2772
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7306769344904702,
      "learning_rate": 3.552809045639123e-06,
      "loss": 0.2109,
      "step": 2773
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8370925267562046,
      "learning_rate": 3.5494230093841144e-06,
      "loss": 0.2322,
      "step": 2774
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7423454132863432,
      "learning_rate": 3.546037699359751e-06,
      "loss": 0.2071,
      "step": 2775
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7117545660164502,
      "learning_rate": 3.5426531172608858e-06,
      "loss": 0.2009,
      "step": 2776
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7384782527029493,
      "learning_rate": 3.5392692647820113e-06,
      "loss": 0.2166,
      "step": 2777
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.770704720247794,
      "learning_rate": 3.5358861436172487e-06,
      "loss": 0.2122,
      "step": 2778
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.747318980723946,
      "learning_rate": 3.5325037554603576e-06,
      "loss": 0.2106,
      "step": 2779
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7726919810413604,
      "learning_rate": 3.5291221020047257e-06,
      "loss": 0.2189,
      "step": 2780
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7432149462067051,
      "learning_rate": 3.525741184943381e-06,
      "loss": 0.2156,
      "step": 2781
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7434184878631972,
      "learning_rate": 3.5223610059689733e-06,
      "loss": 0.2151,
      "step": 2782
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7566381981344732,
      "learning_rate": 3.5189815667737916e-06,
      "loss": 0.2245,
      "step": 2783
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.6770221415948556,
      "learning_rate": 3.515602869049746e-06,
      "loss": 0.1822,
      "step": 2784
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8017110632725811,
      "learning_rate": 3.5122249144883848e-06,
      "loss": 0.2225,
      "step": 2785
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.792510333325396,
      "learning_rate": 3.5088477047808738e-06,
      "loss": 0.2297,
      "step": 2786
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7370481772377873,
      "learning_rate": 3.505471241618018e-06,
      "loss": 0.2136,
      "step": 2787
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7581038292375156,
      "learning_rate": 3.5020955266902344e-06,
      "loss": 0.1941,
      "step": 2788
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7467025578323283,
      "learning_rate": 3.4987205616875803e-06,
      "loss": 0.1978,
      "step": 2789
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7369424086044563,
      "learning_rate": 3.495346348299724e-06,
      "loss": 0.2018,
      "step": 2790
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7628853720878417,
      "learning_rate": 3.4919728882159685e-06,
      "loss": 0.2176,
      "step": 2791
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7499530516788812,
      "learning_rate": 3.4886001831252324e-06,
      "loss": 0.2001,
      "step": 2792
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7527450033143396,
      "learning_rate": 3.485228234716058e-06,
      "loss": 0.2176,
      "step": 2793
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7719911999687904,
      "learning_rate": 3.481857044676613e-06,
      "loss": 0.2323,
      "step": 2794
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7306013716707639,
      "learning_rate": 3.478486614694679e-06,
      "loss": 0.2184,
      "step": 2795
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7876152371064518,
      "learning_rate": 3.475116946457662e-06,
      "loss": 0.2221,
      "step": 2796
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7908087698610968,
      "learning_rate": 3.4717480416525815e-06,
      "loss": 0.2135,
      "step": 2797
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7419503627417827,
      "learning_rate": 3.4683799019660834e-06,
      "loss": 0.2229,
      "step": 2798
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7397528089025972,
      "learning_rate": 3.4650125290844207e-06,
      "loss": 0.1925,
      "step": 2799
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7254749697793577,
      "learning_rate": 3.4616459246934686e-06,
      "loss": 0.2111,
      "step": 2800
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7663672343355904,
      "learning_rate": 3.4582800904787138e-06,
      "loss": 0.2213,
      "step": 2801
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7517600756579654,
      "learning_rate": 3.4549150281252635e-06,
      "loss": 0.207,
      "step": 2802
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7848931758902179,
      "learning_rate": 3.4515507393178316e-06,
      "loss": 0.2172,
      "step": 2803
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7096861413345619,
      "learning_rate": 3.448187225740749e-06,
      "loss": 0.2017,
      "step": 2804
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7442026089320399,
      "learning_rate": 3.4448244890779544e-06,
      "loss": 0.2116,
      "step": 2805
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7532894722110832,
      "learning_rate": 3.4414625310130038e-06,
      "loss": 0.2199,
      "step": 2806
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7301426491012325,
      "learning_rate": 3.4381013532290575e-06,
      "loss": 0.2057,
      "step": 2807
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7784031585318149,
      "learning_rate": 3.4347409574088896e-06,
      "loss": 0.2167,
      "step": 2808
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7921284084972923,
      "learning_rate": 3.4313813452348774e-06,
      "loss": 0.2214,
      "step": 2809
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8449613950487045,
      "learning_rate": 3.4280225183890124e-06,
      "loss": 0.223,
      "step": 2810
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7356869426252647,
      "learning_rate": 3.424664478552887e-06,
      "loss": 0.2129,
      "step": 2811
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7662834007186323,
      "learning_rate": 3.4213072274077047e-06,
      "loss": 0.2236,
      "step": 2812
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7357047983692523,
      "learning_rate": 3.417950766634268e-06,
      "loss": 0.2164,
      "step": 2813
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7128125112141384,
      "learning_rate": 3.4145950979129916e-06,
      "loss": 0.2167,
      "step": 2814
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.822582681034181,
      "learning_rate": 3.411240222923886e-06,
      "loss": 0.2458,
      "step": 2815
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7322895862572911,
      "learning_rate": 3.4078861433465693e-06,
      "loss": 0.2006,
      "step": 2816
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7920625517361299,
      "learning_rate": 3.404532860860258e-06,
      "loss": 0.2278,
      "step": 2817
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7615926108703902,
      "learning_rate": 3.401180377143774e-06,
      "loss": 0.2201,
      "step": 2818
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7681066243055282,
      "learning_rate": 3.3978286938755357e-06,
      "loss": 0.2151,
      "step": 2819
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7675273112779496,
      "learning_rate": 3.39447781273356e-06,
      "loss": 0.229,
      "step": 2820
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.741124795075042,
      "learning_rate": 3.3911277353954665e-06,
      "loss": 0.2222,
      "step": 2821
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7776504299226756,
      "learning_rate": 3.3877784635384668e-06,
      "loss": 0.2199,
      "step": 2822
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.718825997705026,
      "learning_rate": 3.3844299988393757e-06,
      "loss": 0.2138,
      "step": 2823
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7209242486117718,
      "learning_rate": 3.3810823429745955e-06,
      "loss": 0.207,
      "step": 2824
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.763911535459145,
      "learning_rate": 3.3777354976201338e-06,
      "loss": 0.2109,
      "step": 2825
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7387782840300134,
      "learning_rate": 3.374389464451583e-06,
      "loss": 0.2054,
      "step": 2826
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7442698256321959,
      "learning_rate": 3.3710442451441374e-06,
      "loss": 0.2032,
      "step": 2827
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8108490268083132,
      "learning_rate": 3.3676998413725726e-06,
      "loss": 0.2233,
      "step": 2828
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7715507284001714,
      "learning_rate": 3.36435625481127e-06,
      "loss": 0.2226,
      "step": 2829
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.6866444159905544,
      "learning_rate": 3.361013487134188e-06,
      "loss": 0.1945,
      "step": 2830
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7629041855308828,
      "learning_rate": 3.3576715400148863e-06,
      "loss": 0.2161,
      "step": 2831
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7846695672160146,
      "learning_rate": 3.354330415126508e-06,
      "loss": 0.2242,
      "step": 2832
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7772914380693281,
      "learning_rate": 3.3509901141417845e-06,
      "loss": 0.2256,
      "step": 2833
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7308044622114671,
      "learning_rate": 3.3476506387330355e-06,
      "loss": 0.2016,
      "step": 2834
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7886902334627769,
      "learning_rate": 3.34431199057217e-06,
      "loss": 0.2079,
      "step": 2835
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7792122631086218,
      "learning_rate": 3.3409741713306777e-06,
      "loss": 0.2209,
      "step": 2836
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8091822987402468,
      "learning_rate": 3.3376371826796393e-06,
      "loss": 0.2346,
      "step": 2837
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7896095328143884,
      "learning_rate": 3.3343010262897125e-06,
      "loss": 0.2215,
      "step": 2838
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7530010079842792,
      "learning_rate": 3.330965703831146e-06,
      "loss": 0.2077,
      "step": 2839
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7541201470748966,
      "learning_rate": 3.327631216973766e-06,
      "loss": 0.2164,
      "step": 2840
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7509375037078024,
      "learning_rate": 3.324297567386979e-06,
      "loss": 0.2174,
      "step": 2841
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7425391646214463,
      "learning_rate": 3.320964756739781e-06,
      "loss": 0.192,
      "step": 2842
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7338985605507684,
      "learning_rate": 3.3176327867007376e-06,
      "loss": 0.2088,
      "step": 2843
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7243634465809808,
      "learning_rate": 3.3143016589379984e-06,
      "loss": 0.1955,
      "step": 2844
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7306125352881694,
      "learning_rate": 3.310971375119291e-06,
      "loss": 0.2217,
      "step": 2845
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.737833348053032,
      "learning_rate": 3.3076419369119216e-06,
      "loss": 0.2046,
      "step": 2846
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7715599906786684,
      "learning_rate": 3.3043133459827703e-06,
      "loss": 0.2159,
      "step": 2847
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.702427479831748,
      "learning_rate": 3.300985603998296e-06,
      "loss": 0.2053,
      "step": 2848
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7285154838573015,
      "learning_rate": 3.2976587126245275e-06,
      "loss": 0.2005,
      "step": 2849
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7487852598150886,
      "learning_rate": 3.2943326735270766e-06,
      "loss": 0.1828,
      "step": 2850
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.779433248010856,
      "learning_rate": 3.2910074883711186e-06,
      "loss": 0.2251,
      "step": 2851
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7334743906801234,
      "learning_rate": 3.2876831588214093e-06,
      "loss": 0.2142,
      "step": 2852
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7294476854155139,
      "learning_rate": 3.2843596865422687e-06,
      "loss": 0.2071,
      "step": 2853
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7641434989541902,
      "learning_rate": 3.2810370731975956e-06,
      "loss": 0.2082,
      "step": 2854
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.766002300396962,
      "learning_rate": 3.2777153204508516e-06,
      "loss": 0.238,
      "step": 2855
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.714623162424816,
      "learning_rate": 3.2743944299650733e-06,
      "loss": 0.1977,
      "step": 2856
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7655420471821798,
      "learning_rate": 3.2710744034028598e-06,
      "loss": 0.2067,
      "step": 2857
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.757826587835975,
      "learning_rate": 3.2677552424263836e-06,
      "loss": 0.2327,
      "step": 2858
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7579244519545622,
      "learning_rate": 3.2644369486973786e-06,
      "loss": 0.216,
      "step": 2859
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7709593122413221,
      "learning_rate": 3.2611195238771494e-06,
      "loss": 0.2256,
      "step": 2860
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7104590579912448,
      "learning_rate": 3.2578029696265596e-06,
      "loss": 0.1945,
      "step": 2861
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7365318947835864,
      "learning_rate": 3.254487287606044e-06,
      "loss": 0.2149,
      "step": 2862
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7866649655003538,
      "learning_rate": 3.251172479475595e-06,
      "loss": 0.2206,
      "step": 2863
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7781654889472394,
      "learning_rate": 3.247858546894772e-06,
      "loss": 0.2129,
      "step": 2864
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7413429962700112,
      "learning_rate": 3.244545491522689e-06,
      "loss": 0.1958,
      "step": 2865
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7275601677369954,
      "learning_rate": 3.2412333150180298e-06,
      "loss": 0.1977,
      "step": 2866
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7700097435656555,
      "learning_rate": 3.2379220190390337e-06,
      "loss": 0.2163,
      "step": 2867
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.6972945087629594,
      "learning_rate": 3.234611605243496e-06,
      "loss": 0.1783,
      "step": 2868
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7953651010529027,
      "learning_rate": 3.231302075288778e-06,
      "loss": 0.2362,
      "step": 2869
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7700593307517787,
      "learning_rate": 3.2279934308317905e-06,
      "loss": 0.2089,
      "step": 2870
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7632865532372899,
      "learning_rate": 3.2246856735290096e-06,
      "loss": 0.2318,
      "step": 2871
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7723901135944682,
      "learning_rate": 3.2213788050364557e-06,
      "loss": 0.2179,
      "step": 2872
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7971957577383568,
      "learning_rate": 3.2180728270097163e-06,
      "loss": 0.2317,
      "step": 2873
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7599985785221427,
      "learning_rate": 3.2147677411039236e-06,
      "loss": 0.212,
      "step": 2874
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7307187710208437,
      "learning_rate": 3.2114635489737715e-06,
      "loss": 0.2153,
      "step": 2875
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.6877151421313641,
      "learning_rate": 3.2081602522734987e-06,
      "loss": 0.1912,
      "step": 2876
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7641387427861976,
      "learning_rate": 3.204857852656902e-06,
      "loss": 0.216,
      "step": 2877
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7349150490340243,
      "learning_rate": 3.2015563517773214e-06,
      "loss": 0.1861,
      "step": 2878
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7047034203557203,
      "learning_rate": 3.1982557512876567e-06,
      "loss": 0.1985,
      "step": 2879
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7502592853939682,
      "learning_rate": 3.1949560528403487e-06,
      "loss": 0.2096,
      "step": 2880
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7253872939511965,
      "learning_rate": 3.1916572580873927e-06,
      "loss": 0.2065,
      "step": 2881
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7270963058507677,
      "learning_rate": 3.1883593686803244e-06,
      "loss": 0.2044,
      "step": 2882
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7716277369649109,
      "learning_rate": 3.1850623862702344e-06,
      "loss": 0.2244,
      "step": 2883
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7386433419687107,
      "learning_rate": 3.1817663125077534e-06,
      "loss": 0.201,
      "step": 2884
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7682224820498185,
      "learning_rate": 3.178471149043061e-06,
      "loss": 0.2386,
      "step": 2885
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7661815726395621,
      "learning_rate": 3.1751768975258745e-06,
      "loss": 0.2035,
      "step": 2886
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7572492729697257,
      "learning_rate": 3.171883559605465e-06,
      "loss": 0.2142,
      "step": 2887
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7080446640149306,
      "learning_rate": 3.1685911369306364e-06,
      "loss": 0.1935,
      "step": 2888
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7121432601769538,
      "learning_rate": 3.16529963114974e-06,
      "loss": 0.1941,
      "step": 2889
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7530476379952651,
      "learning_rate": 3.162009043910668e-06,
      "loss": 0.2061,
      "step": 2890
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.6985771364958654,
      "learning_rate": 3.158719376860848e-06,
      "loss": 0.1932,
      "step": 2891
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7317684388717102,
      "learning_rate": 3.155430631647254e-06,
      "loss": 0.2124,
      "step": 2892
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7308574158555201,
      "learning_rate": 3.1521428099163897e-06,
      "loss": 0.2082,
      "step": 2893
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8255711560708742,
      "learning_rate": 3.1488559133143065e-06,
      "loss": 0.2309,
      "step": 2894
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.751775556499258,
      "learning_rate": 3.1455699434865844e-06,
      "loss": 0.2082,
      "step": 2895
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7239999975197442,
      "learning_rate": 3.1422849020783442e-06,
      "loss": 0.1942,
      "step": 2896
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7672573631007173,
      "learning_rate": 3.139000790734238e-06,
      "loss": 0.2259,
      "step": 2897
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7493523717441656,
      "learning_rate": 3.1357176110984578e-06,
      "loss": 0.221,
      "step": 2898
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7558928778597231,
      "learning_rate": 3.132435364814723e-06,
      "loss": 0.2256,
      "step": 2899
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8009401260213764,
      "learning_rate": 3.1291540535262905e-06,
      "loss": 0.2206,
      "step": 2900
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.6978840330100905,
      "learning_rate": 3.1258736788759443e-06,
      "loss": 0.2036,
      "step": 2901
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7094971028127064,
      "learning_rate": 3.122594242506006e-06,
      "loss": 0.2102,
      "step": 2902
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7456463890617143,
      "learning_rate": 3.1193157460583217e-06,
      "loss": 0.224,
      "step": 2903
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7630308886177005,
      "learning_rate": 3.11603819117427e-06,
      "loss": 0.2105,
      "step": 2904
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7438446159215542,
      "learning_rate": 3.1127615794947546e-06,
      "loss": 0.2272,
      "step": 2905
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.719279578616853,
      "learning_rate": 3.1094859126602132e-06,
      "loss": 0.185,
      "step": 2906
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7146915970911634,
      "learning_rate": 3.1062111923106044e-06,
      "loss": 0.1962,
      "step": 2907
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7428942530811562,
      "learning_rate": 3.1029374200854167e-06,
      "loss": 0.2047,
      "step": 2908
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7509643842628356,
      "learning_rate": 3.0996645976236593e-06,
      "loss": 0.2355,
      "step": 2909
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7188622670849905,
      "learning_rate": 3.096392726563874e-06,
      "loss": 0.2114,
      "step": 2910
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.756829938668132,
      "learning_rate": 3.0931218085441174e-06,
      "loss": 0.2254,
      "step": 2911
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.711929652282253,
      "learning_rate": 3.089851845201976e-06,
      "loss": 0.2195,
      "step": 2912
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7047861151097911,
      "learning_rate": 3.0865828381745515e-06,
      "loss": 0.2013,
      "step": 2913
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7497570170367941,
      "learning_rate": 3.0833147890984735e-06,
      "loss": 0.2316,
      "step": 2914
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7408999861484271,
      "learning_rate": 3.08004769960989e-06,
      "loss": 0.2051,
      "step": 2915
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7471740024214546,
      "learning_rate": 3.0767815713444655e-06,
      "loss": 0.2109,
      "step": 2916
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7395480718717482,
      "learning_rate": 3.0735164059373872e-06,
      "loss": 0.208,
      "step": 2917
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7293038194046265,
      "learning_rate": 3.070252205023356e-06,
      "loss": 0.2265,
      "step": 2918
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7114387238809267,
      "learning_rate": 3.066988970236596e-06,
      "loss": 0.2009,
      "step": 2919
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7543334823262237,
      "learning_rate": 3.063726703210841e-06,
      "loss": 0.216,
      "step": 2920
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7745238855470195,
      "learning_rate": 3.0604654055793446e-06,
      "loss": 0.2119,
      "step": 2921
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7722564436668407,
      "learning_rate": 3.0572050789748732e-06,
      "loss": 0.2125,
      "step": 2922
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.724825278162283,
      "learning_rate": 3.0539457250297095e-06,
      "loss": 0.208,
      "step": 2923
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7997840965723905,
      "learning_rate": 3.050687345375645e-06,
      "loss": 0.2469,
      "step": 2924
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7270705228540097,
      "learning_rate": 3.0474299416439885e-06,
      "loss": 0.2125,
      "step": 2925
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7709740720833669,
      "learning_rate": 3.044173515465554e-06,
      "loss": 0.2177,
      "step": 2926
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8108245555811933,
      "learning_rate": 3.0409180684706728e-06,
      "loss": 0.2645,
      "step": 2927
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7840252283364654,
      "learning_rate": 3.0376636022891813e-06,
      "loss": 0.2446,
      "step": 2928
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7688776222891252,
      "learning_rate": 3.034410118550428e-06,
      "loss": 0.2204,
      "step": 2929
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.747549053671096,
      "learning_rate": 3.031157618883265e-06,
      "loss": 0.2222,
      "step": 2930
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7022651156017385,
      "learning_rate": 3.027906104916057e-06,
      "loss": 0.195,
      "step": 2931
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7019429935340361,
      "learning_rate": 3.0246555782766723e-06,
      "loss": 0.1989,
      "step": 2932
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8221885616766246,
      "learning_rate": 3.0214060405924863e-06,
      "loss": 0.2322,
      "step": 2933
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.736864233410832,
      "learning_rate": 3.0181574934903745e-06,
      "loss": 0.1988,
      "step": 2934
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.72977863662266,
      "learning_rate": 3.0149099385967262e-06,
      "loss": 0.2142,
      "step": 2935
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7745797854596633,
      "learning_rate": 3.0116633775374237e-06,
      "loss": 0.2154,
      "step": 2936
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.762003901722593,
      "learning_rate": 3.0084178119378584e-06,
      "loss": 0.2031,
      "step": 2937
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7094678757226672,
      "learning_rate": 3.0051732434229185e-06,
      "loss": 0.2044,
      "step": 2938
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7052626130686762,
      "learning_rate": 3.001929673616999e-06,
      "loss": 0.2217,
      "step": 2939
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7233406567995976,
      "learning_rate": 2.998687104143991e-06,
      "loss": 0.2038,
      "step": 2940
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7102887364062569,
      "learning_rate": 2.995445536627282e-06,
      "loss": 0.1758,
      "step": 2941
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7210692436148456,
      "learning_rate": 2.992204972689765e-06,
      "loss": 0.2155,
      "step": 2942
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.752948051670872,
      "learning_rate": 2.988965413953825e-06,
      "loss": 0.2171,
      "step": 2943
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7638260329177536,
      "learning_rate": 2.985726862041346e-06,
      "loss": 0.213,
      "step": 2944
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7171603352707192,
      "learning_rate": 2.9824893185737046e-06,
      "loss": 0.2031,
      "step": 2945
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7843567050070495,
      "learning_rate": 2.9792527851717805e-06,
      "loss": 0.2287,
      "step": 2946
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7507985934832759,
      "learning_rate": 2.976017263455939e-06,
      "loss": 0.1998,
      "step": 2947
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7554017286582617,
      "learning_rate": 2.972782755046043e-06,
      "loss": 0.2057,
      "step": 2948
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7439385942468512,
      "learning_rate": 2.969549261561447e-06,
      "loss": 0.2115,
      "step": 2949
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7288077432749234,
      "learning_rate": 2.966316784621e-06,
      "loss": 0.2099,
      "step": 2950
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.744521999948039,
      "learning_rate": 2.9630853258430376e-06,
      "loss": 0.2026,
      "step": 2951
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7109440706909476,
      "learning_rate": 2.9598548868453904e-06,
      "loss": 0.2025,
      "step": 2952
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7405222712303039,
      "learning_rate": 2.956625469245372e-06,
      "loss": 0.2104,
      "step": 2953
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7479849477233048,
      "learning_rate": 2.9533970746597935e-06,
      "loss": 0.2195,
      "step": 2954
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7728766827032058,
      "learning_rate": 2.950169704704945e-06,
      "loss": 0.2018,
      "step": 2955
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7182631908288256,
      "learning_rate": 2.9469433609966125e-06,
      "loss": 0.2177,
      "step": 2956
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7359567826077259,
      "learning_rate": 2.943718045150058e-06,
      "loss": 0.2133,
      "step": 2957
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7866878937052147,
      "learning_rate": 2.9404937587800374e-06,
      "loss": 0.2066,
      "step": 2958
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7011446857195588,
      "learning_rate": 2.9372705035007863e-06,
      "loss": 0.2018,
      "step": 2959
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.721346092434293,
      "learning_rate": 2.9340482809260294e-06,
      "loss": 0.1988,
      "step": 2960
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7709926118205965,
      "learning_rate": 2.9308270926689663e-06,
      "loss": 0.2207,
      "step": 2961
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7292951829540297,
      "learning_rate": 2.9276069403422846e-06,
      "loss": 0.2072,
      "step": 2962
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7651796358298097,
      "learning_rate": 2.924387825558155e-06,
      "loss": 0.2185,
      "step": 2963
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7154917903696221,
      "learning_rate": 2.9211697499282223e-06,
      "loss": 0.2014,
      "step": 2964
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7702313934402162,
      "learning_rate": 2.9179527150636155e-06,
      "loss": 0.2223,
      "step": 2965
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7298017024396053,
      "learning_rate": 2.9147367225749425e-06,
      "loss": 0.1989,
      "step": 2966
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.740586198687809,
      "learning_rate": 2.9115217740722866e-06,
      "loss": 0.2108,
      "step": 2967
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7487938792933494,
      "learning_rate": 2.90830787116521e-06,
      "loss": 0.2081,
      "step": 2968
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7227465337425859,
      "learning_rate": 2.9050950154627546e-06,
      "loss": 0.2112,
      "step": 2969
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7302843356729529,
      "learning_rate": 2.90188320857343e-06,
      "loss": 0.2065,
      "step": 2970
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7505093267174204,
      "learning_rate": 2.898672452105231e-06,
      "loss": 0.2222,
      "step": 2971
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7185915016053566,
      "learning_rate": 2.895462747665617e-06,
      "loss": 0.2008,
      "step": 2972
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7980338781000522,
      "learning_rate": 2.892254096861529e-06,
      "loss": 0.215,
      "step": 2973
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.757914983313988,
      "learning_rate": 2.889046501299373e-06,
      "loss": 0.2125,
      "step": 2974
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7489259254355982,
      "learning_rate": 2.885839962585032e-06,
      "loss": 0.2233,
      "step": 2975
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7425079675403825,
      "learning_rate": 2.882634482323856e-06,
      "loss": 0.2181,
      "step": 2976
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7483553956225262,
      "learning_rate": 2.8794300621206695e-06,
      "loss": 0.2052,
      "step": 2977
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7632396968331354,
      "learning_rate": 2.8762267035797607e-06,
      "loss": 0.221,
      "step": 2978
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7566481032106919,
      "learning_rate": 2.873024408304894e-06,
      "loss": 0.2178,
      "step": 2979
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.765843949208539,
      "learning_rate": 2.8698231778992925e-06,
      "loss": 0.2108,
      "step": 2980
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7592604129818873,
      "learning_rate": 2.866623013965655e-06,
      "loss": 0.206,
      "step": 2981
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7129284368316611,
      "learning_rate": 2.8634239181061384e-06,
      "loss": 0.2064,
      "step": 2982
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7748333876234029,
      "learning_rate": 2.8602258919223703e-06,
      "loss": 0.2168,
      "step": 2983
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7599184404900995,
      "learning_rate": 2.857028937015438e-06,
      "loss": 0.206,
      "step": 2984
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7199346463706653,
      "learning_rate": 2.8538330549858993e-06,
      "loss": 0.2048,
      "step": 2985
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7798895445422152,
      "learning_rate": 2.850638247433768e-06,
      "loss": 0.2162,
      "step": 2986
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.800645232996667,
      "learning_rate": 2.8474445159585235e-06,
      "loss": 0.2298,
      "step": 2987
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7693735859770735,
      "learning_rate": 2.8442518621591085e-06,
      "loss": 0.224,
      "step": 2988
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7832670688810665,
      "learning_rate": 2.8410602876339214e-06,
      "loss": 0.2141,
      "step": 2989
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7556564120453176,
      "learning_rate": 2.837869793980823e-06,
      "loss": 0.2041,
      "step": 2990
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7431568299780705,
      "learning_rate": 2.834680382797129e-06,
      "loss": 0.2077,
      "step": 2991
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7425883898464428,
      "learning_rate": 2.831492055679622e-06,
      "loss": 0.2177,
      "step": 2992
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7370144290269919,
      "learning_rate": 2.828304814224532e-06,
      "loss": 0.2066,
      "step": 2993
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7686704545284864,
      "learning_rate": 2.8251186600275533e-06,
      "loss": 0.2071,
      "step": 2994
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7256580530173088,
      "learning_rate": 2.821933594683829e-06,
      "loss": 0.2082,
      "step": 2995
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7375059938982201,
      "learning_rate": 2.8187496197879637e-06,
      "loss": 0.1963,
      "step": 2996
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7191731747306637,
      "learning_rate": 2.815566736934009e-06,
      "loss": 0.1806,
      "step": 2997
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7510259159683232,
      "learning_rate": 2.8123849477154808e-06,
      "loss": 0.2158,
