{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 16323,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 11.474336167914327,
      "learning_rate": 1.0204081632653062e-07,
      "loss": 1.7929,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.504469027906618,
      "learning_rate": 2.0408163265306124e-07,
      "loss": 1.7871,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.465005788236182,
      "learning_rate": 3.0612244897959183e-07,
      "loss": 2.1951,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.738673948018356,
      "learning_rate": 4.081632653061225e-07,
      "loss": 1.8262,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.267634032956286,
      "learning_rate": 5.10204081632653e-07,
      "loss": 1.9988,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.045449234301678,
      "learning_rate": 6.122448979591837e-07,
      "loss": 2.2725,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.560826357760906,
      "learning_rate": 7.142857142857143e-07,
      "loss": 1.9363,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 32.82299252645454,
      "learning_rate": 8.16326530612245e-07,
      "loss": 1.7638,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.442254442286046,
      "learning_rate": 9.183673469387756e-07,
      "loss": 1.6046,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.362818460740879,
      "learning_rate": 1.020408163265306e-06,
      "loss": 1.4041,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 15.776511254916763,
      "learning_rate": 1.1224489795918367e-06,
      "loss": 1.7579,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.975095634761272,
      "learning_rate": 1.2244897959183673e-06,
      "loss": 1.5309,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.963266329218884,
      "learning_rate": 1.326530612244898e-06,
      "loss": 1.6067,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.746342769794806,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 2.0865,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.16993290610841,
      "learning_rate": 1.5306122448979593e-06,
      "loss": 1.3216,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.757870954349098,
      "learning_rate": 1.63265306122449e-06,
      "loss": 1.487,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.566077458522056,
      "learning_rate": 1.7346938775510206e-06,
      "loss": 1.6459,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.314489616500606,
      "learning_rate": 1.8367346938775512e-06,
      "loss": 1.6044,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.19831542960961,
      "learning_rate": 1.938775510204082e-06,
      "loss": 1.4032,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.28159049499897,
      "learning_rate": 2.040816326530612e-06,
      "loss": 1.1407,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.994745438203915,
      "learning_rate": 2.142857142857143e-06,
      "loss": 1.3191,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.316287123741859,
      "learning_rate": 2.2448979591836734e-06,
      "loss": 1.7476,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.473681025323938,
      "learning_rate": 2.346938775510204e-06,
      "loss": 1.2224,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 64.88659580877683,
      "learning_rate": 2.4489795918367347e-06,
      "loss": 1.4578,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.189249479191513,
      "learning_rate": 2.5510204081632653e-06,
      "loss": 1.2589,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.073354019692726,
      "learning_rate": 2.653061224489796e-06,
      "loss": 1.3926,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.480881160049335,
      "learning_rate": 2.7551020408163266e-06,
      "loss": 1.0925,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.916319603387308,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 1.6743,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.199352405750797,
      "learning_rate": 2.959183673469388e-06,
      "loss": 1.1979,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.601360042686278,
      "learning_rate": 3.0612244897959185e-06,
      "loss": 0.973,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.395077138803797,
      "learning_rate": 3.1632653061224488e-06,
      "loss": 1.3097,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.232250027239502,
      "learning_rate": 3.26530612244898e-06,
      "loss": 1.0486,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.3191248348203075,
      "learning_rate": 3.3673469387755105e-06,
      "loss": 1.2316,
      "step": 33
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.420154545937291,
      "learning_rate": 3.469387755102041e-06,
      "loss": 1.1947,
      "step": 34
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.197892398479073,
      "learning_rate": 3.5714285714285714e-06,
      "loss": 1.2685,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.301830264761274,
      "learning_rate": 3.6734693877551024e-06,
      "loss": 1.066,
      "step": 36
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.431134796865013,
      "learning_rate": 3.775510204081633e-06,
      "loss": 1.5623,
      "step": 37
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.524840215054727,
      "learning_rate": 3.877551020408164e-06,
      "loss": 1.1811,
      "step": 38
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.377656973730444,
      "learning_rate": 3.979591836734694e-06,
      "loss": 1.0376,
      "step": 39
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.196752744570468,
      "learning_rate": 4.081632653061224e-06,
      "loss": 1.037,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.601163837930134,
      "learning_rate": 4.183673469387755e-06,
      "loss": 1.1721,
      "step": 41
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.8130684168196,
      "learning_rate": 4.285714285714286e-06,
      "loss": 1.1335,
      "step": 42
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.638225495904548,
      "learning_rate": 4.3877551020408165e-06,
      "loss": 1.0529,
      "step": 43
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.4737284079005715,
      "learning_rate": 4.489795918367347e-06,
      "loss": 1.5775,
      "step": 44
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.873707734969654,
      "learning_rate": 4.591836734693878e-06,
      "loss": 1.3164,
      "step": 45
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.103670765439833,
      "learning_rate": 4.693877551020408e-06,
      "loss": 1.3071,
      "step": 46
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.921581785107716,
      "learning_rate": 4.795918367346939e-06,
      "loss": 1.0776,
      "step": 47
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.4552733913037486,
      "learning_rate": 4.897959183673469e-06,
      "loss": 0.9522,
      "step": 48
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.5178644201183156,
      "learning_rate": 5e-06,
      "loss": 0.781,
      "step": 49
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.301975372902402,
      "learning_rate": 5.102040816326531e-06,
      "loss": 1.4782,
      "step": 50
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.701111872519687,
      "learning_rate": 5.204081632653062e-06,
      "loss": 1.2461,
      "step": 51
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.055031527504065,
      "learning_rate": 5.306122448979592e-06,
      "loss": 0.9055,
      "step": 52
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.324423934642495,
      "learning_rate": 5.408163265306123e-06,
      "loss": 0.7069,
      "step": 53
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.818181597574391,
      "learning_rate": 5.510204081632653e-06,
      "loss": 1.3049,
      "step": 54
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.065028531911183,
      "learning_rate": 5.612244897959184e-06,
      "loss": 0.7198,
      "step": 55
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.299279349237072,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 0.9422,
      "step": 56
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.944913749842618,
      "learning_rate": 5.816326530612245e-06,
      "loss": 0.926,
      "step": 57
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.508218360019061,
      "learning_rate": 5.918367346938776e-06,
      "loss": 0.9668,
      "step": 58
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.3256068336580435,
      "learning_rate": 6.020408163265306e-06,
      "loss": 1.0222,
      "step": 59
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.886029556551084,
      "learning_rate": 6.122448979591837e-06,
      "loss": 0.9697,
      "step": 60
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.572490972275901,
      "learning_rate": 6.224489795918367e-06,
      "loss": 1.2411,
      "step": 61
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.0774849162496976,
      "learning_rate": 6.3265306122448975e-06,
      "loss": 1.1302,
      "step": 62
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.202649236826742,
      "learning_rate": 6.428571428571429e-06,
      "loss": 0.968,
      "step": 63
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.773674090321403,
      "learning_rate": 6.53061224489796e-06,
      "loss": 0.822,
      "step": 64
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.888690851890347,
      "learning_rate": 6.632653061224491e-06,
      "loss": 1.103,
      "step": 65
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.776968691485521,
      "learning_rate": 6.734693877551021e-06,
      "loss": 1.0774,
      "step": 66
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.7282697527168414,
      "learning_rate": 6.836734693877551e-06,
      "loss": 0.6615,
      "step": 67
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.979958737218421,
      "learning_rate": 6.938775510204082e-06,
      "loss": 0.8694,
      "step": 68
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.4251235507414233,
      "learning_rate": 7.0408163265306125e-06,
      "loss": 0.6798,
      "step": 69
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.529031033560474,
      "learning_rate": 7.142857142857143e-06,
      "loss": 0.9021,
      "step": 70
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.7018697628712705,
      "learning_rate": 7.244897959183673e-06,
      "loss": 0.9569,
      "step": 71
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.7214886494327724,
      "learning_rate": 7.346938775510205e-06,
      "loss": 0.8707,
      "step": 72
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.671776804313545,
      "learning_rate": 7.448979591836736e-06,
      "loss": 0.7857,
      "step": 73
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.017323856744425,
      "learning_rate": 7.551020408163266e-06,
      "loss": 0.9646,
      "step": 74
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.05034785018141,
      "learning_rate": 7.653061224489797e-06,
      "loss": 0.6804,
      "step": 75
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.115370804135989,
      "learning_rate": 7.755102040816327e-06,
      "loss": 0.7524,
      "step": 76
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.936152729982006,
      "learning_rate": 7.857142857142858e-06,
      "loss": 0.7371,
      "step": 77
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.101693418736244,
      "learning_rate": 7.959183673469388e-06,
      "loss": 1.2134,
      "step": 78
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.481603642559879,
      "learning_rate": 8.061224489795918e-06,
      "loss": 0.5861,
      "step": 79
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.741033970525328,
      "learning_rate": 8.163265306122448e-06,
      "loss": 0.9507,
      "step": 80
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.897162127845057,
      "learning_rate": 8.26530612244898e-06,
      "loss": 1.0196,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.875290283409033,
      "learning_rate": 8.36734693877551e-06,
      "loss": 0.8822,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.581203440590431,
      "learning_rate": 8.46938775510204e-06,
      "loss": 0.8786,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.980624998535415,
      "learning_rate": 8.571428571428573e-06,
      "loss": 0.7861,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0470326162370664,
      "learning_rate": 8.673469387755103e-06,
      "loss": 0.4528,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.188633831774501,
      "learning_rate": 8.775510204081633e-06,
      "loss": 1.0388,
      "step": 86
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.941550220669408,
      "learning_rate": 8.877551020408163e-06,
      "loss": 0.7018,
      "step": 87
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.4349741402543295,
      "learning_rate": 8.979591836734694e-06,
      "loss": 1.038,
      "step": 88
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.3048905964999817,
      "learning_rate": 9.081632653061225e-06,
      "loss": 0.587,
      "step": 89
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.320874946979226,
      "learning_rate": 9.183673469387756e-06,
      "loss": 0.6568,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.5819890506866536,
      "learning_rate": 9.285714285714286e-06,
      "loss": 0.7016,
      "step": 91
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.973534803481136,
      "learning_rate": 9.387755102040816e-06,
      "loss": 0.5337,
      "step": 92
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.166605730957401,
      "learning_rate": 9.489795918367348e-06,
      "loss": 0.4744,
      "step": 93
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.155927966396674,
      "learning_rate": 9.591836734693878e-06,
      "loss": 0.8055,
      "step": 94
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.8344266537983054,
      "learning_rate": 9.693877551020408e-06,
      "loss": 0.626,
      "step": 95
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.8774097701753374,
      "learning_rate": 9.795918367346939e-06,
      "loss": 0.5206,
      "step": 96
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.7377330746400106,
      "learning_rate": 9.897959183673469e-06,
      "loss": 0.6636,
      "step": 97
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.505822674698081,
      "learning_rate": 1e-05,
      "loss": 0.8292,
      "step": 98
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.993736753787066,
      "learning_rate": 1.0102040816326531e-05,
      "loss": 0.8359,
      "step": 99
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.27285839663434,
      "learning_rate": 1.0204081632653061e-05,
      "loss": 0.794,
      "step": 100
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.512210286387246,
      "learning_rate": 1.0306122448979593e-05,
      "loss": 0.9808,
      "step": 101
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.282806716852373,
      "learning_rate": 1.0408163265306123e-05,
      "loss": 1.0332,
      "step": 102
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.958542450994472,
      "learning_rate": 1.0510204081632654e-05,
      "loss": 0.6519,
      "step": 103
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.0534660623069705,
      "learning_rate": 1.0612244897959184e-05,
      "loss": 0.8473,
      "step": 104
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.939840082103936,
      "learning_rate": 1.0714285714285714e-05,
      "loss": 0.5271,
      "step": 105
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.643720528910641,
      "learning_rate": 1.0816326530612246e-05,
      "loss": 0.7195,
      "step": 106
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0224859794872967,
      "learning_rate": 1.0918367346938776e-05,
      "loss": 0.5611,
      "step": 107
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.205070226338764,
      "learning_rate": 1.1020408163265306e-05,
      "loss": 0.7879,
      "step": 108
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.9747434675248585,
      "learning_rate": 1.1122448979591837e-05,
      "loss": 0.8012,
      "step": 109
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.925011241655254,
      "learning_rate": 1.1224489795918369e-05,
      "loss": 0.7154,
      "step": 110
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.149703028992224,
      "learning_rate": 1.1326530612244899e-05,
      "loss": 0.6532,
      "step": 111
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.720931040315669,
      "learning_rate": 1.1428571428571429e-05,
      "loss": 0.5205,
      "step": 112
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8927971173484615,
      "learning_rate": 1.153061224489796e-05,
      "loss": 0.8461,
      "step": 113
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.277626970737179,
      "learning_rate": 1.163265306122449e-05,
      "loss": 0.6701,
      "step": 114
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.855985867901848,
      "learning_rate": 1.1734693877551021e-05,
      "loss": 0.5742,
      "step": 115
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.51512124553738,
      "learning_rate": 1.1836734693877552e-05,
      "loss": 0.7964,
      "step": 116
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.056011463710871,
      "learning_rate": 1.1938775510204082e-05,
      "loss": 0.7993,
      "step": 117
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.233100766552883,
      "learning_rate": 1.2040816326530612e-05,
      "loss": 0.9529,
      "step": 118
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.424148323659408,
      "learning_rate": 1.2142857142857144e-05,
      "loss": 0.6584,
      "step": 119
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.911191699253982,
      "learning_rate": 1.2244897959183674e-05,
      "loss": 0.7792,
      "step": 120
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.301656434784787,
      "learning_rate": 1.2346938775510204e-05,
      "loss": 0.5276,
      "step": 121
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.6025883615474266,
      "learning_rate": 1.2448979591836735e-05,
      "loss": 0.5278,
      "step": 122
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.908119775230391,
      "learning_rate": 1.2551020408163267e-05,
      "loss": 0.5524,
      "step": 123
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.39782415762083,
      "learning_rate": 1.2653061224489795e-05,
      "loss": 1.0719,
      "step": 124
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.156286540319526,
      "learning_rate": 1.2755102040816327e-05,
      "loss": 0.9784,
      "step": 125
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.4676470510581074,
      "learning_rate": 1.2857142857142857e-05,
      "loss": 0.6356,
      "step": 126
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.2117558168196085,
      "learning_rate": 1.2959183673469389e-05,
      "loss": 0.7208,
      "step": 127
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.581717128822217,
      "learning_rate": 1.306122448979592e-05,
      "loss": 0.7981,
      "step": 128
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.1027149135094514,
      "learning_rate": 1.316326530612245e-05,
      "loss": 0.5446,
      "step": 129
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.413781484035469,
      "learning_rate": 1.3265306122448982e-05,
      "loss": 0.6795,
      "step": 130
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.7674647104564096,
      "learning_rate": 1.336734693877551e-05,
      "loss": 0.7011,
      "step": 131
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.176289661247753,
      "learning_rate": 1.3469387755102042e-05,
      "loss": 0.6618,
      "step": 132
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.925994146616722,
      "learning_rate": 1.357142857142857e-05,
      "loss": 0.5021,
      "step": 133
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.5566810256983423,
      "learning_rate": 1.3673469387755102e-05,
      "loss": 0.5443,
      "step": 134
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.4513165552965943,
      "learning_rate": 1.3775510204081633e-05,
      "loss": 0.3869,
      "step": 135
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.220698629298188,
      "learning_rate": 1.3877551020408165e-05,
      "loss": 0.6584,
      "step": 136
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.857713602864085,
      "learning_rate": 1.3979591836734696e-05,
      "loss": 0.5936,
      "step": 137
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.2422831150057614,
      "learning_rate": 1.4081632653061225e-05,
      "loss": 0.4628,
      "step": 138
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.869426949838523,
      "learning_rate": 1.4183673469387757e-05,
      "loss": 0.513,
      "step": 139
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.9289931530430993,
      "learning_rate": 1.4285714285714285e-05,
      "loss": 0.689,
      "step": 140
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.36936595694377,
      "learning_rate": 1.4387755102040817e-05,
      "loss": 0.483,
      "step": 141
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.657544245448382,
      "learning_rate": 1.4489795918367346e-05,
      "loss": 0.2982,
      "step": 142
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.831131234981891,
      "learning_rate": 1.4591836734693878e-05,
      "loss": 0.4274,
      "step": 143
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.442571320676384,
      "learning_rate": 1.469387755102041e-05,
      "loss": 0.631,
      "step": 144
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.186877058789633,
      "learning_rate": 1.479591836734694e-05,
      "loss": 0.5707,
      "step": 145
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.579091883663498,
      "learning_rate": 1.4897959183673472e-05,
      "loss": 0.4819,
      "step": 146
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.5701649401674995,
      "learning_rate": 1.5e-05,
      "loss": 0.6373,
      "step": 147
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.224486089439844,
      "learning_rate": 1.5102040816326532e-05,
      "loss": 0.4451,
      "step": 148
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.249832782323837,
      "learning_rate": 1.520408163265306e-05,
      "loss": 0.5887,
      "step": 149
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.625859360787416,
      "learning_rate": 1.5306122448979594e-05,
      "loss": 0.815,
      "step": 150
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.1835483809505516,
      "learning_rate": 1.5408163265306123e-05,
      "loss": 0.3821,
      "step": 151
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.4355168090830657,
      "learning_rate": 1.5510204081632655e-05,
      "loss": 0.3668,
      "step": 152
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.574246822241648,
      "learning_rate": 1.5612244897959187e-05,
      "loss": 0.5831,
      "step": 153
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.7079263408951793,
      "learning_rate": 1.5714285714285715e-05,
      "loss": 0.6523,
      "step": 154
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.387141274543718,
      "learning_rate": 1.5816326530612247e-05,
      "loss": 0.6489,
      "step": 155
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.7268937429656415,
      "learning_rate": 1.5918367346938776e-05,
      "loss": 0.4352,
      "step": 156
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.5332470630883672,
      "learning_rate": 1.6020408163265308e-05,
      "loss": 0.5502,
      "step": 157
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.582660618875188,
      "learning_rate": 1.6122448979591836e-05,
      "loss": 0.5891,
      "step": 158
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.189313984732749,
      "learning_rate": 1.6224489795918368e-05,
      "loss": 0.8213,
      "step": 159
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.605248940148803,
      "learning_rate": 1.6326530612244897e-05,
      "loss": 0.3406,
      "step": 160
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.27465712906112,
      "learning_rate": 1.642857142857143e-05,
      "loss": 0.5991,
      "step": 161
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.464078407050486,
      "learning_rate": 1.653061224489796e-05,
      "loss": 0.499,
      "step": 162
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.142600079715495,
      "learning_rate": 1.663265306122449e-05,
      "loss": 0.4324,
      "step": 163
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.1521414059614665,
      "learning_rate": 1.673469387755102e-05,
      "loss": 0.3753,
      "step": 164
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.265409957425672,
      "learning_rate": 1.683673469387755e-05,
      "loss": 0.3557,
      "step": 165
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.817006800473023,
      "learning_rate": 1.693877551020408e-05,
      "loss": 0.723,
      "step": 166
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.073202322215028,
      "learning_rate": 1.7040816326530613e-05,
      "loss": 0.3924,
      "step": 167
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.519460185788737,
      "learning_rate": 1.7142857142857145e-05,
      "loss": 0.668,
      "step": 168
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.858807248767048,
      "learning_rate": 1.7244897959183677e-05,
      "loss": 0.5418,
      "step": 169
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.478156518096927,
      "learning_rate": 1.7346938775510206e-05,
      "loss": 0.8856,
      "step": 170
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.344669123184297,
      "learning_rate": 1.7448979591836738e-05,
      "loss": 0.9263,
      "step": 171
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.8835162269374672,
      "learning_rate": 1.7551020408163266e-05,
      "loss": 0.302,
      "step": 172
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.761395221919635,
      "learning_rate": 1.7653061224489798e-05,
      "loss": 0.5218,
      "step": 173
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.5943849726922004,
      "learning_rate": 1.7755102040816327e-05,
      "loss": 0.5196,
      "step": 174
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.7324418608637373,
      "learning_rate": 1.785714285714286e-05,
      "loss": 0.3036,
      "step": 175
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.353983359608475,
      "learning_rate": 1.7959183673469387e-05,
      "loss": 0.5769,
      "step": 176
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.551498661611182,
      "learning_rate": 1.806122448979592e-05,
      "loss": 0.3816,
      "step": 177
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.9638337105061194,
      "learning_rate": 1.816326530612245e-05,
      "loss": 0.5608,
      "step": 178
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.312197342293339,
      "learning_rate": 1.826530612244898e-05,
      "loss": 0.3947,
      "step": 179
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.081876983293785,
      "learning_rate": 1.836734693877551e-05,
      "loss": 0.6029,
      "step": 180
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.3978017599620265,
      "learning_rate": 1.846938775510204e-05,
      "loss": 0.5443,
      "step": 181
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.8412384397410553,
      "learning_rate": 1.8571428571428572e-05,
      "loss": 0.4709,
      "step": 182
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.2107195256136296,
      "learning_rate": 1.8673469387755104e-05,
      "loss": 0.3788,
      "step": 183
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.665341257362767,
      "learning_rate": 1.8775510204081632e-05,
      "loss": 0.7505,
      "step": 184
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.379202985471757,
      "learning_rate": 1.8877551020408164e-05,
      "loss": 0.4378,
      "step": 185
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.5960232980108633,
      "learning_rate": 1.8979591836734696e-05,
      "loss": 0.323,
      "step": 186
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.2390749839899,
      "learning_rate": 1.9081632653061228e-05,
      "loss": 0.5788,
      "step": 187
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.2837582041452595,
      "learning_rate": 1.9183673469387756e-05,
      "loss": 0.4325,
      "step": 188
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.646835279230557,
      "learning_rate": 1.928571428571429e-05,
      "loss": 0.4856,
      "step": 189
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.6076431564256652,
      "learning_rate": 1.9387755102040817e-05,
      "loss": 0.4829,
      "step": 190
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.2237058861365933,
      "learning_rate": 1.948979591836735e-05,
      "loss": 0.4519,
      "step": 191
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.469125464180251,
      "learning_rate": 1.9591836734693877e-05,
      "loss": 0.3636,
      "step": 192
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.9475266619205382,
      "learning_rate": 1.969387755102041e-05,
      "loss": 0.2825,
      "step": 193
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.4234464714492536,
      "learning_rate": 1.9795918367346938e-05,
      "loss": 0.527,
      "step": 194
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.085862532639635,
      "learning_rate": 1.989795918367347e-05,
      "loss": 0.4423,
      "step": 195
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.145754676800841,
      "learning_rate": 2e-05,
      "loss": 0.398,
      "step": 196
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.4519211769571005,
      "learning_rate": 2.010204081632653e-05,
      "loss": 0.4744,
      "step": 197
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.933463047978167,
      "learning_rate": 2.0204081632653062e-05,
      "loss": 0.5319,
      "step": 198
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.229319711302037,
      "learning_rate": 2.030612244897959e-05,
      "loss": 0.5688,
      "step": 199
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0196677696652197,
      "learning_rate": 2.0408163265306123e-05,
      "loss": 0.4515,
      "step": 200
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.3204493302520643,
      "learning_rate": 2.0510204081632654e-05,
      "loss": 0.3957,
      "step": 201
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.3951309267117433,
      "learning_rate": 2.0612244897959186e-05,
      "loss": 0.5953,
      "step": 202
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.146471813688114,
      "learning_rate": 2.0714285714285718e-05,
      "loss": 0.4446,
      "step": 203
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5593638016088764,
      "learning_rate": 2.0816326530612247e-05,
      "loss": 0.3041,
      "step": 204
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5454664219357235,
      "learning_rate": 2.091836734693878e-05,
      "loss": 0.3951,
      "step": 205
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0272147828837674,
      "learning_rate": 2.1020408163265307e-05,
      "loss": 0.4057,
      "step": 206
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.1128846493629427,
      "learning_rate": 2.112244897959184e-05,
      "loss": 0.3139,
      "step": 207
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.3347885900735665,
      "learning_rate": 2.1224489795918368e-05,
      "loss": 0.3786,
      "step": 208
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5296412119489924,
      "learning_rate": 2.13265306122449e-05,
      "loss": 0.2491,
      "step": 209
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.2138093045092395,
      "learning_rate": 2.1428571428571428e-05,
      "loss": 0.3537,
      "step": 210
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.651423462836009,
      "learning_rate": 2.153061224489796e-05,
      "loss": 0.2742,
      "step": 211
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.65620620599498,
      "learning_rate": 2.1632653061224492e-05,
      "loss": 0.4409,
      "step": 212
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.1655405748685017,
      "learning_rate": 2.173469387755102e-05,
      "loss": 0.5723,
      "step": 213
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2477556999535544,
      "learning_rate": 2.1836734693877552e-05,
      "loss": 0.3341,
      "step": 214
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.5934985302924973,
      "learning_rate": 2.193877551020408e-05,
      "loss": 0.3937,
      "step": 215
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.0393924237464125,
      "learning_rate": 2.2040816326530613e-05,
      "loss": 0.7743,
      "step": 216
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.483991283750791,
      "learning_rate": 2.214285714285714e-05,
      "loss": 0.5646,
      "step": 217
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.0217610992300665,
      "learning_rate": 2.2244897959183673e-05,
      "loss": 0.5716,
      "step": 218
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.663639891452499,
      "learning_rate": 2.2346938775510205e-05,
      "loss": 0.263,
      "step": 219
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.7883622942810944,
      "learning_rate": 2.2448979591836737e-05,
      "loss": 0.3158,
      "step": 220
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2636966247706645,
      "learning_rate": 2.255102040816327e-05,
      "loss": 0.2566,
      "step": 221
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.3120897831432576,
      "learning_rate": 2.2653061224489798e-05,
      "loss": 0.2772,
      "step": 222
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8747797665803345,
      "learning_rate": 2.275510204081633e-05,
      "loss": 0.384,
      "step": 223
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.772460987188263,
      "learning_rate": 2.2857142857142858e-05,
      "loss": 0.2346,
      "step": 224
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.691645855470738,
      "learning_rate": 2.295918367346939e-05,
      "loss": 0.7853,
      "step": 225
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.8920200499592736,
      "learning_rate": 2.306122448979592e-05,
      "loss": 0.5527,
      "step": 226
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.4483578855418964,
      "learning_rate": 2.316326530612245e-05,
      "loss": 0.6636,
      "step": 227
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6579401958859052,
      "learning_rate": 2.326530612244898e-05,
      "loss": 0.2716,
      "step": 228
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.5222036200242446,
      "learning_rate": 2.336734693877551e-05,
      "loss": 0.4311,
      "step": 229
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.52825945652028,
      "learning_rate": 2.3469387755102043e-05,
      "loss": 0.45,
      "step": 230
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.769434758433437,
      "learning_rate": 2.357142857142857e-05,
      "loss": 0.3677,
      "step": 231
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.744331177201488,
      "learning_rate": 2.3673469387755103e-05,
      "loss": 0.4202,
      "step": 232
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.713006330556267,
      "learning_rate": 2.3775510204081632e-05,
      "loss": 0.3346,
      "step": 233
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6586712197836517,
      "learning_rate": 2.3877551020408164e-05,
      "loss": 0.4772,
      "step": 234
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.7457514133452414,
      "learning_rate": 2.3979591836734696e-05,
      "loss": 0.5331,
      "step": 235
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.8310406565844364,
      "learning_rate": 2.4081632653061224e-05,
      "loss": 0.7993,
      "step": 236
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1351128165147424,
      "learning_rate": 2.4183673469387756e-05,
      "loss": 0.2723,
      "step": 237
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4981666566602625,
      "learning_rate": 2.4285714285714288e-05,
      "loss": 0.2423,
      "step": 238
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.926035645295326,
      "learning_rate": 2.438775510204082e-05,
      "loss": 0.552,
      "step": 239
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.392084543964776,
      "learning_rate": 2.448979591836735e-05,
      "loss": 0.4239,
      "step": 240
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.8361154262807884,
      "learning_rate": 2.459183673469388e-05,
      "loss": 0.3696,
      "step": 241
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.7488031850214285,
      "learning_rate": 2.469387755102041e-05,
      "loss": 0.4288,
      "step": 242
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.5575727172888,
      "learning_rate": 2.479591836734694e-05,
      "loss": 0.6394,
      "step": 243
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4343484062929592,
      "learning_rate": 2.489795918367347e-05,
      "loss": 0.379,
      "step": 244
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.6184391962951468,
      "learning_rate": 2.5e-05,
      "loss": 0.5064,
      "step": 245
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.323446991668859,
      "learning_rate": 2.5102040816326533e-05,
      "loss": 0.3806,
      "step": 246
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.850740618928301,
      "learning_rate": 2.5204081632653065e-05,
      "loss": 0.3514,
      "step": 247
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.084891412320639,
      "learning_rate": 2.530612244897959e-05,
      "loss": 0.6136,
      "step": 248
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3099058441626577,
      "learning_rate": 2.5408163265306122e-05,
      "loss": 0.2974,
      "step": 249
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.361463862523306,
      "learning_rate": 2.5510204081632654e-05,
      "loss": 0.4925,
      "step": 250
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.8766944526279863,
      "learning_rate": 2.5612244897959186e-05,
      "loss": 0.2913,
      "step": 251
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7102851811891484,
      "learning_rate": 2.5714285714285714e-05,
      "loss": 0.3256,
      "step": 252
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.13461161574286,
      "learning_rate": 2.5816326530612246e-05,
      "loss": 0.4835,
      "step": 253
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.8221650535575886,
      "learning_rate": 2.5918367346938778e-05,
      "loss": 0.4204,
      "step": 254
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.599907717729646,
      "learning_rate": 2.6020408163265307e-05,
      "loss": 0.2662,
      "step": 255
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.3311089383715466,
      "learning_rate": 2.612244897959184e-05,
      "loss": 0.5249,
      "step": 256
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5621561389545597,
      "learning_rate": 2.6224489795918367e-05,
      "loss": 0.3532,
      "step": 257
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.915935960309074,
      "learning_rate": 2.63265306122449e-05,
      "loss": 0.351,
      "step": 258
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8397143030438976,
      "learning_rate": 2.642857142857143e-05,
      "loss": 0.2031,
      "step": 259
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0671670031374965,
      "learning_rate": 2.6530612244897963e-05,
      "loss": 0.3582,
      "step": 260
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.3731120577597804,
      "learning_rate": 2.6632653061224488e-05,
      "loss": 0.4923,
      "step": 261
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.180000071691901,
      "learning_rate": 2.673469387755102e-05,
      "loss": 0.3542,
      "step": 262
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6074147679695505,
      "learning_rate": 2.6836734693877552e-05,
      "loss": 0.4919,
      "step": 263
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0004853602253863,
      "learning_rate": 2.6938775510204084e-05,
      "loss": 0.5295,
      "step": 264
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6902224065246996,
      "learning_rate": 2.7040816326530616e-05,
      "loss": 0.3054,
      "step": 265
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.449481778385189,
      "learning_rate": 2.714285714285714e-05,
      "loss": 0.348,
      "step": 266
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4307235880855966,
      "learning_rate": 2.7244897959183673e-05,
      "loss": 0.337,
      "step": 267
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3428543056456896,
      "learning_rate": 2.7346938775510205e-05,
      "loss": 0.2623,
      "step": 268
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9058019637728392,
      "learning_rate": 2.7448979591836737e-05,
      "loss": 0.2481,
      "step": 269
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.2432237678486446,
      "learning_rate": 2.7551020408163265e-05,
      "loss": 0.4651,
      "step": 270
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7789944512879297,
      "learning_rate": 2.7653061224489797e-05,
      "loss": 0.2877,
      "step": 271
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.243040418504301,
      "learning_rate": 2.775510204081633e-05,
      "loss": 0.2685,
      "step": 272
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.9300678118537435,
      "learning_rate": 2.785714285714286e-05,
      "loss": 0.2492,
      "step": 273
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1724289937364785,
      "learning_rate": 2.7959183673469393e-05,
      "loss": 0.2045,
      "step": 274
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.212735431596714,
      "learning_rate": 2.8061224489795918e-05,
      "loss": 0.5011,
      "step": 275
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3415675120741435,
      "learning_rate": 2.816326530612245e-05,
      "loss": 0.3118,
      "step": 276
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.414813121461509,
      "learning_rate": 2.8265306122448982e-05,
      "loss": 0.1927,
      "step": 277
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.080411274314418,
      "learning_rate": 2.8367346938775514e-05,
      "loss": 0.4024,
      "step": 278
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.5274405251257406,
      "learning_rate": 2.8469387755102046e-05,
      "loss": 0.2893,
      "step": 279
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.336500876927522,
      "learning_rate": 2.857142857142857e-05,
      "loss": 0.7665,
      "step": 280
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.359665767999068,
      "learning_rate": 2.8673469387755103e-05,
      "loss": 0.5192,
      "step": 281
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.2946684202154164,
      "learning_rate": 2.8775510204081635e-05,
      "loss": 0.4469,
      "step": 282
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.781762900226782,
      "learning_rate": 2.8877551020408167e-05,
      "loss": 0.637,
      "step": 283
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.037309387507475,
      "learning_rate": 2.8979591836734692e-05,
      "loss": 0.4831,
      "step": 284
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.574552970810838,
      "learning_rate": 2.9081632653061224e-05,
      "loss": 0.3658,
      "step": 285
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.543218710285574,
      "learning_rate": 2.9183673469387756e-05,
      "loss": 0.9911,
      "step": 286
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.877242420713535,
      "learning_rate": 2.9285714285714288e-05,
      "loss": 0.2308,
      "step": 287
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.14727591590802,
      "learning_rate": 2.938775510204082e-05,
      "loss": 0.3645,
      "step": 288
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5601434778266725,
      "learning_rate": 2.9489795918367348e-05,
      "loss": 0.6393,
      "step": 289
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.032777900124652,
      "learning_rate": 2.959183673469388e-05,
      "loss": 0.2472,
      "step": 290
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.8742194886250823,
      "learning_rate": 2.9693877551020412e-05,
      "loss": 0.4337,
      "step": 291
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.547328919597133,
      "learning_rate": 2.9795918367346944e-05,
      "loss": 0.3522,
      "step": 292
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1405128308721904,
      "learning_rate": 2.989795918367347e-05,
      "loss": 0.3039,
      "step": 293
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0848147669355814,
      "learning_rate": 3e-05,
      "loss": 0.4008,
      "step": 294
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3487886678105765,
      "learning_rate": 3.0102040816326533e-05,
      "loss": 0.2653,
      "step": 295
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6962157148911907,
      "learning_rate": 3.0204081632653065e-05,
      "loss": 0.3127,
      "step": 296
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.942237042276212,
      "learning_rate": 3.0306122448979597e-05,
      "loss": 0.3963,
      "step": 297
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.196803650984296,
      "learning_rate": 3.040816326530612e-05,
      "loss": 0.287,
      "step": 298
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6230729981676366,
      "learning_rate": 3.0510204081632654e-05,
      "loss": 0.284,
      "step": 299
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.205628805035607,
      "learning_rate": 3.061224489795919e-05,
      "loss": 0.4357,
      "step": 300
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.162333922019655,
      "learning_rate": 3.071428571428572e-05,
      "loss": 0.3516,
      "step": 301
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.163270622931779,
      "learning_rate": 3.0816326530612246e-05,
      "loss": 0.4847,
      "step": 302
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7688791606135856,
      "learning_rate": 3.0918367346938774e-05,
      "loss": 0.3273,
      "step": 303
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.331949009897273,
      "learning_rate": 3.102040816326531e-05,
      "loss": 0.2969,
      "step": 304
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1972532621465186,
      "learning_rate": 3.112244897959184e-05,
      "loss": 0.2306,
      "step": 305
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.091046123070687,
      "learning_rate": 3.1224489795918374e-05,
      "loss": 0.4532,
      "step": 306
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.5516352481700544,
      "learning_rate": 3.1326530612244895e-05,
      "loss": 0.5829,
      "step": 307
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.2730184602787893,
      "learning_rate": 3.142857142857143e-05,
      "loss": 0.6158,
      "step": 308
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.548805428208077,
      "learning_rate": 3.153061224489796e-05,
      "loss": 0.3669,
      "step": 309
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.1637195812676393,
      "learning_rate": 3.1632653061224494e-05,
      "loss": 0.2748,
      "step": 310
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.571439479067978,
      "learning_rate": 3.1734693877551016e-05,
      "loss": 0.3867,
      "step": 311
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.415686837193812,
      "learning_rate": 3.183673469387755e-05,
      "loss": 0.2347,
      "step": 312
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.10942283171934,
      "learning_rate": 3.193877551020408e-05,
      "loss": 0.389,
      "step": 313
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.1292254368160948,
      "learning_rate": 3.2040816326530615e-05,
      "loss": 0.7636,
      "step": 314
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.403603286049724,
      "learning_rate": 3.2142857142857144e-05,
      "loss": 0.5234,
      "step": 315
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6365077824907344,
      "learning_rate": 3.224489795918367e-05,
      "loss": 0.3361,
      "step": 316
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6697311046118815,
      "learning_rate": 3.234693877551021e-05,
      "loss": 0.3293,
      "step": 317
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.3909683709463336,
      "learning_rate": 3.2448979591836736e-05,
      "loss": 0.6376,
      "step": 318
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6042696653863726,
      "learning_rate": 3.255102040816327e-05,
      "loss": 0.2438,
      "step": 319
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5942912039473613,
      "learning_rate": 3.265306122448979e-05,
      "loss": 0.3585,
      "step": 320
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.576109420367313,
      "learning_rate": 3.275510204081633e-05,
      "loss": 0.3001,
      "step": 321
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.5702887107121444,
      "learning_rate": 3.285714285714286e-05,
      "loss": 0.5516,
      "step": 322
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6417607883215255,
      "learning_rate": 3.295918367346939e-05,
      "loss": 0.4485,
      "step": 323
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8952400184261384,
      "learning_rate": 3.306122448979592e-05,
      "loss": 0.2021,
      "step": 324
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.2081259332411425,
      "learning_rate": 3.316326530612245e-05,
      "loss": 0.4771,
      "step": 325
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4937045396408277,
      "learning_rate": 3.326530612244898e-05,
      "loss": 0.3697,
      "step": 326
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.787340067248846,
      "learning_rate": 3.336734693877551e-05,
      "loss": 0.7499,
      "step": 327
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.1583575295043103,
      "learning_rate": 3.346938775510204e-05,
      "loss": 0.4114,
      "step": 328
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.1364270045596174,
      "learning_rate": 3.357142857142857e-05,
      "loss": 0.5302,
      "step": 329
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.96593535266715,
      "learning_rate": 3.36734693877551e-05,
      "loss": 0.5594,
      "step": 330
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8198761691179346,
      "learning_rate": 3.3775510204081634e-05,
      "loss": 0.1623,
      "step": 331
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.956404964758462,
      "learning_rate": 3.387755102040816e-05,
      "loss": 0.6836,
      "step": 332
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.495506371908712,
      "learning_rate": 3.39795918367347e-05,
      "loss": 0.4481,
      "step": 333
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2116641724101864,
      "learning_rate": 3.408163265306123e-05,
      "loss": 0.2889,
      "step": 334
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6960938022420238,
      "learning_rate": 3.4183673469387755e-05,
      "loss": 0.4376,
      "step": 335
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1920031131596445,
      "learning_rate": 3.428571428571429e-05,
      "loss": 0.2904,
      "step": 336
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.14105271965708,
      "learning_rate": 3.438775510204082e-05,
      "loss": 0.2842,
      "step": 337
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5815242217993806,
      "learning_rate": 3.4489795918367354e-05,
      "loss": 0.3355,
      "step": 338
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.30559658907026,
      "learning_rate": 3.4591836734693876e-05,
      "loss": 0.6527,
      "step": 339
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9026846315151222,
      "learning_rate": 3.469387755102041e-05,
      "loss": 0.2102,
      "step": 340
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.401327865332335,
      "learning_rate": 3.479591836734694e-05,
      "loss": 0.288,
      "step": 341
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.9244882691314826,
      "learning_rate": 3.4897959183673475e-05,
      "loss": 0.6643,
      "step": 342
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3942575389134957,
      "learning_rate": 3.5e-05,
      "loss": 0.3216,
      "step": 343
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.656003869284268,
      "learning_rate": 3.510204081632653e-05,
      "loss": 0.3674,
      "step": 344
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.209453876463231,
      "learning_rate": 3.520408163265306e-05,
      "loss": 0.3564,
      "step": 345
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.271918354545448,
      "learning_rate": 3.5306122448979596e-05,
      "loss": 0.5556,
      "step": 346
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2103637675322134,
      "learning_rate": 3.5408163265306125e-05,
      "loss": 0.2855,
      "step": 347
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.287113971107436,
      "learning_rate": 3.551020408163265e-05,
      "loss": 0.3207,
      "step": 348
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.986833494519624,
      "learning_rate": 3.561224489795918e-05,
      "loss": 0.2177,
      "step": 349
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.5485519471781353,
      "learning_rate": 3.571428571428572e-05,
      "loss": 0.6473,
      "step": 350
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.852684788105609,
      "learning_rate": 3.5816326530612245e-05,
      "loss": 0.2282,
      "step": 351
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.25949641883457,
      "learning_rate": 3.5918367346938774e-05,
      "loss": 0.2379,
      "step": 352
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.10249345811156,
      "learning_rate": 3.602040816326531e-05,
      "loss": 0.1902,
      "step": 353
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4988370278311915,
      "learning_rate": 3.612244897959184e-05,
      "loss": 0.2616,
      "step": 354
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4742943445122787,
      "learning_rate": 3.622448979591837e-05,
      "loss": 0.2888,
      "step": 355
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.156436092610811,
      "learning_rate": 3.63265306122449e-05,
      "loss": 0.3987,
      "step": 356
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.980368889987992,
      "learning_rate": 3.642857142857143e-05,
      "loss": 0.5217,
      "step": 357
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.145961035735693,
      "learning_rate": 3.653061224489796e-05,
      "loss": 0.2615,
      "step": 358
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.001740112540562,
      "learning_rate": 3.6632653061224494e-05,
      "loss": 0.4244,
      "step": 359
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2225561644882714,
      "learning_rate": 3.673469387755102e-05,
      "loss": 0.4146,
      "step": 360
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2299799674891725,
      "learning_rate": 3.683673469387755e-05,
      "loss": 0.3389,
      "step": 361
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0709064109970723,
      "learning_rate": 3.693877551020408e-05,
      "loss": 0.2608,
      "step": 362
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4433218646819257,
      "learning_rate": 3.7040816326530615e-05,
      "loss": 0.3203,
      "step": 363
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.8104435720939005,
      "learning_rate": 3.7142857142857143e-05,
      "loss": 0.4031,
      "step": 364
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.819285781934326,
      "learning_rate": 3.724489795918368e-05,
      "loss": 0.6592,
      "step": 365
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7138068249888025,
      "learning_rate": 3.734693877551021e-05,
      "loss": 0.6539,
      "step": 366
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.4423504939696765,
      "learning_rate": 3.7448979591836736e-05,
      "loss": 0.5468,
      "step": 367
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.82916093332022,
      "learning_rate": 3.7551020408163264e-05,
      "loss": 0.4209,
      "step": 368
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2700127728009902,
      "learning_rate": 3.76530612244898e-05,
      "loss": 0.3269,
      "step": 369
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7114389913496435,
      "learning_rate": 3.775510204081633e-05,
      "loss": 0.2787,
      "step": 370
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4532085460174815,
      "learning_rate": 3.785714285714286e-05,
      "loss": 0.3796,
      "step": 371
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.805660713784195,
      "learning_rate": 3.795918367346939e-05,
      "loss": 0.584,
      "step": 372
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.3701223007294256,
      "learning_rate": 3.806122448979592e-05,
      "loss": 0.5698,
      "step": 373
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.304962145661763,
      "learning_rate": 3.8163265306122456e-05,
      "loss": 0.2131,
      "step": 374
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.790948386464233,
      "learning_rate": 3.826530612244898e-05,
      "loss": 0.3281,
      "step": 375
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.0862740356502565,
      "learning_rate": 3.836734693877551e-05,
      "loss": 0.3809,
      "step": 376
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5787020811668255,
      "learning_rate": 3.846938775510204e-05,
      "loss": 0.3319,
      "step": 377
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2596928218215786,
      "learning_rate": 3.857142857142858e-05,
      "loss": 0.3136,
      "step": 378
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.104145148959446,
      "learning_rate": 3.86734693877551e-05,
      "loss": 0.3302,
      "step": 379
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.253803049221037,
      "learning_rate": 3.8775510204081634e-05,
      "loss": 0.2884,
      "step": 380
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.534482706834779,
      "learning_rate": 3.887755102040816e-05,
      "loss": 0.2607,
      "step": 381
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1089434267092044,
      "learning_rate": 3.89795918367347e-05,
      "loss": 0.2244,
      "step": 382
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.04402276917921,
      "learning_rate": 3.9081632653061226e-05,
      "loss": 0.4494,
      "step": 383
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7265275653452203,
      "learning_rate": 3.9183673469387755e-05,
      "loss": 0.4598,
      "step": 384
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.571601209556989,
      "learning_rate": 3.928571428571429e-05,
      "loss": 0.4362,
      "step": 385
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2062922801556235,
      "learning_rate": 3.938775510204082e-05,
      "loss": 0.2733,
      "step": 386
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.06662165997914,
      "learning_rate": 3.9489795918367354e-05,
      "loss": 0.6409,
      "step": 387
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.2854942069897266,
      "learning_rate": 3.9591836734693876e-05,
      "loss": 0.6335,
      "step": 388
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7937679300105316,
      "learning_rate": 3.969387755102041e-05,
      "loss": 0.3861,
      "step": 389
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8331450860632805,
      "learning_rate": 3.979591836734694e-05,
      "loss": 0.2145,
      "step": 390
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4075664531379632,
      "learning_rate": 3.9897959183673475e-05,
      "loss": 0.2965,
      "step": 391
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0788398254434535,
      "learning_rate": 4e-05,
      "loss": 0.3002,
      "step": 392
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1173013448081357,
      "learning_rate": 4.010204081632653e-05,
      "loss": 0.2834,
      "step": 393
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.28349308476157,
      "learning_rate": 4.020408163265306e-05,
      "loss": 0.6201,
      "step": 394
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3149177902469584,
      "learning_rate": 4.0306122448979596e-05,
      "loss": 0.3002,
      "step": 395
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8831020143069388,
      "learning_rate": 4.0408163265306124e-05,
      "loss": 0.1975,
      "step": 396