      "step": 2998
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7623146265128755,
      "learning_rate": 2.8092042537253306e-06,
      "loss": 0.2051,
      "step": 2999
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7278045455976656,
      "learning_rate": 2.806024656555978e-06,
      "loss": 0.1987,
      "step": 3000
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.791611722667823,
      "learning_rate": 2.8028461577992837e-06,
      "loss": 0.2253,
      "step": 3001
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7201138394416821,
      "learning_rate": 2.799668759046564e-06,
      "loss": 0.2059,
      "step": 3002
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7333671755740995,
      "learning_rate": 2.796492461888578e-06,
      "loss": 0.2173,
      "step": 3003
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8278320056476344,
      "learning_rate": 2.7933172679155427e-06,
      "loss": 0.217,
      "step": 3004
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7269680930521324,
      "learning_rate": 2.790143178717113e-06,
      "loss": 0.1924,
      "step": 3005
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7179271475678363,
      "learning_rate": 2.786970195882398e-06,
      "loss": 0.2052,
      "step": 3006
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7551478525815276,
      "learning_rate": 2.7837983209999484e-06,
      "loss": 0.2181,
      "step": 3007
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7793288400911674,
      "learning_rate": 2.7806275556577624e-06,
      "loss": 0.2162,
      "step": 3008
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8075467370363487,
      "learning_rate": 2.777457901443279e-06,
      "loss": 0.2263,
      "step": 3009
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8191972307428367,
      "learning_rate": 2.7742893599433875e-06,
      "loss": 0.2134,
      "step": 3010
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.6956753051635305,
      "learning_rate": 2.7711219327444174e-06,
      "loss": 0.1994,
      "step": 3011
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7253897709586433,
      "learning_rate": 2.7679556214321357e-06,
      "loss": 0.2013,
      "step": 3012
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7652404484748827,
      "learning_rate": 2.764790427591759e-06,
      "loss": 0.2194,
      "step": 3013
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7227289340691821,
      "learning_rate": 2.761626352807938e-06,
      "loss": 0.201,
      "step": 3014
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7488057308731549,
      "learning_rate": 2.7584633986647645e-06,
      "loss": 0.1999,
      "step": 3015
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.6938967514298032,
      "learning_rate": 2.7553015667457694e-06,
      "loss": 0.2109,
      "step": 3016
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7226703752898518,
      "learning_rate": 2.752140858633924e-06,
      "loss": 0.1883,
      "step": 3017
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7225662042856644,
      "learning_rate": 2.748981275911633e-06,
      "loss": 0.2088,
      "step": 3018
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.743145897193266,
      "learning_rate": 2.745822820160744e-06,
      "loss": 0.2087,
      "step": 3019
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7350562114149877,
      "learning_rate": 2.7426654929625307e-06,
      "loss": 0.1963,
      "step": 3020
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7776317116563406,
      "learning_rate": 2.739509295897713e-06,
      "loss": 0.2181,
      "step": 3021
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.821772878697512,
      "learning_rate": 2.736354230546437e-06,
      "loss": 0.2378,
      "step": 3022
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7753288266221179,
      "learning_rate": 2.733200298488284e-06,
      "loss": 0.1886,
      "step": 3023
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7517529779546944,
      "learning_rate": 2.7300475013022666e-06,
      "loss": 0.2083,
      "step": 3024
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7950794188245012,
      "learning_rate": 2.7268958405668356e-06,
      "loss": 0.2146,
      "step": 3025
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7771554730346708,
      "learning_rate": 2.7237453178598645e-06,
      "loss": 0.213,
      "step": 3026
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7523535939608917,
      "learning_rate": 2.7205959347586642e-06,
      "loss": 0.1954,
      "step": 3027
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7820472698154853,
      "learning_rate": 2.7174476928399685e-06,
      "loss": 0.2208,
      "step": 3028
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7560057706683606,
      "learning_rate": 2.714300593679946e-06,
      "loss": 0.2029,
      "step": 3029
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8256966082438671,
      "learning_rate": 2.71115463885419e-06,
      "loss": 0.2268,
      "step": 3030
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7174055134256792,
      "learning_rate": 2.70800982993772e-06,
      "loss": 0.2137,
      "step": 3031
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7609382429659649,
      "learning_rate": 2.7048661685049814e-06,
      "loss": 0.2093,
      "step": 3032
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7726863651707669,
      "learning_rate": 2.701723656129851e-06,
      "loss": 0.2144,
      "step": 3033
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.740781983618483,
      "learning_rate": 2.698582294385621e-06,
      "loss": 0.2051,
      "step": 3034
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7591873597705101,
      "learning_rate": 2.695442084845016e-06,
      "loss": 0.1936,
      "step": 3035
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7752108801881654,
      "learning_rate": 2.6923030290801817e-06,
      "loss": 0.226,
      "step": 3036
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7596859454129754,
      "learning_rate": 2.6891651286626823e-06,
      "loss": 0.2085,
      "step": 3037
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7041324117557398,
      "learning_rate": 2.6860283851635067e-06,
      "loss": 0.2023,
      "step": 3038
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7521851438144769,
      "learning_rate": 2.6828928001530617e-06,
      "loss": 0.2009,
      "step": 3039
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7341476328933613,
      "learning_rate": 2.679758375201181e-06,
      "loss": 0.2045,
      "step": 3040
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7716144259769135,
      "learning_rate": 2.676625111877108e-06,
      "loss": 0.2094,
      "step": 3041
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7225665635601957,
      "learning_rate": 2.6734930117495133e-06,
      "loss": 0.1902,
      "step": 3042
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7639895733622228,
      "learning_rate": 2.670362076386478e-06,
      "loss": 0.2175,
      "step": 3043
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.759330257118388,
      "learning_rate": 2.6672323073555074e-06,
      "loss": 0.2103,
      "step": 3044
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8200480665961722,
      "learning_rate": 2.6641037062235143e-06,
      "loss": 0.2333,
      "step": 3045
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.700936660056242,
      "learning_rate": 2.6609762745568358e-06,
      "loss": 0.1975,
      "step": 3046
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7314043549270267,
      "learning_rate": 2.657850013921216e-06,
      "loss": 0.2131,
      "step": 3047
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7346482005760745,
      "learning_rate": 2.6547249258818162e-06,
      "loss": 0.2064,
      "step": 3048
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7420499910987959,
      "learning_rate": 2.6516010120032094e-06,
      "loss": 0.2092,
      "step": 3049
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7719672742251045,
      "learning_rate": 2.6484782738493843e-06,
      "loss": 0.2107,
      "step": 3050
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7164673195362992,
      "learning_rate": 2.6453567129837333e-06,
      "loss": 0.2008,
      "step": 3051
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7410484577679715,
      "learning_rate": 2.6422363309690702e-06,
      "loss": 0.1972,
      "step": 3052
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.721095310899194,
      "learning_rate": 2.6391171293676077e-06,
      "loss": 0.1955,
      "step": 3053
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7476131577097445,
      "learning_rate": 2.6359991097409767e-06,
      "loss": 0.1969,
      "step": 3054
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7291512502939435,
      "learning_rate": 2.6328822736502102e-06,
      "loss": 0.2051,
      "step": 3055
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7885366362242477,
      "learning_rate": 2.6297666226557505e-06,
      "loss": 0.2421,
      "step": 3056
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7470991836478464,
      "learning_rate": 2.6266521583174443e-06,
      "loss": 0.2027,
      "step": 3057
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7647600296289994,
      "learning_rate": 2.6235388821945497e-06,
      "loss": 0.2116,
      "step": 3058
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7601417074591138,
      "learning_rate": 2.6204267958457288e-06,
      "loss": 0.1956,
      "step": 3059
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7512545890887556,
      "learning_rate": 2.6173159008290407e-06,
      "loss": 0.2167,
      "step": 3060
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8101378960882514,
      "learning_rate": 2.614206198701958e-06,
      "loss": 0.2373,
      "step": 3061
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.74844742766,
      "learning_rate": 2.61109769102135e-06,
      "loss": 0.196,
      "step": 3062
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7478599167664598,
      "learning_rate": 2.607990379343489e-06,
      "loss": 0.1986,
      "step": 3063
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7975727110413166,
      "learning_rate": 2.604884265224048e-06,
      "loss": 0.2193,
      "step": 3064
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7375546892986827,
      "learning_rate": 2.6017793502181043e-06,
      "loss": 0.2121,
      "step": 3065
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7161267854537973,
      "learning_rate": 2.598675635880129e-06,
      "loss": 0.1911,
      "step": 3066
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7442683058085093,
      "learning_rate": 2.595573123763998e-06,
      "loss": 0.2011,
      "step": 3067
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.750494277124273,
      "learning_rate": 2.59247181542298e-06,
      "loss": 0.1904,
      "step": 3068
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7938914300359824,
      "learning_rate": 2.5893717124097463e-06,
      "loss": 0.2237,
      "step": 3069
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7885991144271078,
      "learning_rate": 2.5862728162763595e-06,
      "loss": 0.2095,
      "step": 3070
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7163965948241163,
      "learning_rate": 2.5831751285742825e-06,
      "loss": 0.1991,
      "step": 3071
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.769870147717939,
      "learning_rate": 2.580078650854366e-06,
      "loss": 0.2336,
      "step": 3072
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7444651645280621,
      "learning_rate": 2.576983384666867e-06,
      "loss": 0.2104,
      "step": 3073
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7588830550056228,
      "learning_rate": 2.573889331561423e-06,
      "loss": 0.2195,
      "step": 3074
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7468782677603185,
      "learning_rate": 2.5707964930870753e-06,
      "loss": 0.1956,
      "step": 3075
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7427153663812084,
      "learning_rate": 2.5677048707922482e-06,
      "loss": 0.2103,
      "step": 3076
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7558793142577239,
      "learning_rate": 2.564614466224763e-06,
      "loss": 0.2209,
      "step": 3077
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7792648523388702,
      "learning_rate": 2.5615252809318287e-06,
      "loss": 0.2151,
      "step": 3078
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.69124910264825,
      "learning_rate": 2.558437316460044e-06,
      "loss": 0.1783,
      "step": 3079
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7246730498144728,
      "learning_rate": 2.555350574355395e-06,
      "loss": 0.2021,
      "step": 3080
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8007383070002778,
      "learning_rate": 2.5522650561632613e-06,
      "loss": 0.222,
      "step": 3081
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.767196932673918,
      "learning_rate": 2.5491807634284016e-06,
      "loss": 0.1995,
      "step": 3082
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7863105776671981,
      "learning_rate": 2.5460976976949686e-06,
      "loss": 0.2196,
      "step": 3083
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7527846655716884,
      "learning_rate": 2.543015860506499e-06,
      "loss": 0.1841,
      "step": 3084
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7220544118443742,
      "learning_rate": 2.5399352534059087e-06,
      "loss": 0.1967,
      "step": 3085
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8200586604500657,
      "learning_rate": 2.536855877935507e-06,
      "loss": 0.2313,
      "step": 3086
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7994863133584942,
      "learning_rate": 2.5337777356369785e-06,
      "loss": 0.2241,
      "step": 3087
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.731280455095426,
      "learning_rate": 2.5307008280513956e-06,
      "loss": 0.2078,
      "step": 3088
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7610873777546675,
      "learning_rate": 2.5276251567192063e-06,
      "loss": 0.2282,
      "step": 3089
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.70654823166269,
      "learning_rate": 2.524550723180249e-06,
      "loss": 0.1954,
      "step": 3090
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7400083565044152,
      "learning_rate": 2.5214775289737335e-06,
      "loss": 0.2234,
      "step": 3091
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7419986410276243,
      "learning_rate": 2.5184055756382566e-06,
      "loss": 0.2154,
      "step": 3092
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8157188754375421,
      "learning_rate": 2.515334864711786e-06,
      "loss": 0.1897,
      "step": 3093
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7435801576574778,
      "learning_rate": 2.5122653977316758e-06,
      "loss": 0.198,
      "step": 3094
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8074121692850749,
      "learning_rate": 2.5091971762346505e-06,
      "loss": 0.2304,
      "step": 3095
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7713699011069735,
      "learning_rate": 2.5061302017568153e-06,
      "loss": 0.2089,
      "step": 3096
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8142200818850869,
      "learning_rate": 2.503064475833646e-06,
      "loss": 0.2184,
      "step": 3097
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7492998991044925,
      "learning_rate": 2.5000000000000015e-06,
      "loss": 0.2107,
      "step": 3098
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7767552876493822,
      "learning_rate": 2.4969367757901054e-06,
      "loss": 0.2173,
      "step": 3099
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7764476296523197,
      "learning_rate": 2.4938748047375637e-06,
      "loss": 0.2156,
      "step": 3100
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7654837521169429,
      "learning_rate": 2.4908140883753468e-06,
      "loss": 0.2071,
      "step": 3101
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8560403340195889,
      "learning_rate": 2.487754628235805e-06,
      "loss": 0.2265,
      "step": 3102
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7391263130176627,
      "learning_rate": 2.484696425850653e-06,
      "loss": 0.193,
      "step": 3103
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7625052801701498,
      "learning_rate": 2.481639482750979e-06,
      "loss": 0.2099,
      "step": 3104
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7945914108971434,
      "learning_rate": 2.478583800467237e-06,
      "loss": 0.2125,
      "step": 3105
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7325889304551797,
      "learning_rate": 2.4755293805292572e-06,
      "loss": 0.2102,
      "step": 3106
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.809961525217763,
      "learning_rate": 2.4724762244662304e-06,
      "loss": 0.2246,
      "step": 3107
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7498599056155619,
      "learning_rate": 2.469424333806718e-06,
      "loss": 0.2091,
      "step": 3108
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7672965739446735,
      "learning_rate": 2.466373710078651e-06,
      "loss": 0.2088,
      "step": 3109
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7700133630002989,
      "learning_rate": 2.4633243548093195e-06,
      "loss": 0.2166,
      "step": 3110
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.764632148988597,
      "learning_rate": 2.4602762695253825e-06,
      "loss": 0.2047,
      "step": 3111
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7421868510906762,
      "learning_rate": 2.4572294557528604e-06,
      "loss": 0.2107,
      "step": 3112
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8021860357732706,
      "learning_rate": 2.454183915017142e-06,
      "loss": 0.2319,
      "step": 3113
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7670052943247235,
      "learning_rate": 2.4511396488429727e-06,
      "loss": 0.2335,
      "step": 3114
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7708368254809405,
      "learning_rate": 2.448096658754467e-06,
      "loss": 0.2317,
      "step": 3115
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7838410887793326,
      "learning_rate": 2.4450549462750917e-06,
      "loss": 0.2192,
      "step": 3116
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7843604238338782,
      "learning_rate": 2.442014512927682e-06,
      "loss": 0.2197,
      "step": 3117
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7334198085473491,
      "learning_rate": 2.4389753602344298e-06,
      "loss": 0.2065,
      "step": 3118
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7940513941172386,
      "learning_rate": 2.4359374897168823e-06,
      "loss": 0.2255,
      "step": 3119
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.712149841405434,
      "learning_rate": 2.4329009028959484e-06,
      "loss": 0.1964,
      "step": 3120
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.81237605975469,
      "learning_rate": 2.4298656012918963e-06,
      "loss": 0.2311,
      "step": 3121
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7850798099146858,
      "learning_rate": 2.4268315864243447e-06,
      "loss": 0.2219,
      "step": 3122
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7735260198627338,
      "learning_rate": 2.423798859812275e-06,
      "loss": 0.2268,
      "step": 3123
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7073848859853125,
      "learning_rate": 2.420767422974018e-06,
      "loss": 0.2022,
      "step": 3124
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7954569927041096,
      "learning_rate": 2.4177372774272627e-06,
      "loss": 0.2307,
      "step": 3125
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7538614894870772,
      "learning_rate": 2.4147084246890478e-06,
      "loss": 0.2258,
      "step": 3126
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7566494142487223,
      "learning_rate": 2.411680866275772e-06,
      "loss": 0.2053,
      "step": 3127
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7432819933396321,
      "learning_rate": 2.4086546037031734e-06,
      "loss": 0.2065,
      "step": 3128
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7431901489786128,
      "learning_rate": 2.405629638486354e-06,
      "loss": 0.211,
      "step": 3129
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.78719794548637,
      "learning_rate": 2.402605972139759e-06,
      "loss": 0.2053,
      "step": 3130
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7314762403210455,
      "learning_rate": 2.399583606177185e-06,
      "loss": 0.2005,
      "step": 3131
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7398325247650323,
      "learning_rate": 2.396562542111781e-06,
      "loss": 0.1943,
      "step": 3132
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7632220063603623,
      "learning_rate": 2.393542781456038e-06,
      "loss": 0.2065,
      "step": 3133
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7919923234686083,
      "learning_rate": 2.3905243257218007e-06,
      "loss": 0.2171,
      "step": 3134
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8061086472625066,
      "learning_rate": 2.387507176420256e-06,
      "loss": 0.2096,
      "step": 3135
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7372498916567468,
      "learning_rate": 2.3844913350619382e-06,
      "loss": 0.2078,
      "step": 3136
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7224847975644707,
      "learning_rate": 2.381476803156724e-06,
      "loss": 0.2016,
      "step": 3137
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.803385125529838,
      "learning_rate": 2.3784635822138424e-06,
      "loss": 0.2228,
      "step": 3138
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7468836406572271,
      "learning_rate": 2.3754516737418563e-06,
      "loss": 0.2069,
      "step": 3139
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7173470148660341,
      "learning_rate": 2.37244107924868e-06,
      "loss": 0.1929,
      "step": 3140
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7802187163574884,
      "learning_rate": 2.3694318002415627e-06,
      "loss": 0.2299,
      "step": 3141
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7503072986284147,
      "learning_rate": 2.366423838227102e-06,
      "loss": 0.214,
      "step": 3142
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7572651729618118,
      "learning_rate": 2.3634171947112307e-06,
      "loss": 0.2321,
      "step": 3143
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7588168024903805,
      "learning_rate": 2.3604118711992237e-06,
      "loss": 0.2032,
      "step": 3144
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7750939804488411,
      "learning_rate": 2.357407869195692e-06,
      "loss": 0.212,
      "step": 3145
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7210928425275561,
      "learning_rate": 2.354405190204592e-06,
      "loss": 0.1992,
      "step": 3146
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7175525861848969,
      "learning_rate": 2.3514038357292103e-06,
      "loss": 0.2013,
      "step": 3147
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7593574281038484,
      "learning_rate": 2.348403807272176e-06,
      "loss": 0.2129,
      "step": 3148
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7622212145196505,
      "learning_rate": 2.345405106335449e-06,
      "loss": 0.211,
      "step": 3149
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7509364863273627,
      "learning_rate": 2.342407734420331e-06,
      "loss": 0.2123,
      "step": 3150
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7469064405154215,
      "learning_rate": 2.3394116930274523e-06,
      "loss": 0.2179,
      "step": 3151
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7704988713787991,
      "learning_rate": 2.3364169836567805e-06,
      "loss": 0.2113,
      "step": 3152
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.753604500948749,
      "learning_rate": 2.3334236078076126e-06,
      "loss": 0.2171,
      "step": 3153
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7442843630041738,
      "learning_rate": 2.3304315669785853e-06,
      "loss": 0.1993,
      "step": 3154
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7462641482893393,
      "learning_rate": 2.327440862667658e-06,
      "loss": 0.213,
      "step": 3155
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7401522039496596,
      "learning_rate": 2.3244514963721276e-06,
      "loss": 0.2046,
      "step": 3156
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7574184148545041,
      "learning_rate": 2.32146346958862e-06,
      "loss": 0.2129,
      "step": 3157
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7228747898690636,
      "learning_rate": 2.318476783813088e-06,
      "loss": 0.2184,
      "step": 3158
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.746520338997165,
      "learning_rate": 2.315491440540815e-06,
      "loss": 0.2041,
      "step": 3159
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7754640718565532,
      "learning_rate": 2.3125074412664085e-06,
      "loss": 0.2305,
      "step": 3160
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7427562436502468,
      "learning_rate": 2.30952478748381e-06,
      "loss": 0.2145,
      "step": 3161
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7437683971046725,
      "learning_rate": 2.306543480686281e-06,
      "loss": 0.2155,
      "step": 3162
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7907200799012929,
      "learning_rate": 2.3035635223664136e-06,
      "loss": 0.2347,
      "step": 3163
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7515797438511479,
      "learning_rate": 2.30058491401612e-06,
      "loss": 0.2116,
      "step": 3164
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7186329311704311,
      "learning_rate": 2.2976076571266424e-06,
      "loss": 0.1944,
      "step": 3165
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7887853714938992,
      "learning_rate": 2.294631753188541e-06,
      "loss": 0.2317,
      "step": 3166
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7658346156393909,
      "learning_rate": 2.2916572036917e-06,
      "loss": 0.2075,
      "step": 3167
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7144136934231304,
      "learning_rate": 2.288684010125325e-06,
      "loss": 0.2094,
      "step": 3168
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7003455342458662,
      "learning_rate": 2.2857121739779474e-06,
      "loss": 0.1957,
      "step": 3169
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7441309897571764,
      "learning_rate": 2.2827416967374115e-06,
      "loss": 0.2091,
      "step": 3170
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7447138471675074,
      "learning_rate": 2.2797725798908897e-06,
      "loss": 0.2108,
      "step": 3171
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7198576721622232,
      "learning_rate": 2.2768048249248648e-06,
      "loss": 0.1951,
      "step": 3172
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7352757409379683,
      "learning_rate": 2.2738384333251447e-06,
      "loss": 0.2049,
      "step": 3173
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7710878584342173,
      "learning_rate": 2.270873406576849e-06,
      "loss": 0.2087,
      "step": 3174
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7530687665196977,
      "learning_rate": 2.2679097461644204e-06,
      "loss": 0.2081,
      "step": 3175
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7467260172549448,
      "learning_rate": 2.2649474535716128e-06,
      "loss": 0.2038,
      "step": 3176
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7665421535508036,
      "learning_rate": 2.2619865302814955e-06,
      "loss": 0.2189,
      "step": 3177
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7206125491322567,
      "learning_rate": 2.2590269777764516e-06,
      "loss": 0.1964,
      "step": 3178
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7594264633103339,
      "learning_rate": 2.256068797538182e-06,
      "loss": 0.219,
      "step": 3179
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7783615056757933,
      "learning_rate": 2.2531119910476994e-06,
      "loss": 0.2298,
      "step": 3180
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7379562554613774,
      "learning_rate": 2.250156559785324e-06,
      "loss": 0.1986,
      "step": 3181
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7260029586546394,
      "learning_rate": 2.247202505230695e-06,
      "loss": 0.1971,
      "step": 3182
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7372298590913394,
      "learning_rate": 2.2442498288627555e-06,
      "loss": 0.211,
      "step": 3183
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7926418690725786,
      "learning_rate": 2.241298532159763e-06,
      "loss": 0.2203,
      "step": 3184
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7335082565434189,
      "learning_rate": 2.2383486165992796e-06,
      "loss": 0.1893,
      "step": 3185
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7485638644762004,
      "learning_rate": 2.2354000836581834e-06,
      "loss": 0.1906,
      "step": 3186
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7558391913961519,
      "learning_rate": 2.2324529348126523e-06,
      "loss": 0.2046,
      "step": 3187
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7654155218111017,
      "learning_rate": 2.229507171538178e-06,
      "loss": 0.2125,
      "step": 3188
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7270145711073304,
      "learning_rate": 2.226562795309552e-06,
      "loss": 0.193,
      "step": 3189
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7566917440736427,
      "learning_rate": 2.223619807600878e-06,
      "loss": 0.2187,
      "step": 3190
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7955868483743416,
      "learning_rate": 2.2206782098855605e-06,
      "loss": 0.209,
      "step": 3191
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7574899763221778,
      "learning_rate": 2.217738003636309e-06,
      "loss": 0.2151,
      "step": 3192
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.80883744152384,
      "learning_rate": 2.214799190325133e-06,
      "loss": 0.2037,
      "step": 3193
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7221615994988417,
      "learning_rate": 2.2118617714233524e-06,
      "loss": 0.1924,
      "step": 3194
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8706573370273565,
      "learning_rate": 2.208925748401582e-06,
      "loss": 0.2161,
      "step": 3195
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7627983816876147,
      "learning_rate": 2.205991122729742e-06,
      "loss": 0.2208,
      "step": 3196
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8037275356971975,
      "learning_rate": 2.2030578958770484e-06,
      "loss": 0.2236,
      "step": 3197
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7929653807830882,
      "learning_rate": 2.2001260693120236e-06,
      "loss": 0.2231,
      "step": 3198
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7780845595463892,
      "learning_rate": 2.1971956445024826e-06,
      "loss": 0.2196,
      "step": 3199
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7544486389482408,
      "learning_rate": 2.1942666229155424e-06,
      "loss": 0.2117,
      "step": 3200
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7880772140048802,
      "learning_rate": 2.1913390060176135e-06,
      "loss": 0.218,
      "step": 3201
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7528240062252288,
      "learning_rate": 2.1884127952744095e-06,
      "loss": 0.2186,
      "step": 3202
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7601175059483993,
      "learning_rate": 2.185487992150933e-06,
      "loss": 0.2171,
      "step": 3203
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7503465169251279,
      "learning_rate": 2.1825645981114862e-06,
      "loss": 0.2101,
      "step": 3204
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7494596518868024,
      "learning_rate": 2.179642614619668e-06,
      "loss": 0.196,
      "step": 3205
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7229990891866096,
      "learning_rate": 2.1767220431383652e-06,
      "loss": 0.2022,
      "step": 3206
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7484487378550455,
      "learning_rate": 2.173802885129761e-06,
      "loss": 0.1975,
      "step": 3207
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7916464517497531,
      "learning_rate": 2.1708851420553277e-06,
      "loss": 0.2199,
      "step": 3208
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7411610151629401,
      "learning_rate": 2.1679688153758373e-06,
      "loss": 0.209,
      "step": 3209
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8065876930605408,
      "learning_rate": 2.1650539065513415e-06,
      "loss": 0.2241,
      "step": 3210
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7972313573962964,
      "learning_rate": 2.162140417041193e-06,
      "loss": 0.1982,
      "step": 3211
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7861826589991693,
      "learning_rate": 2.159228348304025e-06,
      "loss": 0.2142,
      "step": 3212
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7350811476087701,
      "learning_rate": 2.156317701797766e-06,
      "loss": 0.1979,
      "step": 3213
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7739530932694312,
      "learning_rate": 2.153408478979628e-06,
      "loss": 0.2039,
      "step": 3214
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7414928976740293,
      "learning_rate": 2.150500681306113e-06,
      "loss": 0.2013,
      "step": 3215
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7777723666647328,
      "learning_rate": 2.147594310233009e-06,
      "loss": 0.2068,
      "step": 3216
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7719315027252166,
      "learning_rate": 2.144689367215388e-06,
      "loss": 0.2095,
      "step": 3217
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7335250463418129,
      "learning_rate": 2.141785853707607e-06,
      "loss": 0.2095,
      "step": 3218
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7367516500480795,
      "learning_rate": 2.1388837711633114e-06,
      "loss": 0.1974,
      "step": 3219
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7574804566475072,
      "learning_rate": 2.1359831210354247e-06,
      "loss": 0.2067,
      "step": 3220
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7828414779437692,
      "learning_rate": 2.133083904776159e-06,
      "loss": 0.2286,
      "step": 3221
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7534439909983763,
      "learning_rate": 2.130186123837002e-06,
      "loss": 0.2053,
      "step": 3222
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.763356413134015,
      "learning_rate": 2.12728977966873e-06,
      "loss": 0.2147,
      "step": 3223
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.755905897336596,
      "learning_rate": 2.124394873721394e-06,
      "loss": 0.2184,
      "step": 3224