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1198617039201912,
      "learning_rate": 4.051020408163265e-05,
      "loss": 0.4516,
      "step": 397
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.702779792520783,
      "learning_rate": 4.061224489795918e-05,
      "loss": 0.2791,
      "step": 398
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6805971067818506,
      "learning_rate": 4.0714285714285717e-05,
      "loss": 0.3745,
      "step": 399
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9080767045244535,
      "learning_rate": 4.0816326530612245e-05,
      "loss": 0.2355,
      "step": 400
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.369628182828549,
      "learning_rate": 4.091836734693878e-05,
      "loss": 0.5241,
      "step": 401
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.079287170727681,
      "learning_rate": 4.102040816326531e-05,
      "loss": 0.2954,
      "step": 402
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2543506952016883,
      "learning_rate": 4.112244897959184e-05,
      "loss": 0.2292,
      "step": 403
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.245987840792176,
      "learning_rate": 4.122448979591837e-05,
      "loss": 0.3241,
      "step": 404
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.490046215537596,
      "learning_rate": 4.13265306122449e-05,
      "loss": 0.2863,
      "step": 405
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2474122674044867,
      "learning_rate": 4.1428571428571437e-05,
      "loss": 0.4201,
      "step": 406
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.300550608131465,
      "learning_rate": 4.153061224489796e-05,
      "loss": 0.2854,
      "step": 407
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7702220799256325,
      "learning_rate": 4.1632653061224494e-05,
      "loss": 0.393,
      "step": 408
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6707376422230853,
      "learning_rate": 4.173469387755102e-05,
      "loss": 0.307,
      "step": 409
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.276130238178927,
      "learning_rate": 4.183673469387756e-05,
      "loss": 0.3412,
      "step": 410
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3484352453871264,
      "learning_rate": 4.193877551020408e-05,
      "loss": 0.2943,
      "step": 411
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.225403870135358,
      "learning_rate": 4.2040816326530615e-05,
      "loss": 0.2293,
      "step": 412
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.214852173788171,
      "learning_rate": 4.214285714285714e-05,
      "loss": 0.2229,
      "step": 413
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.418944445342651,
      "learning_rate": 4.224489795918368e-05,
      "loss": 0.3674,
      "step": 414
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.30599009224934,
      "learning_rate": 4.234693877551021e-05,
      "loss": 0.4303,
      "step": 415
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6768339489491955,
      "learning_rate": 4.2448979591836735e-05,
      "loss": 0.9367,
      "step": 416
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.297208163958793,
      "learning_rate": 4.2551020408163264e-05,
      "loss": 0.3555,
      "step": 417
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3290475838049742,
      "learning_rate": 4.26530612244898e-05,
      "loss": 0.3713,
      "step": 418
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9833651523296787,
      "learning_rate": 4.275510204081633e-05,
      "loss": 0.1987,
      "step": 419
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.203065859638541,
      "learning_rate": 4.2857142857142856e-05,
      "loss": 0.3737,
      "step": 420
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.518584654206289,
      "learning_rate": 4.295918367346939e-05,
      "loss": 0.297,
      "step": 421
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6759287676336023,
      "learning_rate": 4.306122448979592e-05,
      "loss": 0.3117,
      "step": 422
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2765712534595273,
      "learning_rate": 4.3163265306122455e-05,
      "loss": 0.236,
      "step": 423
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.578095939758724,
      "learning_rate": 4.3265306122448984e-05,
      "loss": 0.3671,
      "step": 424
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6126955945802472,
      "learning_rate": 4.336734693877551e-05,
      "loss": 0.2718,
      "step": 425
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4791256573523377,
      "learning_rate": 4.346938775510204e-05,
      "loss": 0.3973,
      "step": 426
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6744622997876206,
      "learning_rate": 4.3571428571428576e-05,
      "loss": 0.5063,
      "step": 427
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.8681941995124514,
      "learning_rate": 4.3673469387755105e-05,
      "loss": 0.5878,
      "step": 428
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.739112465566556,
      "learning_rate": 4.377551020408163e-05,
      "loss": 0.4206,
      "step": 429
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.255409188933018,
      "learning_rate": 4.387755102040816e-05,
      "loss": 0.288,
      "step": 430
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0903569711230974,
      "learning_rate": 4.39795918367347e-05,
      "loss": 0.286,
      "step": 431
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6791360738198997,
      "learning_rate": 4.4081632653061226e-05,
      "loss": 0.6778,
      "step": 432
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.406256832673656,
      "learning_rate": 4.418367346938776e-05,
      "loss": 0.4372,
      "step": 433
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.119278861797278,
      "learning_rate": 4.428571428571428e-05,
      "loss": 0.4501,
      "step": 434
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.798184102244453,
      "learning_rate": 4.438775510204082e-05,
      "loss": 0.2924,
      "step": 435
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.9371168543216015,
      "learning_rate": 4.448979591836735e-05,
      "loss": 0.328,
      "step": 436
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.673665443383621,
      "learning_rate": 4.459183673469388e-05,
      "loss": 0.262,
      "step": 437
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.9986604823826517,
      "learning_rate": 4.469387755102041e-05,
      "loss": 0.279,
      "step": 438
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.624464402647075,
      "learning_rate": 4.479591836734694e-05,
      "loss": 0.6523,
      "step": 439
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.720558697844901,
      "learning_rate": 4.4897959183673474e-05,
      "loss": 1.0365,
      "step": 440
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6823367860384657,
      "learning_rate": 4.5e-05,
      "loss": 0.2957,
      "step": 441
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.210973925189798,
      "learning_rate": 4.510204081632654e-05,
      "loss": 0.288,
      "step": 442
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.2375485783083784,
      "learning_rate": 4.520408163265306e-05,
      "loss": 0.3651,
      "step": 443
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.0534749352748913,
      "learning_rate": 4.5306122448979595e-05,
      "loss": 0.5145,
      "step": 444
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.557125961928091,
      "learning_rate": 4.5408163265306124e-05,
      "loss": 0.5243,
      "step": 445
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6995937133026935,
      "learning_rate": 4.551020408163266e-05,
      "loss": 0.2365,
      "step": 446
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.0676703052657417,
      "learning_rate": 4.561224489795918e-05,
      "loss": 0.5435,
      "step": 447
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.9571403795256415,
      "learning_rate": 4.5714285714285716e-05,
      "loss": 0.3512,
      "step": 448
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1721790324374894,
      "learning_rate": 4.5816326530612245e-05,
      "loss": 0.2553,
      "step": 449
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.2909312307901,
      "learning_rate": 4.591836734693878e-05,
      "loss": 0.3811,
      "step": 450
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3897251504237214,
      "learning_rate": 4.602040816326531e-05,
      "loss": 0.3098,
      "step": 451
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1173751413649566,
      "learning_rate": 4.612244897959184e-05,
      "loss": 0.2391,
      "step": 452
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.549006613298891,
      "learning_rate": 4.6224489795918366e-05,
      "loss": 0.2604,
      "step": 453
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.838270387456827,
      "learning_rate": 4.63265306122449e-05,
      "loss": 0.3561,
      "step": 454
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5195117756763277,
      "learning_rate": 4.642857142857143e-05,
      "loss": 0.4001,
      "step": 455
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.383429835125223,
      "learning_rate": 4.653061224489796e-05,
      "loss": 0.3986,
      "step": 456
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.324932254894581,
      "learning_rate": 4.663265306122449e-05,
      "loss": 0.2986,
      "step": 457
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.588455194294807,
      "learning_rate": 4.673469387755102e-05,
      "loss": 0.3604,
      "step": 458
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.636013867838106,
      "learning_rate": 4.683673469387756e-05,
      "loss": 0.4955,
      "step": 459
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.806584963976717,
      "learning_rate": 4.6938775510204086e-05,
      "loss": 0.4839,
      "step": 460
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.3851939527661434,
      "learning_rate": 4.7040816326530614e-05,
      "loss": 0.5662,
      "step": 461
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.189673772588685,
      "learning_rate": 4.714285714285714e-05,
      "loss": 0.6523,
      "step": 462
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.8241103232728073,
      "learning_rate": 4.724489795918368e-05,
      "loss": 0.3776,
      "step": 463
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.212855375894677,
      "learning_rate": 4.7346938775510206e-05,
      "loss": 0.3088,
      "step": 464
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0468097252293482,
      "learning_rate": 4.744897959183674e-05,
      "loss": 0.2285,
      "step": 465
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.6584345835481753,
      "learning_rate": 4.7551020408163263e-05,
      "loss": 0.3843,
      "step": 466
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.819277880387727,
      "learning_rate": 4.76530612244898e-05,
      "loss": 0.941,
      "step": 467
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2206382557532396,
      "learning_rate": 4.775510204081633e-05,
      "loss": 0.3598,
      "step": 468
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6397742132907993,
      "learning_rate": 4.785714285714286e-05,
      "loss": 0.3905,
      "step": 469
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4363394463581862,
      "learning_rate": 4.795918367346939e-05,
      "loss": 0.2732,
      "step": 470
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.354154093317635,
      "learning_rate": 4.806122448979592e-05,
      "loss": 0.4372,
      "step": 471
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.419097063261618,
      "learning_rate": 4.816326530612245e-05,
      "loss": 0.3245,
      "step": 472
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3257972286744546,
      "learning_rate": 4.8265306122448984e-05,
      "loss": 0.2882,
      "step": 473
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.9298236619517475,
      "learning_rate": 4.836734693877551e-05,
      "loss": 0.7805,
      "step": 474
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6382644589390534,
      "learning_rate": 4.846938775510204e-05,
      "loss": 0.2939,
      "step": 475
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0344419620495726,
      "learning_rate": 4.8571428571428576e-05,
      "loss": 0.2741,
      "step": 476
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.565970509543789,
      "learning_rate": 4.8673469387755104e-05,
      "loss": 0.3892,
      "step": 477
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.526749081429444,
      "learning_rate": 4.877551020408164e-05,
      "loss": 0.3888,
      "step": 478
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2070311250964267,
      "learning_rate": 4.887755102040816e-05,
      "loss": 0.2623,
      "step": 479
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3693581439627094,
      "learning_rate": 4.89795918367347e-05,
      "loss": 0.2947,
      "step": 480
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3732098365574896,
      "learning_rate": 4.9081632653061225e-05,
      "loss": 0.2526,
      "step": 481
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6180795609740755,
      "learning_rate": 4.918367346938776e-05,
      "loss": 0.3153,
      "step": 482
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.0797472900690406,
      "learning_rate": 4.928571428571429e-05,
      "loss": 1.2817,
      "step": 483
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.0425238816596964,
      "learning_rate": 4.938775510204082e-05,
      "loss": 0.5956,
      "step": 484
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.881548017380476,
      "learning_rate": 4.9489795918367346e-05,
      "loss": 0.2443,
      "step": 485
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5734249988364766,
      "learning_rate": 4.959183673469388e-05,
      "loss": 0.2404,
      "step": 486
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.9451103777771626,
      "learning_rate": 4.969387755102041e-05,
      "loss": 0.4144,
      "step": 487
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.704786913427208,
      "learning_rate": 4.979591836734694e-05,
      "loss": 0.2537,
      "step": 488
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5022859845285557,
      "learning_rate": 4.9897959183673474e-05,
      "loss": 0.2822,
      "step": 489
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.628166892880242,
      "learning_rate": 5e-05,
      "loss": 0.1294,
      "step": 490
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.700605846357637,
      "learning_rate": 4.999999950786605e-05,
      "loss": 0.9816,
      "step": 491
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3591448400802033,
      "learning_rate": 4.999999803146419e-05,
      "loss": 0.2793,
      "step": 492
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.70173772122823,
      "learning_rate": 4.999999557079449e-05,
      "loss": 0.2604,
      "step": 493
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.349854389015659,
      "learning_rate": 4.9999992125857055e-05,
      "loss": 0.6343,
      "step": 494
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6331944512117578,
      "learning_rate": 4.9999987696652005e-05,
      "loss": 0.3148,
      "step": 495
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7069259870403335,
      "learning_rate": 4.999998228317953e-05,
      "loss": 0.2377,
      "step": 496
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.341210557951612,
      "learning_rate": 4.9999975885439834e-05,
      "loss": 0.3676,
      "step": 497
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9106005028650799,
      "learning_rate": 4.999996850343317e-05,
      "loss": 0.2885,
      "step": 498
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.493028990008482,
      "learning_rate": 4.999996013715983e-05,
      "loss": 0.3605,
      "step": 499
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3580995765243316,
      "learning_rate": 4.9999950786620144e-05,
      "loss": 0.2483,
      "step": 500
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8321003370268987,
      "learning_rate": 4.9999940451814474e-05,
      "loss": 0.1743,
      "step": 501
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.899911421805452,
      "learning_rate": 4.999992913274324e-05,
      "loss": 0.5753,
      "step": 502
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.094574276839171,
      "learning_rate": 4.999991682940687e-05,
      "loss": 0.6497,
      "step": 503
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4833629092444314,
      "learning_rate": 4.999990354180586e-05,
      "loss": 0.3151,
      "step": 504
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.897746792883387,
      "learning_rate": 4.999988926994074e-05,
      "loss": 0.549,
      "step": 505
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.450375600750813,
      "learning_rate": 4.999987401381205e-05,
      "loss": 0.5966,
      "step": 506
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8890053214747684,
      "learning_rate": 4.999985777342041e-05,
      "loss": 0.2494,
      "step": 507
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.6260999822708047,
      "learning_rate": 4.999984054876644e-05,
      "loss": 0.6152,
      "step": 508
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.2494972118260885,
      "learning_rate": 4.9999822339850844e-05,
      "loss": 0.4353,
      "step": 509
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.905086931634075,
      "learning_rate": 4.999980314667433e-05,
      "loss": 0.2174,
      "step": 510
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4648041487186063,
      "learning_rate": 4.999978296923764e-05,
      "loss": 0.2628,
      "step": 511
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.135366784812126,
      "learning_rate": 4.9999761807541576e-05,
      "loss": 0.4744,
      "step": 512
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.555255429979861,
      "learning_rate": 4.999973966158698e-05,
      "loss": 0.4483,
      "step": 513
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.110433606123367,
      "learning_rate": 4.9999716531374714e-05,
      "loss": 0.3318,
      "step": 514
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9818924064889922,
      "learning_rate": 4.999969241690569e-05,
      "loss": 0.3633,
      "step": 515
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.462292284513512,
      "learning_rate": 4.9999667318180864e-05,
      "loss": 0.5172,
      "step": 516
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.859583254995562,
      "learning_rate": 4.999964123520121e-05,
      "loss": 1.1445,
      "step": 517
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8892415611285034,
      "learning_rate": 4.9999614167967765e-05,
      "loss": 0.3515,
      "step": 518
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.693761207203327,
      "learning_rate": 4.999958611648161e-05,
      "loss": 0.5724,
      "step": 519
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5474620484282027,
      "learning_rate": 4.999955708074382e-05,
      "loss": 0.4739,
      "step": 520
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0856042269241266,
      "learning_rate": 4.999952706075555e-05,
      "loss": 0.2631,
      "step": 521
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.3156843828876905,
      "learning_rate": 4.999949605651798e-05,
      "loss": 0.4839,
      "step": 522
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.582773768112357,
      "learning_rate": 4.999946406803234e-05,
      "loss": 0.3347,
      "step": 523
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2595291890207725,
      "learning_rate": 4.9999431095299886e-05,
      "loss": 0.2418,
      "step": 524
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.6545863642887797,
      "learning_rate": 4.999939713832191e-05,
      "loss": 0.4558,
      "step": 525
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.77988313195967,
      "learning_rate": 4.999936219709975e-05,
      "loss": 1.2431,
      "step": 526
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2995182477594533,
      "learning_rate": 4.9999326271634784e-05,
      "loss": 0.3233,
      "step": 527
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.77025153471341,
      "learning_rate": 4.999928936192843e-05,
      "loss": 1.0662,
      "step": 528
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.5261445946697574,
      "learning_rate": 4.999925146798213e-05,
      "loss": 0.5997,
      "step": 529
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9970184100247894,
      "learning_rate": 4.9999212589797396e-05,
      "loss": 0.2688,
      "step": 530
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.825146676528708,
      "learning_rate": 4.999917272737574e-05,
      "loss": 0.3678,
      "step": 531
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4636003642884132,
      "learning_rate": 4.999913188071873e-05,
      "loss": 0.3123,
      "step": 532
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.272165703944856,
      "learning_rate": 4.999909004982799e-05,
      "loss": 0.3406,
      "step": 533
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1048190012211094,
      "learning_rate": 4.9999047234705154e-05,
      "loss": 0.2119,
      "step": 534
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.326919449706403,
      "learning_rate": 4.999900343535192e-05,
      "loss": 0.8271,
      "step": 535
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3588269191612032,
      "learning_rate": 4.999895865177e-05,
      "loss": 0.4812,
      "step": 536
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.797003118862323,
      "learning_rate": 4.999891288396116e-05,
      "loss": 0.4585,
      "step": 537
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.2221136986773398,
      "learning_rate": 4.9998866131927205e-05,
      "loss": 0.5613,
      "step": 538
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.242958117973674,
      "learning_rate": 4.999881839566997e-05,
      "loss": 0.3291,
      "step": 539
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6125295235519808,
      "learning_rate": 4.999876967519135e-05,
      "loss": 0.3711,
      "step": 540
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7140797700662682,
      "learning_rate": 4.9998719970493236e-05,
      "loss": 0.2054,
      "step": 541
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.528892608284554,
      "learning_rate": 4.999866928157761e-05,
      "loss": 0.3813,
      "step": 542
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.7521018246801945,
      "learning_rate": 4.9998617608446466e-05,
      "loss": 0.4215,
      "step": 543
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.759099395715376,
      "learning_rate": 4.9998564951101815e-05,
      "loss": 1.0201,
      "step": 544
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.325671077070099,
      "learning_rate": 4.999851130954576e-05,
      "loss": 0.4177,
      "step": 545
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.723768248368744,
      "learning_rate": 4.999845668378039e-05,
      "loss": 0.5279,
      "step": 546
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.7031382119903844,
      "learning_rate": 4.9998401073807864e-05,
      "loss": 0.4245,
      "step": 547
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4428044588807036,
      "learning_rate": 4.999834447963038e-05,
      "loss": 0.1407,
      "step": 548
    },
    {
      "epoch": 0.03,
      "grad_norm": 26.121757903734455,
      "learning_rate": 4.999828690125015e-05,
      "loss": 2.1518,
      "step": 549
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6066369745423104,
      "learning_rate": 4.999822833866945e-05,
      "loss": 0.2993,
      "step": 550
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.204599169881766,
      "learning_rate": 4.9998168791890585e-05,
      "loss": 0.8317,
      "step": 551
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.02363961704848,
      "learning_rate": 4.999810826091591e-05,
      "loss": 0.2807,
      "step": 552
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.9399660281710953,
      "learning_rate": 4.999804674574779e-05,
      "loss": 0.6157,
      "step": 553
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.986234265729994,
      "learning_rate": 4.999798424638865e-05,
      "loss": 0.7607,
      "step": 554
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.851490117923392,
      "learning_rate": 4.9997920762840955e-05,
      "loss": 0.3949,
      "step": 555
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4157300961880894,
      "learning_rate": 4.999785629510721e-05,
      "loss": 0.3006,
      "step": 556
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.304140726545551,
      "learning_rate": 4.999779084318994e-05,
      "loss": 0.2986,
      "step": 557
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.766588221689472,
      "learning_rate": 4.999772440709174e-05,
      "loss": 0.2975,
      "step": 558
    },
    {
      "epoch": 0.03,
      "grad_norm": 12.055286699942442,
      "learning_rate": 4.99976569868152e-05,
      "loss": 1.2618,
      "step": 559
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.762082710712015,
      "learning_rate": 4.9997588582363e-05,
      "loss": 0.3199,
      "step": 560
    },
    {
      "epoch": 0.03,
      "grad_norm": 15.961435081236859,
      "learning_rate": 4.9997519193737816e-05,
      "loss": 1.3886,
      "step": 561
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.6249876686875373,
      "learning_rate": 4.9997448820942396e-05,
      "loss": 1.0337,
      "step": 562
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.113424142807744,
      "learning_rate": 4.9997377463979496e-05,
      "loss": 1.0266,
      "step": 563
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.9329379133175855,
      "learning_rate": 4.999730512285193e-05,
      "loss": 0.871,
      "step": 564
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5030283836040987,
      "learning_rate": 4.9997231797562545e-05,
      "loss": 0.3374,
      "step": 565
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.3945042423063727,
      "learning_rate": 4.999715748811423e-05,
      "loss": 0.4664,
      "step": 566
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.350957193115633,
      "learning_rate": 4.999708219450991e-05,
      "loss": 1.0724,
      "step": 567
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.614250437709336,
      "learning_rate": 4.999700591675255e-05,
      "loss": 0.3308,
      "step": 568
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.328337380114586,
      "learning_rate": 4.999692865484515e-05,
      "loss": 0.4016,
      "step": 569
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.649851619477541,
      "learning_rate": 4.999685040879075e-05,
      "loss": 0.7966,
      "step": 570
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.099624655628761,
      "learning_rate": 4.999677117859244e-05,
      "loss": 0.7097,
      "step": 571
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.0265914435362795,
      "learning_rate": 4.999669096425334e-05,
      "loss": 0.5377,
      "step": 572
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3586962094904,
      "learning_rate": 4.9996609765776595e-05,
      "loss": 0.3521,
      "step": 573
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.129280226253194,
      "learning_rate": 4.999652758316541e-05,
      "loss": 0.7983,
      "step": 574
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.777685694840515,
      "learning_rate": 4.999644441642302e-05,
      "loss": 0.8379,
      "step": 575
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.9639928249571073,
      "learning_rate": 4.99963602655527e-05,
      "loss": 0.5115,
      "step": 576
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.450180462480459,
      "learning_rate": 4.999627513055776e-05,
      "loss": 0.5037,
      "step": 577
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.468061379875611,
      "learning_rate": 4.9996189011441556e-05,
      "loss": 0.3663,
      "step": 578
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.7221469613598948,
      "learning_rate": 4.999610190820748e-05,
      "loss": 0.4486,
      "step": 579
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3973550248165214,
      "learning_rate": 4.9996013820858954e-05,
      "loss": 0.3362,
      "step": 580
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3896936514246856,
      "learning_rate": 4.999592474939945e-05,
      "loss": 0.3506,
      "step": 581
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.640878830120925,
      "learning_rate": 4.999583469383247e-05,
      "loss": 1.0923,
      "step": 582
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.203090460674757,
      "learning_rate": 4.999574365416157e-05,
      "loss": 0.8185,
      "step": 583
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.535195552774887,
      "learning_rate": 4.9995651630390324e-05,
      "loss": 0.4361,
      "step": 584
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4605195023082027,
      "learning_rate": 4.999555862252236e-05,
      "loss": 0.3277,
      "step": 585
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9679931742676329,
      "learning_rate": 4.999546463056135e-05,
      "loss": 0.2981,
      "step": 586
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.030208840624476,
      "learning_rate": 4.999536965451097e-05,
      "loss": 0.3749,
      "step": 587
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.243888848255742,
      "learning_rate": 4.999527369437498e-05,
      "loss": 0.875,
      "step": 588
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.388739996437854,
      "learning_rate": 4.999517675015715e-05,
      "loss": 0.8157,
      "step": 589
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.030289085183765,
      "learning_rate": 4.99950788218613e-05,
      "loss": 0.711,
      "step": 590
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.162206164118788,
      "learning_rate": 4.999497990949128e-05,
      "loss": 0.2614,
      "step": 591
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.878123732485429,
      "learning_rate": 4.999488001305099e-05,
      "loss": 0.453,
      "step": 592
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.066237819939844,
      "learning_rate": 4.999477913254436e-05,
      "loss": 0.2433,
      "step": 593
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.0221917912346754,
      "learning_rate": 4.999467726797537e-05,
      "loss": 0.6452,
      "step": 594
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.5411458232891033,
      "learning_rate": 4.999457441934801e-05,
      "loss": 0.5281,
      "step": 595
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.391303370712637,
      "learning_rate": 4.999447058666635e-05,
      "loss": 0.6357,
      "step": 596
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.46527685875336,
      "learning_rate": 4.9994365769934465e-05,
      "loss": 0.2862,
      "step": 597
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.178617224168217,
      "learning_rate": 4.999425996915649e-05,
      "loss": 0.3263,
      "step": 598
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.457213478709761,
      "learning_rate": 4.999415318433658e-05,
      "loss": 0.4656,
      "step": 599
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.525051059679397,
      "learning_rate": 4.999404541547896e-05,
      "loss": 0.6035,
      "step": 600
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.892581732890769,
      "learning_rate": 4.999393666258785e-05,
      "loss": 0.5744,
      "step": 601
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2461562733926104,
      "learning_rate": 4.999382692566754e-05,
      "loss": 0.6841,
      "step": 602
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.5841854722027797,
      "learning_rate": 4.999371620472236e-05,
      "loss": 0.4641,
      "step": 603
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.251183671600175,
      "learning_rate": 4.9993604499756655e-05,
      "loss": 0.8969,
      "step": 604
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.106160268094245,
      "learning_rate": 4.999349181077483e-05,
      "loss": 0.4465,
      "step": 605
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.8676727106880118,
      "learning_rate": 4.999337813778132e-05,
      "loss": 0.6298,
      "step": 606
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4461934918994406,
      "learning_rate": 4.99932634807806e-05,
      "loss": 0.2117,
      "step": 607
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6183157695803905,
      "learning_rate": 4.9993147839777184e-05,
      "loss": 0.1813,
      "step": 608
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.154367730270362,
      "learning_rate": 4.9993031214775624e-05,
      "loss": 0.4203,
      "step": 609
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1797287530894662,
      "learning_rate": 4.999291360578052e-05,
      "loss": 0.2642,
      "step": 610
    },
    {
      "epoch": 0.04,
      "grad_norm": 18.994740258143942,
      "learning_rate": 4.999279501279649e-05,
      "loss": 2.0137,
      "step": 611
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4233762558322534,
      "learning_rate": 4.9992675435828204e-05,
      "loss": 0.1302,
      "step": 612
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0830997965385074,
      "learning_rate": 4.999255487488038e-05,
      "loss": 0.4039,
      "step": 613
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1980316979241525,
      "learning_rate": 4.9992433329957764e-05,
      "loss": 0.311,
      "step": 614
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.6678301989643183,
      "learning_rate": 4.999231080106513e-05,
      "loss": 0.3436,
      "step": 615
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.864335309674611,
      "learning_rate": 4.999218728820731e-05,
      "loss": 0.5527,
      "step": 616
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.8294191389182335,
      "learning_rate": 4.9992062791389166e-05,
      "loss": 0.479,
      "step": 617
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4334456017177124,
      "learning_rate": 4.99919373106156e-05,
      "loss": 0.297,
      "step": 618
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.395102446335727,
      "learning_rate": 4.9991810845891545e-05,
      "loss": 0.3061,
      "step": 619
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.017860164433303,
      "learning_rate": 4.999168339722199e-05,
      "loss": 1.7833,
      "step": 620
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.7305058760424186,
      "learning_rate": 4.9991554964611954e-05,
      "loss": 0.5629,
      "step": 621
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1686280824248363,
      "learning_rate": 4.999142554806648e-05,
      "loss": 0.2772,
      "step": 622
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.2990241742658095,
      "learning_rate": 4.999129514759067e-05,
      "loss": 0.6637,
      "step": 623
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.7925474639922414,
      "learning_rate": 4.9991163763189664e-05,
      "loss": 0.3873,
      "step": 624
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8877776007861025,
      "learning_rate": 4.9991031394868635e-05,
      "loss": 0.242,
      "step": 625
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.9064105945245973,
      "learning_rate": 4.9990898042632785e-05,
      "loss": 0.4655,
      "step": 626
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.9031355846939713,
      "learning_rate": 4.999076370648737e-05,
      "loss": 0.2616,
      "step": 627
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4450940094519034,
      "learning_rate": 4.999062838643767e-05,
      "loss": 0.2858,
      "step": 628
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2039104109500425,
      "learning_rate": 4.999049208248903e-05,
      "loss": 0.357,
      "step": 629
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.031736473986355,
      "learning_rate": 4.99903547946468e-05,
      "loss": 0.7452,
      "step": 630
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9168234321721898,
      "learning_rate": 4.999021652291639e-05,
      "loss": 0.313,
      "step": 631
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.528616720813143,
      "learning_rate": 4.9990077267303254e-05,
      "loss": 0.186,
      "step": 632
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.095190461473124,
      "learning_rate": 4.9989937027812864e-05,
      "loss": 0.2585,
      "step": 633
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.293526091075791,
      "learning_rate": 4.9989795804450744e-05,
      "loss": 0.4167,
      "step": 634
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2432286365664487,
      "learning_rate": 4.998965359722245e-05,
      "loss": 0.3263,
      "step": 635
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4505949882683433,
      "learning_rate": 4.998951040613359e-05,
      "loss": 0.4294,
      "step": 636
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.976681262125343,
      "learning_rate": 4.998936623118979e-05,
      "loss": 0.3767,
      "step": 637
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.10022297750849,
      "learning_rate": 4.998922107239673e-05,
      "loss": 0.6962,
      "step": 638
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.38674114145697,
      "learning_rate": 4.9989074929760136e-05,
      "loss": 0.3376,
      "step": 639
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.579782190220901,
      "learning_rate": 4.998892780328575e-05,
      "loss": 0.2572,
      "step": 640
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5599354122585254,
      "learning_rate": 4.998877969297936e-05,
      "loss": 0.2061,
      "step": 641
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1572421195598577,
      "learning_rate": 4.998863059884681e-05,
      "loss": 0.2725,
      "step": 642
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.042474255863456,
      "learning_rate": 4.9988480520893966e-05,
      "loss": 0.2679,
      "step": 643
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6176109895483244,
      "learning_rate": 4.998832945912674e-05,
      "loss": 0.2466,
      "step": 644
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2003046072384005,
      "learning_rate": 4.9988177413551064e-05,
      "loss": 0.5586,
      "step": 645
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0513174661996665,
      "learning_rate": 4.998802438417294e-05,
      "loss": 0.2313,
      "step": 646
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8658525249760165,
      "learning_rate": 4.998787037099839e-05,
      "loss": 0.1697,
      "step": 647
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.674278750185836,
      "learning_rate": 4.998771537403347e-05,
      "loss": 1.1538,
      "step": 648
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.853283594049376,
      "learning_rate": 4.9987559393284286e-05,
      "loss": 0.2834,
      "step": 649
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4301985115161524,
      "learning_rate": 4.998740242875699e-05,
      "loss": 0.3258,
      "step": 650
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3850695195399023,
      "learning_rate": 4.998724448045775e-05,
      "loss": 0.3387,
      "step": 651
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.050597344466583,
      "learning_rate": 4.9987085548392786e-05,
      "loss": 0.3769,
      "step": 652
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.328194900198699,
      "learning_rate": 4.998692563256835e-05,
      "loss": 0.4373,
      "step": 653
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.165260468592831,
      "learning_rate": 4.998676473299075e-05,
      "loss": 0.4133,
      "step": 654
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3101278087959278,
      "learning_rate": 4.998660284966632e-05,
      "loss": 0.1473,
      "step": 655
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.61759712051244,
      "learning_rate": 4.998643998260142e-05,
      "loss": 0.6189,
      "step": 656
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2869930878209943,
      "learning_rate": 4.998627613180248e-05,
      "loss": 0.5256,
      "step": 657
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6847746506745174,
      "learning_rate": 4.9986111297275926e-05,
      "loss": 0.2575,
      "step": 658
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4256904609625978,
      "learning_rate": 4.998594547902828e-05,
      "loss": 0.1653,
      "step": 659
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.259561888012156,
      "learning_rate": 4.998577867706604e-05,
      "loss": 0.5036,
      "step": 660
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.957731568987509,
      "learning_rate": 4.9985610891395795e-05,
      "loss": 0.2675,
      "step": 661
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8264891192872166,
      "learning_rate": 4.998544212202415e-05,
      "loss": 0.2383,
      "step": 662
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5917435234142565,
      "learning_rate": 4.998527236895773e-05,
      "loss": 0.1705,
      "step": 663
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.280692456648373,
      "learning_rate": 4.998510163220323e-05,
      "loss": 0.8773,
      "step": 664
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1396104571481884,
      "learning_rate": 4.9984929911767375e-05,
      "loss": 0.4368,
      "step": 665
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7437880037938025,
      "learning_rate": 4.998475720765693e-05,
      "loss": 0.3245,
      "step": 666
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.826055639124767,
      "learning_rate": 4.998458351987868e-05,
      "loss": 0.2206,
      "step": 667
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4260715942152697,
      "learning_rate": 4.998440884843947e-05,
      "loss": 0.2026,
      "step": 668
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.682757197698045,
      "learning_rate": 4.9984233193346185e-05,
      "loss": 0.2315,
      "step": 669
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.6591459821422365,
      "learning_rate": 4.9984056554605726e-05,
      "loss": 0.688,
      "step": 670
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8306636784415944,
      "learning_rate": 4.998387893222506e-05,
      "loss": 0.2168,
      "step": 671
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9859555176213004,
      "learning_rate": 4.998370032621118e-05,
      "loss": 0.2533,
      "step": 672
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7895908971411487,
      "learning_rate": 4.9983520736571104e-05,
      "loss": 0.3225,
      "step": 673
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4805211351540537,
      "learning_rate": 4.998334016331192e-05,
      "loss": 0.252,
      "step": 674
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.006608939740909,
      "learning_rate": 4.998315860644072e-05,
      "loss": 0.2886,
      "step": 675
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3779252571517207,
      "learning_rate": 4.998297606596466e-05,
      "loss": 0.253,
      "step": 676
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3252114584632366,
      "learning_rate": 4.998279254189093e-05,
      "loss": 0.3565,
      "step": 677
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0173061018866916,
      "learning_rate": 4.9982608034226765e-05,
      "loss": 0.4092,
      "step": 678
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6239977142359574,
      "learning_rate": 4.99824225429794e-05,
      "loss": 0.2257,
      "step": 679
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3559534649531657,
      "learning_rate": 4.998223606815617e-05,
      "loss": 0.1929,
      "step": 680
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5828964532889855,
      "learning_rate": 4.998204860976439e-05,
      "loss": 0.2095,
      "step": 681
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0065421001110857,
      "learning_rate": 4.998186016781146e-05,
      "loss": 0.25,
      "step": 682
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.943381830468981,
      "learning_rate": 4.9981670742304785e-05,
      "loss": 0.4289,
      "step": 683
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8849652863865265,
      "learning_rate": 4.998148033325183e-05,
      "loss": 0.3744,
      "step": 684
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.036966800881291,
      "learning_rate": 4.9981288940660096e-05,
      "loss": 0.3615,
      "step": 685
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.674873175648795,
      "learning_rate": 4.998109656453711e-05,
      "loss": 0.4254,
      "step": 686
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0216468580829154,
      "learning_rate": 4.998090320489045e-05,
      "loss": 0.2952,
      "step": 687
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7393335009364894,
      "learning_rate": 4.9980708861727725e-05,
      "loss": 0.2451,
      "step": 688
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1671220647283076,
      "learning_rate": 4.9980513535056594e-05,
      "loss": 0.2393,
      "step": 689
    },
    {
      "epoch": 0.04,
      "grad_norm": 14.540246224817242,
      "learning_rate": 4.998031722488474e-05,
      "loss": 0.6943,
      "step": 690
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.8795052886653854,
      "learning_rate": 4.9980119931219896e-05,
      "loss": 0.3147,
      "step": 691
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.435275404873924,
      "learning_rate": 4.997992165406982e-05,
      "loss": 0.5008,
      "step": 692
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.344021714556989,
      "learning_rate": 4.997972239344233e-05,
      "loss": 0.1901,
      "step": 693
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.8555336608365316,
      "learning_rate": 4.997952214934527e-05,
      "loss": 0.3211,
      "step": 694
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.0213720247973033,
      "learning_rate": 4.997932092178652e-05,
      "loss": 0.6329,
      "step": 695
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.64204880296433,
      "learning_rate": 4.9979118710773996e-05,
      "loss": 0.2985,
      "step": 696
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9234437117503036,
      "learning_rate": 4.997891551631567e-05,
      "loss": 0.3025,
      "step": 697
    },
    {
      "epoch": 0.04,
      "grad_norm": 70.23332032114614,
      "learning_rate": 4.9978711338419546e-05,
      "loss": 1.5889,
      "step": 698
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.5018298273116737,
      "learning_rate": 4.997850617709365e-05,
      "loss": 0.2433,
      "step": 699
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.188840196429507,
      "learning_rate": 4.9978300032346056e-05,
      "loss": 0.2267,
      "step": 700
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.279907998660516,
      "learning_rate": 4.99780929041849e-05,
      "loss": 1.2534,
      "step": 701
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.257337587597935,
      "learning_rate": 4.997788479261832e-05,
      "loss": 0.4075,
      "step": 702
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.3667265325789266,
      "learning_rate": 4.997767569765452e-05,
      "loss": 0.3835,
      "step": 703
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3491428942750727,
      "learning_rate": 4.997746561930172e-05,
      "loss": 0.3566,
      "step": 704
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.243144499744783,
      "learning_rate": 4.997725455756821e-05,
      "loss": 0.3392,
      "step": 705
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7453410967398955,
      "learning_rate": 4.9977042512462266e-05,
      "loss": 0.3293,
      "step": 706
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.6509164603655195,
      "learning_rate": 4.997682948399228e-05,
      "loss": 0.3524,
      "step": 707
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.5239448744052453,
      "learning_rate": 4.99766154721666e-05,
      "loss": 0.8829,
      "step": 708
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9312271154762652,
      "learning_rate": 4.997640047699368e-05,
      "loss": 0.3306,
      "step": 709
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4320354180245487,
      "learning_rate": 4.997618449848197e-05,
      "loss": 0.3477,
      "step": 710
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.262798290647729,
      "learning_rate": 4.997596753663998e-05,
      "loss": 0.259,
      "step": 711
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.306148838692192,
      "learning_rate": 4.997574959147624e-05,
      "loss": 0.8936,
      "step": 712
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1374955943043377,
      "learning_rate": 4.9975530662999344e-05,
      "loss": 0.2572,
      "step": 713
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2879663119199476,
      "learning_rate": 4.9975310751217905e-05,
      "loss": 0.4304,
      "step": 714
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.538221964489274,
      "learning_rate": 4.997508985614058e-05,
      "loss": 0.3603,
      "step": 715
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.289433178998054,
      "learning_rate": 4.997486797777607e-05,
      "loss": 0.6444,
      "step": 716
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.782900520732679,
      "learning_rate": 4.9974645116133114e-05,
      "loss": 0.2547,
      "step": 717
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.385606913272907,
      "learning_rate": 4.9974421271220476e-05,
      "loss": 0.4902,
      "step": 718
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.9986125910016646,
      "learning_rate": 4.9974196443046976e-05,
      "loss": 0.6028,
      "step": 719
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.970529889113797,
      "learning_rate": 4.997397063162147e-05,
      "loss": 0.6587,
      "step": 720
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8535613289675426,
      "learning_rate": 4.997374383695284e-05,
      "loss": 0.2277,
      "step": 721
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.910112342484708,
      "learning_rate": 4.997351605905001e-05,
      "loss": 0.2768,
      "step": 722
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.9583315146745695,
      "learning_rate": 4.997328729792195e-05,
      "loss": 0.3289,
      "step": 723
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9038550476309617,
      "learning_rate": 4.997305755357768e-05,
      "loss": 0.2882,
      "step": 724
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8331419072950763,
      "learning_rate": 4.9972826826026234e-05,
      "loss": 0.2172,
      "step": 725
    },
    {
      "epoch": 0.04,
      "grad_norm": 23.180099868762596,
      "learning_rate": 4.99725951152767e-05,
      "loss": 1.8865,
      "step": 726
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8675192970185093,
      "learning_rate": 4.99723624213382e-05,
      "loss": 0.3149,
      "step": 727
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.013871622741711,
      "learning_rate": 4.997212874421989e-05,
      "loss": 0.2054,
      "step": 728
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4231811872384115,
      "learning_rate": 4.997189408393098e-05,
      "loss": 0.4866,
      "step": 729
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3311060504920644,
      "learning_rate": 4.99716584404807e-05,
      "loss": 0.4168,
      "step": 730
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4169294075139023,
      "learning_rate": 4.9971421813878334e-05,
      "loss": 0.5385,
      "step": 731
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.879247862247275,
      "learning_rate": 4.9971184204133195e-05,
      "loss": 0.3211,
      "step": 732
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.36632186906302,
      "learning_rate": 4.9970945611254634e-05,
      "loss": 0.6104,
      "step": 733
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.0626333813053623,
      "learning_rate": 4.997070603525204e-05,
      "loss": 0.5326,
      "step": 734
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.704893681422923,
      "learning_rate": 4.997046547613487e-05,
      "loss": 0.313,
      "step": 735
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.751134222083944,
      "learning_rate": 4.9970223933912576e-05,
      "loss": 0.5395,
      "step": 736
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.092913447222655,
      "learning_rate": 4.996998140859466e-05,
      "loss": 0.2553,
      "step": 737
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.075877148626866,
      "learning_rate": 4.996973790019068e-05,
      "loss": 0.3114,
      "step": 738
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.850684628884376,
      "learning_rate": 4.9969493408710233e-05,
      "loss": 0.5115,
      "step": 739
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.5677477955965875,
      "learning_rate": 4.9969247934162935e-05,
      "loss": 0.4516,
      "step": 740
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8617199748277802,
      "learning_rate": 4.9969001476558454e-05,
      "loss": 0.3368,
      "step": 741
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.271584735729905,
      "learning_rate": 4.996875403590648e-05,
      "loss": 0.4484,
      "step": 742
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.466076938170474,
      "learning_rate": 4.996850561221677e-05,
      "loss": 0.7231,
      "step": 743
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.5637391710503286,
      "learning_rate": 4.99682562054991e-05,
      "loss": 0.4123,
      "step": 744
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.155576713319605,
      "learning_rate": 4.996800581576328e-05,
      "loss": 0.4667,
      "step": 745
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7276580605896577,
      "learning_rate": 4.996775444301919e-05,
      "loss": 0.2454,
      "step": 746
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.723223593916733,
      "learning_rate": 4.996750208727671e-05,
      "loss": 0.2683,
      "step": 747
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.24715492183301,
      "learning_rate": 4.996724874854578e-05,
      "loss": 0.8408,
      "step": 748
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3468877986355383,
      "learning_rate": 4.996699442683637e-05,
      "loss": 0.2777,
      "step": 749
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.505832011943894,
      "learning_rate": 4.99667391221585e-05,
      "loss": 0.4338,
      "step": 750
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.671433376236951,
      "learning_rate": 4.996648283452222e-05,
      "loss": 0.3977,
      "step": 751
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.214634807133492,
      "learning_rate": 4.996622556393761e-05,
      "loss": 0.4409,
      "step": 752
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8367155129940906,
      "learning_rate": 4.9965967310414806e-05,
      "loss": 0.3132,
      "step": 753
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4379954556162517,
      "learning_rate": 4.9965708073963984e-05,
      "loss": 0.2707,
      "step": 754
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.622108643095567,
      "learning_rate": 4.996544785459534e-05,
      "loss": 0.9356,
      "step": 755
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2541501234812036,
      "learning_rate": 4.996518665231912e-05,
      "loss": 0.2585,
      "step": 756
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.1949349842634587,
      "learning_rate": 4.9964924467145615e-05,
      "loss": 0.244,
      "step": 757
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7148635943877173,
      "learning_rate": 4.996466129908514e-05,
      "loss": 0.2698,
      "step": 758
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.707178398410849,
      "learning_rate": 4.996439714814806e-05,
      "loss": 0.5895,
      "step": 759
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6676306996132848,
      "learning_rate": 4.9964132014344766e-05,
      "loss": 0.2397,
      "step": 760
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.231131181397122,
      "learning_rate": 4.996386589768571e-05,
      "loss": 0.9845,
      "step": 761
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8104785282232585,
      "learning_rate": 4.996359879818136e-05,
      "loss": 0.4047,
      "step": 762
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.078786726022739,
      "learning_rate": 4.9963330715842226e-05,
      "loss": 0.5678,
      "step": 763
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0834724102065,
      "learning_rate": 4.996306165067888e-05,
      "loss": 0.2631,
      "step": 764
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3275696857478865,
      "learning_rate": 4.996279160270191e-05,
      "loss": 0.3698,
      "step": 765
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9653317118791198,
      "learning_rate": 4.996252057192194e-05,
      "loss": 0.2424,
      "step": 766
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6887355969964994,
      "learning_rate": 4.9962248558349643e-05,
      "loss": 0.2339,
      "step": 767
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.9796283687259684,
      "learning_rate": 4.9961975561995734e-05,
      "loss": 0.4998,
      "step": 768
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2138505157114854,
      "learning_rate": 4.996170158287096e-05,
      "loss": 0.2611,
      "step": 769
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0037454171940854,
      "learning_rate": 4.9961426620986095e-05,
      "loss": 0.1961,
      "step": 770
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8577250989718876,
      "learning_rate": 4.996115067635199e-05,
      "loss": 0.2687,
      "step": 771
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.839806976448472,
      "learning_rate": 4.996087374897948e-05,
      "loss": 0.2374,
      "step": 772
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0925471836539082,
      "learning_rate": 4.9960595838879486e-05,
      "loss": 0.312,
      "step": 773
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7043231346768102,
      "learning_rate": 4.9960316946062944e-05,
      "loss": 0.2467,
      "step": 774
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.1615660565377355,
      "learning_rate": 4.996003707054085e-05,
      "loss": 0.2426,
      "step": 775
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0986074541331514,
      "learning_rate": 4.9959756212324186e-05,
      "loss": 0.3103,
      "step": 776
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5184780952478782,
      "learning_rate": 4.995947437142404e-05,
      "loss": 0.2876,
      "step": 777
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3871516405157163,
      "learning_rate": 4.99591915478515e-05,
      "loss": 0.4249,
      "step": 778
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.069294439935317,
      "learning_rate": 4.9958907741617704e-05,
      "loss": 0.9971,
      "step": 779
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.870790579865526,
      "learning_rate": 4.995862295273382e-05,
      "loss": 0.5869,
      "step": 780
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.018898749014038,
      "learning_rate": 4.9958337181211065e-05,
      "loss": 0.3238,
      "step": 781
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.732802502347244,
      "learning_rate": 4.9958050427060685e-05,
      "loss": 0.2585,
      "step": 782
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.054132762345351,
      "learning_rate": 4.995776269029399e-05,
      "loss": 0.3136,
      "step": 783
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5727122663017816,
      "learning_rate": 4.9957473970922266e-05,
      "loss": 0.4132,
      "step": 784
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0835951694534858,
      "learning_rate": 4.995718426895692e-05,
      "loss": 0.2497,
      "step": 785
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.436960088415468,
      "learning_rate": 4.9956893584409335e-05,
      "loss": 0.5799,
      "step": 786
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6428767996492553,
      "learning_rate": 4.995660191729097e-05,
      "loss": 0.2957,
      "step": 787