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.6993660429751029,
      "learning_rate": 2.121501407444328e-06,
      "loss": 0.1907,
      "step": 3225
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7776704385051363,
      "learning_rate": 2.118609382286142e-06,
      "loss": 0.2102,
      "step": 3226
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7750746489616519,
      "learning_rate": 2.1157187996947293e-06,
      "loss": 0.2116,
      "step": 3227
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.737489266421375,
      "learning_rate": 2.1128296611172593e-06,
      "loss": 0.2157,
      "step": 3228
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7749006556032676,
      "learning_rate": 2.1099419680001754e-06,
      "loss": 0.2158,
      "step": 3229
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8314670052976836,
      "learning_rate": 2.107055721789203e-06,
      "loss": 0.2197,
      "step": 3230
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7383699602784871,
      "learning_rate": 2.104170923929338e-06,
      "loss": 0.1988,
      "step": 3231
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8082579352309258,
      "learning_rate": 2.101287575864853e-06,
      "loss": 0.2234,
      "step": 3232
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7477355644748627,
      "learning_rate": 2.0984056790392926e-06,
      "loss": 0.2075,
      "step": 3233
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8023005117011535,
      "learning_rate": 2.0955252348954806e-06,
      "loss": 0.204,
      "step": 3234
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7634863277821057,
      "learning_rate": 2.092646244875507e-06,
      "loss": 0.2317,
      "step": 3235
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7526575318745047,
      "learning_rate": 2.08976871042074e-06,
      "loss": 0.2166,
      "step": 3236
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7673861661588937,
      "learning_rate": 2.086892632971812e-06,
      "loss": 0.2075,
      "step": 3237
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7218907182884803,
      "learning_rate": 2.0840180139686333e-06,
      "loss": 0.1914,
      "step": 3238
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7993527791934256,
      "learning_rate": 2.0811448548503793e-06,
      "loss": 0.2196,
      "step": 3239
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7656976590319509,
      "learning_rate": 2.0782731570554946e-06,
      "loss": 0.2112,
      "step": 3240
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7807565014641357,
      "learning_rate": 2.0754029220216924e-06,
      "loss": 0.2203,
      "step": 3241
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7667115227331415,
      "learning_rate": 2.0725341511859576e-06,
      "loss": 0.2135,
      "step": 3242
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7442393332870512,
      "learning_rate": 2.0696668459845354e-06,
      "loss": 0.2165,
      "step": 3243
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7714366268264105,
      "learning_rate": 2.066801007852944e-06,
      "loss": 0.2247,
      "step": 3244
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7419964469062659,
      "learning_rate": 2.0639366382259613e-06,
      "loss": 0.2107,
      "step": 3245
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7883286091561063,
      "learning_rate": 2.061073738537635e-06,
      "loss": 0.2223,
      "step": 3246
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7299543553455148,
      "learning_rate": 2.0582123102212726e-06,
      "loss": 0.1957,
      "step": 3247
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7241036882587043,
      "learning_rate": 2.0553523547094473e-06,
      "loss": 0.2064,
      "step": 3248
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7360886549142182,
      "learning_rate": 2.052493873433993e-06,
      "loss": 0.2326,
      "step": 3249
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7463438683258619,
      "learning_rate": 2.0496368678260094e-06,
      "loss": 0.1929,
      "step": 3250
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7956704888094355,
      "learning_rate": 2.0467813393158516e-06,
      "loss": 0.2172,
      "step": 3251
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7449528061893775,
      "learning_rate": 2.043927289333141e-06,
      "loss": 0.2211,
      "step": 3252
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.769193992473196,
      "learning_rate": 2.041074719306757e-06,
      "loss": 0.2175,
      "step": 3253
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7326681371878727,
      "learning_rate": 2.038223630664835e-06,
      "loss": 0.213,
      "step": 3254
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7769161334498391,
      "learning_rate": 2.0353740248347758e-06,
      "loss": 0.2101,
      "step": 3255
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7769508900036981,
      "learning_rate": 2.0325259032432264e-06,
      "loss": 0.2103,
      "step": 3256
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7257999424111584,
      "learning_rate": 2.0296792673161026e-06,
      "loss": 0.1989,
      "step": 3257
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7449899755869913,
      "learning_rate": 2.0268341184785674e-06,
      "loss": 0.2114,
      "step": 3258
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7745140532517841,
      "learning_rate": 2.0239904581550463e-06,
      "loss": 0.2049,
      "step": 3259
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7569423621941099,
      "learning_rate": 2.021148287769214e-06,
      "loss": 0.2156,
      "step": 3260
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7991198285783934,
      "learning_rate": 2.0183076087440044e-06,
      "loss": 0.2077,
      "step": 3261
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7722195567167679,
      "learning_rate": 2.015468422501599e-06,
      "loss": 0.214,
      "step": 3262
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.759663613666585,
      "learning_rate": 2.0126307304634383e-06,
      "loss": 0.206,
      "step": 3263
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7221061850001018,
      "learning_rate": 2.00979453405021e-06,
      "loss": 0.1931,
      "step": 3264
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7709560775719694,
      "learning_rate": 2.0069598346818537e-06,
      "loss": 0.2111,
      "step": 3265
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8038823650536693,
      "learning_rate": 2.004126633777559e-06,
      "loss": 0.2385,
      "step": 3266
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.748016085710794,
      "learning_rate": 2.0012949327557694e-06,
      "loss": 0.1881,
      "step": 3267
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7364641839024344,
      "learning_rate": 1.998464733034172e-06,
      "loss": 0.1844,
      "step": 3268
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8090503633111015,
      "learning_rate": 1.995636036029708e-06,
      "loss": 0.2215,
      "step": 3269
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7701899478343425,
      "learning_rate": 1.992808843158559e-06,
      "loss": 0.2159,
      "step": 3270
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7464219029433902,
      "learning_rate": 1.989983155836162e-06,
      "loss": 0.2031,
      "step": 3271
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7506640032405847,
      "learning_rate": 1.987158975477194e-06,
      "loss": 0.2184,
      "step": 3272
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7000230620703646,
      "learning_rate": 1.98433630349558e-06,
      "loss": 0.1855,
      "step": 3273
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7676231400140388,
      "learning_rate": 1.9815151413044865e-06,
      "loss": 0.2255,
      "step": 3274
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7828131512861543,
      "learning_rate": 1.9786954903163293e-06,
      "loss": 0.2021,
      "step": 3275
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7848657084752456,
      "learning_rate": 1.9758773519427664e-06,
      "loss": 0.2337,
      "step": 3276
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7745221463928189,
      "learning_rate": 1.9730607275946952e-06,
      "loss": 0.215,
      "step": 3277
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8066399158005753,
      "learning_rate": 1.9702456186822595e-06,
      "loss": 0.2147,
      "step": 3278
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7538234560497259,
      "learning_rate": 1.9674320266148404e-06,
      "loss": 0.2001,
      "step": 3279
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.6973934426520807,
      "learning_rate": 1.9646199528010625e-06,
      "loss": 0.1878,
      "step": 3280
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.74935807429014,
      "learning_rate": 1.961809398648787e-06,
      "loss": 0.2158,
      "step": 3281
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.788304651381274,
      "learning_rate": 1.959000365565119e-06,
      "loss": 0.2279,
      "step": 3282
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7656108955615736,
      "learning_rate": 1.956192854956397e-06,
      "loss": 0.2113,
      "step": 3283
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7349645275830896,
      "learning_rate": 1.9533868682282026e-06,
      "loss": 0.1955,
      "step": 3284
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7484801836801322,
      "learning_rate": 1.9505824067853485e-06,
      "loss": 0.2051,
      "step": 3285
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7596927583389118,
      "learning_rate": 1.947779472031891e-06,
      "loss": 0.2116,
      "step": 3286
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7632010650926448,
      "learning_rate": 1.944978065371115e-06,
      "loss": 0.2112,
      "step": 3287
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7911783197797484,
      "learning_rate": 1.9421781882055447e-06,
      "loss": 0.2272,
      "step": 3288
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7517589636644413,
      "learning_rate": 1.9393798419369345e-06,
      "loss": 0.2084,
      "step": 3289
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7721520619233503,
      "learning_rate": 1.936583027966279e-06,
      "loss": 0.2105,
      "step": 3290
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.733049469594586,
      "learning_rate": 1.9337877476937976e-06,
      "loss": 0.1846,
      "step": 3291
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.756208337713892,
      "learning_rate": 1.93099400251895e-06,
      "loss": 0.2146,
      "step": 3292
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7143473575703936,
      "learning_rate": 1.9282017938404202e-06,
      "loss": 0.1904,
      "step": 3293
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7290927948144775,
      "learning_rate": 1.9254111230561282e-06,
      "loss": 0.199,
      "step": 3294
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7920745283599991,
      "learning_rate": 1.922621991563222e-06,
      "loss": 0.2063,
      "step": 3295
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7333512399967124,
      "learning_rate": 1.9198344007580776e-06,
      "loss": 0.2047,
      "step": 3296
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7745327128602181,
      "learning_rate": 1.9170483520362997e-06,
      "loss": 0.2142,
      "step": 3297
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7365139543928448,
      "learning_rate": 1.9142638467927254e-06,
      "loss": 0.1992,
      "step": 3298
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7108221776973611,
      "learning_rate": 1.911480886421413e-06,
      "loss": 0.1986,
      "step": 3299
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7693828054116141,
      "learning_rate": 1.908699472315651e-06,
      "loss": 0.2163,
      "step": 3300
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.731360026923262,
      "learning_rate": 1.9059196058679558e-06,
      "loss": 0.222,
      "step": 3301
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7723308101234473,
      "learning_rate": 1.9031412884700612e-06,
      "loss": 0.2019,
      "step": 3302
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7503103908176669,
      "learning_rate": 1.9003645215129356e-06,
      "loss": 0.2169,
      "step": 3303
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7533877608743828,
      "learning_rate": 1.8975893063867635e-06,
      "loss": 0.1856,
      "step": 3304
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7425431033987205,
      "learning_rate": 1.8948156444809545e-06,
      "loss": 0.1911,
      "step": 3305
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7622979341570318,
      "learning_rate": 1.8920435371841394e-06,
      "loss": 0.2111,
      "step": 3306
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7365985770655021,
      "learning_rate": 1.8892729858841768e-06,
      "loss": 0.2019,
      "step": 3307
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7520266705958047,
      "learning_rate": 1.8865039919681377e-06,
      "loss": 0.212,
      "step": 3308
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7369001386050783,
      "learning_rate": 1.8837365568223204e-06,
      "loss": 0.203,
      "step": 3309
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7216999800599095,
      "learning_rate": 1.8809706818322375e-06,
      "loss": 0.1863,
      "step": 3310
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7725533439377321,
      "learning_rate": 1.8782063683826256e-06,
      "loss": 0.2191,
      "step": 3311
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7410855961405914,
      "learning_rate": 1.8754436178574358e-06,
      "loss": 0.2041,
      "step": 3312
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7195639624748751,
      "learning_rate": 1.8726824316398372e-06,
      "loss": 0.1915,
      "step": 3313
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7600580046381759,
      "learning_rate": 1.8699228111122146e-06,
      "loss": 0.2111,
      "step": 3314
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7514168212126224,
      "learning_rate": 1.8671647576561747e-06,
      "loss": 0.1909,
      "step": 3315
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7449808786698084,
      "learning_rate": 1.8644082726525313e-06,
      "loss": 0.19,
      "step": 3316
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.6968053419745567,
      "learning_rate": 1.8616533574813216e-06,
      "loss": 0.1905,
      "step": 3317
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7648206210298466,
      "learning_rate": 1.8589000135217882e-06,
      "loss": 0.1968,
      "step": 3318
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7721797154262587,
      "learning_rate": 1.8561482421523952e-06,
      "loss": 0.2004,
      "step": 3319
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8203699885824725,
      "learning_rate": 1.8533980447508138e-06,
      "loss": 0.2224,
      "step": 3320
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8053520214272623,
      "learning_rate": 1.8506494226939292e-06,
      "loss": 0.2091,
      "step": 3321
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7910402154867108,
      "learning_rate": 1.8479023773578354e-06,
      "loss": 0.2199,
      "step": 3322
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7403414045527448,
      "learning_rate": 1.845156910117843e-06,
      "loss": 0.2104,
      "step": 3323
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7367652388581323,
      "learning_rate": 1.8424130223484655e-06,
      "loss": 0.1997,
      "step": 3324
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7852910461985908,
      "learning_rate": 1.8396707154234306e-06,
      "loss": 0.22,
      "step": 3325
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7514885289749541,
      "learning_rate": 1.8369299907156745e-06,
      "loss": 0.2077,
      "step": 3326
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7962142476607124,
      "learning_rate": 1.8341908495973381e-06,
      "loss": 0.2137,
      "step": 3327
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.727391843847343,
      "learning_rate": 1.831453293439771e-06,
      "loss": 0.213,
      "step": 3328
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.740731139978489,
      "learning_rate": 1.8287173236135276e-06,
      "loss": 0.1951,
      "step": 3329
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7438333032172136,
      "learning_rate": 1.8259829414883728e-06,
      "loss": 0.2096,
      "step": 3330
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.720441519096238,
      "learning_rate": 1.8232501484332709e-06,
      "loss": 0.1926,
      "step": 3331
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7505893700138231,
      "learning_rate": 1.8205189458163964e-06,
      "loss": 0.2164,
      "step": 3332
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7622314694432534,
      "learning_rate": 1.8177893350051213e-06,
      "loss": 0.2159,
      "step": 3333
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.744664392734183,
      "learning_rate": 1.815061317366027e-06,
      "loss": 0.1913,
      "step": 3334
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7261375026473659,
      "learning_rate": 1.8123348942648928e-06,
      "loss": 0.196,
      "step": 3335
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.740475130588379,
      "learning_rate": 1.809610067066701e-06,
      "loss": 0.197,
      "step": 3336
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7708185262211967,
      "learning_rate": 1.8068868371356335e-06,
      "loss": 0.2153,
      "step": 3337
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7904403952122782,
      "learning_rate": 1.8041652058350768e-06,
      "loss": 0.2179,
      "step": 3338
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7949671051113232,
      "learning_rate": 1.8014451745276107e-06,
      "loss": 0.2143,
      "step": 3339
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7997326492315125,
      "learning_rate": 1.7987267445750224e-06,
      "loss": 0.2177,
      "step": 3340
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7609254287425091,
      "learning_rate": 1.7960099173382873e-06,
      "loss": 0.2058,
      "step": 3341
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7487462199277013,
      "learning_rate": 1.7932946941775882e-06,
      "loss": 0.2238,
      "step": 3342
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7842030698818896,
      "learning_rate": 1.7905810764522963e-06,
      "loss": 0.2239,
      "step": 3343
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7529391197589949,
      "learning_rate": 1.7878690655209862e-06,
      "loss": 0.228,
      "step": 3344
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7777120795526972,
      "learning_rate": 1.7851586627414237e-06,
      "loss": 0.2116,
      "step": 3345
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7689077039638613,
      "learning_rate": 1.78244986947057e-06,
      "loss": 0.223,
      "step": 3346
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7477578400247147,
      "learning_rate": 1.7797426870645802e-06,
      "loss": 0.2049,
      "step": 3347
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7804974527161129,
      "learning_rate": 1.7770371168788042e-06,
      "loss": 0.2293,
      "step": 3348
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7613495811930242,
      "learning_rate": 1.7743331602677877e-06,
      "loss": 0.2038,
      "step": 3349
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7803363347152649,
      "learning_rate": 1.7716308185852605e-06,
      "loss": 0.2277,
      "step": 3350
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.752205670549533,
      "learning_rate": 1.7689300931841525e-06,
      "loss": 0.2091,
      "step": 3351
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7461852510335568,
      "learning_rate": 1.7662309854165788e-06,
      "loss": 0.1982,
      "step": 3352
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7582934273399599,
      "learning_rate": 1.7635334966338463e-06,
      "loss": 0.2111,
      "step": 3353
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7792194820580745,
      "learning_rate": 1.7608376281864502e-06,
      "loss": 0.2022,
      "step": 3354
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7654518894250022,
      "learning_rate": 1.7581433814240783e-06,
      "loss": 0.2164,
      "step": 3355
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.742750375782661,
      "learning_rate": 1.755450757695601e-06,
      "loss": 0.1962,
      "step": 3356
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7426488992120482,
      "learning_rate": 1.7527597583490825e-06,
      "loss": 0.1898,
      "step": 3357
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.759933457195289,
      "learning_rate": 1.7500703847317663e-06,
      "loss": 0.2047,
      "step": 3358
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7043530368068058,
      "learning_rate": 1.74738263819009e-06,
      "loss": 0.1845,
      "step": 3359
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7731137973224645,
      "learning_rate": 1.7446965200696702e-06,
      "loss": 0.2126,
      "step": 3360
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.782822228644815,
      "learning_rate": 1.742012031715311e-06,
      "loss": 0.2203,
      "step": 3361
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7305836584499052,
      "learning_rate": 1.739329174470999e-06,
      "loss": 0.2026,
      "step": 3362
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7257393559529294,
      "learning_rate": 1.7366479496799076e-06,
      "loss": 0.1889,
      "step": 3363
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7892976021641295,
      "learning_rate": 1.733968358684388e-06,
      "loss": 0.2125,
      "step": 3364
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7438553198325273,
      "learning_rate": 1.7312904028259797e-06,
      "loss": 0.2004,
      "step": 3365
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7636459765922823,
      "learning_rate": 1.7286140834453958e-06,
      "loss": 0.1921,
      "step": 3366
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.6966280135635537,
      "learning_rate": 1.7259394018825382e-06,
      "loss": 0.1863,
      "step": 3367
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7855128608950677,
      "learning_rate": 1.723266359476483e-06,
      "loss": 0.2145,
      "step": 3368
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7556102176371221,
      "learning_rate": 1.7205949575654874e-06,
      "loss": 0.2054,
      "step": 3369
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7477346952999584,
      "learning_rate": 1.7179251974869858e-06,
      "loss": 0.1962,
      "step": 3370
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7849838082980598,
      "learning_rate": 1.7152570805775965e-06,
      "loss": 0.1978,
      "step": 3371
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7615293405331304,
      "learning_rate": 1.7125906081731064e-06,
      "loss": 0.1974,
      "step": 3372
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7757285000242223,
      "learning_rate": 1.7099257816084851e-06,
      "loss": 0.2245,
      "step": 3373
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7616952937251171,
      "learning_rate": 1.7072626022178796e-06,
      "loss": 0.2071,
      "step": 3374
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8191314469775978,
      "learning_rate": 1.7046010713346067e-06,
      "loss": 0.2162,
      "step": 3375
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7992458351425747,
      "learning_rate": 1.7019411902911609e-06,
      "loss": 0.2244,
      "step": 3376
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7803135191342511,
      "learning_rate": 1.6992829604192078e-06,
      "loss": 0.2077,
      "step": 3377
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7769185440578696,
      "learning_rate": 1.6966263830495939e-06,
      "loss": 0.2077,
      "step": 3378
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8024215879518829,
      "learning_rate": 1.6939714595123286e-06,
      "loss": 0.2221,
      "step": 3379
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7836679877863733,
      "learning_rate": 1.6913181911366017e-06,
      "loss": 0.2308,
      "step": 3380
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7784693599396283,
      "learning_rate": 1.6886665792507673e-06,
      "loss": 0.1939,
      "step": 3381
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7855181208830591,
      "learning_rate": 1.6860166251823574e-06,
      "loss": 0.2292,
      "step": 3382
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7788481143036874,
      "learning_rate": 1.6833683302580661e-06,
      "loss": 0.2078,
      "step": 3383
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7811469520698112,
      "learning_rate": 1.6807216958037664e-06,
      "loss": 0.2097,
      "step": 3384
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7965165134747373,
      "learning_rate": 1.6780767231444878e-06,
      "loss": 0.2243,
      "step": 3385
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7642874385387348,
      "learning_rate": 1.6754334136044398e-06,
      "loss": 0.2083,
      "step": 3386
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7697618449259657,
      "learning_rate": 1.67279176850699e-06,
      "loss": 0.2097,
      "step": 3387
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7513924925155115,
      "learning_rate": 1.6701517891746805e-06,
      "loss": 0.2047,
      "step": 3388
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7840777887744541,
      "learning_rate": 1.6675134769292118e-06,
      "loss": 0.2081,
      "step": 3389
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7565180230612211,
      "learning_rate": 1.6648768330914578e-06,
      "loss": 0.2045,
      "step": 3390
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.720254394880242,
      "learning_rate": 1.6622418589814488e-06,
      "loss": 0.1951,
      "step": 3391
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7281459231676277,
      "learning_rate": 1.6596085559183872e-06,
      "loss": 0.2024,
      "step": 3392
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8348018685882456,
      "learning_rate": 1.656976925220633e-06,
      "loss": 0.2295,
      "step": 3393
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8286454001365877,
      "learning_rate": 1.6543469682057105e-06,
      "loss": 0.2218,
      "step": 3394
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7509209780030723,
      "learning_rate": 1.651718686190305e-06,
      "loss": 0.1947,
      "step": 3395
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7569346473409185,
      "learning_rate": 1.649092080490266e-06,
      "loss": 0.194,
      "step": 3396
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.77528890794805,
      "learning_rate": 1.6464671524206045e-06,
      "loss": 0.2169,
      "step": 3397
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.777334175319625,
      "learning_rate": 1.6438439032954857e-06,
      "loss": 0.2167,
      "step": 3398
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7540860226113226,
      "learning_rate": 1.6412223344282408e-06,
      "loss": 0.2145,
      "step": 3399
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7390923989691414,
      "learning_rate": 1.6386024471313561e-06,
      "loss": 0.1899,
      "step": 3400
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.756620795098248,
      "learning_rate": 1.635984242716477e-06,
      "loss": 0.1996,
      "step": 3401
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8048261003026606,
      "learning_rate": 1.633367722494404e-06,
      "loss": 0.2234,
      "step": 3402
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.747283855775182,
      "learning_rate": 1.6307528877751e-06,
      "loss": 0.1959,
      "step": 3403
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7770261829509555,
      "learning_rate": 1.6281397398676774e-06,
      "loss": 0.2109,
      "step": 3404
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.783331270777519,
      "learning_rate": 1.6255282800804106e-06,
      "loss": 0.2045,
      "step": 3405
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7424834980758301,
      "learning_rate": 1.6229185097207228e-06,
      "loss": 0.1965,
      "step": 3406
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7606449078034517,
      "learning_rate": 1.6203104300951977e-06,
      "loss": 0.2207,
      "step": 3407
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7146715270579124,
      "learning_rate": 1.6177040425095664e-06,
      "loss": 0.1835,
      "step": 3408
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7710859923476642,
      "learning_rate": 1.615099348268716e-06,
      "loss": 0.204,
      "step": 3409
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7792511790375609,
      "learning_rate": 1.6124963486766844e-06,
      "loss": 0.2214,
      "step": 3410
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.769049103830543,
      "learning_rate": 1.6098950450366646e-06,
      "loss": 0.21,
      "step": 3411
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7282746270644688,
      "learning_rate": 1.6072954386509948e-06,
      "loss": 0.1934,
      "step": 3412
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7884535637178154,
      "learning_rate": 1.6046975308211699e-06,
      "loss": 0.2035,
      "step": 3413
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7472053979834276,
      "learning_rate": 1.6021013228478277e-06,
      "loss": 0.2042,
      "step": 3414
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7553009682093845,
      "learning_rate": 1.5995068160307626e-06,
      "loss": 0.2223,
      "step": 3415
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.773919724842456,
      "learning_rate": 1.5969140116689102e-06,
      "loss": 0.2227,
      "step": 3416
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8398792833731847,
      "learning_rate": 1.594322911060358e-06,
      "loss": 0.215,
      "step": 3417
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7972386686497948,
      "learning_rate": 1.5917335155023368e-06,
      "loss": 0.2126,
      "step": 3418
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7315008650612653,
      "learning_rate": 1.589145826291229e-06,
      "loss": 0.2044,
      "step": 3419
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7019533110985439,
      "learning_rate": 1.5865598447225576e-06,
      "loss": 0.1905,
      "step": 3420
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7194736059371561,
      "learning_rate": 1.583975572090995e-06,
      "loss": 0.1905,
      "step": 3421
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7663910039631101,
      "learning_rate": 1.5813930096903562e-06,
      "loss": 0.2089,
      "step": 3422
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7488498287125647,
      "learning_rate": 1.5788121588135975e-06,
      "loss": 0.2077,
      "step": 3423
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7823968273892347,
      "learning_rate": 1.5762330207528249e-06,
      "loss": 0.2003,
      "step": 3424
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.779158240826112,
      "learning_rate": 1.5736555967992767e-06,
      "loss": 0.2028,
      "step": 3425
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7837092639884713,
      "learning_rate": 1.5710798882433432e-06,
      "loss": 0.2341,
      "step": 3426
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7458058327051587,
      "learning_rate": 1.5685058963745485e-06,
      "loss": 0.1909,
      "step": 3427
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7941217842197864,
      "learning_rate": 1.5659336224815642e-06,
      "loss": 0.237,
      "step": 3428
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7808269994268048,
      "learning_rate": 1.5633630678521944e-06,
      "loss": 0.2154,
      "step": 3429
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7099992119827047,
      "learning_rate": 1.5607942337733884e-06,
      "loss": 0.2107,
      "step": 3430
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7748811798790564,
      "learning_rate": 1.5582271215312294e-06,
      "loss": 0.2045,
      "step": 3431
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7594728709563282,
      "learning_rate": 1.555661732410944e-06,
      "loss": 0.2121,
      "step": 3432
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7811025645779477,
      "learning_rate": 1.553098067696891e-06,
      "loss": 0.2245,
      "step": 3433
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7355266421190476,
      "learning_rate": 1.5505361286725679e-06,
      "loss": 0.1926,
      "step": 3434
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.748269388260839,
      "learning_rate": 1.5479759166206066e-06,
      "loss": 0.2034,
      "step": 3435
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7778234131939812,
      "learning_rate": 1.5454174328227794e-06,
      "loss": 0.2188,
      "step": 3436
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7246771059167461,
      "learning_rate": 1.542860678559987e-06,
      "loss": 0.2122,
      "step": 3437
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7273592912365245,
      "learning_rate": 1.5403056551122697e-06,
      "loss": 0.19,
      "step": 3438
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.766197537812733,
      "learning_rate": 1.537752363758796e-06,
      "loss": 0.208,
      "step": 3439
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7902874272189736,
      "learning_rate": 1.5352008057778723e-06,
      "loss": 0.214,
      "step": 3440
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7409925559161525,
      "learning_rate": 1.5326509824469344e-06,
      "loss": 0.2005,
      "step": 3441
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8200224628693398,
      "learning_rate": 1.5301028950425494e-06,
      "loss": 0.2141,
      "step": 3442
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.736651695159385,
      "learning_rate": 1.5275565448404146e-06,
      "loss": 0.1987,
      "step": 3443
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7777803561267989,
      "learning_rate": 1.5250119331153606e-06,
      "loss": 0.2223,
      "step": 3444
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8141777221880667,
      "learning_rate": 1.5224690611413473e-06,
      "loss": 0.2201,
      "step": 3445
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7707924813036107,
      "learning_rate": 1.5199279301914592e-06,
      "loss": 0.2133,
      "step": 3446
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7699378486111828,
      "learning_rate": 1.5173885415379158e-06,
      "loss": 0.213,
      "step": 3447
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7573205647099805,
      "learning_rate": 1.5148508964520586e-06,
      "loss": 0.2103,
      "step": 3448
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8359192142764326,
      "learning_rate": 1.5123149962043586e-06,
      "loss": 0.2452,
      "step": 3449
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7843171835174788,
      "learning_rate": 1.5097808420644117e-06,
      "loss": 0.2095,
      "step": 3450
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7627151383709239,
      "learning_rate": 1.5072484353009432e-06,
      "loss": 0.2076,
      "step": 3451
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7710134936224705,
      "learning_rate": 1.5047177771817989e-06,
      "loss": 0.201,
      "step": 3452
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7537888497789195,
      "learning_rate": 1.502188868973955e-06,
      "loss": 0.2024,
      "step": 3453
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7896154190625198,
      "learning_rate": 1.4996617119435036e-06,
      "loss": 0.2135,
      "step": 3454
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7541917742428912,
      "learning_rate": 1.4971363073556695e-06,
      "loss": 0.2003,
      "step": 3455
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7527298702665981,
      "learning_rate": 1.494612656474792e-06,
      "loss": 0.2011,
      "step": 3456
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7495048592321802,
      "learning_rate": 1.492090760564337e-06,
      "loss": 0.2074,
      "step": 3457
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.771908567774284,
      "learning_rate": 1.4895706208868876e-06,
      "loss": 0.2113,
      "step": 3458
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7646519232580798,
      "learning_rate": 1.4870522387041547e-06,
      "loss": 0.2057,
      "step": 3459
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.868616288113341,
      "learning_rate": 1.4845356152769618e-06,
      "loss": 0.2262,
      "step": 3460
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7165791282365929,
      "learning_rate": 1.4820207518652579e-06,
      "loss": 0.1922,
      "step": 3461
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7775952983319899,
      "learning_rate": 1.4795076497281052e-06,
      "loss": 0.2112,
      "step": 3462
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7696539084119345,
      "learning_rate": 1.4769963101236894e-06,
      "loss": 0.1956,
      "step": 3463
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7625539464818303,
      "learning_rate": 1.4744867343093112e-06,
      "loss": 0.2098,
      "step": 3464
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7353881175612376,
      "learning_rate": 1.4719789235413878e-06,
      "loss": 0.2069,
      "step": 3465
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8059054913720926,
      "learning_rate": 1.4694728790754514e-06,
      "loss": 0.2204,
      "step": 3466
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7406211744875948,
      "learning_rate": 1.466968602166156e-06,
      "loss": 0.2075,
      "step": 3467
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7428550096667149,
      "learning_rate": 1.4644660940672628e-06,
      "loss": 0.2016,
      "step": 3468
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7504805345332239,
      "learning_rate": 1.4619653560316526e-06,
      "loss": 0.2068,
      "step": 3469
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7630267129536031,
      "learning_rate": 1.459466389311321e-06,
      "loss": 0.2028,
      "step": 3470
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7182027853319627,
      "learning_rate": 1.4569691951573706e-06,
      "loss": 0.1948,
      "step": 3471
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.723519496937015,
      "learning_rate": 1.4544737748200238e-06,
      "loss": 0.2034,
      "step": 3472
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7060048990572201,
      "learning_rate": 1.4519801295486102e-06,
      "loss": 0.186,
      "step": 3473
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7633028393180397,
      "learning_rate": 1.4494882605915717e-06,
      "loss": 0.2074,
      "step": 3474
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7757502560434657,
      "learning_rate": 1.4469981691964597e-06,
      "loss": 0.1998,
      "step": 3475
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7663865353180795,
      "learning_rate": 1.4445098566099409e-06,
      "loss": 0.2028,
      "step": 3476
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7689774836735426,
      "learning_rate": 1.4420233240777837e-06,
      "loss": 0.219,
      "step": 3477
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.757939527208888,
      "learning_rate": 1.439538572844873e-06,
      "loss": 0.2034,
      "step": 3478
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7468139478365173,
      "learning_rate": 1.4370556041551954e-06,
      "loss": 0.2049,
      "step": 3479
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8024249423266238,
      "learning_rate": 1.4345744192518507e-06,
      "loss": 0.2244,
      "step": 3480
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7493786682990621,
      "learning_rate": 1.432095019377041e-06,
      "loss": 0.201,
      "step": 3481
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8151809044124575,
      "learning_rate": 1.4296174057720775e-06,
      "loss": 0.2078,
      "step": 3482
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7699524277321648,
      "learning_rate": 1.4271415796773742e-06,
      "loss": 0.2065,
      "step": 3483
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7485396074325271,
      "learning_rate": 1.4246675423324546e-06,
      "loss": 0.2049,
      "step": 3484
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7686248976115138,
      "learning_rate": 1.422195294975942e-06,
      "loss": 0.2153,
      "step": 3485
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.773392600298359,
      "learning_rate": 1.4197248388455693e-06,
      "loss": 0.2141,
      "step": 3486
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7966510137683703,
      "learning_rate": 1.417256175178165e-06,
      "loss": 0.2088,
      "step": 3487
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7417919900079218,
      "learning_rate": 1.4147893052096684e-06,
      "loss": 0.1976,
      "step": 3488
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7481367731407025,
      "learning_rate": 1.412324230175114e-06,
      "loss": 0.2134,
      "step": 3489
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7759057033680987,
      "learning_rate": 1.4098609513086414e-06,
      "loss": 0.1868,
      "step": 3490
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7924589330370754,
      "learning_rate": 1.4073994698434878e-06,
      "loss": 0.2194,
      "step": 3491
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7992261086404171,
      "learning_rate": 1.4049397870119962e-06,
      "loss": 0.2227,
      "step": 3492
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7676739544628021,
      "learning_rate": 1.4024819040456023e-06,
      "loss": 0.2104,
      "step": 3493
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.79822093630648,
      "learning_rate": 1.4000258221748452e-06,
      "loss": 0.2113,
      "step": 3494
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7629421476314435,
      "learning_rate": 1.3975715426293636e-06,
      "loss": 0.2054,
      "step": 3495
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7808850501307297,
      "learning_rate": 1.3951190666378884e-06,
      "loss": 0.2003,
      "step": 3496
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8092825179861874,
      "learning_rate": 1.392668395428251e-06,
      "loss": 0.2167,
      "step": 3497
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7594322322894717,
      "learning_rate": 1.390219530227378e-06,
      "loss": 0.2062,
      "step": 3498
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7972151251553741,
      "learning_rate": 1.3877724722612946e-06,
      "loss": 0.2096,
      "step": 3499
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7841085690336281,
      "learning_rate": 1.3853272227551167e-06,
      "loss": 0.2188,
      "step": 3500
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7925504064507048,
      "learning_rate": 1.38288378293306e-06,
      "loss": 0.2075,
      "step": 3501
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7459511078703398,
      "learning_rate": 1.3804421540184282e-06,
      "loss": 0.2047,
      "step": 3502
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7688954522342727,
      "learning_rate": 1.378002337233625e-06,
      "loss": 0.215,
      "step": 3503
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7722261915893252,
      "learning_rate": 1.3755643338001411e-06,
      "loss": 0.2079,
      "step": 3504
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7592686270684904,
      "learning_rate": 1.373128144938563e-06,
      "loss": 0.1874,
      "step": 3505
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.747954801046188,
      "learning_rate": 1.3706937718685654e-06,
      "loss": 0.1965,
      "step": 3506
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7668302328739977,
      "learning_rate": 1.3682612158089193e-06,
      "loss": 0.2138,
      "step": 3507
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7788729146731793,
      "learning_rate": 1.3658304779774784e-06,
      "loss": 0.2128,
      "step": 3508
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7870878622679695,
      "learning_rate": 1.3634015595911949e-06,
      "loss": 0.224,
      "step": 3509
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8226435048567022,
      "learning_rate": 1.3609744618661013e-06,
      "loss": 0.2232,
      "step": 3510
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8457098815996796,
      "learning_rate": 1.3585491860173266e-06,
      "loss": 0.2322,
      "step": 3511
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7558675357317847,
      "learning_rate": 1.3561257332590816e-06,
      "loss": 0.2082,
      "step": 3512
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7744638406807817,
      "learning_rate": 1.3537041048046696e-06,
      "loss": 0.205,
      "step": 3513
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7938710759081128,
      "learning_rate": 1.3512843018664728e-06,
      "loss": 0.2203,
      "step": 3514
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7560320583367852,
      "learning_rate": 1.3488663256559686e-06,
      "loss": 0.2077,
      "step": 3515
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7460931248401209,
      "learning_rate": 1.3464501773837124e-06,
      "loss": 0.2028,
      "step": 3516
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7885450263600878,
      "learning_rate": 1.3440358582593494e-06,
      "loss": 0.2125,
      "step": 3517
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.743137013362281,
      "learning_rate": 1.3416233694916086e-06,
      "loss": 0.1991,
      "step": 3518
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7714834270185682,
      "learning_rate": 1.339212712288298e-06,
      "loss": 0.2151,
      "step": 3519
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7405020759816529,
      "learning_rate": 1.3368038878563154e-06,
      "loss": 0.1974,
      "step": 3520
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7934648442481316,
      "learning_rate": 1.3343968974016363e-06,
      "loss": 0.2277,
      "step": 3521
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7754721340143574,
      "learning_rate": 1.3319917421293182e-06,
      "loss": 0.2137,
      "step": 3522
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7863438938159247,
      "learning_rate": 1.3295884232435008e-06,
      "loss": 0.2052,
      "step": 3523
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7627508405248926,
      "learning_rate": 1.3271869419474059e-06,
      "loss": 0.1944,
      "step": 3524
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7542115493924426,
      "learning_rate": 1.3247872994433326e-06,
      "loss": 0.2036,
      "step": 3525
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7754645423191144,
      "learning_rate": 1.3223894969326623e-06,
      "loss": 0.209,
      "step": 3526
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8245968129040898,
      "learning_rate": 1.319993535615851e-06,
      "loss": 0.2221,
      "step": 3527
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.739513047398736,
      "learning_rate": 1.3175994166924394e-06,
      "loss": 0.201,
      "step": 3528
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7516521549897963,
      "learning_rate": 1.315207141361039e-06,
      "loss": 0.1918,
      "step": 3529
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7573524403069841,
      "learning_rate": 1.312816710819343e-06,
      "loss": 0.2228,
      "step": 3530
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7459395862481006,
      "learning_rate": 1.310428126264116e-06,
      "loss": 0.2013,
      "step": 3531
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7469553158051919,
      "learning_rate": 1.3080413888912057e-06,
      "loss": 0.1924,
      "step": 3532
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7892723427064837,
      "learning_rate": 1.3056564998955274e-06,
      "loss": 0.2198,
      "step": 3533
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.768582055452969,
      "learning_rate": 1.3032734604710784e-06,
      "loss": 0.2081,
      "step": 3534
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7803541677318185,
      "learning_rate": 1.3008922718109224e-06,
      "loss": 0.2032,
      "step": 3535
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8049137765798424,
      "learning_rate": 1.298512935107204e-06,
      "loss": 0.2037,
      "step": 3536
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8060571081697472,
      "learning_rate": 1.2961354515511342e-06,
      "loss": 0.2311,
      "step": 3537
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7749848418992554,
      "learning_rate": 1.2937598223330006e-06,
      "loss": 0.2206,
      "step": 3538
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7652623474878673,
      "learning_rate": 1.2913860486421582e-06,
      "loss": 0.1964,
      "step": 3539
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.748297945985138,
      "learning_rate": 1.289014131667039e-06,
      "loss": 0.2066,
      "step": 3540
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7937175789610145,
      "learning_rate": 1.286644072595139e-06,
      "loss": 0.2107,
      "step": 3541
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.771861052169987,
      "learning_rate": 1.2842758726130283e-06,
      "loss": 0.2076,
      "step": 3542
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7632754409022927,
      "learning_rate": 1.2819095329063469e-06,
      "loss": 0.2048,
      "step": 3543
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7598114028823019,
      "learning_rate": 1.2795450546597999e-06,
      "loss": 0.2249,
      "step": 3544
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7365863173772723,
      "learning_rate": 1.277182439057162e-06,
      "loss": 0.1983,
      "step": 3545
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7618751232644536,
      "learning_rate": 1.2748216872812747e-06,
      "loss": 0.2189,
      "step": 3546
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7551519915668968,
      "learning_rate": 1.2724628005140493e-06,
      "loss": 0.202,
      "step": 3547
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7593239109697136,
      "learning_rate": 1.2701057799364591e-06,
      "loss": 0.2063,
      "step": 3548
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7135534180417807,
      "learning_rate": 1.2677506267285479e-06,
      "loss": 0.177,
      "step": 3549
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7177788056523978,
      "learning_rate": 1.2653973420694182e-06,
      "loss": 0.1736,
      "step": 3550
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7669397798891535,
      "learning_rate": 1.2630459271372452e-06,
      "loss": 0.1958,
      "step": 3551
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.76568853515967,
      "learning_rate": 1.26069638310926e-06,
      "loss": 0.1933,
      "step": 3552
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7631263956546349,
      "learning_rate": 1.2583487111617647e-06,
      "loss": 0.2155,
      "step": 3553
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7878942400752369,
      "learning_rate": 1.256002912470115e-06,
      "loss": 0.2028,
      "step": 3554
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7737750273384006,
      "learning_rate": 1.2536589882087373e-06,
      "loss": 0.2031,
      "step": 3555
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7827216370881891,
      "learning_rate": 1.2513169395511143e-06,
      "loss": 0.2054,
      "step": 3556
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7660790589234228,
      "learning_rate": 1.2489767676697933e-06,
      "loss": 0.2174,
      "step": 3557
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7846415460804566,
      "learning_rate": 1.246638473736378e-06,
      "loss": 0.2247,
      "step": 3558
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7491286670012065,
      "learning_rate": 1.244302058921537e-06,
      "loss": 0.2005,
      "step": 3559
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7823770636934292,
      "learning_rate": 1.2419675243949914e-06,
      "loss": 0.212,
      "step": 3560
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7850925103686315,
      "learning_rate": 1.239634871325529e-06,
      "loss": 0.2134,
      "step": 3561
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7581158254040632,
      "learning_rate": 1.2373041008809888e-06,
      "loss": 0.217,
      "step": 3562
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.818861065198619,
      "learning_rate": 1.2349752142282706e-06,
      "loss": 0.2119,
      "step": 3563
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7786506915535019,
      "learning_rate": 1.2326482125333284e-06,
      "loss": 0.2165,
      "step": 3564
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7815367497896726,
      "learning_rate": 1.2303230969611762e-06,
      "loss": 0.2158,
      "step": 3565
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7323063030013038,
      "learning_rate": 1.2279998686758833e-06,
      "loss": 0.1912,
      "step": 3566
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8158678790411352,
      "learning_rate": 1.2256785288405699e-06,
      "loss": 0.2156,
      "step": 3567
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7627338243185372,
      "learning_rate": 1.223359078617416e-06,
      "loss": 0.1988,
      "step": 3568
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7804667800398641,
      "learning_rate": 1.2210415191676528e-06,
      "loss": 0.2034,
      "step": 3569
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7508576598497682,
      "learning_rate": 1.2187258516515644e-06,
      "loss": 0.1907,
      "step": 3570
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7557606157411436,
      "learning_rate": 1.2164120772284875e-06,
      "loss": 0.2062,
      "step": 3571
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8197701192227276,
      "learning_rate": 1.214100197056815e-06,
      "loss": 0.2096,
      "step": 3572
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.795580670731467,
      "learning_rate": 1.2117902122939861e-06,
      "loss": 0.2168,
      "step": 3573
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7414082314618914,
      "learning_rate": 1.2094821240964955e-06,
      "loss": 0.1974,
      "step": 3574
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7631811621097102,
      "learning_rate": 1.207175933619884e-06,
      "loss": 0.1961,
      "step": 3575
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7112774005044114,
      "learning_rate": 1.2048716420187473e-06,
      "loss": 0.1906,
      "step": 3576
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7471387137946094,
      "learning_rate": 1.2025692504467269e-06,
      "loss": 0.1969,
      "step": 3577
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8006425362727476,
      "learning_rate": 1.2002687600565138e-06,
      "loss": 0.2113,
      "step": 3578
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7695441586094296,
      "learning_rate": 1.1979701719998454e-06,
      "loss": 0.2115,
      "step": 3579
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7836018660690515,
      "learning_rate": 1.1956734874275127e-06,
      "loss": 0.2267,
      "step": 3580
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.778614549134335,
      "learning_rate": 1.1933787074893456e-06,
      "loss": 0.2177,
      "step": 3581
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7818738974759352,
      "learning_rate": 1.1910858333342279e-06,
      "loss": 0.2071,
      "step": 3582
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7710885827805933,
      "learning_rate": 1.1887948661100833e-06,
      "loss": 0.2062,
      "step": 3583
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7913601011808653,
      "learning_rate": 1.1865058069638857e-06,
      "loss": 0.218,
      "step": 3584
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7757561425759196,
      "learning_rate": 1.1842186570416508e-06,
      "loss": 0.2088,
      "step": 3585
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7779701504715362,
      "learning_rate": 1.1819334174884378e-06,
      "loss": 0.2265,
      "step": 3586
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.753929694508387,
      "learning_rate": 1.1796500894483509e-06,
      "loss": 0.2041,
      "step": 3587
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8046592993792274,
      "learning_rate": 1.1773686740645384e-06,
      "loss": 0.2086,
      "step": 3588
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7859639758077426,
      "learning_rate": 1.175089172479188e-06,
      "loss": 0.2095,
      "step": 3589
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7559091063415572,
      "learning_rate": 1.172811585833532e-06,
      "loss": 0.2085,
      "step": 3590
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7852005155101397,
      "learning_rate": 1.1705359152678447e-06,
      "loss": 0.2113,
      "step": 3591
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7778322328983314,
      "learning_rate": 1.168262161921438e-06,
      "loss": 0.2187,
      "step": 3592
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7890618003244507,
      "learning_rate": 1.165990326932665e-06,
      "loss": 0.2127,
      "step": 3593
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7750609123109042,
      "learning_rate": 1.1637204114389179e-06,
      "loss": 0.211,
      "step": 3594
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7237152813965614,
      "learning_rate": 1.161452416576631e-06,
      "loss": 0.2073,
      "step": 3595
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.795407261750728,
      "learning_rate": 1.1591863434812722e-06,
      "loss": 0.2241,
      "step": 3596
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.798153667352529,
      "learning_rate": 1.1569221932873532e-06,
      "loss": 0.2086,
      "step": 3597
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7678486881399474,
      "learning_rate": 1.1546599671284158e-06,
      "loss": 0.2041,
      "step": 3598
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7544659329174286,
      "learning_rate": 1.1523996661370461e-06,
      "loss": 0.2041,
      "step": 3599
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7313238446306423,
      "learning_rate": 1.1501412914448596e-06,
      "loss": 0.1686,
      "step": 3600
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7018047542758712,
      "learning_rate": 1.147884844182513e-06,
      "loss": 0.1856,
      "step": 3601
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7534153136236214,
      "learning_rate": 1.1456303254796941e-06,
      "loss": 0.2018,
      "step": 3602
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7978930586958141,
      "learning_rate": 1.1433777364651272e-06,
      "loss": 0.2238,
      "step": 3603
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7469982115548495,
      "learning_rate": 1.141127078266568e-06,
      "loss": 0.2089,
      "step": 3604
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.756906070148747,
      "learning_rate": 1.1388783520108099e-06,
      "loss": 0.1925,
      "step": 3605
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.744403840818599,
      "learning_rate": 1.1366315588236743e-06,
      "loss": 0.1879,
      "step": 3606
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7631880597546541,
      "learning_rate": 1.1343866998300197e-06,
      "loss": 0.2019,
      "step": 3607
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7631500093382515,
      "learning_rate": 1.1321437761537307e-06,
      "loss": 0.2019,
      "step": 3608
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7739569205929795,
      "learning_rate": 1.1299027889177294e-06,
      "loss": 0.2322,
      "step": 3609
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.763706864722938,
      "learning_rate": 1.127663739243962e-06,
      "loss": 0.2075,
      "step": 3610
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7067144368086159,
      "learning_rate": 1.1254266282534093e-06,
      "loss": 0.1978,
      "step": 3611
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7360213190560845,
      "learning_rate": 1.1231914570660774e-06,
      "loss": 0.1978,
      "step": 3612
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7328340569736588,
      "learning_rate": 1.1209582268010056e-06,
      "loss": 0.1974,
      "step": 3613
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7609582807737956,
      "learning_rate": 1.1187269385762605e-06,
      "loss": 0.199,
      "step": 3614
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7583558800583723,
      "learning_rate": 1.1164975935089328e-06,
      "loss": 0.2182,
      "step": 3615
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7973639940798337,
      "learning_rate": 1.1142701927151456e-06,
      "loss": 0.2102,
      "step": 3616
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7429815267591996,
      "learning_rate": 1.1120447373100446e-06,
      "loss": 0.1948,
      "step": 3617
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7479616559629203,
      "learning_rate": 1.1098212284078037e-06,
      "loss": 0.2059,
      "step": 3618
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7989239843604841,
      "learning_rate": 1.1075996671216188e-06,
      "loss": 0.2185,
      "step": 3619
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7637409192926969,
      "learning_rate": 1.1053800545637173e-06,
      "loss": 0.2042,
      "step": 3620
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7747428415380758,
      "learning_rate": 1.1031623918453443e-06,
      "loss": 0.2137,
      "step": 3621
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7599772869866341,
      "learning_rate": 1.1009466800767738e-06,
      "loss": 0.2114,
      "step": 3622
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7909205527070541,
      "learning_rate": 1.098732920367298e-06,
      "loss": 0.216,
      "step": 3623
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7606974663712975,
      "learning_rate": 1.0965211138252373e-06,
      "loss": 0.2087,
      "step": 3624
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7564011533589083,
      "learning_rate": 1.094311261557931e-06,
      "loss": 0.2143,
      "step": 3625
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7537511804291013,
      "learning_rate": 1.09210336467174e-06,
      "loss": 0.1991,
      "step": 3626
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7459705716900331,
      "learning_rate": 1.089897424272045e-06,
      "loss": 0.2144,
      "step": 3627
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7663821193717232,
      "learning_rate": 1.0876934414632523e-06,
      "loss": 0.2025,
      "step": 3628
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7653200619111279,
      "learning_rate": 1.0854914173487818e-06,
      "loss": 0.2172,
      "step": 3629
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.809506277353454,
      "learning_rate": 1.0832913530310784e-06,
      "loss": 0.2074,
      "step": 3630
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8248793155745996,
      "learning_rate": 1.081093249611601e-06,
      "loss": 0.2142,
      "step": 3631
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7564809792301832,
      "learning_rate": 1.0788971081908306e-06,
      "loss": 0.1984,
      "step": 3632
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7575264511787883,
      "learning_rate": 1.0767029298682642e-06,
      "loss": 0.2117,
      "step": 3633
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7889810629402946,
      "learning_rate": 1.0745107157424155e-06,
      "loss": 0.2262,
      "step": 3634
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7410479808791454,
      "learning_rate": 1.072320466910814e-06,
      "loss": 0.1937,
      "step": 3635
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7519656264592359,
      "learning_rate": 1.0701321844700097e-06,
      "loss": 0.1954,
      "step": 3636
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7212423270154834,
      "learning_rate": 1.0679458695155615e-06,
      "loss": 0.193,
      "step": 3637
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.762123892773918,
      "learning_rate": 1.0657615231420492e-06,
      "loss": 0.2071,
      "step": 3638
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7784801304383725,
      "learning_rate": 1.0635791464430662e-06,
      "loss": 0.2034,
      "step": 3639
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7697281995986297,
      "learning_rate": 1.0613987405112153e-06,
      "loss": 0.1989,
      "step": 3640
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7805926290849218,
      "learning_rate": 1.0592203064381173e-06,
      "loss": 0.2093,
      "step": 3641
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7977863295321542,
      "learning_rate": 1.0570438453144044e-06,
      "loss": 0.2179,
      "step": 3642
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7510806864505057,
      "learning_rate": 1.0548693582297203e-06,
      "loss": 0.2129,
      "step": 3643
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7882201353103497,
      "learning_rate": 1.0526968462727189e-06,
      "loss": 0.2065,
      "step": 3644
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7558889877448349,
      "learning_rate": 1.0505263105310703e-06,
      "loss": 0.1904,
      "step": 3645
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7387300397643796,
      "learning_rate": 1.0483577520914495e-06,
      "loss": 0.1978,
      "step": 3646
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7947194832861381,
      "learning_rate": 1.0461911720395473e-06,
      "loss": 0.2107,
      "step": 3647
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8235732525016292,
      "learning_rate": 1.0440265714600573e-06,
      "loss": 0.2152,
      "step": 3648
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7256021039507795,
      "learning_rate": 1.041863951436689e-06,
      "loss": 0.1942,
      "step": 3649
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7967080228247753,
      "learning_rate": 1.0397033130521554e-06,
      "loss": 0.2028,
      "step": 3650
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7768416306925182,
      "learning_rate": 1.0375446573881799e-06,
      "loss": 0.1968,
      "step": 3651
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7004246805536788,
      "learning_rate": 1.0353879855254895e-06,
      "loss": 0.1935,
      "step": 3652
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7784379054031814,
      "learning_rate": 1.0332332985438248e-06,
      "loss": 0.2232,
      "step": 3653
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.744347994198608,
      "learning_rate": 1.0310805975219256e-06,
      "loss": 0.1976,
      "step": 3654
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.759385501714776,
      "learning_rate": 1.028929883537544e-06,
      "loss": 0.197,
      "step": 3655
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7431919807507137,
      "learning_rate": 1.0267811576674298e-06,
      "loss": 0.181,
      "step": 3656
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7656027873353545,
      "learning_rate": 1.0246344209873448e-06,
      "loss": 0.1978,
      "step": 3657
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7979828553645472,
      "learning_rate": 1.0224896745720513e-06,
      "loss": 0.2275,
      "step": 3658
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8075903043278654,
      "learning_rate": 1.0203469194953142e-06,
      "loss": 0.2177,
      "step": 3659
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7544417345630047,
      "learning_rate": 1.0182061568299019e-06,
      "loss": 0.191,
      "step": 3660
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7702163110570736,
      "learning_rate": 1.0160673876475892e-06,
      "loss": 0.2041,
      "step": 3661
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7761651564095681,
      "learning_rate": 1.0139306130191463e-06,
      "loss": 0.202,
      "step": 3662
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7701070810551842,
      "learning_rate": 1.0117958340143508e-06,
      "loss": 0.2067,
      "step": 3663
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7838225372270218,
      "learning_rate": 1.009663051701979e-06,
      "loss": 0.2037,
      "step": 3664
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7683354016743121,
      "learning_rate": 1.0075322671498066e-06,
      "loss": 0.205,
      "step": 3665
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7526474677000123,
      "learning_rate": 1.0054034814246093e-06,
      "loss": 0.2093,
      "step": 3666
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7761187757394454,
      "learning_rate": 1.0032766955921618e-06,
      "loss": 0.1998,
      "step": 3667
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8023301430020022,
      "learning_rate": 1.0011519107172413e-06,
      "loss": 0.2069,
      "step": 3668
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8387167025458908,
      "learning_rate": 9.990291278636172e-07,
      "loss": 0.2203,
      "step": 3669
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7992234682446984,
      "learning_rate": 9.969083480940617e-07,
      "loss": 0.216,
      "step": 3670
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7600605735792129,
      "learning_rate": 9.947895724703405e-07,
      "loss": 0.2213,
      "step": 3671
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7864477421810002,
      "learning_rate": 9.926728020532196e-07,
      "loss": 0.2163,
      "step": 3672
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7522319537789918,
      "learning_rate": 9.905580379024581e-07,
      "loss": 0.2147,
      "step": 3673
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7140657291742326,
      "learning_rate": 9.884452810768115e-07,
      "loss": 0.1952,
      "step": 3674
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.791391607833475,
      "learning_rate": 9.86334532634029e-07,
      "loss": 0.1963,
      "step": 3675
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.757663513805937,