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.486440755405283,
      "learning_rate": 4.9956309267613296e-05,
      "loss": 0.1582,
      "step": 788
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.370916598205426,
      "learning_rate": 4.9956015635387856e-05,
      "loss": 0.3458,
      "step": 789
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.856452302404927,
      "learning_rate": 4.995572102062618e-05,
      "loss": 0.5566,
      "step": 790
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5541861443255822,
      "learning_rate": 4.9955425423339885e-05,
      "loss": 0.227,
      "step": 791
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3793267278756498,
      "learning_rate": 4.995512884354061e-05,
      "loss": 0.1426,
      "step": 792
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7692819620273632,
      "learning_rate": 4.995483128124002e-05,
      "loss": 0.3537,
      "step": 793
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.706733656915361,
      "learning_rate": 4.995453273644985e-05,
      "loss": 0.316,
      "step": 794
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0754257447180806,
      "learning_rate": 4.995423320918183e-05,
      "loss": 0.4716,
      "step": 795
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.603614739836298,
      "learning_rate": 4.995393269944777e-05,
      "loss": 0.2694,
      "step": 796
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7043672806550585,
      "learning_rate": 4.9953631207259496e-05,
      "loss": 0.2336,
      "step": 797
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.9047728634466736,
      "learning_rate": 4.9953328732628874e-05,
      "loss": 0.398,
      "step": 798
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.8754608587153037,
      "learning_rate": 4.995302527556782e-05,
      "loss": 0.6085,
      "step": 799
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7220120370124938,
      "learning_rate": 4.9952720836088274e-05,
      "loss": 0.2381,
      "step": 800
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.354479569531251,
      "learning_rate": 4.995241541420223e-05,
      "loss": 0.3309,
      "step": 801
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.186273264831017,
      "learning_rate": 4.995210900992171e-05,
      "loss": 0.1341,
      "step": 802
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4018091117165736,
      "learning_rate": 4.995180162325877e-05,
      "loss": 0.2455,
      "step": 803
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8298440818019817,
      "learning_rate": 4.9951493254225526e-05,
      "loss": 0.2092,
      "step": 804
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.00413492858743,
      "learning_rate": 4.99511839028341e-05,
      "loss": 0.9457,
      "step": 805
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.686587866521381,
      "learning_rate": 4.9950873569096686e-05,
      "loss": 0.3271,
      "step": 806
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.7122552911340154,
      "learning_rate": 4.9950562253025504e-05,
      "loss": 0.658,
      "step": 807
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5489077532437274,
      "learning_rate": 4.995024995463279e-05,
      "loss": 0.1399,
      "step": 808
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.547836683100218,
      "learning_rate": 4.994993667393086e-05,
      "loss": 0.5346,
      "step": 809
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8731397752161203,
      "learning_rate": 4.994962241093205e-05,
      "loss": 0.2512,
      "step": 810
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9017657568549298,
      "learning_rate": 4.994930716564872e-05,
      "loss": 0.4743,
      "step": 811
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7985886508359292,
      "learning_rate": 4.994899093809329e-05,
      "loss": 0.2473,
      "step": 812
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3563417062374263,
      "learning_rate": 4.994867372827821e-05,
      "loss": 0.1477,
      "step": 813
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8302070740784546,
      "learning_rate": 4.994835553621595e-05,
      "loss": 0.3462,
      "step": 814
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5359097957539156,
      "learning_rate": 4.994803636191906e-05,
      "loss": 0.2635,
      "step": 815
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2991832485202162,
      "learning_rate": 4.99477162054001e-05,
      "loss": 0.1955,
      "step": 816
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.54013451352761,
      "learning_rate": 4.994739506667167e-05,
      "loss": 2.0326,
      "step": 817
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5170122468752552,
      "learning_rate": 4.994707294574643e-05,
      "loss": 0.19,
      "step": 818
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6369558706957905,
      "learning_rate": 4.994674984263704e-05,
      "loss": 0.2277,
      "step": 819
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2192498630526738,
      "learning_rate": 4.994642575735623e-05,
      "loss": 0.1173,
      "step": 820
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.112286573403404,
      "learning_rate": 4.994610068991675e-05,
      "loss": 0.2447,
      "step": 821
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9537407008880767,
      "learning_rate": 4.9945774640331424e-05,
      "loss": 0.2333,
      "step": 822
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4374267886442387,
      "learning_rate": 4.994544760861306e-05,
      "loss": 0.2052,
      "step": 823
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7616287913680964,
      "learning_rate": 4.994511959477456e-05,
      "loss": 0.203,
      "step": 824
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8202439532207877,
      "learning_rate": 4.994479059882881e-05,
      "loss": 0.2517,
      "step": 825
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.161271411913809,
      "learning_rate": 4.994446062078878e-05,
      "loss": 0.4333,
      "step": 826
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9688254190428216,
      "learning_rate": 4.994412966066746e-05,
      "loss": 0.4777,
      "step": 827
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.1787386219039,
      "learning_rate": 4.994379771847788e-05,
      "loss": 0.1775,
      "step": 828
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.75766289383427,
      "learning_rate": 4.994346479423311e-05,
      "loss": 0.2412,
      "step": 829
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.425292462278961,
      "learning_rate": 4.9943130887946246e-05,
      "loss": 0.2916,
      "step": 830
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9649337985289053,
      "learning_rate": 4.994279599963045e-05,
      "loss": 0.3535,
      "step": 831
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.5367061259609907,
      "learning_rate": 4.99424601292989e-05,
      "loss": 0.6359,
      "step": 832
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8020108127856633,
      "learning_rate": 4.994212327696481e-05,
      "loss": 0.2529,
      "step": 833
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8912490644117121,
      "learning_rate": 4.9941785442641464e-05,
      "loss": 0.3058,
      "step": 834
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.6461910167311435,
      "learning_rate": 4.994144662634214e-05,
      "loss": 0.5033,
      "step": 835
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9212462093300129,
      "learning_rate": 4.994110682808019e-05,
      "loss": 0.3859,
      "step": 836
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6754805002235102,
      "learning_rate": 4.994076604786899e-05,
      "loss": 0.2501,
      "step": 837
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.022491693749079,
      "learning_rate": 4.994042428572196e-05,
      "loss": 0.41,
      "step": 838
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.6565521073848357,
      "learning_rate": 4.994008154165255e-05,
      "loss": 0.5385,
      "step": 839
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.92669271164786,
      "learning_rate": 4.993973781567425e-05,
      "loss": 0.283,
      "step": 840
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7060942602976772,
      "learning_rate": 4.99393931078006e-05,
      "loss": 0.1862,
      "step": 841
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2894341068184954,
      "learning_rate": 4.9939047418045175e-05,
      "loss": 0.2422,
      "step": 842
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.863302552505379,
      "learning_rate": 4.993870074642158e-05,
      "loss": 0.4528,
      "step": 843
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.403668538616977,
      "learning_rate": 4.9938353092943456e-05,
      "loss": 0.1755,
      "step": 844
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2005128299561405,
      "learning_rate": 4.993800445762451e-05,
      "loss": 0.4358,
      "step": 845
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7995017941673557,
      "learning_rate": 4.9937654840478446e-05,
      "loss": 0.2174,
      "step": 846
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6366701561770576,
      "learning_rate": 4.993730424151904e-05,
      "loss": 0.4305,
      "step": 847
    },
    {
      "epoch": 0.05,
      "grad_norm": 31.910993769239365,
      "learning_rate": 4.99369526607601e-05,
      "loss": 2.9916,
      "step": 848
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1721582762006149,
      "learning_rate": 4.993660009821546e-05,
      "loss": 0.23,
      "step": 849
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.178614716037442,
      "learning_rate": 4.9936246553899e-05,
      "loss": 0.2815,
      "step": 850
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.835923569844924,
      "learning_rate": 4.9935892027824635e-05,
      "loss": 0.5592,
      "step": 851
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.355557222762847,
      "learning_rate": 4.993553652000634e-05,
      "loss": 0.3322,
      "step": 852
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.334512397714535,
      "learning_rate": 4.9935180030458107e-05,
      "loss": 0.2763,
      "step": 853
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.2352111509374932,
      "learning_rate": 4.9934822559193956e-05,
      "loss": 0.5186,
      "step": 854
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.6061389331259015,
      "learning_rate": 4.993446410622798e-05,
      "loss": 0.78,
      "step": 855
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3895358073181803,
      "learning_rate": 4.9934104671574265e-05,
      "loss": 0.3069,
      "step": 856
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.77794101139786,
      "learning_rate": 4.9933744255246995e-05,
      "loss": 0.3494,
      "step": 857
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.525077163077914,
      "learning_rate": 4.993338285726034e-05,
      "loss": 0.4702,
      "step": 858
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0266469230454933,
      "learning_rate": 4.993302047762853e-05,
      "loss": 0.4182,
      "step": 859
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.208268478882623,
      "learning_rate": 4.993265711636583e-05,
      "loss": 0.4255,
      "step": 860
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.7821577019124404,
      "learning_rate": 4.993229277348656e-05,
      "loss": 0.5972,
      "step": 861
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.295015959845844,
      "learning_rate": 4.993192744900505e-05,
      "loss": 0.7039,
      "step": 862
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.020628744746083,
      "learning_rate": 4.993156114293569e-05,
      "loss": 0.4395,
      "step": 863
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.04038855910574,
      "learning_rate": 4.99311938552929e-05,
      "loss": 0.2442,
      "step": 864
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0791826454761675,
      "learning_rate": 4.993082558609115e-05,
      "loss": 0.393,
      "step": 865
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1576402626286475,
      "learning_rate": 4.9930456335344917e-05,
      "loss": 0.2594,
      "step": 866
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0289049547543447,
      "learning_rate": 4.993008610306875e-05,
      "loss": 0.2638,
      "step": 867
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5526271521546715,
      "learning_rate": 4.9929714889277234e-05,
      "loss": 0.1819,
      "step": 868
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0961431141114413,
      "learning_rate": 4.992934269398497e-05,
      "loss": 0.3249,
      "step": 869
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3271663712207507,
      "learning_rate": 4.992896951720662e-05,
      "loss": 0.4637,
      "step": 870
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.064714392079645,
      "learning_rate": 4.9928595358956884e-05,
      "loss": 0.5027,
      "step": 871
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4959242795716707,
      "learning_rate": 4.992822021925047e-05,
      "loss": 0.2045,
      "step": 872
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6351037909252775,
      "learning_rate": 4.9927844098102164e-05,
      "loss": 0.2896,
      "step": 873
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8335788387038698,
      "learning_rate": 4.9927466995526774e-05,
      "loss": 0.2099,
      "step": 874
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2421988842103127,
      "learning_rate": 4.992708891153915e-05,
      "loss": 0.4418,
      "step": 875
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.418789331225249,
      "learning_rate": 4.992670984615415e-05,
      "loss": 0.949,
      "step": 876
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.329270101366789,
      "learning_rate": 4.992632979938674e-05,
      "loss": 0.1784,
      "step": 877
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8324758316097625,
      "learning_rate": 4.9925948771251844e-05,
      "loss": 0.3098,
      "step": 878
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9820004607399722,
      "learning_rate": 4.9925566761764494e-05,
      "loss": 0.3969,
      "step": 879
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8700051544324663,
      "learning_rate": 4.992518377093971e-05,
      "loss": 0.3048,
      "step": 880
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.420345603086133,
      "learning_rate": 4.992479979879258e-05,
      "loss": 0.2362,
      "step": 881
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5049419004347055,
      "learning_rate": 4.992441484533822e-05,
      "loss": 0.2092,
      "step": 882
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1839268795253877,
      "learning_rate": 4.992402891059179e-05,
      "loss": 0.2896,
      "step": 883
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5907731831786485,
      "learning_rate": 4.992364199456847e-05,
      "loss": 0.2052,
      "step": 884
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2278071154299186,
      "learning_rate": 4.992325409728351e-05,
      "loss": 0.1766,
      "step": 885
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4848436514847505,
      "learning_rate": 4.992286521875217e-05,
      "loss": 0.1627,
      "step": 886
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4126676347657068,
      "learning_rate": 4.992247535898976e-05,
      "loss": 0.187,
      "step": 887
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3631428008805453,
      "learning_rate": 4.992208451801164e-05,
      "loss": 0.2308,
      "step": 888
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7519487384887062,
      "learning_rate": 4.992169269583319e-05,
      "loss": 0.5285,
      "step": 889
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.39958216476298,
      "learning_rate": 4.9921299892469844e-05,
      "loss": 1.2241,
      "step": 890
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1511162859738757,
      "learning_rate": 4.992090610793705e-05,
      "loss": 0.1809,
      "step": 891
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.740798095574385,
      "learning_rate": 4.992051134225033e-05,
      "loss": 0.7929,
      "step": 892
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.619739665847113,
      "learning_rate": 4.9920115595425216e-05,
      "loss": 0.2629,
      "step": 893
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.665396095238753,
      "learning_rate": 4.991971886747729e-05,
      "loss": 1.1545,
      "step": 894
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.815664263916652,
      "learning_rate": 4.991932115842217e-05,
      "loss": 0.2581,
      "step": 895
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.9291260845061533,
      "learning_rate": 4.991892246827553e-05,
      "loss": 0.3855,
      "step": 896
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.931928323713152,
      "learning_rate": 4.991852279705304e-05,
      "loss": 0.8856,
      "step": 897
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3913347662086815,
      "learning_rate": 4.991812214477045e-05,
      "loss": 0.4673,
      "step": 898
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7699975385224884,
      "learning_rate": 4.991772051144354e-05,
      "loss": 0.3257,
      "step": 899
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9746358124999954,
      "learning_rate": 4.991731789708811e-05,
      "loss": 0.3874,
      "step": 900
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1498229528899118,
      "learning_rate": 4.991691430172002e-05,
      "loss": 0.4154,
      "step": 901
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.5248026842567555,
      "learning_rate": 4.991650972535516e-05,
      "loss": 0.3396,
      "step": 902
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.619043627677799,
      "learning_rate": 4.991610416800945e-05,
      "loss": 0.2206,
      "step": 903
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1284409654549052,
      "learning_rate": 4.991569762969887e-05,
      "loss": 0.3381,
      "step": 904
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.5174953462743086,
      "learning_rate": 4.991529011043941e-05,
      "loss": 0.2708,
      "step": 905
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1668883904861946,
      "learning_rate": 4.991488161024712e-05,
      "loss": 0.4015,
      "step": 906
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.823987245832203,
      "learning_rate": 4.991447212913809e-05,
      "loss": 0.197,
      "step": 907
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.2935280185110183,
      "learning_rate": 4.991406166712843e-05,
      "loss": 0.4229,
      "step": 908
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8169098570693474,
      "learning_rate": 4.991365022423431e-05,
      "loss": 0.2449,
      "step": 909
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.114216259327152,
      "learning_rate": 4.9913237800471935e-05,
      "loss": 0.2557,
      "step": 910
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7220870810426923,
      "learning_rate": 4.991282439585753e-05,
      "loss": 0.3145,
      "step": 911
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.872896051724137,
      "learning_rate": 4.991241001040736e-05,
      "loss": 0.7643,
      "step": 912
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.821339418970765,
      "learning_rate": 4.9911994644137775e-05,
      "loss": 0.3287,
      "step": 913
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9373765399516731,
      "learning_rate": 4.9911578297065085e-05,
      "loss": 0.2783,
      "step": 914
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6697230026088843,
      "learning_rate": 4.9911160969205715e-05,
      "loss": 0.2591,
      "step": 915
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8532191290724462,
      "learning_rate": 4.991074266057609e-05,
      "loss": 0.2646,
      "step": 916
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.7238733089749783,
      "learning_rate": 4.991032337119267e-05,
      "loss": 0.5845,
      "step": 917
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.91134546843062,
      "learning_rate": 4.990990310107196e-05,
      "loss": 1.2223,
      "step": 918
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.022096316227199,
      "learning_rate": 4.990948185023052e-05,
      "loss": 2.8732,
      "step": 919
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6172797215263992,
      "learning_rate": 4.9909059618684914e-05,
      "loss": 0.2445,
      "step": 920
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.383322962650388,
      "learning_rate": 4.990863640645179e-05,
      "loss": 0.247,
      "step": 921
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8503600742312005,
      "learning_rate": 4.99082122135478e-05,
      "loss": 0.2529,
      "step": 922
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.7881055178046634,
      "learning_rate": 4.990778703998965e-05,
      "loss": 0.6697,
      "step": 923
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6468197866464904,
      "learning_rate": 4.990736088579406e-05,
      "loss": 0.2211,
      "step": 924
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.81722966436699,
      "learning_rate": 4.990693375097783e-05,
      "loss": 0.2987,
      "step": 925
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.00759021975293,
      "learning_rate": 4.990650563555776e-05,
      "loss": 0.8772,
      "step": 926
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1337987100676568,
      "learning_rate": 4.9906076539550725e-05,
      "loss": 0.4799,
      "step": 927
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1614029460850297,
      "learning_rate": 4.99056464629736e-05,
      "loss": 0.4296,
      "step": 928
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.616552286604228,
      "learning_rate": 4.990521540584333e-05,
      "loss": 0.4334,
      "step": 929
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1503429767068756,
      "learning_rate": 4.9904783368176874e-05,
      "loss": 0.3664,
      "step": 930
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6767221407338364,
      "learning_rate": 4.990435034999125e-05,
      "loss": 0.2075,
      "step": 931
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7893274760956397,
      "learning_rate": 4.9903916351303505e-05,
      "loss": 0.6308,
      "step": 932
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7987788567354555,
      "learning_rate": 4.990348137213072e-05,
      "loss": 0.2107,
      "step": 933
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.554994327058922,
      "learning_rate": 4.990304541249003e-05,
      "loss": 0.6594,
      "step": 934
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.759851053373344,
      "learning_rate": 4.9902608472398604e-05,
      "loss": 0.2585,
      "step": 935
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.0479819173820437,
      "learning_rate": 4.990217055187362e-05,
      "loss": 0.6605,
      "step": 936
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.331807581127279,
      "learning_rate": 4.990173165093235e-05,
      "loss": 0.5604,
      "step": 937
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.057480944321783,
      "learning_rate": 4.990129176959204e-05,
      "loss": 0.2561,
      "step": 938
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7410614106931632,
      "learning_rate": 4.9900850907870044e-05,
      "loss": 0.2008,
      "step": 939
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.923992561822031,
      "learning_rate": 4.9900409065783695e-05,
      "loss": 0.523,
      "step": 940
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.8598243276373534,
      "learning_rate": 4.989996624335039e-05,
      "loss": 0.6382,
      "step": 941
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.245890457887188,
      "learning_rate": 4.9899522440587575e-05,
      "loss": 0.3109,
      "step": 942
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8546931930301314,
      "learning_rate": 4.9899077657512715e-05,
      "loss": 0.2668,
      "step": 943
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3807312830104403,
      "learning_rate": 4.989863189414332e-05,
      "loss": 0.4866,
      "step": 944
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8475948017162618,
      "learning_rate": 4.989818515049695e-05,
      "loss": 0.275,
      "step": 945
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5257611470080956,
      "learning_rate": 4.9897737426591186e-05,
      "loss": 0.2268,
      "step": 946
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.682077677317074,
      "learning_rate": 4.989728872244366e-05,
      "loss": 0.6924,
      "step": 947
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.163728748794103,
      "learning_rate": 4.9896839038072017e-05,
      "loss": 0.2931,
      "step": 948
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3739367392054267,
      "learning_rate": 4.9896388373493994e-05,
      "loss": 0.5282,
      "step": 949
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3879821310155473,
      "learning_rate": 4.989593672872731e-05,
      "loss": 0.2516,
      "step": 950
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4592776974422699,
      "learning_rate": 4.989548410378976e-05,
      "loss": 0.2572,
      "step": 951
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.967958980686945,
      "learning_rate": 4.9895030498699157e-05,
      "loss": 0.3689,
      "step": 952
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1477909148558163,
      "learning_rate": 4.9894575913473365e-05,
      "loss": 0.156,
      "step": 953
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.300198786365102,
      "learning_rate": 4.9894120348130265e-05,
      "loss": 1.0161,
      "step": 954
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.0631392821396566,
      "learning_rate": 4.9893663802687816e-05,
      "loss": 0.6443,
      "step": 955
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1064848706956507,
      "learning_rate": 4.989320627716398e-05,
      "loss": 0.5346,
      "step": 956
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3732247199688334,
      "learning_rate": 4.989274777157678e-05,
      "loss": 0.2517,
      "step": 957
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.1885233539133755,
      "learning_rate": 4.989228828594425e-05,
      "loss": 0.5942,
      "step": 958
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.197424110344677,
      "learning_rate": 4.98918278202845e-05,
      "loss": 0.1088,
      "step": 959
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5495304700657002,
      "learning_rate": 4.989136637461564e-05,
      "loss": 0.3332,
      "step": 960
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.907861948499954,
      "learning_rate": 4.989090394895585e-05,
      "loss": 0.3774,
      "step": 961
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.558068964679458,
      "learning_rate": 4.9890440543323336e-05,
      "loss": 0.1933,
      "step": 962
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.293551667787647,
      "learning_rate": 4.988997615773633e-05,
      "loss": 0.2759,
      "step": 963
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.5339743255695315,
      "learning_rate": 4.9889510792213136e-05,
      "loss": 0.516,
      "step": 964
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1700310309184347,
      "learning_rate": 4.988904444677206e-05,
      "loss": 0.4964,
      "step": 965
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.603774290885298,
      "learning_rate": 4.988857712143147e-05,
      "loss": 0.3653,
      "step": 966
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6518284067135205,
      "learning_rate": 4.988810881620975e-05,
      "loss": 0.3678,
      "step": 967
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4261233373333173,
      "learning_rate": 4.988763953112536e-05,
      "loss": 0.2259,
      "step": 968
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2958101657500969,
      "learning_rate": 4.988716926619677e-05,
      "loss": 0.1741,
      "step": 969
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2812822791768335,
      "learning_rate": 4.988669802144248e-05,
      "loss": 0.1648,
      "step": 970
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.2806883266363664,
      "learning_rate": 4.988622579688105e-05,
      "loss": 0.5423,
      "step": 971
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4314516283111427,
      "learning_rate": 4.988575259253109e-05,
      "loss": 0.3011,
      "step": 972
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7171388766170659,
      "learning_rate": 4.98852784084112e-05,
      "loss": 0.3964,
      "step": 973
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3021586904005165,
      "learning_rate": 4.9884803244540074e-05,
      "loss": 0.2042,
      "step": 974
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.334811679161844,
      "learning_rate": 4.98843271009364e-05,
      "loss": 0.3501,
      "step": 975
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.8753698878237417,
      "learning_rate": 4.9883849977618945e-05,
      "loss": 0.6046,
      "step": 976
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2686607229960798,
      "learning_rate": 4.9883371874606483e-05,
      "loss": 0.1452,
      "step": 977
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3437875021412078,
      "learning_rate": 4.9882892791917836e-05,
      "loss": 0.1388,
      "step": 978
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7985013814115438,
      "learning_rate": 4.988241272957186e-05,
      "loss": 0.1741,
      "step": 979
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.286766793996886,
      "learning_rate": 4.988193168758747e-05,
      "loss": 1.1028,
      "step": 980
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2515563918949806,
      "learning_rate": 4.988144966598359e-05,
      "loss": 0.2326,
      "step": 981
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0050747015747596,
      "learning_rate": 4.9880966664779205e-05,
      "loss": 0.4218,
      "step": 982
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7044785574148946,
      "learning_rate": 4.988048268399334e-05,
      "loss": 0.2176,
      "step": 983
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3794850793909779,
      "learning_rate": 4.9879997723645036e-05,
      "loss": 0.1868,
      "step": 984
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5777344048579391,
      "learning_rate": 4.9879511783753385e-05,
      "loss": 0.2642,
      "step": 985
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5413778856267177,
      "learning_rate": 4.987902486433753e-05,
      "loss": 0.2251,
      "step": 986
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9007130653803948,
      "learning_rate": 4.9878536965416636e-05,
      "loss": 0.5036,
      "step": 987
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.663013077053618,
      "learning_rate": 4.987804808700991e-05,
      "loss": 0.2948,
      "step": 988
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.382027750209713,
      "learning_rate": 4.98775582291366e-05,
      "loss": 0.5661,
      "step": 989
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.188981970142155,
      "learning_rate": 4.987706739181599e-05,
      "loss": 0.1682,
      "step": 990
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7152970695438454,
      "learning_rate": 4.987657557506742e-05,
      "loss": 0.2533,
      "step": 991
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3635120851951688,
      "learning_rate": 4.9876082778910234e-05,
      "loss": 0.3639,
      "step": 992
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5918345988410636,
      "learning_rate": 4.987558900336384e-05,
      "loss": 0.2026,
      "step": 993
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5178605147413353,
      "learning_rate": 4.987509424844768e-05,
      "loss": 0.2143,
      "step": 994
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8883950766421425,
      "learning_rate": 4.987459851418123e-05,
      "loss": 0.3611,
      "step": 995
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.646844168206758,
      "learning_rate": 4.987410180058402e-05,
      "loss": 0.2196,
      "step": 996
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2335605747706218,
      "learning_rate": 4.987360410767558e-05,
      "loss": 0.1614,
      "step": 997
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8115978064105775,
      "learning_rate": 4.987310543547553e-05,
      "loss": 0.6528,
      "step": 998
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.364780838074598,
      "learning_rate": 4.987260578400349e-05,
      "loss": 0.1386,
      "step": 999
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8656907808994727,
      "learning_rate": 4.9872105153279136e-05,
      "loss": 0.4428,
      "step": 1000
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.559858500037014,
      "learning_rate": 4.9871603543322176e-05,
      "loss": 0.2363,
      "step": 1001
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3916238716886697,
      "learning_rate": 4.9871100954152363e-05,
      "loss": 0.1948,
      "step": 1002
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7424939740383258,
      "learning_rate": 4.987059738578947e-05,
      "loss": 0.2975,
      "step": 1003
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7203860367147037,
      "learning_rate": 4.987009283825335e-05,
      "loss": 0.4419,
      "step": 1004
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3687068645575058,
      "learning_rate": 4.9869587311563836e-05,
      "loss": 0.2024,
      "step": 1005
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4614524134162867,
      "learning_rate": 4.9869080805740855e-05,
      "loss": 0.2748,
      "step": 1006
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9366910546826042,
      "learning_rate": 4.986857332080434e-05,
      "loss": 0.4177,
      "step": 1007
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.172586813824488,
      "learning_rate": 4.9868064856774266e-05,
      "loss": 0.2087,
      "step": 1008
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3235389532178359,
      "learning_rate": 4.9867555413670665e-05,
      "loss": 0.1556,
      "step": 1009
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.161697839260603,
      "learning_rate": 4.9867044991513575e-05,
      "loss": 0.4956,
      "step": 1010
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2197052781169304,
      "learning_rate": 4.9866533590323104e-05,
      "loss": 0.1819,
      "step": 1011
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4259995204494216,
      "learning_rate": 4.986602121011939e-05,
      "loss": 0.1759,
      "step": 1012
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0708788138318397,
      "learning_rate": 4.98655078509226e-05,
      "loss": 0.4472,
      "step": 1013
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.276512692149199,
      "learning_rate": 4.986499351275294e-05,
      "loss": 0.2191,
      "step": 1014
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2764180164050496,
      "learning_rate": 4.986447819563067e-05,
      "loss": 0.3771,
      "step": 1015
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3309793003172992,
      "learning_rate": 4.986396189957607e-05,
      "loss": 0.2422,
      "step": 1016
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2433411529154657,
      "learning_rate": 4.986344462460947e-05,
      "loss": 0.1163,
      "step": 1017
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1626072172357602,
      "learning_rate": 4.986292637075124e-05,
      "loss": 0.1642,
      "step": 1018
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8015345401573295,
      "learning_rate": 4.9862407138021774e-05,
      "loss": 0.5691,
      "step": 1019
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2003198414319238,
      "learning_rate": 4.9861886926441527e-05,
      "loss": 0.1492,
      "step": 1020
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8395641711425827,
      "learning_rate": 4.9861365736030975e-05,
      "loss": 0.2594,
      "step": 1021
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2454138907248602,
      "learning_rate": 4.986084356681063e-05,
      "loss": 0.1818,
      "step": 1022
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9213283070767255,
      "learning_rate": 4.986032041880106e-05,
      "loss": 0.2553,
      "step": 1023
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.4475001400095593,
      "learning_rate": 4.985979629202285e-05,
      "loss": 0.4515,
      "step": 1024
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.137885570489185,
      "learning_rate": 4.985927118649666e-05,
      "loss": 0.1968,
      "step": 1025
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1484087219870398,
      "learning_rate": 4.985874510224314e-05,
      "loss": 0.199,
      "step": 1026
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.192487585856956,
      "learning_rate": 4.985821803928301e-05,
      "loss": 0.3687,
      "step": 1027
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1374419718068831,
      "learning_rate": 4.985768999763702e-05,
      "loss": 0.0961,
      "step": 1028
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.151033036113619,
      "learning_rate": 4.985716097732596e-05,
      "loss": 0.1745,
      "step": 1029
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2640608320002733,
      "learning_rate": 4.985663097837065e-05,
      "loss": 0.1767,
      "step": 1030
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5508176316754196,
      "learning_rate": 4.9856100000791974e-05,
      "loss": 0.3348,
      "step": 1031
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.241896321312421,
      "learning_rate": 4.985556804461082e-05,
      "loss": 0.2043,
      "step": 1032
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4425069002173891,
      "learning_rate": 4.9855035109848144e-05,
      "loss": 0.2148,
      "step": 1033
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6868984016208457,
      "learning_rate": 4.985450119652493e-05,
      "loss": 0.3688,
      "step": 1034
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0178440986061386,
      "learning_rate": 4.985396630466218e-05,
      "loss": 0.5602,
      "step": 1035
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.177964572947333,
      "learning_rate": 4.985343043428097e-05,
      "loss": 0.158,
      "step": 1036
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2130126496369333,
      "learning_rate": 4.9852893585402385e-05,
      "loss": 0.2311,
      "step": 1037
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4042478823144555,
      "learning_rate": 4.9852355758047577e-05,
      "loss": 0.2109,
      "step": 1038
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.715529848526034,
      "learning_rate": 4.9851816952237705e-05,
      "loss": 0.274,
      "step": 1039
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1744863661267,
      "learning_rate": 4.9851277167993995e-05,
      "loss": 0.3158,
      "step": 1040
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0238568385748648,
      "learning_rate": 4.985073640533768e-05,
      "loss": 0.1245,
      "step": 1041
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8792726183046244,
      "learning_rate": 4.985019466429007e-05,
      "loss": 0.1703,
      "step": 1042
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6607178744078355,
      "learning_rate": 4.9849651944872494e-05,
      "loss": 0.1978,
      "step": 1043
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5557987105014164,
      "learning_rate": 4.984910824710631e-05,
      "loss": 0.4595,
      "step": 1044
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.565205170864824,
      "learning_rate": 4.984856357101292e-05,
      "loss": 0.7115,
      "step": 1045
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8766565603542387,
      "learning_rate": 4.984801791661378e-05,
      "loss": 0.3617,
      "step": 1046
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4554069959781777,
      "learning_rate": 4.984747128393036e-05,
      "loss": 0.1744,
      "step": 1047
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.256772868508413,
      "learning_rate": 4.984692367298419e-05,
      "loss": 0.1266,
      "step": 1048
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3190791940738253,
      "learning_rate": 4.984637508379682e-05,
      "loss": 0.3118,
      "step": 1049
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.600116947035674,
      "learning_rate": 4.9845825516389874e-05,
      "loss": 0.2292,
      "step": 1050
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.592146945233789,
      "learning_rate": 4.984527497078496e-05,
      "loss": 0.2425,
      "step": 1051
    },
    {
      "epoch": 0.06,
      "grad_norm": 10.280246054082678,
      "learning_rate": 4.984472344700376e-05,
      "loss": 0.9406,
      "step": 1052
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7811445519499427,
      "learning_rate": 4.9844170945068e-05,
      "loss": 0.2862,
      "step": 1053
    },
    {
      "epoch": 0.06,
      "grad_norm": 45.271539330222005,
      "learning_rate": 4.984361746499943e-05,
      "loss": 2.1061,
      "step": 1054
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.30966723593193,
      "learning_rate": 4.9843063006819824e-05,
      "loss": 0.2414,
      "step": 1055
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.959666008945564,
      "learning_rate": 4.984250757055103e-05,
      "loss": 0.2591,
      "step": 1056
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.038929688621527,
      "learning_rate": 4.9841951156214906e-05,
      "loss": 0.3522,
      "step": 1057
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.817929614193551,
      "learning_rate": 4.984139376383336e-05,
      "loss": 0.26,
      "step": 1058
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8499639900370841,
      "learning_rate": 4.984083539342834e-05,
      "loss": 0.2889,
      "step": 1059
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7256222309893992,
      "learning_rate": 4.9840276045021826e-05,
      "loss": 0.229,
      "step": 1060
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.7582246053309847,
      "learning_rate": 4.983971571863585e-05,
      "loss": 0.7948,
      "step": 1061
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9067746230935168,
      "learning_rate": 4.983915441429246e-05,
      "loss": 0.482,
      "step": 1062
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1323131876412154,
      "learning_rate": 4.983859213201375e-05,
      "loss": 0.556,
      "step": 1063
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3823347755433082,
      "learning_rate": 4.983802887182187e-05,
      "loss": 0.1751,
      "step": 1064
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4909012999284001,
      "learning_rate": 4.9837464633739e-05,
      "loss": 0.387,
      "step": 1065
    },
    {
      "epoch": 0.07,
      "grad_norm": 14.42844626368046,
      "learning_rate": 4.983689941778734e-05,
      "loss": 0.5716,
      "step": 1066
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9196062604597133,
      "learning_rate": 4.9836333223989155e-05,
      "loss": 0.2114,
      "step": 1067
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.693911134208612,
      "learning_rate": 4.983576605236673e-05,
      "loss": 0.7056,
      "step": 1068
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.518552305257714,
      "learning_rate": 4.98351979029424e-05,
      "loss": 0.3992,
      "step": 1069
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9399906654911219,
      "learning_rate": 4.9834628775738525e-05,
      "loss": 0.2453,
      "step": 1070
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6705984720251814,
      "learning_rate": 4.983405867077752e-05,
      "loss": 0.1906,
      "step": 1071
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3193266589690458,
      "learning_rate": 4.983348758808182e-05,
      "loss": 0.2007,
      "step": 1072
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1027515986277274,
      "learning_rate": 4.9832915527673916e-05,
      "loss": 0.3426,
      "step": 1073
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.5518857360418754,
      "learning_rate": 4.983234248957634e-05,
      "loss": 0.5019,
      "step": 1074
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.260919209734482,
      "learning_rate": 4.9831768473811635e-05,
      "loss": 0.2988,
      "step": 1075
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8127604316448562,
      "learning_rate": 4.983119348040242e-05,
      "loss": 0.2905,
      "step": 1076
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0573628360591094,
      "learning_rate": 4.9830617509371317e-05,
      "loss": 0.2931,
      "step": 1077
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0568487165188514,
      "learning_rate": 4.9830040560741e-05,
      "loss": 0.3329,
      "step": 1078
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9286880996102262,
      "learning_rate": 4.98294626345342e-05,
      "loss": 0.2825,
      "step": 1079
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9394612342888906,
      "learning_rate": 4.982888373077365e-05,
      "loss": 0.3475,
      "step": 1080
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1834727069229896,
      "learning_rate": 4.982830384948216e-05,
      "loss": 0.3958,
      "step": 1081
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8778423202610635,
      "learning_rate": 4.9827722990682557e-05,
      "loss": 0.3172,
      "step": 1082
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.843253377709878,
      "learning_rate": 4.98271411543977e-05,
      "loss": 0.2942,
      "step": 1083
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.618573383495456,
      "learning_rate": 4.982655834065051e-05,
      "loss": 0.3039,
      "step": 1084
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5595831503574236,
      "learning_rate": 4.982597454946392e-05,
      "loss": 0.3714,
      "step": 1085
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0028842437575656,
      "learning_rate": 4.982538978086092e-05,
      "loss": 0.3288,
      "step": 1086
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5229124483698466,
      "learning_rate": 4.9824804034864535e-05,
      "loss": 0.2793,
      "step": 1087
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7159593928044,
      "learning_rate": 4.982421731149782e-05,
      "loss": 0.3129,
      "step": 1088
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9746018953875168,
      "learning_rate": 4.9823629610783886e-05,
      "loss": 0.4853,
      "step": 1089
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7729021163739092,
      "learning_rate": 4.9823040932745854e-05,
      "loss": 0.2959,
      "step": 1090
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3662141913448802,
      "learning_rate": 4.982245127740692e-05,
      "loss": 0.1672,
      "step": 1091
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2815137187372154,
      "learning_rate": 4.982186064479029e-05,
      "loss": 0.1654,
      "step": 1092
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.013029017696469,
      "learning_rate": 4.98212690349192e-05,
      "loss": 0.4357,
      "step": 1093
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3231402816244608,
      "learning_rate": 4.9820676447816985e-05,
      "loss": 0.1314,
      "step": 1094
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.224122148511048,
      "learning_rate": 4.9820082883506936e-05,
      "loss": 0.3696,
      "step": 1095
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.338882360636723,
      "learning_rate": 4.981948834201243e-05,
      "loss": 0.496,
      "step": 1096
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6068798267221454,
      "learning_rate": 4.981889282335689e-05,
      "loss": 0.2989,
      "step": 1097
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8677746686337995,
      "learning_rate": 4.9818296327563754e-05,
      "loss": 0.2885,
      "step": 1098
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2904630586461427,
      "learning_rate": 4.98176988546565e-05,
      "loss": 0.1531,
      "step": 1099
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.803937695567104,
      "learning_rate": 4.9817100404658655e-05,
      "loss": 0.4741,
      "step": 1100
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6681209437913036,
      "learning_rate": 4.9816500977593786e-05,
      "loss": 0.2379,
      "step": 1101
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3892642313485397,
      "learning_rate": 4.981590057348548e-05,
      "loss": 0.1908,
      "step": 1102
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5659080245533947,
      "learning_rate": 4.981529919235739e-05,
      "loss": 0.1845,
      "step": 1103
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.5744209981963757,
      "learning_rate": 4.981469683423318e-05,
      "loss": 0.4405,
      "step": 1104
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.419516483504763,
      "learning_rate": 4.981409349913658e-05,
      "loss": 0.2093,
      "step": 1105
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7135089217074593,
      "learning_rate": 4.981348918709133e-05,
      "loss": 0.2532,
      "step": 1106
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.8916718420375176,
      "learning_rate": 4.981288389812123e-05,
      "loss": 0.5351,
      "step": 1107
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.7160913749420423,
      "learning_rate": 4.9812277632250106e-05,
      "loss": 0.5587,
      "step": 1108
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4437936773789994,
      "learning_rate": 4.981167038950183e-05,
      "loss": 0.1678,
      "step": 1109
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4430269979971042,
      "learning_rate": 4.981106216990031e-05,
      "loss": 0.2556,
      "step": 1110
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7482150921566708,
      "learning_rate": 4.981045297346949e-05,
      "loss": 0.4132,
      "step": 1111
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.594968966201651,
      "learning_rate": 4.980984280023336e-05,
      "loss": 0.2908,
      "step": 1112
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6472836758181013,
      "learning_rate": 4.9809231650215933e-05,
      "loss": 0.2864,
      "step": 1113
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1042948405922433,
      "learning_rate": 4.980861952344128e-05,
      "loss": 0.4648,
      "step": 1114
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.0346680917724904,
      "learning_rate": 4.980800641993349e-05,
      "loss": 0.3839,
      "step": 1115
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3785826235771697,
      "learning_rate": 4.980739233971671e-05,
      "loss": 0.1602,
      "step": 1116
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2533009225058254,
      "learning_rate": 4.980677728281512e-05,
      "loss": 0.1358,
      "step": 1117
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4041631390607816,
      "learning_rate": 4.980616124925293e-05,
      "loss": 0.1964,
      "step": 1118
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5558506106188275,
      "learning_rate": 4.980554423905438e-05,
      "loss": 0.2087,
      "step": 1119
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2253121070496107,
      "learning_rate": 4.980492625224379e-05,
      "loss": 0.1859,
      "step": 1120
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.401849536664951,
      "learning_rate": 4.980430728884548e-05,
      "loss": 0.8314,
      "step": 1121
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.416512876555562,
      "learning_rate": 4.98036873488838e-05,
      "loss": 0.1464,
      "step": 1122
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4826334429769261,
      "learning_rate": 4.9803066432383185e-05,
      "loss": 0.261,
      "step": 1123
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3373863426064454,
      "learning_rate": 4.980244453936806e-05,
      "loss": 0.1958,
      "step": 1124
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9464595127873183,
      "learning_rate": 4.980182166986293e-05,
      "loss": 0.3507,
      "step": 1125
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8083207320164383,
      "learning_rate": 4.980119782389229e-05,
      "loss": 0.2381,
      "step": 1126
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4100022178522589,
      "learning_rate": 4.980057300148073e-05,
      "loss": 0.3035,
      "step": 1127
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6285977888545649,
      "learning_rate": 4.979994720265283e-05,
      "loss": 0.3327,
      "step": 1128
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.14501233282493,
      "learning_rate": 4.979932042743324e-05,
      "loss": 0.1777,
      "step": 1129
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8721247532775438,
      "learning_rate": 4.9798692675846626e-05,
      "loss": 0.4204,
      "step": 1130
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9694028317015222,
      "learning_rate": 4.9798063947917714e-05,
      "loss": 0.3033,
      "step": 1131
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.118131833140533,
      "learning_rate": 4.9797434243671245e-05,
      "loss": 0.6025,
      "step": 1132
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4934718850965376,
      "learning_rate": 4.979680356313203e-05,
      "loss": 0.2151,
      "step": 1133
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1960294619557315,
      "learning_rate": 4.9796171906324884e-05,
      "loss": 0.1462,
      "step": 1134
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.325731908380887,
      "learning_rate": 4.979553927327467e-05,
      "loss": 0.2609,
      "step": 1135
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.662120663914877,
      "learning_rate": 4.979490566400631e-05,
      "loss": 0.2503,
      "step": 1136
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.379032887424554,
      "learning_rate": 4.9794271078544755e-05,
      "loss": 0.1978,
      "step": 1137
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4654280037845633,
      "learning_rate": 4.979363551691496e-05,
      "loss": 0.1859,
      "step": 1138
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7473843761293715,
      "learning_rate": 4.979299897914198e-05,
      "loss": 0.159,
      "step": 1139
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7913735525380547,
      "learning_rate": 4.979236146525085e-05,
      "loss": 0.3117,
      "step": 1140
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0473965755927406,
      "learning_rate": 4.979172297526669e-05,
      "loss": 0.2448,
      "step": 1141
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.852729392488185,
      "learning_rate": 4.9791083509214624e-05,
      "loss": 0.1657,
      "step": 1142
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1971022573412167,
      "learning_rate": 4.979044306711984e-05,
      "loss": 0.2028,
      "step": 1143
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6890702204691486,
      "learning_rate": 4.978980164900754e-05,
      "loss": 0.5522,
      "step": 1144
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3165306585282308,
      "learning_rate": 4.978915925490299e-05,
      "loss": 0.2283,
      "step": 1145
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8838200227316731,
      "learning_rate": 4.978851588483146e-05,
      "loss": 0.1611,
      "step": 1146
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1597992225545144,
      "learning_rate": 4.97878715388183e-05,
      "loss": 0.1763,
      "step": 1147
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2672027012372222,
      "learning_rate": 4.978722621688888e-05,
      "loss": 0.3908,
      "step": 1148
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5228630518796213,
      "learning_rate": 4.978657991906859e-05,
      "loss": 0.1915,
      "step": 1149
    },
    {
      "epoch": 0.07,
      "grad_norm": 14.90578610624724,
      "learning_rate": 4.978593264538289e-05,
      "loss": 0.7236,
      "step": 1150
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3299879913945047,
      "learning_rate": 4.978528439585726e-05,
      "loss": 0.517,
      "step": 1151
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9611697240082753,
      "learning_rate": 4.978463517051721e-05,
      "loss": 0.1153,
      "step": 1152
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7600990366266773,
      "learning_rate": 4.978398496938832e-05,
      "loss": 0.2038,
      "step": 1153
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8370063802717245,
      "learning_rate": 4.978333379249618e-05,
      "loss": 0.4409,
      "step": 1154
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4452533363008704,
      "learning_rate": 4.9782681639866423e-05,
      "loss": 0.1968,
      "step": 1155
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3954455204729832,
      "learning_rate": 4.978202851152474e-05,
      "loss": 0.3026,
      "step": 1156
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9993867841052921,
      "learning_rate": 4.978137440749682e-05,
      "loss": 0.1487,
      "step": 1157
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2671719090220384,
      "learning_rate": 4.978071932780843e-05,
      "loss": 0.2326,
      "step": 1158
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.8313487844311562,
      "learning_rate": 4.978006327248537e-05,
      "loss": 0.1956,
      "step": 1159
    },
    {
      "epoch": 0.07,
      "grad_norm": 25.009234780984176,
      "learning_rate": 4.977940624155345e-05,
      "loss": 1.8758,
      "step": 1160
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.576689825034545,
      "learning_rate": 4.9778748235038554e-05,
      "loss": 0.1989,
      "step": 1161
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.254413765530563,
      "learning_rate": 4.977808925296657e-05,
      "loss": 1.4009,
      "step": 1162
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4084771092131751,
      "learning_rate": 4.9777429295363473e-05,
      "loss": 0.206,
      "step": 1163
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7902279484577837,
      "learning_rate": 4.977676836225522e-05,
      "loss": 0.2665,
      "step": 1164
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7919455505713662,
      "learning_rate": 4.9776106453667834e-05,
      "loss": 0.1867,
      "step": 1165
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9164599449152508,
      "learning_rate": 4.977544356962739e-05,
      "loss": 0.3953,
      "step": 1166
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7231413556518749,
      "learning_rate": 4.977477971015997e-05,
      "loss": 0.2264,
      "step": 1167
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8336666383726898,
      "learning_rate": 4.977411487529171e-05,
      "loss": 0.4273,
      "step": 1168
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6576821737888294,
      "learning_rate": 4.9773449065048804e-05,
      "loss": 0.2123,
      "step": 1169
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2553841283489622,
      "learning_rate": 4.977278227945745e-05,
      "loss": 0.1485,
      "step": 1170
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.228835645901867,
      "learning_rate": 4.97721145185439e-05,
      "loss": 0.1703,
      "step": 1171
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7953333130530829,
      "learning_rate": 4.9771445782334456e-05,
      "loss": 0.2966,
      "step": 1172
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.4407684406196797,
      "learning_rate": 4.977077607085543e-05,
      "loss": 0.2168,
      "step": 1173
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1695271193136554,
      "learning_rate": 4.977010538413321e-05,
      "loss": 0.5599,
      "step": 1174
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7059997062966072,
      "learning_rate": 4.9769433722194184e-05,
      "loss": 0.259,
      "step": 1175
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8376094160212268,
      "learning_rate": 4.97687610850648e-05,
      "loss": 0.4714,
      "step": 1176
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3540570294822767,
      "learning_rate": 4.976808747277153e-05,
      "loss": 0.1796,
      "step": 1177
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5200316959597027,