      "learning_rate": 9.842257936308586e-07,
      "loss": 0.2103,
      "step": 3676
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7065060015014906,
      "learning_rate": 9.82119065123036e-07,
      "loss": 0.1916,
      "step": 3677
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7695031352330204,
      "learning_rate": 9.80014348165298e-07,
      "loss": 0.2128,
      "step": 3678
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7921327957809725,
      "learning_rate": 9.77911643811366e-07,
      "loss": 0.2105,
      "step": 3679
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8122416216971017,
      "learning_rate": 9.758109531139614e-07,
      "loss": 0.2112,
      "step": 3680
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7604226004976817,
      "learning_rate": 9.737122771247902e-07,
      "loss": 0.2187,
      "step": 3681
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.713256977142277,
      "learning_rate": 9.71615616894559e-07,
      "loss": 0.1906,
      "step": 3682
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7654097555467749,
      "learning_rate": 9.695209734729533e-07,
      "loss": 0.201,
      "step": 3683
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7792980192766747,
      "learning_rate": 9.6742834790866e-07,
      "loss": 0.213,
      "step": 3684
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8029478717379582,
      "learning_rate": 9.653377412493475e-07,
      "loss": 0.2143,
      "step": 3685
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7964661821567686,
      "learning_rate": 9.632491545416788e-07,
      "loss": 0.2216,
      "step": 3686
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8003732403343939,
      "learning_rate": 9.611625888313053e-07,
      "loss": 0.2292,
      "step": 3687
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7194499395538997,
      "learning_rate": 9.590780451628617e-07,
      "loss": 0.2045,
      "step": 3688
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7464933529858724,
      "learning_rate": 9.569955245799772e-07,
      "loss": 0.1971,
      "step": 3689
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7940101184767144,
      "learning_rate": 9.549150281252633e-07,
      "loss": 0.22,
      "step": 3690
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7989636846851057,
      "learning_rate": 9.528365568403192e-07,
      "loss": 0.2052,
      "step": 3691
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7986899596202351,
      "learning_rate": 9.507601117657289e-07,
      "loss": 0.2327,
      "step": 3692
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7351615361337609,
      "learning_rate": 9.486856939410672e-07,
      "loss": 0.189,
      "step": 3693
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7641153252011887,
      "learning_rate": 9.466133044048865e-07,
      "loss": 0.2011,
      "step": 3694
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7901669004079301,
      "learning_rate": 9.445429441947312e-07,
      "loss": 0.1918,
      "step": 3695
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7417959599906694,
      "learning_rate": 9.424746143471231e-07,
      "loss": 0.1944,
      "step": 3696
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7981768852379154,
      "learning_rate": 9.40408315897573e-07,
      "loss": 0.2125,
      "step": 3697
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7373525574205112,
      "learning_rate": 9.383440498805712e-07,
      "loss": 0.1896,
      "step": 3698
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7579567933775245,
      "learning_rate": 9.362818173295913e-07,
      "loss": 0.2007,
      "step": 3699
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7483863089490176,
      "learning_rate": 9.34221619277087e-07,
      "loss": 0.1931,
      "step": 3700
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8083609679808917,
      "learning_rate": 9.321634567544991e-07,
      "loss": 0.2081,
      "step": 3701
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7615720514782495,
      "learning_rate": 9.301073307922432e-07,
      "loss": 0.2098,
      "step": 3702
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7590174388737766,
      "learning_rate": 9.280532424197192e-07,
      "loss": 0.2084,
      "step": 3703
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7194145384829358,
      "learning_rate": 9.260011926653034e-07,
      "loss": 0.2028,
      "step": 3704
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8389749289816829,
      "learning_rate": 9.239511825563557e-07,
      "loss": 0.2199,
      "step": 3705
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8199174625895844,
      "learning_rate": 9.219032131192124e-07,
      "loss": 0.2261,
      "step": 3706
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.797901844446785,
      "learning_rate": 9.19857285379187e-07,
      "loss": 0.2189,
      "step": 3707
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7566781042198386,
      "learning_rate": 9.178134003605721e-07,
      "loss": 0.1915,
      "step": 3708
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7707807455833047,
      "learning_rate": 9.157715590866396e-07,
      "loss": 0.2082,
      "step": 3709
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7620965307780829,
      "learning_rate": 9.137317625796338e-07,
      "loss": 0.1967,
      "step": 3710
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7689905614721343,
      "learning_rate": 9.116940118607792e-07,
      "loss": 0.2082,
      "step": 3711
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7856020231408674,
      "learning_rate": 9.096583079502763e-07,
      "loss": 0.2179,
      "step": 3712
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7718528736105987,
      "learning_rate": 9.076246518672971e-07,
      "loss": 0.2012,
      "step": 3713
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.782785588311739,
      "learning_rate": 9.055930446299916e-07,
      "loss": 0.2039,
      "step": 3714
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7914999280328854,
      "learning_rate": 9.035634872554805e-07,
      "loss": 0.2145,
      "step": 3715
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7802157273563766,
      "learning_rate": 9.015359807598645e-07,
      "loss": 0.2098,
      "step": 3716
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7850968703291091,
      "learning_rate": 8.995105261582099e-07,
      "loss": 0.2027,
      "step": 3717
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8009123240146959,
      "learning_rate": 8.974871244645628e-07,
      "loss": 0.2097,
      "step": 3718
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7190598324420489,
      "learning_rate": 8.954657766919361e-07,
      "loss": 0.1947,
      "step": 3719
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7703232522564193,
      "learning_rate": 8.934464838523183e-07,
      "loss": 0.2128,
      "step": 3720
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7146611769264761,
      "learning_rate": 8.91429246956666e-07,
      "loss": 0.1846,
      "step": 3721
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7424728968867788,
      "learning_rate": 8.894140670149082e-07,
      "loss": 0.2105,
      "step": 3722
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8054171063770955,
      "learning_rate": 8.874009450359428e-07,
      "loss": 0.2237,
      "step": 3723
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7362633340293888,
      "learning_rate": 8.853898820276402e-07,
      "loss": 0.1904,
      "step": 3724
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.738527929525012,
      "learning_rate": 8.833808789968362e-07,
      "loss": 0.2124,
      "step": 3725
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7545616635521347,
      "learning_rate": 8.813739369493395e-07,
      "loss": 0.1913,
      "step": 3726
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8078699446281565,
      "learning_rate": 8.793690568899216e-07,
      "loss": 0.216,
      "step": 3727
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7422594826586738,
      "learning_rate": 8.773662398223276e-07,
      "loss": 0.1979,
      "step": 3728
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8053088273960809,
      "learning_rate": 8.753654867492645e-07,
      "loss": 0.2084,
      "step": 3729
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7394051472669031,
      "learning_rate": 8.733667986724109e-07,
      "loss": 0.2108,
      "step": 3730
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8086728180523498,
      "learning_rate": 8.713701765924071e-07,
      "loss": 0.2219,
      "step": 3731
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7374959226868493,
      "learning_rate": 8.693756215088617e-07,
      "loss": 0.1885,
      "step": 3732
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7379993843081575,
      "learning_rate": 8.673831344203454e-07,
      "loss": 0.2004,
      "step": 3733
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7501187670859342,
      "learning_rate": 8.653927163243986e-07,
      "loss": 0.2015,
      "step": 3734
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7592842728785878,
      "learning_rate": 8.634043682175231e-07,
      "loss": 0.202,
      "step": 3735
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7571119029188337,
      "learning_rate": 8.61418091095183e-07,
      "loss": 0.2065,
      "step": 3736
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7845150609491834,
      "learning_rate": 8.594338859518081e-07,
      "loss": 0.2254,
      "step": 3737
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7222718944934212,
      "learning_rate": 8.574517537807897e-07,
      "loss": 0.192,
      "step": 3738
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7740563615471231,
      "learning_rate": 8.554716955744808e-07,
      "loss": 0.212,
      "step": 3739
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.729680726268439,
      "learning_rate": 8.534937123241949e-07,
      "loss": 0.2009,
      "step": 3740
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7760440331544337,
      "learning_rate": 8.515178050202116e-07,
      "loss": 0.2009,
      "step": 3741
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7486031531683077,
      "learning_rate": 8.495439746517648e-07,
      "loss": 0.1953,
      "step": 3742
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7747395378482759,
      "learning_rate": 8.475722222070542e-07,
      "loss": 0.2157,
      "step": 3743
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7194260729000117,
      "learning_rate": 8.45602548673235e-07,
      "loss": 0.1875,
      "step": 3744
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7843732823277773,
      "learning_rate": 8.436349550364248e-07,
      "loss": 0.2186,
      "step": 3745
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8192120617751015,
      "learning_rate": 8.416694422816974e-07,
      "loss": 0.2178,
      "step": 3746
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7765598302733261,
      "learning_rate": 8.39706011393086e-07,
      "loss": 0.2063,
      "step": 3747
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8365498450812161,
      "learning_rate": 8.377446633535797e-07,
      "loss": 0.2304,
      "step": 3748
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7405500665799853,
      "learning_rate": 8.357853991451292e-07,
      "loss": 0.1924,
      "step": 3749
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7814175748676045,
      "learning_rate": 8.338282197486364e-07,
      "loss": 0.2052,
      "step": 3750
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7841258267109303,
      "learning_rate": 8.31873126143965e-07,
      "loss": 0.2035,
      "step": 3751
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7461980523252878,
      "learning_rate": 8.299201193099287e-07,
      "loss": 0.1928,
      "step": 3752
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7509839141143676,
      "learning_rate": 8.279692002243028e-07,
      "loss": 0.2103,
      "step": 3753
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7477315636786586,
      "learning_rate": 8.260203698638114e-07,
      "loss": 0.195,
      "step": 3754
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7540131304165281,
      "learning_rate": 8.240736292041368e-07,
      "loss": 0.2071,
      "step": 3755
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7497714498687541,
      "learning_rate": 8.221289792199117e-07,
      "loss": 0.2043,
      "step": 3756
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7465909128944268,
      "learning_rate": 8.20186420884726e-07,
      "loss": 0.1981,
      "step": 3757
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.762037707997367,
      "learning_rate": 8.182459551711197e-07,
      "loss": 0.199,
      "step": 3758
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7720783989419915,
      "learning_rate": 8.163075830505862e-07,
      "loss": 0.2137,
      "step": 3759
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7543367060729267,
      "learning_rate": 8.143713054935715e-07,
      "loss": 0.2102,
      "step": 3760
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7916633928295919,
      "learning_rate": 8.124371234694716e-07,
      "loss": 0.2191,
      "step": 3761
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7822168898863536,
      "learning_rate": 8.105050379466334e-07,
      "loss": 0.2121,
      "step": 3762
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7656819000575368,
      "learning_rate": 8.085750498923528e-07,
      "loss": 0.2096,
      "step": 3763
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7679996885199054,
      "learning_rate": 8.066471602728804e-07,
      "loss": 0.2152,
      "step": 3764
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7925243923575197,
      "learning_rate": 8.047213700534113e-07,
      "loss": 0.2127,
      "step": 3765
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7728377553479,
      "learning_rate": 8.027976801980924e-07,
      "loss": 0.2086,
      "step": 3766
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7912158416949301,
      "learning_rate": 8.008760916700164e-07,
      "loss": 0.2095,
      "step": 3767
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7638475570962862,
      "learning_rate": 7.989566054312286e-07,
      "loss": 0.1969,
      "step": 3768
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7560529055804616,
      "learning_rate": 7.970392224427154e-07,
      "loss": 0.2077,
      "step": 3769
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7853357051171036,
      "learning_rate": 7.95123943664417e-07,
      "loss": 0.1944,
      "step": 3770
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.826860419080168,
      "learning_rate": 7.932107700552145e-07,
      "loss": 0.2182,
      "step": 3771
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7194458913962135,
      "learning_rate": 7.91299702572938e-07,
      "loss": 0.1876,
      "step": 3772
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7194928300485914,
      "learning_rate": 7.893907421743613e-07,
      "loss": 0.1948,
      "step": 3773
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7853742635538132,
      "learning_rate": 7.87483889815207e-07,
      "loss": 0.2059,
      "step": 3774
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7873987642976027,
      "learning_rate": 7.855791464501367e-07,
      "loss": 0.2316,
      "step": 3775
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7234643902147747,
      "learning_rate": 7.836765130327622e-07,
      "loss": 0.196,
      "step": 3776
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7350268132421258,
      "learning_rate": 7.817759905156335e-07,
      "loss": 0.1939,
      "step": 3777
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7674127570732878,
      "learning_rate": 7.798775798502484e-07,
      "loss": 0.202,
      "step": 3778
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7931030843158331,
      "learning_rate": 7.779812819870442e-07,
      "loss": 0.2084,
      "step": 3779
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7584347836829588,
      "learning_rate": 7.760870978754015e-07,
      "loss": 0.2197,
      "step": 3780
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7596258094643521,
      "learning_rate": 7.741950284636418e-07,
      "loss": 0.2098,
      "step": 3781
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7931768598810862,
      "learning_rate": 7.723050746990291e-07,
      "loss": 0.2279,
      "step": 3782
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7892469326294439,
      "learning_rate": 7.704172375277691e-07,
      "loss": 0.2169,
      "step": 3783
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7893448327569474,
      "learning_rate": 7.685315178950048e-07,
      "loss": 0.2006,
      "step": 3784
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7685612752875797,
      "learning_rate": 7.666479167448232e-07,
      "loss": 0.1946,
      "step": 3785
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7332823989998085,
      "learning_rate": 7.647664350202461e-07,
      "loss": 0.1861,
      "step": 3786
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.735093877392984,
      "learning_rate": 7.628870736632371e-07,
      "loss": 0.1893,
      "step": 3787
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7560135902224459,
      "learning_rate": 7.610098336146965e-07,
      "loss": 0.2158,
      "step": 3788
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7864690781102583,
      "learning_rate": 7.591347158144657e-07,
      "loss": 0.2045,
      "step": 3789
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.755094760282024,
      "learning_rate": 7.572617212013189e-07,
      "loss": 0.2129,
      "step": 3790
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7519964143312511,
      "learning_rate": 7.553908507129726e-07,
      "loss": 0.1982,
      "step": 3791
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7893845516968343,
      "learning_rate": 7.535221052860747e-07,
      "loss": 0.212,
      "step": 3792
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.726477254804541,
      "learning_rate": 7.516554858562142e-07,
      "loss": 0.1944,
      "step": 3793
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7992597598012801,
      "learning_rate": 7.497909933579117e-07,
      "loss": 0.2146,
      "step": 3794
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8200332318594592,
      "learning_rate": 7.479286287246251e-07,
      "loss": 0.2036,
      "step": 3795
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8129217371679206,
      "learning_rate": 7.460683928887441e-07,
      "loss": 0.2124,
      "step": 3796
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7611421118500098,
      "learning_rate": 7.442102867815976e-07,
      "loss": 0.2046,
      "step": 3797
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8163735940865139,
      "learning_rate": 7.423543113334436e-07,
      "loss": 0.2174,
      "step": 3798
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7773098541456359,
      "learning_rate": 7.405004674734773e-07,
      "loss": 0.2033,
      "step": 3799
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7602268101025859,
      "learning_rate": 7.386487561298223e-07,
      "loss": 0.1986,
      "step": 3800
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7840568241079902,
      "learning_rate": 7.367991782295392e-07,
      "loss": 0.2117,
      "step": 3801
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7985046423826075,
      "learning_rate": 7.349517346986173e-07,
      "loss": 0.2012,
      "step": 3802
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7751186128953454,
      "learning_rate": 7.331064264619786e-07,
      "loss": 0.2004,
      "step": 3803
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7589142372286611,
      "learning_rate": 7.312632544434739e-07,
      "loss": 0.2082,
      "step": 3804
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7526775463128181,
      "learning_rate": 7.29422219565889e-07,
      "loss": 0.1959,
      "step": 3805
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.772493108263315,
      "learning_rate": 7.275833227509344e-07,
      "loss": 0.2157,
      "step": 3806
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7416567666233843,
      "learning_rate": 7.257465649192547e-07,
      "loss": 0.2028,
      "step": 3807
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7966739896741277,
      "learning_rate": 7.239119469904227e-07,
      "loss": 0.2181,
      "step": 3808
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8108923118455363,
      "learning_rate": 7.220794698829364e-07,
      "loss": 0.2097,
      "step": 3809
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7503186787715664,
      "learning_rate": 7.202491345142288e-07,
      "loss": 0.1972,
      "step": 3810
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7500198921486868,
      "learning_rate": 7.184209418006505e-07,
      "loss": 0.2062,
      "step": 3811
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7735387120034286,
      "learning_rate": 7.165948926574895e-07,
      "loss": 0.2053,
      "step": 3812
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7645862855176688,
      "learning_rate": 7.147709879989539e-07,
      "loss": 0.1989,
      "step": 3813
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7720685894991374,
      "learning_rate": 7.129492287381828e-07,
      "loss": 0.1963,
      "step": 3814
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7577111962719258,
      "learning_rate": 7.111296157872377e-07,
      "loss": 0.2147,
      "step": 3815
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.758072661353426,
      "learning_rate": 7.093121500571082e-07,
      "loss": 0.1996,
      "step": 3816
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8026126233478904,
      "learning_rate": 7.074968324577053e-07,
      "loss": 0.213,
      "step": 3817
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7206276040086295,
      "learning_rate": 7.056836638978698e-07,
      "loss": 0.1892,
      "step": 3818
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7942652950467015,
      "learning_rate": 7.038726452853617e-07,
      "loss": 0.2184,
      "step": 3819
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7701547802337665,
      "learning_rate": 7.020637775268674e-07,
      "loss": 0.2059,
      "step": 3820
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7556216477025021,
      "learning_rate": 7.002570615279942e-07,
      "loss": 0.2134,
      "step": 3821
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8002624361475621,
      "learning_rate": 6.984524981932756e-07,
      "loss": 0.2234,
      "step": 3822
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7702273593026053,
      "learning_rate": 6.966500884261635e-07,
      "loss": 0.1976,
      "step": 3823
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7793181748050403,
      "learning_rate": 6.948498331290349e-07,
      "loss": 0.2069,
      "step": 3824
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.804834269178791,
      "learning_rate": 6.93051733203185e-07,
      "loss": 0.2186,
      "step": 3825
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7808715449836559,
      "learning_rate": 6.912557895488342e-07,
      "loss": 0.2103,
      "step": 3826
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7718104297072786,
      "learning_rate": 6.894620030651194e-07,
      "loss": 0.1967,
      "step": 3827
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7612903812248829,
      "learning_rate": 6.876703746500984e-07,
      "loss": 0.1955,
      "step": 3828
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7352826229448427,
      "learning_rate": 6.858809052007487e-07,
      "loss": 0.1921,
      "step": 3829
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7674376729476768,
      "learning_rate": 6.84093595612968e-07,
      "loss": 0.1859,
      "step": 3830
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7732634336660945,
      "learning_rate": 6.823084467815732e-07,
      "loss": 0.2027,
      "step": 3831
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7784665680366748,
      "learning_rate": 6.805254596002959e-07,
      "loss": 0.1948,
      "step": 3832
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7779774934440101,
      "learning_rate": 6.787446349617899e-07,
      "loss": 0.2167,
      "step": 3833
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7454958768896152,
      "learning_rate": 6.769659737576229e-07,
      "loss": 0.1724,
      "step": 3834
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7902796479115399,
      "learning_rate": 6.751894768782813e-07,
      "loss": 0.1968,
      "step": 3835
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7647348808912086,
      "learning_rate": 6.73415145213166e-07,
      "loss": 0.203,
      "step": 3836
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.758439556346983,
      "learning_rate": 6.716429796505974e-07,
      "loss": 0.1913,
      "step": 3837
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7952290222830792,
      "learning_rate": 6.698729810778065e-07,
      "loss": 0.2148,
      "step": 3838
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7923208244247107,
      "learning_rate": 6.681051503809455e-07,
      "loss": 0.2019,
      "step": 3839
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7856927460773588,
      "learning_rate": 6.663394884450752e-07,
      "loss": 0.1994,
      "step": 3840
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7423859207219062,
      "learning_rate": 6.645759961541759e-07,
      "loss": 0.1791,
      "step": 3841
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7885357356354573,
      "learning_rate": 6.628146743911374e-07,
      "loss": 0.2101,
      "step": 3842
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7810295938412946,
      "learning_rate": 6.610555240377653e-07,
      "loss": 0.1938,
      "step": 3843
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7497170134889142,
      "learning_rate": 6.592985459747753e-07,
      "loss": 0.1977,
      "step": 3844
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7545131797512679,
      "learning_rate": 6.575437410818008e-07,
      "loss": 0.2072,
      "step": 3845
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7694270119256894,
      "learning_rate": 6.55791110237381e-07,
      "loss": 0.2028,
      "step": 3846
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7751714201220586,
      "learning_rate": 6.540406543189726e-07,
      "loss": 0.202,
      "step": 3847
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.802730343628184,
      "learning_rate": 6.522923742029374e-07,
      "loss": 0.2173,
      "step": 3848
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7060989404131566,
      "learning_rate": 6.505462707645538e-07,
      "loss": 0.1843,
      "step": 3849
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8027977813242296,
      "learning_rate": 6.488023448780068e-07,
      "loss": 0.2104,
      "step": 3850
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8057539392470482,
      "learning_rate": 6.470605974163913e-07,
      "loss": 0.2065,
      "step": 3851
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8091968651695132,
      "learning_rate": 6.453210292517114e-07,
      "loss": 0.2057,
      "step": 3852
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8018496780759629,
      "learning_rate": 6.435836412548835e-07,
      "loss": 0.2047,
      "step": 3853
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7291783104630749,
      "learning_rate": 6.418484342957277e-07,
      "loss": 0.1988,
      "step": 3854
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8249483481516167,
      "learning_rate": 6.401154092429751e-07,
      "loss": 0.2111,
      "step": 3855
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.795938015765473,
      "learning_rate": 6.383845669642657e-07,
      "loss": 0.2181,
      "step": 3856
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7946161387709898,
      "learning_rate": 6.36655908326142e-07,
      "loss": 0.2155,
      "step": 3857
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7729292394175726,
      "learning_rate": 6.349294341940593e-07,
      "loss": 0.193,
      "step": 3858
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.723737882789714,
      "learning_rate": 6.332051454323734e-07,
      "loss": 0.1999,
      "step": 3859
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7737600291494644,
      "learning_rate": 6.3148304290435e-07,
      "loss": 0.2196,
      "step": 3860
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7622621187582135,
      "learning_rate": 6.297631274721567e-07,
      "loss": 0.2039,
      "step": 3861
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.772332046200433,
      "learning_rate": 6.280453999968711e-07,
      "loss": 0.2077,
      "step": 3862
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8125442575667065,
      "learning_rate": 6.263298613384705e-07,
      "loss": 0.2136,
      "step": 3863
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8057989625913227,
      "learning_rate": 6.246165123558401e-07,
      "loss": 0.205,
      "step": 3864
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8025700186209845,
      "learning_rate": 6.229053539067659e-07,
      "loss": 0.207,
      "step": 3865
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.83003404523124,
      "learning_rate": 6.211963868479393e-07,
      "loss": 0.2312,
      "step": 3866
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7939672557282214,
      "learning_rate": 6.194896120349542e-07,
      "loss": 0.2151,
      "step": 3867
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8001447276306493,
      "learning_rate": 6.177850303223059e-07,
      "loss": 0.2138,
      "step": 3868
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8059983369031503,
      "learning_rate": 6.160826425633914e-07,
      "loss": 0.2307,
      "step": 3869
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7676319351342902,
      "learning_rate": 6.143824496105122e-07,
      "loss": 0.1951,
      "step": 3870
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7411827625788316,
      "learning_rate": 6.126844523148672e-07,
      "loss": 0.2014,
      "step": 3871
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.732907369630689,
      "learning_rate": 6.109886515265595e-07,
      "loss": 0.184,
      "step": 3872
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7575927244044426,
      "learning_rate": 6.092950480945897e-07,
      "loss": 0.224,
      "step": 3873
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7886793774731354,
      "learning_rate": 6.076036428668613e-07,
      "loss": 0.2089,
      "step": 3874
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7683192782322438,
      "learning_rate": 6.059144366901737e-07,
      "loss": 0.2189,
      "step": 3875
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8142459266170049,
      "learning_rate": 6.042274304102281e-07,
      "loss": 0.2296,
      "step": 3876
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7544412443580772,
      "learning_rate": 6.025426248716216e-07,
      "loss": 0.207,
      "step": 3877
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.740883283501701,
      "learning_rate": 6.008600209178539e-07,
      "loss": 0.1947,
      "step": 3878
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7778835327676111,
      "learning_rate": 5.991796193913175e-07,
      "loss": 0.2108,
      "step": 3879
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7592624555903269,
      "learning_rate": 5.975014211333053e-07,
      "loss": 0.2114,
      "step": 3880
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7721400225961037,
      "learning_rate": 5.95825426984008e-07,
      "loss": 0.2161,
      "step": 3881
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7567923413211762,
      "learning_rate": 5.941516377825102e-07,
      "loss": 0.2124,
      "step": 3882
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7917095251782685,
      "learning_rate": 5.92480054366793e-07,
      "loss": 0.2085,
      "step": 3883
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7615094521605279,
      "learning_rate": 5.908106775737332e-07,
      "loss": 0.2064,
      "step": 3884
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8253716374865768,
      "learning_rate": 5.891435082391056e-07,
      "loss": 0.2268,
      "step": 3885
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.761559247323199,
      "learning_rate": 5.874785471975753e-07,
      "loss": 0.2118,
      "step": 3886
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7807554024498157,
      "learning_rate": 5.858157952827064e-07,
      "loss": 0.2048,
      "step": 3887
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7894173844550096,
      "learning_rate": 5.841552533269534e-07,
      "loss": 0.2011,
      "step": 3888
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7470887051091237,
      "learning_rate": 5.824969221616666e-07,
      "loss": 0.1833,
      "step": 3889
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.721479037695857,
      "learning_rate": 5.808408026170892e-07,
      "loss": 0.1993,
      "step": 3890
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7346476544902097,
      "learning_rate": 5.791868955223551e-07,
      "loss": 0.2018,
      "step": 3891
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.828618462637648,
      "learning_rate": 5.775352017054919e-07,
      "loss": 0.2275,
      "step": 3892
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7483580230935315,
      "learning_rate": 5.75885721993421e-07,
      "loss": 0.2061,
      "step": 3893
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7502290180236034,
      "learning_rate": 5.742384572119519e-07,
      "loss": 0.2126,
      "step": 3894
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7376918912347482,
      "learning_rate": 5.725934081857887e-07,
      "loss": 0.1841,
      "step": 3895
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.759457605606271,
      "learning_rate": 5.70950575738522e-07,
      "loss": 0.1966,
      "step": 3896
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7535073190746758,
      "learning_rate": 5.693099606926378e-07,
      "loss": 0.215,
      "step": 3897
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7781948853914666,
      "learning_rate": 5.676715638695063e-07,
      "loss": 0.2093,
      "step": 3898
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7231510887330568,
      "learning_rate": 5.66035386089393e-07,
      "loss": 0.1956,
      "step": 3899
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.810668082376125,
      "learning_rate": 5.644014281714483e-07,
      "loss": 0.2297,
      "step": 3900
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7745681891400702,
      "learning_rate": 5.627696909337116e-07,
      "loss": 0.201,
      "step": 3901
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7914598535097331,
      "learning_rate": 5.611401751931112e-07,
      "loss": 0.1922,
      "step": 3902
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7771242144714838,
      "learning_rate": 5.595128817654638e-07,
      "loss": 0.2072,
      "step": 3903
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.782573662531737,
      "learning_rate": 5.578878114654746e-07,
      "loss": 0.2051,
      "step": 3904
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8004628557596253,
      "learning_rate": 5.56264965106732e-07,
      "loss": 0.2058,
      "step": 3905
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7980876861742112,
      "learning_rate": 5.546443435017146e-07,
      "loss": 0.2183,
      "step": 3906
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7576405289337355,
      "learning_rate": 5.530259474617855e-07,
      "loss": 0.1938,
      "step": 3907
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7241907361306976,
      "learning_rate": 5.514097777971939e-07,
      "loss": 0.1874,
      "step": 3908
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7649618188399769,
      "learning_rate": 5.49795835317073e-07,
      "loss": 0.2048,
      "step": 3909
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7540749674876462,
      "learning_rate": 5.481841208294447e-07,
      "loss": 0.2048,
      "step": 3910
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7917540826579184,
      "learning_rate": 5.465746351412104e-07,
      "loss": 0.2252,
      "step": 3911
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7740828106470536,
      "learning_rate": 5.449673790581611e-07,
      "loss": 0.1971,
      "step": 3912
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7902374468109484,
      "learning_rate": 5.433623533849658e-07,
      "loss": 0.203,
      "step": 3913
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7612632350869316,
      "learning_rate": 5.417595589251828e-07,
      "loss": 0.2092,
      "step": 3914
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7562155270514018,
      "learning_rate": 5.401589964812487e-07,
      "loss": 0.2004,
      "step": 3915
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7689183921014295,
      "learning_rate": 5.385606668544846e-07,
      "loss": 0.1965,
      "step": 3916
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7618263193604587,
      "learning_rate": 5.369645708450921e-07,
      "loss": 0.1931,
      "step": 3917
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7260568496864086,
      "learning_rate": 5.353707092521581e-07,
      "loss": 0.1852,
      "step": 3918
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7378089551474638,
      "learning_rate": 5.337790828736466e-07,
      "loss": 0.194,
      "step": 3919
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7865248826903543,
      "learning_rate": 5.321896925064069e-07,
      "loss": 0.2008,
      "step": 3920
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8012289982473807,
      "learning_rate": 5.306025389461639e-07,
      "loss": 0.2103,
      "step": 3921
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.802104697208873,
      "learning_rate": 5.290176229875271e-07,
      "loss": 0.2145,
      "step": 3922
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.773537166885186,
      "learning_rate": 5.274349454239836e-07,
      "loss": 0.2082,
      "step": 3923
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7901801053080739,
      "learning_rate": 5.258545070478999e-07,
      "loss": 0.2256,
      "step": 3924
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7861695772215312,
      "learning_rate": 5.242763086505204e-07,
      "loss": 0.2043,
      "step": 3925
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7427129169427101,
      "learning_rate": 5.227003510219714e-07,
      "loss": 0.1829,
      "step": 3926
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7371453362591904,
      "learning_rate": 5.211266349512534e-07,
      "loss": 0.1967,
      "step": 3927
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7709432866217139,
      "learning_rate": 5.195551612262478e-07,
      "loss": 0.2027,
      "step": 3928
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.796280497095809,
      "learning_rate": 5.179859306337126e-07,
      "loss": 0.2226,
      "step": 3929
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7497847899471237,
      "learning_rate": 5.16418943959282e-07,
      "loss": 0.1949,
      "step": 3930
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.8160315088013836,
      "learning_rate": 5.148542019874664e-07,
      "loss": 0.2144,
      "step": 3931
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7377282024475221,
      "learning_rate": 5.132917055016523e-07,
      "loss": 0.1919,
      "step": 3932
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7692229438304479,
      "learning_rate": 5.117314552841052e-07,
      "loss": 0.2128,
      "step": 3933
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.738888461835676,
      "learning_rate": 5.101734521159613e-07,
      "loss": 0.1949,
      "step": 3934
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7468103184416978,
      "learning_rate": 5.086176967772355e-07,
      "loss": 0.1852,
      "step": 3935
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7535961145079114,
      "learning_rate": 5.070641900468149e-07,
      "loss": 0.204,
      "step": 3936
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7938559127722714,
      "learning_rate": 5.055129327024638e-07,
      "loss": 0.2051,
      "step": 3937
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7463970896544987,
      "learning_rate": 5.039639255208156e-07,
      "loss": 0.1862,
      "step": 3938
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.8050319779379017,
      "learning_rate": 5.024171692773838e-07,
      "loss": 0.2116,
      "step": 3939
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7607418599031393,
      "learning_rate": 5.008726647465468e-07,
      "loss": 0.2048,
      "step": 3940
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7974390693095229,
      "learning_rate": 4.993304127015625e-07,
      "loss": 0.1974,
      "step": 3941
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7603490355589176,
      "learning_rate": 4.977904139145579e-07,
      "loss": 0.2114,
      "step": 3942
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7869725250981948,
      "learning_rate": 4.962526691565333e-07,
      "loss": 0.2117,
      "step": 3943
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7442860970420757,
      "learning_rate": 4.947171791973587e-07,
      "loss": 0.1992,
      "step": 3944
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7258181549791155,
      "learning_rate": 4.931839448057779e-07,
      "loss": 0.1873,
      "step": 3945
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7660144297288481,
      "learning_rate": 4.916529667494024e-07,
      "loss": 0.2153,
      "step": 3946
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7827843393460074,
      "learning_rate": 4.90124245794717e-07,
      "loss": 0.2022,
      "step": 3947
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.793080777189271,
      "learning_rate": 4.885977827070748e-07,
      "loss": 0.2082,
      "step": 3948
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7608741675569402,
      "learning_rate": 4.87073578250698e-07,
      "loss": 0.2031,
      "step": 3949
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7661059194897608,
      "learning_rate": 4.855516331886783e-07,
      "loss": 0.2035,
      "step": 3950
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7980048871809721,
      "learning_rate": 4.840319482829775e-07,
      "loss": 0.2021,
      "step": 3951
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7907218722762239,
      "learning_rate": 4.825145242944262e-07,
      "loss": 0.214,
      "step": 3952
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7103129932414661,
      "learning_rate": 4.809993619827203e-07,
      "loss": 0.1874,
      "step": 3953
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7959372205060216,
      "learning_rate": 4.794864621064266e-07,
      "loss": 0.2329,
      "step": 3954
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7704970631655895,
      "learning_rate": 4.779758254229766e-07,
      "loss": 0.2099,
      "step": 3955
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7713100672523374,
      "learning_rate": 4.7646745268867034e-07,
      "loss": 0.2088,
      "step": 3956
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7938494676467208,
      "learning_rate": 4.749613446586726e-07,
      "loss": 0.2091,
      "step": 3957
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7169788812686507,
      "learning_rate": 4.734575020870169e-07,
      "loss": 0.1894,
      "step": 3958
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.8422559727695784,
      "learning_rate": 4.719559257266004e-07,
      "loss": 0.2175,
      "step": 3959
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.713802064633633,
      "learning_rate": 4.704566163291879e-07,
      "loss": 0.1911,
      "step": 3960
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7559327761697222,
      "learning_rate": 4.689595746454062e-07,
      "loss": 0.1709,
      "step": 3961
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7709181250724749,
      "learning_rate": 4.6746480142475015e-07,
      "loss": 0.2051,
      "step": 3962
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7483726760073884,
      "learning_rate": 4.659722974155767e-07,
      "loss": 0.2057,
      "step": 3963
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7337677209208546,
      "learning_rate": 4.6448206336510646e-07,
      "loss": 0.1823,
      "step": 3964
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7473854030380628,
      "learning_rate": 4.6299410001942436e-07,
      "loss": 0.1991,
      "step": 3965
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.8193401615000145,
      "learning_rate": 4.6150840812348e-07,
      "loss": 0.2111,
      "step": 3966
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7618328745580929,
      "learning_rate": 4.600249884210822e-07,
      "loss": 0.2064,
      "step": 3967
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7963349006392402,
      "learning_rate": 4.5854384165490596e-07,
      "loss": 0.2205,
      "step": 3968
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.785578462296396,
      "learning_rate": 4.5706496856648517e-07,
      "loss": 0.2208,
      "step": 3969
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7089675926612904,
      "learning_rate": 4.5558836989621933e-07,
      "loss": 0.1796,
      "step": 3970
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7752120634677302,
      "learning_rate": 4.541140463833649e-07,
      "loss": 0.2108,
      "step": 3971
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.8085384110759591,
      "learning_rate": 4.526419987660419e-07,
      "loss": 0.2119,
      "step": 3972
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7747716138896887,
      "learning_rate": 4.511722277812286e-07,
      "loss": 0.2146,
      "step": 3973
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7704516413023849,
      "learning_rate": 4.497047341647676e-07,
      "loss": 0.2082,
      "step": 3974
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.8227453760551346,
      "learning_rate": 4.4823951865135663e-07,
      "loss": 0.2198,
      "step": 3975
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.766740301591599,
      "learning_rate": 4.467765819745562e-07,
      "loss": 0.1974,
      "step": 3976
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7328816794093591,
      "learning_rate": 4.4531592486678597e-07,
      "loss": 0.1809,
      "step": 3977
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7806776546362794,
      "learning_rate": 4.43857548059321e-07,
      "loss": 0.1987,
      "step": 3978
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7947626798547669,
      "learning_rate": 4.424014522822995e-07,
      "loss": 0.2052,
      "step": 3979
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.782609406050509,
      "learning_rate": 4.409476382647121e-07,
      "loss": 0.2023,
      "step": 3980
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.8005650901365922,
      "learning_rate": 4.3949610673441223e-07,
      "loss": 0.1961,
      "step": 3981
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7936295105498583,
      "learning_rate": 4.380468584181069e-07,
      "loss": 0.2072,
      "step": 3982
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7815029546587268,
      "learning_rate": 4.365998940413629e-07,
      "loss": 0.2078,
      "step": 3983
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.8004506794193795,
      "learning_rate": 4.351552143286014e-07,
      "loss": 0.2003,
      "step": 3984
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.8092259195723482,
      "learning_rate": 4.33712820003101e-07,
      "loss": 0.2024,
      "step": 3985
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7626618603963052,
      "learning_rate": 4.322727117869951e-07,
      "loss": 0.221,
      "step": 3986
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7979460966637706,
      "learning_rate": 4.3083489040127435e-07,
      "loss": 0.222,
      "step": 3987
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.8297020651804404,
      "learning_rate": 4.293993565657828e-07,
      "loss": 0.209,
      "step": 3988
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.8099419318299228,
      "learning_rate": 4.2796611099921947e-07,
      "loss": 0.2056,
      "step": 3989
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7844834849672034,
      "learning_rate": 4.265351544191365e-07,
      "loss": 0.2032,
      "step": 3990
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.8097687317229884,
      "learning_rate": 4.2510648754194474e-07,
      "loss": 0.2018,
      "step": 3991
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.8159166098424289,
      "learning_rate": 4.2368011108290306e-07,
      "loss": 0.2185,
      "step": 3992
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7275316734295816,
      "learning_rate": 4.222560257561276e-07,
      "loss": 0.1971,
      "step": 3993
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7610871144342101,
      "learning_rate": 4.208342322745845e-07,
      "loss": 0.2076,
      "step": 3994
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7418413575043772,
      "learning_rate": 4.1941473135009627e-07,
      "loss": 0.1889,
      "step": 3995
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7265291326020024,
      "learning_rate": 4.1799752369333313e-07,
      "loss": 0.2007,
      "step": 3996
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7444341132712461,
      "learning_rate": 4.1658261001382096e-07,
      "loss": 0.1808,
      "step": 3997
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7379903254732929,
      "learning_rate": 4.151699910199336e-07,
      "loss": 0.1846,
      "step": 3998
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7875160720118242,
      "learning_rate": 4.1375966741889874e-07,
      "loss": 0.2166,
      "step": 3999
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.786724354998197,
      "learning_rate": 4.123516399167965e-07,
      "loss": 0.2083,
      "step": 4000
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7615003499992956,
      "learning_rate": 4.109459092185519e-07,
      "loss": 0.1993,
      "step": 4001
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7708701105335135,
      "learning_rate": 4.095424760279454e-07,
      "loss": 0.2118,
      "step": 4002
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7529965184366311,
      "learning_rate": 4.0814134104760483e-07,
      "loss": 0.204,
      "step": 4003
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.772474960456954,
      "learning_rate": 4.067425049790075e-07,
      "loss": 0.2059,
      "step": 4004
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.8320324938898096,
      "learning_rate": 4.0534596852247867e-07,
      "loss": 0.2241,
      "step": 4005
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7619123336796962,
      "learning_rate": 4.039517323771958e-07,
      "loss": 0.2045,
      "step": 4006
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7361876899878159,
      "learning_rate": 4.0255979724118043e-07,
      "loss": 0.1998,
      "step": 4007
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7232938892087786,
      "learning_rate": 4.0117016381130636e-07,
      "loss": 0.1881,
      "step": 4008
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7394641609020189,
      "learning_rate": 3.997828327832909e-07,
      "loss": 0.1933,
      "step": 4009
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7554030645785942,
      "learning_rate": 3.983978048517029e-07,
      "loss": 0.202,
      "step": 4010
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.799575557695373,
      "learning_rate": 3.970150807099538e-07,
      "loss": 0.2059,
      "step": 4011
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7900839948512779,
      "learning_rate": 3.956346610503048e-07,
      "loss": 0.2088,
      "step": 4012
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7904155625882897,
      "learning_rate": 3.9425654656386094e-07,
      "loss": 0.2284,
      "step": 4013
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7733062878212277,
      "learning_rate": 3.9288073794057634e-07,
      "loss": 0.2001,
      "step": 4014
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7389904402929933,
      "learning_rate": 3.915072358692479e-07,
      "loss": 0.2054,
      "step": 4015
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.742549452521906,
      "learning_rate": 3.9013604103751933e-07,
      "loss": 0.186,
      "step": 4016
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7351558825636288,
      "learning_rate": 3.887671541318777e-07,
      "loss": 0.1893,
      "step": 4017
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7822065865186768,
      "learning_rate": 3.87400575837657e-07,
      "loss": 0.2017,
      "step": 4018
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7971709610225405,
      "learning_rate": 3.860363068390327e-07,
      "loss": 0.2086,
      "step": 4019
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7911656518010272,
      "learning_rate": 3.8467434781902624e-07,
      "loss": 0.2305,
      "step": 4020
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7805356542202795,
      "learning_rate": 3.833146994595005e-07,
      "loss": 0.2104,
      "step": 4021
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7013658824956588,
      "learning_rate": 3.819573624411643e-07,
      "loss": 0.1886,
      "step": 4022
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.774607864176088,
      "learning_rate": 3.8060233744356634e-07,
      "loss": 0.2074,
      "step": 4023
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.8044625825981996,
      "learning_rate": 3.792496251451e-07,
      "loss": 0.2021,
      "step": 4024
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.8418228026812473,
      "learning_rate": 3.7789922622300023e-07,
      "loss": 0.234,
      "step": 4025
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7896343525881236,
      "learning_rate": 3.765511413533429e-07,
      "loss": 0.2111,
      "step": 4026
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7577155471939302,
      "learning_rate": 3.7520537121104707e-07,
      "loss": 0.1996,
      "step": 4027
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7629543869688551,
      "learning_rate": 3.7386191646987094e-07,
      "loss": 0.2092,
      "step": 4028
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7906497999948122,
      "learning_rate": 3.725207778024148e-07,
      "loss": 0.2019,
      "step": 4029
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.767693627912254,
      "learning_rate": 3.711819558801183e-07,
      "loss": 0.2035,
      "step": 4030
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.707530591489142,
      "learning_rate": 3.69845451373263e-07,
      "loss": 0.1717,
      "step": 4031
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.810005728289289,
      "learning_rate": 3.685112649509681e-07,
      "loss": 0.193,
      "step": 4032
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.762733810843418,
      "learning_rate": 3.671793972811954e-07,
      "loss": 0.2089,
      "step": 4033
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.8035741391716891,
      "learning_rate": 3.6584984903074104e-07,
      "loss": 0.2291,
      "step": 4034
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7445783264579398,
      "learning_rate": 3.645226208652458e-07,
      "loss": 0.2072,
      "step": 4035
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7560760549446365,
      "learning_rate": 3.631977134491832e-07,
      "loss": 0.1986,
      "step": 4036
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7737403050853112,
      "learning_rate": 3.6187512744586873e-07,
      "loss": 0.2205,
      "step": 4037
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7172183555676852,
      "learning_rate": 3.6055486351745327e-07,
      "loss": 0.1972,
      "step": 4038
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.8175606003826346,
      "learning_rate": 3.592369223249281e-07,
      "loss": 0.2289,
      "step": 4039
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7829856719148054,
      "learning_rate": 3.5792130452811825e-07,
      "loss": 0.2026,
      "step": 4040
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7448406269318041,
      "learning_rate": 3.5660801078568785e-07,
      "loss": 0.2,
      "step": 4041
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.754263464884007,
      "learning_rate": 3.5529704175513604e-07,
      "loss": 0.2069,
      "step": 4042
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7756915278383583,
      "learning_rate": 3.539883980928005e-07,
      "loss": 0.198,
      "step": 4043
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.8001908957814998,
      "learning_rate": 3.5268208045385167e-07,
      "loss": 0.2069,
      "step": 4044
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7570532213560428,
      "learning_rate": 3.513780894922969e-07,
      "loss": 0.1924,
      "step": 4045
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7711001754955722,
      "learning_rate": 3.500764258609779e-07,
      "loss": 0.2026,
      "step": 4046
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7684409199443745,
      "learning_rate": 3.487770902115739e-07,
      "loss": 0.1991,
      "step": 4047
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7809821773597568,
      "learning_rate": 3.4748008319459457e-07,
      "loss": 0.2044,
      "step": 4048
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7985931510578306,
      "learning_rate": 3.4618540545938664e-07,
      "loss": 0.2333,
      "step": 4049
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.8019952218998693,
      "learning_rate": 3.4489305765413097e-07,
      "loss": 0.2123,
      "step": 4050
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7558988241493807,
      "learning_rate": 3.4360304042583946e-07,
      "loss": 0.2062,
      "step": 4051
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7597582593556096,
      "learning_rate": 3.4231535442035947e-07,
      "loss": 0.2061,
      "step": 4052
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7892909788969609,
      "learning_rate": 3.410300002823691e-07,
      "loss": 0.2057,
      "step": 4053
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7598437694481458,
      "learning_rate": 3.3974697865538195e-07,
      "loss": 0.1954,
      "step": 4054
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7808849402890604,
      "learning_rate": 3.3846629018174037e-07,
      "loss": 0.2122,
      "step": 4055
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.8040438904973748,
      "learning_rate": 3.37187935502622e-07,
      "loss": 0.2124,
      "step": 4056
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7336131755131836,
      "learning_rate": 3.3591191525803255e-07,
      "loss": 0.1897,
      "step": 4057
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.84786820229128,
      "learning_rate": 3.346382300868134e-07,
      "loss": 0.2008,
      "step": 4058
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.8068512551627965,
      "learning_rate": 3.3336688062663327e-07,
      "loss": 0.2203,
      "step": 4059
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7892389049535463,
      "learning_rate": 3.320978675139919e-07,
      "loss": 0.2199,
      "step": 4060
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7888784356005265,
      "learning_rate": 3.308311913842205e-07,
      "loss": 0.2113,
      "step": 4061
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.789142527506934,
      "learning_rate": 3.2956685287148014e-07,
      "loss": 0.218,
      "step": 4062
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7317937680260385,
      "learning_rate": 3.2830485260876064e-07,
      "loss": 0.2003,
      "step": 4063
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.8052792808239903,
      "learning_rate": 3.2704519122788324e-07,
      "loss": 0.194,
      "step": 4064
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.8010406303764093,
      "learning_rate": 3.2578786935949514e-07,
      "loss": 0.2076,
      "step": 4065
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7895513017625382,
      "learning_rate": 3.24532887633075e-07,
      "loss": 0.1943,
      "step": 4066
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7598304529585908,
      "learning_rate": 3.2328024667692804e-07,
      "loss": 0.199,
      "step": 4067
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.802581704696027,
      "learning_rate": 3.220299471181898e-07,
      "loss": 0.2069,
      "step": 4068
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.783097800548459,
      "learning_rate": 3.207819895828196e-07,
      "loss": 0.2071,
      "step": 4069
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.8072602429404699,
      "learning_rate": 3.195363746956087e-07,
      "loss": 0.2227,
      "step": 4070
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7725558519893608,
      "learning_rate": 3.182931030801728e-07,
      "loss": 0.2044,
      "step": 4071
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7571478173262329,
      "learning_rate": 3.1705217535895504e-07,
      "loss": 0.1985,
      "step": 4072
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.8038296139127521,
      "learning_rate": 3.158135921532268e-07,
      "loss": 0.2191,
      "step": 4073
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7373647951876839,
      "learning_rate": 3.1457735408308153e-07,
      "loss": 0.1922,
      "step": 4074
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7810518237334947,
      "learning_rate": 3.133434617674441e-07,
      "loss": 0.2148,
      "step": 4075
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.758092182789381,
      "learning_rate": 3.121119158240604e-07,
      "loss": 0.2104,
      "step": 4076
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7976989141289751,
      "learning_rate": 3.108827168695039e-07,
      "loss": 0.2034,
      "step": 4077
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7518833909256039,
      "learning_rate": 3.096558655191706e-07,
      "loss": 0.1905,
      "step": 4078
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7780176690657118,
      "learning_rate": 3.084313623872859e-07,
      "loss": 0.2024,
      "step": 4079
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7588963520790166,
      "learning_rate": 3.0720920808689435e-07,
      "loss": 0.1962,
      "step": 4080
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7985424489360149,
      "learning_rate": 3.059894032298688e-07,
      "loss": 0.2079,
      "step": 4081
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7716220934357508,
      "learning_rate": 3.047719484269018e-07,
      "loss": 0.1998,
      "step": 4082
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.817119952487458,
      "learning_rate": 3.035568442875136e-07,
      "loss": 0.2067,
      "step": 4083
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7294353406128303,
      "learning_rate": 3.023440914200448e-07,
      "loss": 0.1973,
      "step": 4084
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7526696974963335,
      "learning_rate": 3.011336904316592e-07,
      "loss": 0.1947,
      "step": 4085
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7528901787194671,
      "learning_rate": 2.9992564192834253e-07,
      "loss": 0.1888,
      "step": 4086
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7656378317372033,
      "learning_rate": 2.9871994651490543e-07,
      "loss": 0.2029,
      "step": 4087
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7626509903275484,
      "learning_rate": 2.9751660479497737e-07,
      "loss": 0.1874,
      "step": 4088
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7679765605957137,
      "learning_rate": 2.963156173710119e-07,
      "loss": 0.1906,
      "step": 4089
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7760109602588635,
      "learning_rate": 2.9511698484428144e-07,
      "loss": 0.2039,
      "step": 4090
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7707351789526306,
      "learning_rate": 2.93920707814882e-07,
      "loss": 0.1942,
      "step": 4091
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7181517485463093,
      "learning_rate": 2.9272678688172905e-07,
      "loss": 0.191,
      "step": 4092
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7608205995652737,
      "learning_rate": 2.915352226425583e-07,
      "loss": 0.2245,
      "step": 4093
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.837941636349983,
      "learning_rate": 2.9034601569392417e-07,
      "loss": 0.2147,
      "step": 4094
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7608804516808214,
      "learning_rate": 2.891591666312049e-07,
      "loss": 0.2038,
      "step": 4095
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7082821821712684,
      "learning_rate": 2.879746760485946e-07,
      "loss": 0.1925,
      "step": 4096
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8429817336192562,
      "learning_rate": 2.867925445391079e-07,
      "loss": 0.225,
      "step": 4097
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7500561721703212,
      "learning_rate": 2.85612772694579e-07,
      "loss": 0.1867,
      "step": 4098
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7319651377120722,
      "learning_rate": 2.8443536110565993e-07,
      "loss": 0.1967,
      "step": 4099
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7711129020330675,
      "learning_rate": 2.832603103618214e-07,
      "loss": 0.1932,
      "step": 4100
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7881787980336648,
      "learning_rate": 2.820876210513507e-07,
      "loss": 0.2006,
      "step": 4101
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7543695060120207,
      "learning_rate": 2.809172937613552e-07,
      "loss": 0.1998,
      "step": 4102
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.797100424199389,
      "learning_rate": 2.7974932907775863e-07,
      "loss": 0.2176,
      "step": 4103
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7774544985975386,
      "learning_rate": 2.7858372758530206e-07,
      "loss": 0.2101,
      "step": 4104
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7923328734533642,
      "learning_rate": 2.77420489867542e-07,
      "loss": 0.2023,
      "step": 4105
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8391517052938328,
      "learning_rate": 2.762596165068548e-07,
      "loss": 0.2212,
      "step": 4106
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8611938622513461,
      "learning_rate": 2.7510110808443014e-07,
      "loss": 0.2348,
      "step": 4107
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.877949916209416,
      "learning_rate": 2.739449651802756e-07,
      "loss": 0.2258,
      "step": 4108
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7633855661156186,
      "learning_rate": 2.7279118837321094e-07,
      "loss": 0.2118,
      "step": 4109
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.765751310908599,
      "learning_rate": 2.7163977824087696e-07,
      "loss": 0.2013,
      "step": 4110
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7913753742785347,
      "learning_rate": 2.7049073535972415e-07,
      "loss": 0.2006,
      "step": 4111
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7635734486759279,
      "learning_rate": 2.6934406030502235e-07,
      "loss": 0.2079,
      "step": 4112
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7702417496309857,
      "learning_rate": 2.6819975365085237e-07,
      "loss": 0.2172,
      "step": 4113
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7757122657705645,
      "learning_rate": 2.6705781597011147e-07,
      "loss": 0.2176,
      "step": 4114
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7720944337650785,
      "learning_rate": 2.659182478345096e-07,
      "loss": 0.2012,
      "step": 4115
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.721073336721064,
      "learning_rate": 2.6478104981457254e-07,
      "loss": 0.1839,
      "step": 4116
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7230684312586587,
      "learning_rate": 2.63646222479636e-07,
      "loss": 0.1976,
      "step": 4117
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7557079808853059,
      "learning_rate": 2.6251376639785163e-07,
      "loss": 0.1971,
      "step": 4118
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7585325367505716,
      "learning_rate": 2.613836821361826e-07,
      "loss": 0.2039,
      "step": 4119
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.8107172017710597,
      "learning_rate": 2.6025597026040463e-07,
      "loss": 0.2154,
      "step": 4120
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.8098377488651566,
      "learning_rate": 2.5913063133510776e-07,
      "loss": 0.2102,
      "step": 4121
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7679521937884232,
      "learning_rate": 2.5800766592369077e-07,
      "loss": 0.1876,
      "step": 4122
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7436005706222404,
      "learning_rate": 2.5688707458836724e-07,
      "loss": 0.2046,
      "step": 4123
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7714648677497543,
      "learning_rate": 2.5576885789015905e-07,
      "loss": 0.1797,
      "step": 4124
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.79220853026333,
      "learning_rate": 2.5465301638890216e-07,
      "loss": 0.2156,
      "step": 4125
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7904534520186347,
      "learning_rate": 2.535395506432403e-07,
      "loss": 0.21,
      "step": 4126
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.813254788053044,
      "learning_rate": 2.524284612106315e-07,
      "loss": 0.22,
      "step": 4127
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7892176276508218,
      "learning_rate": 2.5131974864734063e-07,
      "loss": 0.2283,
      "step": 4128
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.8107259131034734,
      "learning_rate": 2.50213413508445e-07,
      "loss": 0.2167,