      "learning_rate": 4.9767412885340914e-05,
      "loss": 0.3252,
      "step": 1178
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7734972228095216,
      "learning_rate": 4.9766737322799506e-05,
      "loss": 0.3595,
      "step": 1179
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.214382540795412,
      "learning_rate": 4.9766060785173896e-05,
      "loss": 0.1705,
      "step": 1180
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.292067284074449,
      "learning_rate": 4.9765383272490726e-05,
      "loss": 0.3571,
      "step": 1181
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3720538455306424,
      "learning_rate": 4.976470478477666e-05,
      "loss": 0.199,
      "step": 1182
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.072335747458976,
      "learning_rate": 4.976402532205843e-05,
      "loss": 0.371,
      "step": 1183
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2987519775642438,
      "learning_rate": 4.9763344884362765e-05,
      "loss": 0.1832,
      "step": 1184
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.392767030537254,
      "learning_rate": 4.976266347171647e-05,
      "loss": 0.132,
      "step": 1185
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9749338899100737,
      "learning_rate": 4.976198108414637e-05,
      "loss": 0.5152,
      "step": 1186
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1921557316692033,
      "learning_rate": 4.976129772167932e-05,
      "loss": 0.167,
      "step": 1187
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2641279326618542,
      "learning_rate": 4.9760613384342245e-05,
      "loss": 0.2113,
      "step": 1188
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.104158946682645,
      "learning_rate": 4.9759928072162065e-05,
      "loss": 0.5696,
      "step": 1189
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.662646977315846,
      "learning_rate": 4.975924178516578e-05,
      "loss": 0.1797,
      "step": 1190
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9331569001015503,
      "learning_rate": 4.9758554523380393e-05,
      "loss": 0.2271,
      "step": 1191
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3731287630910385,
      "learning_rate": 4.9757866286832976e-05,
      "loss": 0.1929,
      "step": 1192
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4772925295167494,
      "learning_rate": 4.975717707555062e-05,
      "loss": 0.3383,
      "step": 1193
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5537364477154096,
      "learning_rate": 4.975648688956046e-05,
      "loss": 0.3541,
      "step": 1194
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6670433418989148,
      "learning_rate": 4.975579572888967e-05,
      "loss": 0.2637,
      "step": 1195
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2124667360614394,
      "learning_rate": 4.9755103593565455e-05,
      "loss": 0.1518,
      "step": 1196
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6909207794995407,
      "learning_rate": 4.975441048361507e-05,
      "loss": 0.2754,
      "step": 1197
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.7168444295498357,
      "learning_rate": 4.9753716399065805e-05,
      "loss": 0.6892,
      "step": 1198
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1869760868845205,
      "learning_rate": 4.975302133994499e-05,
      "loss": 0.1762,
      "step": 1199
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4870637323468074,
      "learning_rate": 4.975232530627998e-05,
      "loss": 0.3348,
      "step": 1200
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3148819103986789,
      "learning_rate": 4.9751628298098185e-05,
      "loss": 0.2226,
      "step": 1201
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3790095703530654,
      "learning_rate": 4.975093031542705e-05,
      "loss": 0.294,
      "step": 1202
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6279871298105966,
      "learning_rate": 4.9750231358294044e-05,
      "loss": 0.1956,
      "step": 1203
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2357684952277435,
      "learning_rate": 4.974953142672669e-05,
      "loss": 0.3267,
      "step": 1204
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3125757456120786,
      "learning_rate": 4.9748830520752545e-05,
      "loss": 0.2756,
      "step": 1205
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5534977255531708,
      "learning_rate": 4.974812864039921e-05,
      "loss": 0.2431,
      "step": 1206
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9083015770006593,
      "learning_rate": 4.974742578569431e-05,
      "loss": 0.5334,
      "step": 1207
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5297010809944867,
      "learning_rate": 4.9746721956665526e-05,
      "loss": 0.2062,
      "step": 1208
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3507654697610718,
      "learning_rate": 4.9746017153340564e-05,
      "loss": 0.1715,
      "step": 1209
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0918531426569424,
      "learning_rate": 4.9745311375747166e-05,
      "loss": 0.1259,
      "step": 1210
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.747571784916271,
      "learning_rate": 4.974460462391313e-05,
      "loss": 0.2533,
      "step": 1211
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.968384542614489,
      "learning_rate": 4.974389689786627e-05,
      "loss": 0.4061,
      "step": 1212
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0233915861874605,
      "learning_rate": 4.974318819763446e-05,
      "loss": 0.5307,
      "step": 1213
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4504215146264277,
      "learning_rate": 4.9742478523245604e-05,
      "loss": 0.2389,
      "step": 1214
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4588989830550776,
      "learning_rate": 4.9741767874727626e-05,
      "loss": 0.2317,
      "step": 1215
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9919336406008246,
      "learning_rate": 4.974105625210852e-05,
      "loss": 0.1146,
      "step": 1216
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.362821519409011,
      "learning_rate": 4.97403436554163e-05,
      "loss": 0.655,
      "step": 1217
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3502079059485812,
      "learning_rate": 4.973963008467901e-05,
      "loss": 0.2103,
      "step": 1218
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4351153875902882,
      "learning_rate": 4.973891553992477e-05,
      "loss": 0.1881,
      "step": 1219
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8899535638525722,
      "learning_rate": 4.973820002118168e-05,
      "loss": 0.2396,
      "step": 1220
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5177329472305403,
      "learning_rate": 4.973748352847793e-05,
      "loss": 0.2384,
      "step": 1221
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9328288615265876,
      "learning_rate": 4.9736766061841735e-05,
      "loss": 0.3152,
      "step": 1222
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7129793724467692,
      "learning_rate": 4.973604762130132e-05,
      "loss": 0.3458,
      "step": 1223
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1519532070387692,
      "learning_rate": 4.973532820688498e-05,
      "loss": 0.1425,
      "step": 1224
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1534291012093325,
      "learning_rate": 4.9734607818621045e-05,
      "loss": 0.189,
      "step": 1225
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.192119291974832,
      "learning_rate": 4.973388645653788e-05,
      "loss": 0.1851,
      "step": 1226
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3836691811026187,
      "learning_rate": 4.973316412066388e-05,
      "loss": 0.1735,
      "step": 1227
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4469897618668526,
      "learning_rate": 4.973244081102747e-05,
      "loss": 0.4083,
      "step": 1228
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1465160381471942,
      "learning_rate": 4.973171652765715e-05,
      "loss": 0.1909,
      "step": 1229
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.3230572688291327,
      "learning_rate": 4.973099127058142e-05,
      "loss": 1.009,
      "step": 1230
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2055997651512884,
      "learning_rate": 4.973026503982884e-05,
      "loss": 0.1981,
      "step": 1231
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6113631125886916,
      "learning_rate": 4.9729537835428006e-05,
      "loss": 0.2185,
      "step": 1232
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.919077011938624,
      "learning_rate": 4.972880965740755e-05,
      "loss": 1.6327,
      "step": 1233
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.522414312879853,
      "learning_rate": 4.9728080505796124e-05,
      "loss": 0.1896,
      "step": 1234
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9954742256295542,
      "learning_rate": 4.972735038062245e-05,
      "loss": 0.116,
      "step": 1235
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0404570231963435,
      "learning_rate": 4.972661928191528e-05,
      "loss": 0.1471,
      "step": 1236
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2963645909472739,
      "learning_rate": 4.972588720970337e-05,
      "loss": 0.1386,
      "step": 1237
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4103842514706693,
      "learning_rate": 4.9725154164015574e-05,
      "loss": 0.1661,
      "step": 1238
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.80662720610632,
      "learning_rate": 4.9724420144880726e-05,
      "loss": 0.7953,
      "step": 1239
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8914368337687852,
      "learning_rate": 4.972368515232775e-05,
      "loss": 0.4109,
      "step": 1240
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9028574835152443,
      "learning_rate": 4.972294918638557e-05,
      "loss": 0.2133,
      "step": 1241
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.013029140802961,
      "learning_rate": 4.972221224708315e-05,
      "loss": 0.2729,
      "step": 1242
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.357480721002134,
      "learning_rate": 4.9721474334449526e-05,
      "loss": 0.4223,
      "step": 1243
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5711362220671048,
      "learning_rate": 4.972073544851374e-05,
      "loss": 0.1665,
      "step": 1244
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6243634270398524,
      "learning_rate": 4.971999558930488e-05,
      "loss": 0.4571,
      "step": 1245
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4573594126655907,
      "learning_rate": 4.971925475685208e-05,
      "loss": 0.1939,
      "step": 1246
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2031195710077816,
      "learning_rate": 4.9718512951184506e-05,
      "loss": 0.2946,
      "step": 1247
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.607060569581511,
      "learning_rate": 4.971777017233136e-05,
      "loss": 0.2617,
      "step": 1248
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2713695406159096,
      "learning_rate": 4.971702642032188e-05,
      "loss": 0.1582,
      "step": 1249
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6759624832114723,
      "learning_rate": 4.971628169518536e-05,
      "loss": 0.2321,
      "step": 1250
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5234445576014117,
      "learning_rate": 4.971553599695112e-05,
      "loss": 0.2122,
      "step": 1251
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.290395799978097,
      "learning_rate": 4.9714789325648515e-05,
      "loss": 0.263,
      "step": 1252
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4910371647131593,
      "learning_rate": 4.9714041681306944e-05,
      "loss": 0.2895,
      "step": 1253
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1562332332343843,
      "learning_rate": 4.971329306395584e-05,
      "loss": 0.1698,
      "step": 1254
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.360901712819297,
      "learning_rate": 4.9712543473624666e-05,
      "loss": 0.2396,
      "step": 1255
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.680233658622869,
      "learning_rate": 4.971179291034295e-05,
      "loss": 0.244,
      "step": 1256
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6201600958817717,
      "learning_rate": 4.9711041374140244e-05,
      "loss": 0.3701,
      "step": 1257
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6068426975926124,
      "learning_rate": 4.971028886504612e-05,
      "loss": 0.3022,
      "step": 1258
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0459924402741614,
      "learning_rate": 4.9709535383090225e-05,
      "loss": 0.1189,
      "step": 1259
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.590461825328669,
      "learning_rate": 4.97087809283022e-05,
      "loss": 0.2351,
      "step": 1260
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4571169072783063,
      "learning_rate": 4.970802550071177e-05,
      "loss": 0.2247,
      "step": 1261
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7567229552188526,
      "learning_rate": 4.9707269100348664e-05,
      "loss": 0.4356,
      "step": 1262
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.294636833312829,
      "learning_rate": 4.970651172724267e-05,
      "loss": 0.3734,
      "step": 1263
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4817808295200834,
      "learning_rate": 4.9705753381423604e-05,
      "loss": 0.195,
      "step": 1264
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.446373243130398,
      "learning_rate": 4.970499406292132e-05,
      "loss": 0.2456,
      "step": 1265
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1866174013536537,
      "learning_rate": 4.970423377176571e-05,
      "loss": 0.3216,
      "step": 1266
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0992228904797854,
      "learning_rate": 4.970347250798671e-05,
      "loss": 0.1228,
      "step": 1267
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6006293085199104,
      "learning_rate": 4.97027102716143e-05,
      "loss": 0.3395,
      "step": 1268
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4029755877777734,
      "learning_rate": 4.970194706267848e-05,
      "loss": 0.2444,
      "step": 1269
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.439305621139756,
      "learning_rate": 4.97011828812093e-05,
      "loss": 0.5378,
      "step": 1270
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6886704608543002,
      "learning_rate": 4.970041772723685e-05,
      "loss": 0.1433,
      "step": 1271
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.320211327403198,
      "learning_rate": 4.9699651600791254e-05,
      "loss": 0.2221,
      "step": 1272
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0178723353679353,
      "learning_rate": 4.969888450190267e-05,
      "loss": 0.3983,
      "step": 1273
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3386591473307492,
      "learning_rate": 4.9698116430601296e-05,
      "loss": 0.1626,
      "step": 1274
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0916238564086311,
      "learning_rate": 4.9697347386917384e-05,
      "loss": 0.2001,
      "step": 1275
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4628478669183778,
      "learning_rate": 4.96965773708812e-05,
      "loss": 0.3169,
      "step": 1276
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.359942678255517,
      "learning_rate": 4.969580638252308e-05,
      "loss": 0.6356,
      "step": 1277
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.609791671387979,
      "learning_rate": 4.969503442187335e-05,
      "loss": 0.3276,
      "step": 1278
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7047486243173975,
      "learning_rate": 4.969426148896242e-05,
      "loss": 0.2332,
      "step": 1279
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.795474669001534,
      "learning_rate": 4.969348758382072e-05,
      "loss": 0.3241,
      "step": 1280
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2894422820468563,
      "learning_rate": 4.969271270647871e-05,
      "loss": 0.2268,
      "step": 1281
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.277084343754402,
      "learning_rate": 4.969193685696691e-05,
      "loss": 0.1933,
      "step": 1282
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0196811736870202,
      "learning_rate": 4.969116003531586e-05,
      "loss": 0.1423,
      "step": 1283
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8788266425162936,
      "learning_rate": 4.969038224155614e-05,
      "loss": 0.5632,
      "step": 1284
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2017171526119812,
      "learning_rate": 4.968960347571838e-05,
      "loss": 0.1795,
      "step": 1285
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0573197564305898,
      "learning_rate": 4.968882373783324e-05,
      "loss": 0.1945,
      "step": 1286
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.881493292531261,
      "learning_rate": 4.9688043027931406e-05,
      "loss": 0.1338,
      "step": 1287
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3215771485591725,
      "learning_rate": 4.968726134604363e-05,
      "loss": 0.126,
      "step": 1288
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0053792336408405,
      "learning_rate": 4.968647869220068e-05,
      "loss": 0.8607,
      "step": 1289
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.025729949396765,
      "learning_rate": 4.9685695066433374e-05,
      "loss": 0.4333,
      "step": 1290
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.169351779962753,
      "learning_rate": 4.968491046877256e-05,
      "loss": 0.1608,
      "step": 1291
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2693618146232715,
      "learning_rate": 4.968412489924913e-05,
      "loss": 0.1947,
      "step": 1292
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0665548090331631,
      "learning_rate": 4.9683338357894015e-05,
      "loss": 0.1009,
      "step": 1293
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.22741744656006,
      "learning_rate": 4.968255084473817e-05,
      "loss": 0.4938,
      "step": 1294
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0847262456457507,
      "learning_rate": 4.968176235981262e-05,
      "loss": 0.3424,
      "step": 1295
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2626849499799213,
      "learning_rate": 4.968097290314839e-05,
      "loss": 0.1601,
      "step": 1296
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5518151550596948,
      "learning_rate": 4.968018247477658e-05,
      "loss": 0.2574,
      "step": 1297
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5129369856624135,
      "learning_rate": 4.967939107472829e-05,
      "loss": 0.2803,
      "step": 1298
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1826205489639263,
      "learning_rate": 4.967859870303469e-05,
      "loss": 0.222,
      "step": 1299
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4612126205150606,
      "learning_rate": 4.967780535972696e-05,
      "loss": 0.2203,
      "step": 1300
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.534298628835779,
      "learning_rate": 4.9677011044836357e-05,
      "loss": 0.3419,
      "step": 1301
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0468348208973612,
      "learning_rate": 4.967621575839414e-05,
      "loss": 0.1727,
      "step": 1302
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.472777016713078,
      "learning_rate": 4.967541950043163e-05,
      "loss": 0.2866,
      "step": 1303
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9731960749940224,
      "learning_rate": 4.9674622270980165e-05,
      "loss": 0.1789,
      "step": 1304
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.656898581017471,
      "learning_rate": 4.967382407007114e-05,
      "loss": 0.2289,
      "step": 1305
    },
    {
      "epoch": 0.08,
      "grad_norm": 59.2184499915668,
      "learning_rate": 4.9673024897735976e-05,
      "loss": 1.4305,
      "step": 1306
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3328269293356763,
      "learning_rate": 4.967222475400614e-05,
      "loss": 0.2041,
      "step": 1307
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8448246612365842,
      "learning_rate": 4.9671423638913133e-05,
      "loss": 0.3049,
      "step": 1308
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8183292297429348,
      "learning_rate": 4.967062155248849e-05,
      "loss": 0.4258,
      "step": 1309
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.660712055076484,
      "learning_rate": 4.966981849476381e-05,
      "loss": 0.2196,
      "step": 1310
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.205770997175105,
      "learning_rate": 4.9669014465770685e-05,
      "loss": 0.5697,
      "step": 1311
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1318570326973063,
      "learning_rate": 4.9668209465540785e-05,
      "loss": 0.1842,
      "step": 1312
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.212531529798557,
      "learning_rate": 4.96674034941058e-05,
      "loss": 0.3587,
      "step": 1313
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4183473171111616,
      "learning_rate": 4.966659655149746e-05,
      "loss": 0.5822,
      "step": 1314
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5552810407813895,
      "learning_rate": 4.9665788637747534e-05,
      "loss": 0.3651,
      "step": 1315
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2199247321402107,
      "learning_rate": 4.966497975288783e-05,
      "loss": 0.2396,
      "step": 1316
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1301353767193543,
      "learning_rate": 4.96641698969502e-05,
      "loss": 0.1736,
      "step": 1317
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4077663384915302,
      "learning_rate": 4.9663359069966525e-05,
      "loss": 0.1866,
      "step": 1318
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.617238899989246,
      "learning_rate": 4.966254727196873e-05,
      "loss": 0.4942,
      "step": 1319
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3121182599157821,
      "learning_rate": 4.966173450298876e-05,
      "loss": 0.1877,
      "step": 1320
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4308489669457654,
      "learning_rate": 4.966092076305865e-05,
      "loss": 0.2599,
      "step": 1321
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.723518833523077,
      "learning_rate": 4.96601060522104e-05,
      "loss": 0.2451,
      "step": 1322
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.420764365253715,
      "learning_rate": 4.965929037047611e-05,
      "loss": 0.1469,
      "step": 1323
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2166481041052903,
      "learning_rate": 4.965847371788788e-05,
      "loss": 0.1482,
      "step": 1324
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.709752134629159,
      "learning_rate": 4.965765609447787e-05,
      "loss": 0.6566,
      "step": 1325
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.5588962275692575,
      "learning_rate": 4.965683750027826e-05,
      "loss": 0.2863,
      "step": 1326
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4135164933762352,
      "learning_rate": 4.96560179353213e-05,
      "loss": 0.2749,
      "step": 1327
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.043404234979078,
      "learning_rate": 4.9655197399639235e-05,
      "loss": 0.4937,
      "step": 1328
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.294487495625922,
      "learning_rate": 4.965437589326438e-05,
      "loss": 0.2148,
      "step": 1329
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5055279348741641,
      "learning_rate": 4.965355341622908e-05,
      "loss": 0.171,
      "step": 1330
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8403154923974807,
      "learning_rate": 4.9652729968565715e-05,
      "loss": 0.3684,
      "step": 1331
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.520345649329764,
      "learning_rate": 4.96519055503067e-05,
      "loss": 0.5176,
      "step": 1332
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.516651430170447,
      "learning_rate": 4.9651080161484495e-05,
      "loss": 0.355,
      "step": 1333
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.560098544763247,
      "learning_rate": 4.9650253802131595e-05,
      "loss": 0.6952,
      "step": 1334
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.628539831718541,
      "learning_rate": 4.964942647228054e-05,
      "loss": 0.3108,
      "step": 1335
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2485974078838833,
      "learning_rate": 4.96485981719639e-05,
      "loss": 0.1476,
      "step": 1336
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5264637696092158,
      "learning_rate": 4.964776890121428e-05,
      "loss": 0.2724,
      "step": 1337
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.161581700623837,
      "learning_rate": 4.964693866006434e-05,
      "loss": 0.9372,
      "step": 1338
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3232757545304064,
      "learning_rate": 4.9646107448546756e-05,
      "loss": 0.199,
      "step": 1339
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4524796717749444,
      "learning_rate": 4.964527526669426e-05,
      "loss": 0.6479,
      "step": 1340
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7899499637792164,
      "learning_rate": 4.964444211453962e-05,
      "loss": 0.489,
      "step": 1341
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8726010151913501,
      "learning_rate": 4.964360799211562e-05,
      "loss": 0.5947,
      "step": 1342
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9108203927623093,
      "learning_rate": 4.964277289945513e-05,
      "loss": 0.1977,
      "step": 1343
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0273813984130458,
      "learning_rate": 4.9641936836591e-05,
      "loss": 0.3831,
      "step": 1344
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.348238312483372,
      "learning_rate": 4.964109980355615e-05,
      "loss": 0.2122,
      "step": 1345
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1297527509807495,
      "learning_rate": 4.964026180038355e-05,
      "loss": 0.1989,
      "step": 1346
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9808549614632555,
      "learning_rate": 4.963942282710618e-05,
      "loss": 0.1389,
      "step": 1347
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1958453656237695,
      "learning_rate": 4.963858288375707e-05,
      "loss": 0.2193,
      "step": 1348
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.631231095366703,
      "learning_rate": 4.96377419703693e-05,
      "loss": 0.2397,
      "step": 1349
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2198297836499847,
      "learning_rate": 4.963690008697597e-05,
      "loss": 0.1504,
      "step": 1350
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8647710421432941,
      "learning_rate": 4.9636057233610236e-05,
      "loss": 0.2779,
      "step": 1351
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1938800967920247,
      "learning_rate": 4.963521341030526e-05,
      "loss": 0.1803,
      "step": 1352
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5118400460804782,
      "learning_rate": 4.963436861709428e-05,
      "loss": 0.1722,
      "step": 1353
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.935421087029981,
      "learning_rate": 4.963352285401055e-05,
      "loss": 0.6045,
      "step": 1354
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.741575534658268,
      "learning_rate": 4.963267612108738e-05,
      "loss": 0.3704,
      "step": 1355
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3070144377758373,
      "learning_rate": 4.9631828418358095e-05,
      "loss": 0.1436,
      "step": 1356
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.660131465838028,
      "learning_rate": 4.963097974585606e-05,
      "loss": 0.2868,
      "step": 1357
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9420621799414093,
      "learning_rate": 4.9630130103614715e-05,
      "loss": 0.5147,
      "step": 1358
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9217827421877394,
      "learning_rate": 4.962927949166749e-05,
      "loss": 0.3606,
      "step": 1359
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.150027040813671,
      "learning_rate": 4.962842791004788e-05,
      "loss": 0.1871,
      "step": 1360
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.1618637436259904,
      "learning_rate": 4.962757535878941e-05,
      "loss": 0.49,
      "step": 1361
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9231778423178154,
      "learning_rate": 4.9626721837925664e-05,
      "loss": 0.2158,
      "step": 1362
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5861225634712286,
      "learning_rate": 4.962586734749021e-05,
      "loss": 0.3108,
      "step": 1363
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.186694395710347,
      "learning_rate": 4.9625011887516714e-05,
      "loss": 0.1339,
      "step": 1364
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1512680785393432,
      "learning_rate": 4.9624155458038855e-05,
      "loss": 0.1851,
      "step": 1365
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.251535563419548,
      "learning_rate": 4.962329805909035e-05,
      "loss": 0.3268,
      "step": 1366
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7551940529474455,
      "learning_rate": 4.9622439690704954e-05,
      "loss": 0.3151,
      "step": 1367
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8128812464154416,
      "learning_rate": 4.962158035291646e-05,
      "loss": 0.307,
      "step": 1368
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2503339987964208,
      "learning_rate": 4.96207200457587e-05,
      "loss": 0.1639,
      "step": 1369
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1229048143678686,
      "learning_rate": 4.9619858769265545e-05,
      "loss": 0.1547,
      "step": 1370
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.385130057246838,
      "learning_rate": 4.9618996523470904e-05,
      "loss": 0.2805,
      "step": 1371
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6191653703503506,
      "learning_rate": 4.961813330840873e-05,