      "step": 4129
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7470206453820231,
      "learning_rate": 2.4910945634783024e-07,
      "loss": 0.1931,
      "step": 4130
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7854194032488793,
      "learning_rate": 2.4800787771819266e-07,
      "loss": 0.214,
      "step": 4131
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.8121236448599444,
      "learning_rate": 2.469086781710367e-07,
      "loss": 0.2119,
      "step": 4132
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.785184320355899,
      "learning_rate": 2.45811858256676e-07,
      "loss": 0.2005,
      "step": 4133
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7982240563933859,
      "learning_rate": 2.447174185242324e-07,
      "loss": 0.2069,
      "step": 4134
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.743057005578591,
      "learning_rate": 2.436253595216376e-07,
      "loss": 0.1924,
      "step": 4135
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7963311457985538,
      "learning_rate": 2.425356817956287e-07,
      "loss": 0.2236,
      "step": 4136
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7683525715895655,
      "learning_rate": 2.4144838589175433e-07,
      "loss": 0.2062,
      "step": 4137
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7537637670342082,
      "learning_rate": 2.403634723543674e-07,
      "loss": 0.1966,
      "step": 4138
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7865323255001937,
      "learning_rate": 2.3928094172663005e-07,
      "loss": 0.2065,
      "step": 4139
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7799177751372975,
      "learning_rate": 2.3820079455051048e-07,
      "loss": 0.2008,
      "step": 4140
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7699031215427693,
      "learning_rate": 2.3712303136678383e-07,
      "loss": 0.2079,
      "step": 4141
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7820364238318862,
      "learning_rate": 2.3604765271503073e-07,
      "loss": 0.2117,
      "step": 4142
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7354593062254687,
      "learning_rate": 2.3497465913364047e-07,
      "loss": 0.1981,
      "step": 4143
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7694407146236549,
      "learning_rate": 2.3390405115980563e-07,
      "loss": 0.2026,
      "step": 4144
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.8065033140063605,
      "learning_rate": 2.3283582932952575e-07,
      "loss": 0.2084,
      "step": 4145
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7630202374350333,
      "learning_rate": 2.3176999417760637e-07,
      "loss": 0.2049,
      "step": 4146
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7677056185688477,
      "learning_rate": 2.3070654623765677e-07,
      "loss": 0.1927,
      "step": 4147
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7520979636120412,
      "learning_rate": 2.2964548604209214e-07,
      "loss": 0.1953,
      "step": 4148
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.741203165459052,
      "learning_rate": 2.2858681412213034e-07,
      "loss": 0.198,
      "step": 4149
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.8215641052910667,
      "learning_rate": 2.2753053100779632e-07,
      "loss": 0.2059,
      "step": 4150
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7673634754294941,
      "learning_rate": 2.2647663722791645e-07,
      "loss": 0.1997,
      "step": 4151
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7543362283867263,
      "learning_rate": 2.2542513331012372e-07,
      "loss": 0.2065,
      "step": 4152
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7321105794686666,
      "learning_rate": 2.2437601978085144e-07,
      "loss": 0.1998,
      "step": 4153
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7607884044961705,
      "learning_rate": 2.2332929716533947e-07,
      "loss": 0.2016,
      "step": 4154
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.736901192598569,
      "learning_rate": 2.2228496598762696e-07,
      "loss": 0.193,
      "step": 4155
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7542831169437886,
      "learning_rate": 2.2124302677055954e-07,
      "loss": 0.2091,
      "step": 4156
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.8161382910320536,
      "learning_rate": 2.202034800357833e-07,
      "loss": 0.2364,
      "step": 4157
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7744368094667119,
      "learning_rate": 2.1916632630374579e-07,
      "loss": 0.1891,
      "step": 4158
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7974809464608472,
      "learning_rate": 2.1813156609369778e-07,
      "loss": 0.2211,
      "step": 4159
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7720537495326883,
      "learning_rate": 2.1709919992369211e-07,
      "loss": 0.2138,
      "step": 4160
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.8220197895634637,
      "learning_rate": 2.1606922831058198e-07,
      "loss": 0.2329,
      "step": 4161
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.8211963782595186,
      "learning_rate": 2.1504165177002212e-07,
      "loss": 0.2092,
      "step": 4162
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7807695950222003,
      "learning_rate": 2.1401647081646825e-07,
      "loss": 0.2132,
      "step": 4163
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.779507186452254,
      "learning_rate": 2.1299368596317705e-07,
      "loss": 0.2165,
      "step": 4164
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7784052438142887,
      "learning_rate": 2.1197329772220555e-07,
      "loss": 0.1982,
      "step": 4165
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7124381268625513,
      "learning_rate": 2.1095530660440954e-07,
      "loss": 0.1793,
      "step": 4166
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7695903904736504,
      "learning_rate": 2.0993971311944528e-07,
      "loss": 0.2191,
      "step": 4167
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.778478049090742,
      "learning_rate": 2.0892651777577045e-07,
      "loss": 0.2051,
      "step": 4168
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7663269695418208,
      "learning_rate": 2.0791572108064152e-07,
      "loss": 0.1959,
      "step": 4169
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7929910326696494,
      "learning_rate": 2.069073235401109e-07,
      "loss": 0.212,
      "step": 4170
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7794613752828358,
      "learning_rate": 2.0590132565903475e-07,
      "loss": 0.2052,
      "step": 4171
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7748687571374892,
      "learning_rate": 2.0489772794106411e-07,
      "loss": 0.2175,
      "step": 4172
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.8282039099856301,
      "learning_rate": 2.0389653088865035e-07,
      "loss": 0.2062,
      "step": 4173
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.8371533627558525,
      "learning_rate": 2.028977350030409e-07,
      "loss": 0.2293,
      "step": 4174
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7578591805949878,
      "learning_rate": 2.0190134078428402e-07,
      "loss": 0.1926,
      "step": 4175
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.785286975100647,
      "learning_rate": 2.0090734873122243e-07,
      "loss": 0.2142,
      "step": 4176
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7306791190539608,
      "learning_rate": 1.9991575934149965e-07,
      "loss": 0.1749,
      "step": 4177
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7877485241422459,
      "learning_rate": 1.989265731115525e-07,
      "loss": 0.2129,
      "step": 4178
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7583701470072386,
      "learning_rate": 1.9793979053661815e-07,
      "loss": 0.2082,
      "step": 4179
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7554391069481399,
      "learning_rate": 1.969554121107281e-07,
      "loss": 0.1937,
      "step": 4180
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7790217550220512,
      "learning_rate": 1.9597343832671144e-07,
      "loss": 0.2172,
      "step": 4181
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7916085672553715,
      "learning_rate": 1.9499386967619104e-07,
      "loss": 0.2092,
      "step": 4182
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7143163080898066,
      "learning_rate": 1.940167066495896e-07,
      "loss": 0.1982,
      "step": 4183
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7390194546318596,
      "learning_rate": 1.9304194973612244e-07,
      "loss": 0.1822,
      "step": 4184
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7341586233675971,
      "learning_rate": 1.920695994238009e-07,
      "loss": 0.1929,
      "step": 4185
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7617563613267784,
      "learning_rate": 1.910996561994316e-07,
      "loss": 0.1964,
      "step": 4186
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7855538323292904,
      "learning_rate": 1.901321205486173e-07,
      "loss": 0.1972,
      "step": 4187
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.8201843805736307,
      "learning_rate": 1.8916699295575324e-07,
      "loss": 0.2234,
      "step": 4188
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7870666938969759,
      "learning_rate": 1.8820427390403074e-07,
      "loss": 0.2091,
      "step": 4189
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7650931078501292,
      "learning_rate": 1.872439638754331e-07,
      "loss": 0.191,
      "step": 4190
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7601651091015967,
      "learning_rate": 1.8628606335074017e-07,
      "loss": 0.2053,
      "step": 4191
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7839256963134195,
      "learning_rate": 1.853305728095245e-07,
      "loss": 0.2108,
      "step": 4192
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7912197206822676,
      "learning_rate": 1.8437749273015116e-07,
      "loss": 0.2151,
      "step": 4193
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7113025008351793,
      "learning_rate": 1.834268235897807e-07,
      "loss": 0.1827,
      "step": 4194
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7579945809013869,
      "learning_rate": 1.8247856586436352e-07,
      "loss": 0.1898,
      "step": 4195
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.730415467266727,
      "learning_rate": 1.8153272002864542e-07,
      "loss": 0.1903,
      "step": 4196
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7293428242878267,
      "learning_rate": 1.805892865561637e-07,
      "loss": 0.1925,
      "step": 4197
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7389210108238391,
      "learning_rate": 1.7964826591924722e-07,
      "loss": 0.1951,
      "step": 4198
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7727442318382269,
      "learning_rate": 1.787096585890169e-07,
      "loss": 0.2066,
      "step": 4199
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7647889988730195,
      "learning_rate": 1.7777346503538796e-07,
      "loss": 0.2148,
      "step": 4200
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7831686834892116,
      "learning_rate": 1.7683968572706323e-07,
      "loss": 0.2071,
      "step": 4201
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.750416614349471,
      "learning_rate": 1.7590832113154045e-07,
      "loss": 0.1902,
      "step": 4202
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7731779123081873,
      "learning_rate": 1.749793717151055e-07,
      "loss": 0.2047,
      "step": 4203
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7740146581725315,
      "learning_rate": 1.7405283794283745e-07,
      "loss": 0.1964,
      "step": 4204
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7418129788969433,
      "learning_rate": 1.731287202786053e-07,
      "loss": 0.2089,
      "step": 4205
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.8145131070949655,
      "learning_rate": 1.7220701918506666e-07,
      "loss": 0.2198,
      "step": 4206
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7117484190746248,
      "learning_rate": 1.7128773512367137e-07,
      "loss": 0.1803,
      "step": 4207
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.8168215780887461,
      "learning_rate": 1.7037086855465902e-07,
      "loss": 0.2164,
      "step": 4208
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.8059490884038136,
      "learning_rate": 1.6945641993705743e-07,
      "loss": 0.2267,
      "step": 4209
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7837680478489482,
      "learning_rate": 1.685443897286859e-07,
      "loss": 0.2006,
      "step": 4210
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7731731574539111,
      "learning_rate": 1.6763477838615093e-07,
      "loss": 0.2029,
      "step": 4211
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.770726951641516,
      "learning_rate": 1.6672758636485043e-07,
      "loss": 0.2065,
      "step": 4212
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7924657670485571,
      "learning_rate": 1.6582281411896827e-07,
      "loss": 0.2218,
      "step": 4213
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7613353382153476,
      "learning_rate": 1.6492046210147884e-07,
      "loss": 0.21,
      "step": 4214
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.767290598660112,
      "learning_rate": 1.6402053076414294e-07,
      "loss": 0.1984,
      "step": 4215
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.8107247763047389,
      "learning_rate": 1.6312302055751295e-07,
      "loss": 0.2022,
      "step": 4216
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7371782088088853,
      "learning_rate": 1.62227931930925e-07,
      "loss": 0.1895,
      "step": 4217
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7366661868924376,
      "learning_rate": 1.6133526533250566e-07,
      "loss": 0.1836,
      "step": 4218
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.766135531215323,
      "learning_rate": 1.604450212091685e-07,
      "loss": 0.2153,
      "step": 4219
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7597876952145319,
      "learning_rate": 1.5955720000661312e-07,
      "loss": 0.197,
      "step": 4220
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7602654853053262,
      "learning_rate": 1.5867180216932676e-07,
      "loss": 0.1995,
      "step": 4221
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7482290790733929,
      "learning_rate": 1.577888281405826e-07,
      "loss": 0.1857,
      "step": 4222
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7700659271567152,
      "learning_rate": 1.5690827836244317e-07,
      "loss": 0.2042,
      "step": 4223
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7757270817531734,
      "learning_rate": 1.5603015327575354e-07,
      "loss": 0.2048,
      "step": 4224
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7471028279386743,
      "learning_rate": 1.551544533201471e-07,
      "loss": 0.1977,
      "step": 4225
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7716057869957753,
      "learning_rate": 1.5428117893404315e-07,
      "loss": 0.2092,
      "step": 4226
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.757122227037295,
      "learning_rate": 1.5341033055464527e-07,
      "loss": 0.1973,
      "step": 4227
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7534579076641588,
      "learning_rate": 1.5254190861794415e-07,
      "loss": 0.1989,
      "step": 4228
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7905178984118182,
      "learning_rate": 1.5167591355871425e-07,
      "loss": 0.2144,
      "step": 4229
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7799755200764374,
      "learning_rate": 1.5081234581051484e-07,
      "loss": 0.2165,
      "step": 4230
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7758900589468746,
      "learning_rate": 1.4995120580569178e-07,
      "loss": 0.2069,
      "step": 4231
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.8143376740641987,
      "learning_rate": 1.490924939753735e-07,
      "loss": 0.2268,
      "step": 4232
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7641788477423427,
      "learning_rate": 1.4823621074947503e-07,
      "loss": 0.2145,
      "step": 4233
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.74805357223315,
      "learning_rate": 1.473823565566923e-07,
      "loss": 0.1989,
      "step": 4234
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7743405470443272,
      "learning_rate": 1.4653093182450896e-07,
      "loss": 0.2029,
      "step": 4235
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7744667192837427,
      "learning_rate": 1.4568193697918843e-07,
      "loss": 0.2138,
      "step": 4236
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7451286510825526,
      "learning_rate": 1.4483537244578183e-07,
      "loss": 0.1937,
      "step": 4237
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7467102265435055,
      "learning_rate": 1.4399123864811904e-07,
      "loss": 0.1993,
      "step": 4238
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7108741716925544,
      "learning_rate": 1.4314953600881641e-07,
      "loss": 0.2001,
      "step": 4239
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7521918294835498,
      "learning_rate": 1.4231026494927082e-07,
      "loss": 0.2139,
      "step": 4240
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.833252690939737,
      "learning_rate": 1.414734258896644e-07,
      "loss": 0.2424,
      "step": 4241
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.81688768249614,
      "learning_rate": 1.4063901924895985e-07,
      "loss": 0.2168,
      "step": 4242
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7980809245117071,
      "learning_rate": 1.398070454449013e-07,
      "loss": 0.2058,
      "step": 4243
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.8020584533216626,
      "learning_rate": 1.3897750489401785e-07,
      "loss": 0.2052,
      "step": 4244
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7712973982307767,
      "learning_rate": 1.3815039801161723e-07,
      "loss": 0.2054,
      "step": 4245
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7757441857195778,
      "learning_rate": 1.3732572521179043e-07,
      "loss": 0.225,
      "step": 4246
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7933039537398766,
      "learning_rate": 1.3650348690740888e-07,
      "loss": 0.2063,
      "step": 4247
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7861120349423525,
      "learning_rate": 1.3568368351012718e-07,
      "loss": 0.2147,
      "step": 4248
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7705089851888045,
      "learning_rate": 1.348663154303781e-07,
      "loss": 0.2197,
      "step": 4249
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7951792943782844,
      "learning_rate": 1.3405138307737765e-07,
      "loss": 0.2159,
      "step": 4250
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7501078362487948,
      "learning_rate": 1.3323888685912e-07,
      "loss": 0.2059,
      "step": 4251
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7960767527859273,
      "learning_rate": 1.324288271823826e-07,
      "loss": 0.2249,
      "step": 4252
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.780274447770514,
      "learning_rate": 1.3162120445272096e-07,
      "loss": 0.1905,
      "step": 4253
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7328557823235766,
      "learning_rate": 1.3081601907447007e-07,
      "loss": 0.1956,
      "step": 4254
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.8133330492029422,
      "learning_rate": 1.3001327145074572e-07,
      "loss": 0.2008,
      "step": 4255
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7482304957535925,
      "learning_rate": 1.2921296198344424e-07,
      "loss": 0.2172,
      "step": 4256
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7896191517022075,
      "learning_rate": 1.2841509107323845e-07,
      "loss": 0.1935,
      "step": 4257
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7386018801871933,
      "learning_rate": 1.2761965911958385e-07,
      "loss": 0.1905,
      "step": 4258
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7453255621956824,
      "learning_rate": 1.2682666652071184e-07,
      "loss": 0.2036,
      "step": 4259
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.8080819879191392,
      "learning_rate": 1.260361136736349e-07,
      "loss": 0.217,
      "step": 4260
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7478551158041037,
      "learning_rate": 1.2524800097414193e-07,
      "loss": 0.1987,
      "step": 4261
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7643126702094263,
      "learning_rate": 1.244623288168012e-07,
      "loss": 0.2053,
      "step": 4262
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.8279808135264742,
      "learning_rate": 1.236790975949592e-07,
      "loss": 0.2212,
      "step": 4263
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7342585281734587,
      "learning_rate": 1.2289830770074064e-07,
      "loss": 0.1829,
      "step": 4264
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7713646947240512,
      "learning_rate": 1.2211995952504663e-07,
      "loss": 0.2183,
      "step": 4265
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7667644771420201,
      "learning_rate": 1.2134405345755773e-07,
      "loss": 0.2103,
      "step": 4266
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7686612577083308,
      "learning_rate": 1.20570589886731e-07,
      "loss": 0.2062,
      "step": 4267
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7727971241547719,
      "learning_rate": 1.1979956919979996e-07,
      "loss": 0.1997,
      "step": 4268
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7898337041250693,
      "learning_rate": 1.1903099178277644e-07,
      "loss": 0.211,
      "step": 4269
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7649174022294443,
      "learning_rate": 1.1826485802044707e-07,
      "loss": 0.1882,
      "step": 4270
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7216385122528296,
      "learning_rate": 1.1750116829637837e-07,
      "loss": 0.1859,
      "step": 4271
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.834623283099054,
      "learning_rate": 1.1673992299290893e-07,
      "loss": 0.2047,
      "step": 4272
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7789570465560631,
      "learning_rate": 1.1598112249115723e-07,
      "loss": 0.2031,
      "step": 4273
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8027529402041206,
      "learning_rate": 1.1522476717101605e-07,
      "loss": 0.2087,
      "step": 4274
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7777469016606495,
      "learning_rate": 1.144708574111547e-07,
      "loss": 0.2115,
      "step": 4275
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8461206625136993,
      "learning_rate": 1.137193935890174e-07,
      "loss": 0.2099,
      "step": 4276
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7651433938026521,
      "learning_rate": 1.1297037608082372e-07,
      "loss": 0.2053,
      "step": 4277
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.772817621839532,
      "learning_rate": 1.1222380526156929e-07,
      "loss": 0.2076,
      "step": 4278
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7909073649320799,
      "learning_rate": 1.1147968150502453e-07,
      "loss": 0.2053,
      "step": 4279
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7940188177148867,
      "learning_rate": 1.1073800518373423e-07,
      "loss": 0.1977,
      "step": 4280
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7391642780604318,
      "learning_rate": 1.099987766690186e-07,
      "loss": 0.1986,
      "step": 4281
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.753949307941398,
      "learning_rate": 1.0926199633097156e-07,
      "loss": 0.2001,
      "step": 4282
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7444095291101427,
      "learning_rate": 1.0852766453846308e-07,
      "loss": 0.1892,
      "step": 4283
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7513034608879484,
      "learning_rate": 1.0779578165913463e-07,
      "loss": 0.1892,
      "step": 4284
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7460594116063586,
      "learning_rate": 1.0706634805940363e-07,
      "loss": 0.1895,
      "step": 4285
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7753992350337001,
      "learning_rate": 1.0633936410446077e-07,
      "loss": 0.2182,
      "step": 4286
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7771069473350695,
      "learning_rate": 1.056148301582699e-07,
      "loss": 0.2162,
      "step": 4287
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8079470520412888,
      "learning_rate": 1.0489274658356808e-07,
      "loss": 0.204,
      "step": 4288
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8210500732893999,
      "learning_rate": 1.0417311374186668e-07,
      "loss": 0.2272,
      "step": 4289
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7762271753893906,
      "learning_rate": 1.0345593199344972e-07,
      "loss": 0.1978,
      "step": 4290
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8123818264596853,
      "learning_rate": 1.0274120169737334e-07,
      "loss": 0.2097,
      "step": 4291
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7873613520870432,
      "learning_rate": 1.0202892321146796e-07,
      "loss": 0.2032,
      "step": 4292
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7697488123775446,
      "learning_rate": 1.0131909689233444e-07,
      "loss": 0.2061,
      "step": 4293
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7363634054483832,
      "learning_rate": 1.0061172309534739e-07,
      "loss": 0.1946,
      "step": 4294
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7318288854675193,
      "learning_rate": 9.990680217465243e-08,
      "loss": 0.1813,
      "step": 4295
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7722694239178702,
      "learning_rate": 9.92043344831689e-08,
      "loss": 0.2049,
      "step": 4296
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7664975068420893,
      "learning_rate": 9.850432037258607e-08,
      "loss": 0.2028,
      "step": 4297
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.8101581680602439,
      "learning_rate": 9.780676019336632e-08,
      "loss": 0.2199,
      "step": 4298
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.749961807088771,
      "learning_rate": 9.711165429474256e-08,
      "loss": 0.2057,
      "step": 4299
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7478009300338998,
      "learning_rate": 9.641900302471918e-08,
      "loss": 0.1808,
      "step": 4300
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7650151636771054,
      "learning_rate": 9.572880673007156e-08,
      "loss": 0.2009,
      "step": 4301
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.8502433931167203,
      "learning_rate": 9.504106575634664e-08,
      "loss": 0.2154,
      "step": 4302
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7508020581049631,
      "learning_rate": 9.435578044786009e-08,
      "loss": 0.1959,
      "step": 4303
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7809747983457873,
      "learning_rate": 9.367295114770191e-08,
      "loss": 0.2008,
      "step": 4304
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.8136136435442755,
      "learning_rate": 9.299257819772811e-08,
      "loss": 0.2182,
      "step": 4305
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7713545729760453,
      "learning_rate": 9.231466193856897e-08,
      "loss": 0.2172,
      "step": 4306
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7533959710711974,
      "learning_rate": 9.163920270962134e-08,
      "loss": 0.208,
      "step": 4307
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.788994214952612,
      "learning_rate": 9.096620084905472e-08,
      "loss": 0.2106,
      "step": 4308
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7786202389561554,
      "learning_rate": 9.029565669380625e-08,
      "loss": 0.2167,
      "step": 4309
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7322901826936732,
      "learning_rate": 8.96275705795846e-08,
      "loss": 0.1956,
      "step": 4310
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.764310393186081,
      "learning_rate": 8.896194284086501e-08,
      "loss": 0.1956,
      "step": 4311
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.781395882722771,
      "learning_rate": 8.829877381089536e-08,
      "loss": 0.2093,
      "step": 4312
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7348480637158914,
      "learning_rate": 8.763806382169005e-08,
      "loss": 0.1813,
      "step": 4313
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7720100450154491,
      "learning_rate": 8.697981320403337e-08,
      "loss": 0.2198,
      "step": 4314
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7992795833759949,
      "learning_rate": 8.632402228747893e-08,
      "loss": 0.2125,
      "step": 4315
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7595331725377376,
      "learning_rate": 8.56706914003469e-08,
      "loss": 0.1897,
      "step": 4316
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7440820308659608,
      "learning_rate": 8.501982086972782e-08,
      "loss": 0.1935,
      "step": 4317
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7514251905627038,
      "learning_rate": 8.437141102147883e-08,
      "loss": 0.1823,
      "step": 4318
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7773739267496101,
      "learning_rate": 8.372546218022747e-08,
      "loss": 0.2023,
      "step": 4319
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7601911243096247,
      "learning_rate": 8.308197466936619e-08,
      "loss": 0.2171,
      "step": 4320
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7713164532124522,
      "learning_rate": 8.244094881105779e-08,
      "loss": 0.213,
      "step": 4321
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7544858730259909,
      "learning_rate": 8.18023849262306e-08,
      "loss": 0.1914,
      "step": 4322
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7458757732339502,
      "learning_rate": 8.11662833345822e-08,
      "loss": 0.1867,
      "step": 4323
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7867161093212314,
      "learning_rate": 8.05326443545762e-08,
      "loss": 0.1995,
      "step": 4324
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7772803762996126,
      "learning_rate": 7.990146830344325e-08,
      "loss": 0.2028,
      "step": 4325
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7546337197768481,
      "learning_rate": 7.927275549718228e-08,
      "loss": 0.1884,
      "step": 4326
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.720673917455633,
      "learning_rate": 7.864650625055703e-08,
      "loss": 0.1855,
      "step": 4327
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7873576359037958,
      "learning_rate": 7.802272087709951e-08,
      "loss": 0.2048,
      "step": 4328
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.795148190718808,
      "learning_rate": 7.740139968910709e-08,
      "loss": 0.2059,
      "step": 4329
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7849317381248139,
      "learning_rate": 7.678254299764431e-08,
      "loss": 0.2046,
      "step": 4330
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7674919037117569,
      "learning_rate": 7.616615111254111e-08,
      "loss": 0.2055,
      "step": 4331
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7231942456948386,
      "learning_rate": 7.555222434239395e-08,
      "loss": 0.1863,
      "step": 4332
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7477858400301153,
      "learning_rate": 7.494076299456531e-08,
      "loss": 0.2053,
      "step": 4333
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7958948152429696,
      "learning_rate": 7.433176737518255e-08,
      "loss": 0.2142,
      "step": 4334
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7602349956479654,
      "learning_rate": 7.372523778913898e-08,
      "loss": 0.2056,
      "step": 4335
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7325325023651572,
      "learning_rate": 7.312117454009337e-08,
      "loss": 0.2046,
      "step": 4336
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7766659333825777,
      "learning_rate": 7.251957793046993e-08,
      "loss": 0.2025,
      "step": 4337
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7837417228191458,
      "learning_rate": 7.192044826145772e-08,
      "loss": 0.1929,
      "step": 4338
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7969144843946206,
      "learning_rate": 7.132378583301014e-08,
      "loss": 0.2136,
      "step": 4339
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7926727394674603,
      "learning_rate": 7.072959094384713e-08,
      "loss": 0.1921,
      "step": 4340
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7550685590428533,
      "learning_rate": 7.013786389145127e-08,
      "loss": 0.1984,
      "step": 4341
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8028501870331461,
      "learning_rate": 6.954860497207006e-08,
      "loss": 0.2192,
      "step": 4342
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7790349000225801,
      "learning_rate": 6.896181448071582e-08,
      "loss": 0.2029,
      "step": 4343
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7599963391455786,
      "learning_rate": 6.837749271116578e-08,
      "loss": 0.2027,
      "step": 4344
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8028205118397572,
      "learning_rate": 6.779563995595927e-08,
      "loss": 0.207,
      "step": 4345
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7596981214984934,
      "learning_rate": 6.721625650640163e-08,
      "loss": 0.2191,
      "step": 4346
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7629879859458296,
      "learning_rate": 6.663934265255966e-08,
      "loss": 0.1951,
      "step": 4347
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7983555444688892,
      "learning_rate": 6.606489868326571e-08,
      "loss": 0.1928,
      "step": 4348
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7455150600401651,
      "learning_rate": 6.549292488611525e-08,
      "loss": 0.1875,
      "step": 4349
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7514232095001572,
      "learning_rate": 6.49234215474659e-08,
      "loss": 0.191,
      "step": 4350
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7977361960366868,
      "learning_rate": 6.435638895243957e-08,
      "loss": 0.2129,
      "step": 4351
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7679916427075145,
      "learning_rate": 6.379182738492085e-08,
      "loss": 0.1903,
      "step": 4352
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7944870772096758,
      "learning_rate": 6.322973712755698e-08,
      "loss": 0.2045,
      "step": 4353
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7846802253044177,
      "learning_rate": 6.267011846175896e-08,
      "loss": 0.2017,
      "step": 4354
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7852207685151176,
      "learning_rate": 6.211297166769825e-08,
      "loss": 0.2136,
      "step": 4355
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8015281773834498,
      "learning_rate": 6.15582970243117e-08,
      "loss": 0.2071,
      "step": 4356
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7413876019751526,
      "learning_rate": 6.100609480929553e-08,
      "loss": 0.1854,
      "step": 4357
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8194123450028844,
      "learning_rate": 6.045636529911025e-08,
      "loss": 0.2095,
      "step": 4358
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7656390085191285,
      "learning_rate": 5.990910876897682e-08,
      "loss": 0.2054,
      "step": 4359
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8471290865779272,
      "learning_rate": 5.9364325492880514e-08,
      "loss": 0.2197,
      "step": 4360
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7825979021558925,
      "learning_rate": 5.882201574356483e-08,
      "loss": 0.2044,
      "step": 4361
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8237807490406509,
      "learning_rate": 5.82821797925387e-08,
      "loss": 0.2303,
      "step": 4362
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7678821602830088,
      "learning_rate": 5.7744817910069804e-08,
      "loss": 0.2019,
      "step": 4363
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7115338093080045,
      "learning_rate": 5.720993036518796e-08,
      "loss": 0.1904,
      "step": 4364
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7999639324475672,
      "learning_rate": 5.667751742568506e-08,
      "loss": 0.222,
      "step": 4365
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7777318758574613,
      "learning_rate": 5.6147579358112324e-08,
      "loss": 0.2109,
      "step": 4366
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7992405298328921,
      "learning_rate": 5.56201164277842e-08,
      "loss": 0.2124,
      "step": 4367
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.8290795775056734,
      "learning_rate": 5.509512889877333e-08,
      "loss": 0.2265,
      "step": 4368
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.8173301568295889,
      "learning_rate": 5.457261703391614e-08,
      "loss": 0.2094,
      "step": 4369
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7901215671158355,
      "learning_rate": 5.4052581094806713e-08,
      "loss": 0.2191,
      "step": 4370
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7325734611218774,
      "learning_rate": 5.353502134180122e-08,
      "loss": 0.1765,
      "step": 4371
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7686648362912347,
      "learning_rate": 5.301993803401517e-08,
      "loss": 0.212,
      "step": 4372
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7968752997537657,
      "learning_rate": 5.250733142932562e-08,
      "loss": 0.2054,
      "step": 4373
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7427295178920796,
      "learning_rate": 5.19972017843684e-08,
      "loss": 0.1865,
      "step": 4374
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7687601918150403,
      "learning_rate": 5.148954935453976e-08,
      "loss": 0.1973,
      "step": 4375
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7401056045396123,
      "learning_rate": 5.098437439399473e-08,
      "loss": 0.1871,
      "step": 4376
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7632073375958514,
      "learning_rate": 5.0481677155650445e-08,
      "loss": 0.2064,
      "step": 4377
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7537450005373051,
      "learning_rate": 4.998145789118114e-08,
      "loss": 0.1941,
      "step": 4378
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7747713472827749,
      "learning_rate": 4.948371685102149e-08,
      "loss": 0.206,
      "step": 4379
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7644375425004244,
      "learning_rate": 4.898845428436494e-08,
      "loss": 0.201,
      "step": 4380
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7648561901601362,
      "learning_rate": 4.849567043916481e-08,
      "loss": 0.2206,
      "step": 4381
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7225433382759175,
      "learning_rate": 4.800536556213375e-08,
      "loss": 0.2009,
      "step": 4382
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.8231689013493229,
      "learning_rate": 4.751753989874153e-08,
      "loss": 0.2245,
      "step": 4383
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7998298630927809,
      "learning_rate": 4.7032193693217765e-08,
      "loss": 0.1872,
      "step": 4384
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7299153234111516,
      "learning_rate": 4.654932718855143e-08,
      "loss": 0.1915,
      "step": 4385
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7776440128663066,
      "learning_rate": 4.60689406264897e-08,
      "loss": 0.2064,
      "step": 4386
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7665535837329102,
      "learning_rate": 4.559103424753742e-08,
      "loss": 0.2158,
      "step": 4387
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7928050222408521,
      "learning_rate": 4.511560829095818e-08,
      "loss": 0.2067,
      "step": 4388
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7750391207169047,
      "learning_rate": 4.4642662994773264e-08,
      "loss": 0.2026,
      "step": 4389
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7140059609315803,
      "learning_rate": 4.417219859576383e-08,
      "loss": 0.1948,
      "step": 4390
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7688175527243958,
      "learning_rate": 4.370421532946589e-08,
      "loss": 0.2085,
      "step": 4391
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.8016891554145306,
      "learning_rate": 4.3238713430176494e-08,
      "loss": 0.2255,
      "step": 4392
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7630411579474543,
      "learning_rate": 4.2775693130948094e-08,
      "loss": 0.1989,
      "step": 4393
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7329539694107651,
      "learning_rate": 4.231515466359193e-08,
      "loss": 0.1957,
      "step": 4394
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7678208095777755,
      "learning_rate": 4.1857098258676345e-08,
      "loss": 0.2018,
      "step": 4395
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7302058982406552,
      "learning_rate": 4.140152414552678e-08,
      "loss": 0.1934,
      "step": 4396
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7989601450976026,
      "learning_rate": 4.094843255222691e-08,
      "loss": 0.2173,
      "step": 4397
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.8160903628583412,
      "learning_rate": 4.0497823705615836e-08,
      "loss": 0.2212,
      "step": 4398
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7781217496788301,
      "learning_rate": 4.004969783129087e-08,
      "loss": 0.2064,
      "step": 4399
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7216826004683567,
      "learning_rate": 3.960405515360699e-08,
      "loss": 0.1879,
      "step": 4400
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7184801554385618,
      "learning_rate": 3.9160895895673515e-08,
      "loss": 0.1979,
      "step": 4401
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7522659579041928,
      "learning_rate": 3.8720220279359065e-08,
      "loss": 0.1874,
      "step": 4402
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7900778578372126,
      "learning_rate": 3.828202852528717e-08,
      "loss": 0.1945,
      "step": 4403
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.8021196427508989,
      "learning_rate": 3.784632085283846e-08,
      "loss": 0.1925,
      "step": 4404
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7860458734919237,
      "learning_rate": 3.741309748014954e-08,
      "loss": 0.2023,
      "step": 4405
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7753853680293479,
      "learning_rate": 3.698235862411359e-08,
      "loss": 0.207,
      "step": 4406
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7544450195605186,
      "learning_rate": 3.6554104500379775e-08,
      "loss": 0.2177,
      "step": 4407
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7719095451676459,
      "learning_rate": 3.6128335323353804e-08,
      "loss": 0.1949,
      "step": 4408
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7429399914530471,
      "learning_rate": 3.570505130619573e-08,
      "loss": 0.1795,
      "step": 4409
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7878967396212754,
      "learning_rate": 3.528425266082325e-08,
      "loss": 0.2113,
      "step": 4410
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7684081103409208,
      "learning_rate": 3.486593959790896e-08,
      "loss": 0.2053,
      "step": 4411
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.8051499280203382,
      "learning_rate": 3.445011232688089e-08,
      "loss": 0.1942,
      "step": 4412
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7753225479856776,
      "learning_rate": 3.4036771055923066e-08,
      "loss": 0.2182,
      "step": 4413
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.724287859908868,
      "learning_rate": 3.36259159919744e-08,
      "loss": 0.1811,
      "step": 4414
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.787239007128672,
      "learning_rate": 3.321754734072924e-08,
      "loss": 0.2237,
      "step": 4415
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7650179227657968,
      "learning_rate": 3.281166530663682e-08,
      "loss": 0.1936,
      "step": 4416
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7774177283763857,
      "learning_rate": 3.240827009290237e-08,
      "loss": 0.2126,
      "step": 4417
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7575394985300242,
      "learning_rate": 3.2007361901485455e-08,
      "loss": 0.1997,
      "step": 4418
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7543668085031319,
      "learning_rate": 3.160894093309996e-08,
      "loss": 0.2003,
      "step": 4419
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7424577438053133,
      "learning_rate": 3.121300738721578e-08,
      "loss": 0.1945,
      "step": 4420
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.786755953341638,
      "learning_rate": 3.0819561462056555e-08,
      "loss": 0.2056,
      "step": 4421
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.8154007626594151,
      "learning_rate": 3.042860335460085e-08,
      "loss": 0.2401,
      "step": 4422
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.8222509442616518,
      "learning_rate": 3.004013326058153e-08,
      "loss": 0.2135,
      "step": 4423
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7615051848136337,
      "learning_rate": 2.9654151374485797e-08,
      "loss": 0.2104,
      "step": 4424
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7661570463262186,
      "learning_rate": 2.92706578895563e-08,
      "loss": 0.1926,
      "step": 4425
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7943124526518226,
      "learning_rate": 2.8889652997787233e-08,
      "loss": 0.2028,
      "step": 4426
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7604621675406193,
      "learning_rate": 2.8511136889929902e-08,
      "loss": 0.2035,
      "step": 4427
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.732200332912149,
      "learning_rate": 2.8135109755487723e-08,
      "loss": 0.2027,
      "step": 4428
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7871846619255617,
      "learning_rate": 2.7761571782718433e-08,
      "loss": 0.2262,
      "step": 4429
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7090905820748571,
      "learning_rate": 2.7390523158633552e-08,
      "loss": 0.1806,
      "step": 4430
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7815111922660083,
      "learning_rate": 2.7021964068998375e-08,
      "loss": 0.1917,
      "step": 4431
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7644376901561837,
      "learning_rate": 2.6655894698331962e-08,
      "loss": 0.2015,
      "step": 4432
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7545570176026194,
      "learning_rate": 2.629231522990716e-08,
      "loss": 0.1965,
      "step": 4433
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7537300209233407,
      "learning_rate": 2.593122584574892e-08,
      "loss": 0.182,
      "step": 4434
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.8201078003346534,
      "learning_rate": 2.5572626726637074e-08,
      "loss": 0.2153,
      "step": 4435
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7668496185493342,
      "learning_rate": 2.5216518052104122e-08,
      "loss": 0.2116,
      "step": 4436
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.82845998124489,
      "learning_rate": 2.486290000043523e-08,
      "loss": 0.2282,
      "step": 4437
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7546131494119764,
      "learning_rate": 2.4511772748669894e-08,
      "loss": 0.1919,
      "step": 4438
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7723916054978528,
      "learning_rate": 2.4163136472599157e-08,
      "loss": 0.2122,
      "step": 4439
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7656649117885437,
      "learning_rate": 2.381699134676785e-08,
      "loss": 0.1899,
      "step": 4440
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7760766695407973,
      "learning_rate": 2.3473337544472897e-08,
      "loss": 0.2112,
      "step": 4441
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7400393635101186,
      "learning_rate": 2.3132175237765564e-08,
      "loss": 0.1988,
      "step": 4442
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7883123559257683,
      "learning_rate": 2.2793504597447003e-08,
      "loss": 0.2084,
      "step": 4443
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7411041619579481,
      "learning_rate": 2.2457325793073803e-08,
      "loss": 0.1988,
      "step": 4444
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.732855180922827,
      "learning_rate": 2.2123638992953e-08,
      "loss": 0.2125,
      "step": 4445
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7837028803343771,
      "learning_rate": 2.179244436414485e-08,
      "loss": 0.2113,
      "step": 4446
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.758081781739502,
      "learning_rate": 2.146374207246227e-08,
      "loss": 0.1984,
      "step": 4447
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7167817857176783,
      "learning_rate": 2.1137532282469176e-08,
      "loss": 0.1902,
      "step": 4448
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7210921162098929,
      "learning_rate": 2.081381515748271e-08,
      "loss": 0.1766,
      "step": 4449
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7440456577483008,
      "learning_rate": 2.0492590859571558e-08,
      "loss": 0.1815,
      "step": 4450
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7860012904431641,
      "learning_rate": 2.017385954955653e-08,
      "loss": 0.2113,
      "step": 4451
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.803749315479159,
      "learning_rate": 1.9857621387010528e-08,
      "loss": 0.2229,
      "step": 4452
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.796444353692773,
      "learning_rate": 1.954387653025802e-08,
      "loss": 0.2199,
      "step": 4453
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7254775783978192,
      "learning_rate": 1.9232625136374472e-08,
      "loss": 0.1869,
      "step": 4454
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7393476941917582,
      "learning_rate": 1.8923867361189118e-08,
      "loss": 0.1996,
      "step": 4455
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7554041555932561,
      "learning_rate": 1.8617603359279977e-08,
      "loss": 0.1867,
      "step": 4456
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7942623542660384,
      "learning_rate": 1.8313833283978844e-08,
      "loss": 0.2163,
      "step": 4457
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7515929550983769,
      "learning_rate": 1.8012557287367394e-08,
      "loss": 0.2061,
      "step": 4458
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.8159611913912423,
      "learning_rate": 1.7713775520279976e-08,
      "loss": 0.2001,
      "step": 4459
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7702672145018112,
      "learning_rate": 1.7417488132301375e-08,
      "loss": 0.1911,
      "step": 4460
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7603203502649829,
      "learning_rate": 1.712369527176738e-08,
      "loss": 0.1922,
      "step": 4461
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7790132952453471,
      "learning_rate": 1.6832397085765893e-08,
      "loss": 0.2027,
      "step": 4462
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7660766685992981,
      "learning_rate": 1.6543593720134142e-08,
      "loss": 0.2121,
      "step": 4463
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7661614298074334,
      "learning_rate": 1.6257285319462578e-08,
      "loss": 0.1969,
      "step": 4464
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.8225525899768487,
      "learning_rate": 1.597347202709043e-08,
      "loss": 0.2184,
      "step": 4465
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.8113569320031445,
      "learning_rate": 1.5692153985109037e-08,
      "loss": 0.2096,
      "step": 4466
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7492531527011896,
      "learning_rate": 1.541333133436018e-08,
      "loss": 0.1952,
      "step": 4467
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.8020079780591471,
      "learning_rate": 1.513700421443609e-08,
      "loss": 0.2227,
      "step": 4468
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7731190854744109,
      "learning_rate": 1.4863172763680545e-08,
      "loss": 0.1914,
      "step": 4469
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7810948632364072,
      "learning_rate": 1.4591837119186104e-08,
      "loss": 0.2005,
      "step": 4470
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7254402410955911,
      "learning_rate": 1.4322997416797435e-08,
      "loss": 0.1955,
      "step": 4471
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7713019243524399,
      "learning_rate": 1.4056653791108543e-08,
      "loss": 0.2013,
      "step": 4472
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7828050237075176,
      "learning_rate": 1.379280637546443e-08,
      "loss": 0.2075,
      "step": 4473
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7705228401898997,
      "learning_rate": 1.3531455301960539e-08,
      "loss": 0.2038,
      "step": 4474
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.750385495478384,
      "learning_rate": 1.3272600701441651e-08,
      "loss": 0.2067,
      "step": 4475
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7359080569379826,
      "learning_rate": 1.3016242703503545e-08,
      "loss": 0.1967,
      "step": 4476
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7990505859722992,
      "learning_rate": 1.2762381436491888e-08,
      "loss": 0.2186,
      "step": 4477
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7724634234141625,
      "learning_rate": 1.2511017027501682e-08,
      "loss": 0.2104,
      "step": 4478
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7540396797173139,
      "learning_rate": 1.226214960237837e-08,
      "loss": 0.2057,
      "step": 4479
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.8007140556387763,
      "learning_rate": 1.2015779285717844e-08,
      "loss": 0.2128,
      "step": 4480
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7784881377570229,
      "learning_rate": 1.1771906200864769e-08,
      "loss": 0.2066,
      "step": 4481
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.8069476183935115,
      "learning_rate": 1.1530530469914258e-08,
      "loss": 0.2041,
      "step": 4482
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7639552787152982,
      "learning_rate": 1.1291652213710758e-08,
      "loss": 0.2181,
      "step": 4483
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7566998427388063,
      "learning_rate": 1.1055271551849156e-08,
      "loss": 0.1791,
      "step": 4484
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7535784135367068,
      "learning_rate": 1.0821388602673121e-08,
      "loss": 0.2123,
      "step": 4485
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.823469380086295,
      "learning_rate": 1.0590003483275657e-08,
      "loss": 0.2164,
      "step": 4486
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7773378188629515,
      "learning_rate": 1.0361116309499653e-08,
      "loss": 0.2131,
      "step": 4487
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7575532392421571,
      "learning_rate": 1.0134727195937332e-08,
      "loss": 0.1976,
      "step": 4488
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7421507871038614,
      "learning_rate": 9.91083625593081e-09,
      "loss": 0.2,
      "step": 4489
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7650903106264623,
      "learning_rate": 9.689443601570425e-09,
      "loss": 0.196,
      "step": 4490
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7498743652615781,
      "learning_rate": 9.470549343696401e-09,
      "loss": 0.1974,
      "step": 4491
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7591484848227226,
      "learning_rate": 9.254153591898852e-09,
      "loss": 0.2077,
      "step": 4492
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.758162619196813,
      "learning_rate": 9.04025645451445e-09,
      "loss": 0.2086,
      "step": 4493
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.8017655482620958,
      "learning_rate": 8.828858038632538e-09,
      "loss": 0.2029,
      "step": 4494
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7596668849236821,
      "learning_rate": 8.61995845008845e-09,
      "loss": 0.204,
      "step": 4495
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7602587462734808,
      "learning_rate": 8.413557793467974e-09,
      "loss": 0.2138,
      "step": 4496
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7526056236420481,
      "learning_rate": 8.209656172105673e-09,
      "loss": 0.1903,
      "step": 4497
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7377712984500105,
      "learning_rate": 8.008253688084888e-09,
      "loss": 0.2071,
      "step": 4498
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7972732985862786,
      "learning_rate": 7.80935044223663e-09,
      "loss": 0.2055,
      "step": 4499
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7593582605935041,
      "learning_rate": 7.612946534143461e-09,
      "loss": 0.1995,
      "step": 4500
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7429116442987398,
      "learning_rate": 7.4190420621333965e-09,
      "loss": 0.1944,
      "step": 4501
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7860858573372854,
      "learning_rate": 7.227637123285447e-09,
      "loss": 0.2142,
      "step": 4502
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7260792673605904,
      "learning_rate": 7.038731813426292e-09,
      "loss": 0.1866,
      "step": 4503
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.795605958795444,
      "learning_rate": 6.852326227130835e-09,
      "loss": 0.2154,
      "step": 4504
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.8405019990450766,
      "learning_rate": 6.668420457723312e-09,
      "loss": 0.2246,
      "step": 4505
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7321901473589074,
      "learning_rate": 6.487014597275631e-09,
      "loss": 0.1793,
      "step": 4506
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7885629863184404,
      "learning_rate": 6.308108736609031e-09,
      "loss": 0.2224,
      "step": 4507
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7705702574036786,
      "learning_rate": 6.1317029652929734e-09,
      "loss": 0.2082,
      "step": 4508
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7730428833580067,
      "learning_rate": 5.957797371644036e-09,
      "loss": 0.2113,
      "step": 4509
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.733431452943101,
      "learning_rate": 5.7863920427286834e-09,
      "loss": 0.1985,
      "step": 4510
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7299108197607228,
      "learning_rate": 5.617487064360494e-09,
      "loss": 0.2029,
      "step": 4511
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.763884833951739,
      "learning_rate": 5.45108252110127e-09,
      "loss": 0.199,
      "step": 4512
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7944162606386447,
      "learning_rate": 5.2871784962627015e-09,
      "loss": 0.2248,
      "step": 4513
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.8124161043310352,
      "learning_rate": 5.125775071901373e-09,
      "loss": 0.2024,
      "step": 4514
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7553762990934692,
      "learning_rate": 4.966872328825978e-09,
      "loss": 0.1988,
      "step": 4515
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7449809568027312,
      "learning_rate": 4.810470346588992e-09,
      "loss": 0.21,
      "step": 4516
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7894178202825154,
      "learning_rate": 4.6565692034944435e-09,
      "loss": 0.2075,
      "step": 4517
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7379078955807418,
      "learning_rate": 4.505168976592922e-09,
      "loss": 0.181,
      "step": 4518
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7884427236098112,
      "learning_rate": 4.356269741681573e-09,
      "loss": 0.211,
      "step": 4519
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7494586172350168,
      "learning_rate": 4.209871573307989e-09,
      "loss": 0.2048,
      "step": 4520
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7683490048917736,
      "learning_rate": 4.065974544765761e-09,
      "loss": 0.2148,
      "step": 4521
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.789343454786431,
      "learning_rate": 3.924578728096707e-09,
      "loss": 0.2152,
      "step": 4522
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7588628054651702,
      "learning_rate": 3.785684194090866e-09,
      "loss": 0.1988,
      "step": 4523
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7973156712478708,
      "learning_rate": 3.6492910122859493e-09,
      "loss": 0.2096,
      "step": 4524
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.774523062318031,
      "learning_rate": 3.515399250966223e-09,
      "loss": 0.2036,
      "step": 4525
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.8020476584307383,
      "learning_rate": 3.3840089771658424e-09,
      "loss": 0.2098,
      "step": 4526
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7712641392526925,
      "learning_rate": 3.2551202566644125e-09,
      "loss": 0.2097,
      "step": 4527
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.761175863517502,
      "learning_rate": 3.1287331539903155e-09,
      "loss": 0.2145,
      "step": 4528
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7844031867427002,
      "learning_rate": 3.004847732419047e-09,
      "loss": 0.2031,
      "step": 4529
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7896383401549832,
      "learning_rate": 2.8834640539737723e-09,
      "loss": 0.1943,
      "step": 4530
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7727874040511089,
      "learning_rate": 2.7645821794247684e-09,
      "loss": 0.2109,
      "step": 4531
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7596116676674939,
      "learning_rate": 2.648202168291092e-09,
      "loss": 0.194,
      "step": 4532
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7753182488798499,
      "learning_rate": 2.534324078837802e-09,
      "loss": 0.212,
      "step": 4533
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.759372232425253,
      "learning_rate": 2.422947968078182e-09,
      "loss": 0.2122,
      "step": 4534
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7540904975492201,
      "learning_rate": 2.314073891772073e-09,
      "loss": 0.1988,
      "step": 4535
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.8089846920350543,
      "learning_rate": 2.2077019044280946e-09,
      "loss": 0.2228,
      "step": 4536
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7607901019109335,
      "learning_rate": 2.1038320593003146e-09,
      "loss": 0.2157,
      "step": 4537
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7559119781703244,
      "learning_rate": 2.002464408392135e-09,
      "loss": 0.1982,
      "step": 4538
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.750250232805629,
      "learning_rate": 1.90359900245185e-09,
      "loss": 0.1918,
      "step": 4539
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7645024479371121,
      "learning_rate": 1.8072358909781983e-09,
      "loss": 0.2158,
      "step": 4540
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7710284364019145,
      "learning_rate": 1.7133751222137007e-09,
      "loss": 0.2031,
      "step": 4541
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7497225137732092,
      "learning_rate": 1.622016743150212e-09,
      "loss": 0.2075,
      "step": 4542
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7801858473770927,
      "learning_rate": 1.5331607995267006e-09,
      "loss": 0.2196,
      "step": 4543
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7202686212257948,
      "learning_rate": 1.4468073358286926e-09,
      "loss": 0.1863,
      "step": 4544
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.803019390042267,
      "learning_rate": 1.362956395288828e-09,
      "loss": 0.2033,
      "step": 4545
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.7574873947921831,
      "learning_rate": 1.2816080198868596e-09,
      "loss": 0.1942,
      "step": 4546
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.7883648086014026,
      "learning_rate": 1.202762250350209e-09,
      "loss": 0.2017,
      "step": 4547
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.8151306713562403,
      "learning_rate": 1.1264191261528557e-09,
      "loss": 0.2144,
      "step": 4548
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.8115900722571238,
      "learning_rate": 1.0525786855158926e-09,
      "loss": 0.2159,
      "step": 4549
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.7841058251218473,
      "learning_rate": 9.812409654075261e-10,
      "loss": 0.2175,
      "step": 4550
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.7857186943537182,
      "learning_rate": 9.124060015425206e-10,
      "loss": 0.2144,
      "step": 4551
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.7556526721642087,
      "learning_rate": 8.460738283838643e-10,
      "loss": 0.1988,
      "step": 4552
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.7639970467824269,
      "learning_rate": 7.82244479139993e-10,
      "loss": 0.1866,
      "step": 4553
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.7882122373608186,
      "learning_rate": 7.209179857675664e-10,
      "loss": 0.2077,
      "step": 4554
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.768953777540721,
      "learning_rate": 6.620943789692469e-10,
      "loss": 0.2062,
      "step": 4555
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.8182066077886316,
      "learning_rate": 6.057736881959209e-10,
      "loss": 0.2291,
      "step": 4556
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.8173565378892621,
      "learning_rate": 5.519559416433673e-10,
      "loss": 0.2224,
      "step": 4557
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.783650167552557,
      "learning_rate": 5.006411662555888e-10,
      "loss": 0.2063,
      "step": 4558
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.7562520611096921,
      "learning_rate": 4.518293877242563e-10,
      "loss": 0.2061,
      "step": 4559
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.8014109058647851,
      "learning_rate": 4.0552063048593384e-10,
      "loss": 0.2053,
      "step": 4560
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.8186895538338581,
      "learning_rate": 3.6171491772540865e-10,
      "loss": 0.2266,
      "step": 4561
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.7294052295907465,
      "learning_rate": 3.204122713740265e-10,
      "loss": 0.1942,
      "step": 4562
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.7292918262673972,
      "learning_rate": 2.816127121102463e-10,
      "loss": 0.1839,
      "step": 4563
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.783898868472993,
      "learning_rate": 2.453162593585301e-10,
      "loss": 0.2092,
      "step": 4564
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.7782547058370585,
      "learning_rate": 2.1152293129100832e-10,
      "loss": 0.1915,
      "step": 4565
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.7701868073635726,
      "learning_rate": 1.8023274482636965e-10,
      "loss": 0.2126,
      "step": 4566
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.7757447016368472,
      "learning_rate": 1.5144571562930587e-10,
      "loss": 0.2015,
      "step": 4567
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.7675100966024702,
      "learning_rate": 1.251618581127323e-10,
      "loss": 0.2073,
      "step": 4568
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7273899533005658,
      "learning_rate": 1.013811854361224e-10,
      "loss": 0.198,
      "step": 4569
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7531764141399457,
      "learning_rate": 8.01037095038426e-11,
      "loss": 0.2158,
      "step": 4570
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7659246190195881,
      "learning_rate": 6.132944096959304e-11,
      "loss": 0.2055,
      "step": 4571
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7515363142593082,
      "learning_rate": 4.505838923252181e-11,
      "loss": 0.1932,
      "step": 4572
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.8061820746903433,
      "learning_rate": 3.129056243833528e-11,
      "loss": 0.203,
      "step": 4573
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7053678601284087,
      "learning_rate": 2.0025967479853082e-11,
      "loss": 0.1873,
      "step": 4574
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7229000029123474,
      "learning_rate": 1.126460999700818e-11,
      "loss": 0.1899,
      "step": 4575
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.767189015400315,
      "learning_rate": 5.006494376291748e-12,
      "loss": 0.2024,
      "step": 4576
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7678935867729011,
      "learning_rate": 1.2516237507531615e-12,
      "loss": 0.2094,
      "step": 4577
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7747822251686244,
      "learning_rate": 0.0,
      "loss": 0.2033,
      "step": 4578
    },
    {
      "epoch": 2.0,
      "step": 4578,
      "total_flos": 7126615895048192.0,
      "train_loss": 0.2577521303475762,
      "train_runtime": 70236.9804,
      "train_samples_per_second": 8.346,
      "train_steps_per_second": 0.065
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 4578,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 1000,
  "total_flos": 7126615895048192.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}