      "loss": 0.2807,
      "step": 1372
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8287928640025923,
      "learning_rate": 4.9617269124113006e-05,
      "loss": 0.5213,
      "step": 1373
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3078344944209999,
      "learning_rate": 4.961640397061776e-05,
      "loss": 0.1995,
      "step": 1374
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.748182243803167,
      "learning_rate": 4.961553784795704e-05,
      "loss": 0.4958,
      "step": 1375
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5172645632619655,
      "learning_rate": 4.961467075616495e-05,
      "loss": 0.3281,
      "step": 1376
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4928707875245275,
      "learning_rate": 4.961380269527563e-05,
      "loss": 0.3423,
      "step": 1377
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0595096839462412,
      "learning_rate": 4.961293366532327e-05,
      "loss": 0.145,
      "step": 1378
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4150095406619247,
      "learning_rate": 4.961206366634207e-05,
      "loss": 0.1907,
      "step": 1379
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.315898496218054,
      "learning_rate": 4.961119269836628e-05,
      "loss": 0.4867,
      "step": 1380
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9678964371953732,
      "learning_rate": 4.9610320761430195e-05,
      "loss": 0.1444,
      "step": 1381
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.422336987368297,
      "learning_rate": 4.960944785556814e-05,
      "loss": 0.2038,
      "step": 1382
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5934866166776749,
      "learning_rate": 4.9608573980814494e-05,
      "loss": 0.119,
      "step": 1383
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.301497896320832,
      "learning_rate": 4.960769913720365e-05,
      "loss": 0.2146,
      "step": 1384
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.7066873132368707,
      "learning_rate": 4.960682332477006e-05,
      "loss": 0.3843,
      "step": 1385
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.623997443335778,
      "learning_rate": 4.9605946543548196e-05,
      "loss": 0.6943,
      "step": 1386
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8249904773679746,
      "learning_rate": 4.960506879357258e-05,
      "loss": 0.2667,
      "step": 1387
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2246350993671178,
      "learning_rate": 4.9604190074877775e-05,
      "loss": 0.1865,
      "step": 1388
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1105811854353216,
      "learning_rate": 4.960331038749837e-05,
      "loss": 0.1779,
      "step": 1389
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8861601629618794,
      "learning_rate": 4.9602429731469e-05,
      "loss": 0.2038,
      "step": 1390
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.8748847596444507,
      "learning_rate": 4.960154810682435e-05,
      "loss": 0.9312,
      "step": 1391
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3028328509807554,
      "learning_rate": 4.960066551359912e-05,
      "loss": 0.1883,
      "step": 1392
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3663831622810143,
      "learning_rate": 4.9599781951828054e-05,
      "loss": 0.1904,
      "step": 1393
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9553167651215396,
      "learning_rate": 4.959889742154594e-05,
      "loss": 0.176,
      "step": 1394
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.077135857756438,
      "learning_rate": 4.95980119227876e-05,
      "loss": 0.4935,
      "step": 1395
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.895953359894328,
      "learning_rate": 4.9597125455587914e-05,
      "loss": 0.3412,
      "step": 1396
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5664101562715524,
      "learning_rate": 4.959623801998176e-05,
      "loss": 0.2481,
      "step": 1397
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.529312907813981,
      "learning_rate": 4.959534961600409e-05,
      "loss": 0.6591,
      "step": 1398
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1711625189988166,
      "learning_rate": 4.9594460243689886e-05,
      "loss": 0.4702,
      "step": 1399
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7193361121967157,
      "learning_rate": 4.959356990307415e-05,
      "loss": 0.2529,
      "step": 1400
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9828836646648426,
      "learning_rate": 4.959267859419194e-05,
      "loss": 0.1267,
      "step": 1401
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5793725075483958,
      "learning_rate": 4.959178631707835e-05,
      "loss": 0.3157,
      "step": 1402
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0502738539351488,
      "learning_rate": 4.959089307176851e-05,
      "loss": 0.1385,
      "step": 1403
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1051765142177796,
      "learning_rate": 4.958999885829759e-05,
      "loss": 0.1201,
      "step": 1404
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5388618427971652,
      "learning_rate": 4.9589103676700785e-05,
      "loss": 0.1599,
      "step": 1405
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2337883268006071,
      "learning_rate": 4.9588207527013356e-05,
      "loss": 0.2304,
      "step": 1406
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2186842664872293,
      "learning_rate": 4.958731040927056e-05,
      "loss": 0.2451,
      "step": 1407
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3149970095276011,
      "learning_rate": 4.9586412323507744e-05,
      "loss": 0.2607,
      "step": 1408
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.965918814845177,
      "learning_rate": 4.9585513269760245e-05,
      "loss": 0.3476,
      "step": 1409
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7965433515991045,
      "learning_rate": 4.958461324806348e-05,
      "loss": 0.6186,
      "step": 1410
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5742943465290085,
      "learning_rate": 4.958371225845286e-05,
      "loss": 0.2679,
      "step": 1411
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.2971446073666346,
      "learning_rate": 4.958281030096388e-05,
      "loss": 0.664,
      "step": 1412
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.257701628267232,
      "learning_rate": 4.9581907375632034e-05,
      "loss": 0.1621,
      "step": 1413
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.0046925918355614,
      "learning_rate": 4.9581003482492884e-05,
      "loss": 0.6995,
      "step": 1414
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9584730232678221,
      "learning_rate": 4.958009862158201e-05,
      "loss": 0.1137,
      "step": 1415
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4454933811053507,
      "learning_rate": 4.9579192792935026e-05,
      "loss": 0.1998,
      "step": 1416
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1836770440371223,
      "learning_rate": 4.957828599658762e-05,
      "loss": 0.1642,
      "step": 1417
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.405787702948859,
      "learning_rate": 4.9577378232575465e-05,
      "loss": 0.2043,
      "step": 1418
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5562824342250428,
      "learning_rate": 4.957646950093433e-05,
      "loss": 0.2912,
      "step": 1419
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9571654617493387,
      "learning_rate": 4.957555980169997e-05,
      "loss": 0.1786,
      "step": 1420
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2578966574320591,
      "learning_rate": 4.95746491349082e-05,
      "loss": 0.1976,
      "step": 1421
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.686636382330687,
      "learning_rate": 4.95737375005949e-05,
      "loss": 0.3902,
      "step": 1422
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7752279455436815,
      "learning_rate": 4.957282489879593e-05,
      "loss": 0.2969,
      "step": 1423
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1140599238381852,
      "learning_rate": 4.957191132954724e-05,
      "loss": 0.1509,
      "step": 1424
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9176945018503495,
      "learning_rate": 4.957099679288478e-05,
      "loss": 0.2087,
      "step": 1425
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0686544506276128,
      "learning_rate": 4.957008128884458e-05,
      "loss": 0.1575,
      "step": 1426
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3661534634524743,
      "learning_rate": 4.956916481746266e-05,
      "loss": 0.18,
      "step": 1427
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.287165126022429,
      "learning_rate": 4.956824737877512e-05,
      "loss": 0.2122,
      "step": 1428
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.114060572573419,
      "learning_rate": 4.956732897281807e-05,
      "loss": 0.2315,
      "step": 1429
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2623455254668634,
      "learning_rate": 4.956640959962768e-05,
      "loss": 0.2466,
      "step": 1430
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.678869060720941,
      "learning_rate": 4.956548925924013e-05,
      "loss": 0.2358,
      "step": 1431
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5191217307082532,
      "learning_rate": 4.956456795169167e-05,
      "loss": 0.5119,
      "step": 1432
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7433942739752792,
      "learning_rate": 4.956364567701855e-05,
      "loss": 0.6772,
      "step": 1433
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1481575087636475,
      "learning_rate": 4.95627224352571e-05,
      "loss": 0.1706,
      "step": 1434
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.570974190731787,
      "learning_rate": 4.956179822644367e-05,
      "loss": 0.6595,
      "step": 1435
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5276967582808596,
      "learning_rate": 4.9560873050614645e-05,
      "loss": 0.1816,
      "step": 1436
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.09821960284701,
      "learning_rate": 4.9559946907806435e-05,
      "loss": 0.1678,
      "step": 1437
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0143455621710287,
      "learning_rate": 4.955901979805552e-05,
      "loss": 0.1806,
      "step": 1438
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.869632155640692,
      "learning_rate": 4.955809172139839e-05,
      "loss": 0.4233,
      "step": 1439
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.759079394462291,
      "learning_rate": 4.9557162677871594e-05,
      "loss": 0.1951,
      "step": 1440
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7439739318773357,
      "learning_rate": 4.9556232667511706e-05,
      "loss": 0.4481,
      "step": 1441
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3006673228699006,
      "learning_rate": 4.955530169035534e-05,
      "loss": 0.1349,
      "step": 1442
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3171750285891757,
      "learning_rate": 4.955436974643914e-05,
      "loss": 0.1976,
      "step": 1443
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.654648816667812,
      "learning_rate": 4.9553436835799804e-05,
      "loss": 0.3064,
      "step": 1444
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.1136617525718915,
      "learning_rate": 4.955250295847407e-05,
      "loss": 0.6117,
      "step": 1445
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3225696017856052,
      "learning_rate": 4.9551568114498696e-05,
      "loss": 0.2277,
      "step": 1446
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4163326381107053,
      "learning_rate": 4.9550632303910496e-05,
      "loss": 0.2552,
      "step": 1447
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.196069261023554,
      "learning_rate": 4.95496955267463e-05,
      "loss": 0.2234,
      "step": 1448
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2512851987338751,
      "learning_rate": 4.9548757783043e-05,
      "loss": 0.1646,
      "step": 1449
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5422453477025613,
      "learning_rate": 4.954781907283751e-05,
      "loss": 0.2527,
      "step": 1450
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5712113654596274,
      "learning_rate": 4.954687939616679e-05,
      "loss": 0.5947,
      "step": 1451
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4500334724060893,
      "learning_rate": 4.954593875306784e-05,
      "loss": 0.2357,
      "step": 1452
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2266191349022326,
      "learning_rate": 4.954499714357769e-05,
      "loss": 0.1783,
      "step": 1453
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9832702783620155,
      "learning_rate": 4.954405456773341e-05,
      "loss": 0.1717,
      "step": 1454
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7204998478633238,
      "learning_rate": 4.954311102557211e-05,
      "loss": 0.387,
      "step": 1455
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9711014395199855,
      "learning_rate": 4.9542166517130936e-05,
      "loss": 0.1346,
      "step": 1456
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5634195852027604,
      "learning_rate": 4.954122104244708e-05,
      "loss": 0.2796,
      "step": 1457
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3414358863986853,
      "learning_rate": 4.954027460155777e-05,
      "loss": 0.2601,
      "step": 1458
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.774535001513422,
      "learning_rate": 4.9539327194500255e-05,
      "loss": 0.902,
      "step": 1459
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1179423211581554,
      "learning_rate": 4.953837882131184e-05,
      "loss": 0.1848,
      "step": 1460
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9776857553883698,
      "learning_rate": 4.953742948202987e-05,
      "loss": 0.1165,
      "step": 1461
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7734395955239277,
      "learning_rate": 4.9536479176691715e-05,
      "loss": 0.1129,
      "step": 1462
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5771497219399389,
      "learning_rate": 4.953552790533479e-05,
      "loss": 0.3362,
      "step": 1463
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3354897476504533,
      "learning_rate": 4.953457566799654e-05,
      "loss": 0.2079,
      "step": 1464
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4126210632871365,
      "learning_rate": 4.953362246471447e-05,
      "loss": 0.4053,
      "step": 1465
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1637052974043358,
      "learning_rate": 4.953266829552611e-05,
      "loss": 0.1936,
      "step": 1466
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.799435720425659,
      "learning_rate": 4.9531713160469004e-05,
      "loss": 0.2558,
      "step": 1467
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9759704790422689,
      "learning_rate": 4.953075705958077e-05,
      "loss": 0.0983,
      "step": 1468
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.1308039407157473,
      "learning_rate": 4.952979999289905e-05,
      "loss": 0.6191,
      "step": 1469
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5564317062272899,
      "learning_rate": 4.952884196046153e-05,
      "loss": 0.2795,
      "step": 1470
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.3380521859347483,
      "learning_rate": 4.952788296230592e-05,
      "loss": 0.2578,
      "step": 1471
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2796383203132988,
      "learning_rate": 4.9526922998469985e-05,
      "loss": 0.3546,
      "step": 1472
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4815438112117472,
      "learning_rate": 4.952596206899151e-05,
      "loss": 0.2616,
      "step": 1473
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1843897896514866,
      "learning_rate": 4.952500017390832e-05,
      "loss": 0.1616,
      "step": 1474
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.281813482389721,
      "learning_rate": 4.952403731325831e-05,
      "loss": 0.3848,
      "step": 1475
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5184339052058728,
      "learning_rate": 4.952307348707938e-05,
      "loss": 0.1387,
      "step": 1476
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3183497831572384,
      "learning_rate": 4.9522108695409455e-05,
      "loss": 0.2936,
      "step": 1477
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2193658726043044,
      "learning_rate": 4.952114293828655e-05,
      "loss": 0.1871,
      "step": 1478
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4259519394423703,
      "learning_rate": 4.952017621574866e-05,
      "loss": 0.2999,
      "step": 1479
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8072877588890988,
      "learning_rate": 4.9519208527833874e-05,
      "loss": 0.2514,
      "step": 1480
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1019815075806072,
      "learning_rate": 4.9518239874580264e-05,
      "loss": 0.1161,
      "step": 1481
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.763974510376292,
      "learning_rate": 4.951727025602598e-05,
      "loss": 0.7426,
      "step": 1482
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4026180160434913,
      "learning_rate": 4.951629967220921e-05,
      "loss": 0.2781,
      "step": 1483
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.4076917890140277,
      "learning_rate": 4.951532812316814e-05,
      "loss": 0.2091,
      "step": 1484
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2357956249342754,
      "learning_rate": 4.951435560894103e-05,
      "loss": 0.1454,
      "step": 1485
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4898301887110648,
      "learning_rate": 4.9513382129566175e-05,
      "loss": 0.2368,
      "step": 1486
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7523032163173369,
      "learning_rate": 4.95124076850819e-05,
      "loss": 0.3239,
      "step": 1487
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.5651621527161983,
      "learning_rate": 4.951143227552656e-05,
      "loss": 0.6558,
      "step": 1488
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0698895142561593,
      "learning_rate": 4.951045590093857e-05,
      "loss": 0.2435,
      "step": 1489
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1222231003215417,
      "learning_rate": 4.950947856135636e-05,
      "loss": 0.3457,
      "step": 1490
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3874536739945638,
      "learning_rate": 4.950850025681842e-05,
      "loss": 0.216,
      "step": 1491
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8847801128263983,
      "learning_rate": 4.9507520987363255e-05,
      "loss": 0.3368,
      "step": 1492
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3995758560616647,
      "learning_rate": 4.950654075302943e-05,
      "loss": 0.2067,
      "step": 1493
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3784960232487484,
      "learning_rate": 4.950555955385553e-05,
      "loss": 0.1727,
      "step": 1494
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5538620610406917,
      "learning_rate": 4.950457738988018e-05,
      "loss": 0.335,
      "step": 1495
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1763636764905445,
      "learning_rate": 4.9503594261142075e-05,
      "loss": 0.1281,
      "step": 1496
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9188499573697003,
      "learning_rate": 4.9502610167679886e-05,
      "loss": 0.1342,
      "step": 1497
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.239778237326337,
      "learning_rate": 4.950162510953238e-05,
      "loss": 0.4002,
      "step": 1498
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2804530315136038,
      "learning_rate": 4.9500639086738334e-05,
      "loss": 0.1815,
      "step": 1499
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0914340086596441,
      "learning_rate": 4.9499652099336567e-05,
      "loss": 0.1519,
      "step": 1500
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4547525159760997,
      "learning_rate": 4.9498664147365934e-05,
      "loss": 0.357,
      "step": 1501
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1497238126246736,
      "learning_rate": 4.9497675230865346e-05,
      "loss": 0.2165,
      "step": 1502
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.42369674430443,
      "learning_rate": 4.9496685349873724e-05,
      "loss": 0.1983,
      "step": 1503
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.4480704625257417,
      "learning_rate": 4.949569450443004e-05,
      "loss": 0.2339,
      "step": 1504
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.231189505313219,
      "learning_rate": 4.9494702694573316e-05,
      "loss": 0.603,
      "step": 1505
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1321064353101493,
      "learning_rate": 4.949370992034258e-05,
      "loss": 0.1771,
      "step": 1506
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6041891344821326,
      "learning_rate": 4.9492716181776935e-05,
      "loss": 0.2732,
      "step": 1507
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3832420214720655,
      "learning_rate": 4.949172147891551e-05,
      "loss": 0.2006,
      "step": 1508
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4274978077297353,
      "learning_rate": 4.9490725811797446e-05,
      "loss": 0.2667,
      "step": 1509
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.317277081238402,
      "learning_rate": 4.9489729180461956e-05,
      "loss": 0.1497,
      "step": 1510
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7697584958696875,
      "learning_rate": 4.9488731584948286e-05,
      "loss": 0.3498,
      "step": 1511
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.995995101906425,
      "learning_rate": 4.948773302529569e-05,
      "loss": 0.8611,
      "step": 1512
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.984676346999385,
      "learning_rate": 4.94867335015435e-05,
      "loss": 0.1492,
      "step": 1513
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1245159786386356,
      "learning_rate": 4.948573301373107e-05,
      "loss": 0.5057,
      "step": 1514
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0191048688819986,
      "learning_rate": 4.948473156189778e-05,
      "loss": 0.0995,
      "step": 1515
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9961167584484275,
      "learning_rate": 4.948372914608306e-05,
      "loss": 0.0985,
      "step": 1516
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0175344737955423,
      "learning_rate": 4.9482725766326376e-05,
      "loss": 0.1177,
      "step": 1517
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1299859394168539,
      "learning_rate": 4.9481721422667227e-05,
      "loss": 0.1965,
      "step": 1518
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.514756605711359,
      "learning_rate": 4.948071611514516e-05,
      "loss": 0.2033,
      "step": 1519
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.720652538833679,
      "learning_rate": 4.947970984379977e-05,
      "loss": 0.7181,
      "step": 1520
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.991624676255742,
      "learning_rate": 4.947870260867065e-05,
      "loss": 0.6643,
      "step": 1521
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6240375754501686,
      "learning_rate": 4.947769440979746e-05,
      "loss": 0.3219,
      "step": 1522
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0517329434000573,
      "learning_rate": 4.94766852472199e-05,
      "loss": 0.218,
      "step": 1523
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.1056710794919993,
      "learning_rate": 4.947567512097771e-05,
      "loss": 0.5023,
      "step": 1524
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3769289233456006,
      "learning_rate": 4.947466403111064e-05,
      "loss": 0.3591,
      "step": 1525
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.144723866693252,
      "learning_rate": 4.9473651977658506e-05,
      "loss": 0.1022,
      "step": 1526
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0337660467060092,
      "learning_rate": 4.947263896066116e-05,
      "loss": 0.476,
      "step": 1527
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.7120863870167318,
      "learning_rate": 4.9471624980158474e-05,
      "loss": 0.4955,
      "step": 1528
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2273982056477728,
      "learning_rate": 4.947061003619038e-05,
      "loss": 0.1628,
      "step": 1529
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.051656824462093,
      "learning_rate": 4.946959412879682e-05,
      "loss": 0.1109,
      "step": 1530
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2686410977726976,
      "learning_rate": 4.9468577258017816e-05,
      "loss": 0.196,
      "step": 1531
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4251700368222706,
      "learning_rate": 4.9467559423893385e-05,
      "loss": 0.161,
      "step": 1532
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6562303618652812,
      "learning_rate": 4.9466540626463604e-05,
      "loss": 0.4471,
      "step": 1533
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.379306709397865,
      "learning_rate": 4.946552086576858e-05,
      "loss": 0.244,
      "step": 1534
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1733685275238412,
      "learning_rate": 4.946450014184847e-05,
      "loss": 0.1669,
      "step": 1535
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3890494232511454,
      "learning_rate": 4.946347845474345e-05,
      "loss": 0.2273,
      "step": 1536
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6669269764390262,
      "learning_rate": 4.946245580449376e-05,
      "loss": 0.2838,
      "step": 1537
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4877605445626971,
      "learning_rate": 4.9461432191139654e-05,
      "loss": 0.1549,
      "step": 1538
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.426574786538839,
      "learning_rate": 4.9460407614721426e-05,
      "loss": 0.1634,
      "step": 1539
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8633836868967809,
      "learning_rate": 4.945938207527942e-05,
      "loss": 0.1415,
      "step": 1540
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.071541037650824,
      "learning_rate": 4.945835557285402e-05,
      "loss": 1.0474,
      "step": 1541
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0049949699859941,
      "learning_rate": 4.945732810748562e-05,
      "loss": 0.1909,
      "step": 1542
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7807926039851742,
      "learning_rate": 4.9456299679214694e-05,
      "loss": 0.4012,
      "step": 1543
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3329445347813396,
      "learning_rate": 4.945527028808171e-05,
      "loss": 0.2425,
      "step": 1544
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6604554064529842,
      "learning_rate": 4.9454239934127225e-05,
      "loss": 0.2668,
      "step": 1545
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4522456468284266,
      "learning_rate": 4.9453208617391776e-05,
      "loss": 0.2101,
      "step": 1546
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5440448236049906,
      "learning_rate": 4.945217633791598e-05,
      "loss": 0.273,
      "step": 1547
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.145887054936929,
      "learning_rate": 4.945114309574048e-05,
      "loss": 0.1401,
      "step": 1548
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.74858589300868,
      "learning_rate": 4.945010889090595e-05,
      "loss": 0.3177,
      "step": 1549
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9017066613715343,
      "learning_rate": 4.9449073723453124e-05,
      "loss": 0.1274,
      "step": 1550
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0438888798815196,
      "learning_rate": 4.9448037593422724e-05,
      "loss": 0.1652,
      "step": 1551
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0366885741848606,
      "learning_rate": 4.944700050085558e-05,
      "loss": 0.171,
      "step": 1552
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7883432150398971,
      "learning_rate": 4.9445962445792495e-05,
      "loss": 0.1201,
      "step": 1553
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7396106796788466,
      "learning_rate": 4.944492342827435e-05,
      "loss": 0.0917,
      "step": 1554
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.030054045471028,
      "learning_rate": 4.944388344834205e-05,
      "loss": 0.1476,
      "step": 1555
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7201585000733737,
      "learning_rate": 4.944284250603654e-05,
      "loss": 0.2511,
      "step": 1556
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3212915927448916,
      "learning_rate": 4.9441800601398805e-05,
      "loss": 0.2216,
      "step": 1557
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3893309738802522,
      "learning_rate": 4.944075773446987e-05,
      "loss": 0.2513,
      "step": 1558
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6148094888168276,
      "learning_rate": 4.943971390529077e-05,
      "loss": 0.2474,
      "step": 1559
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2492779651587098,
      "learning_rate": 4.9438669113902635e-05,
      "loss": 0.1449,
      "step": 1560
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0866203675743396,
      "learning_rate": 4.943762336034657e-05,
      "loss": 0.138,
      "step": 1561
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7326026755902824,
      "learning_rate": 4.943657664466377e-05,
      "loss": 0.2967,
      "step": 1562
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3958571040428382,
      "learning_rate": 4.943552896689543e-05,
      "loss": 0.2207,
      "step": 1563
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3202470705696272,
      "learning_rate": 4.94344803270828e-05,
      "loss": 0.2201,
      "step": 1564
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4892020732093314,
      "learning_rate": 4.943343072526717e-05,
      "loss": 0.258,
      "step": 1565
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.010492541604253,
      "learning_rate": 4.9432380161489864e-05,
      "loss": 0.1863,
      "step": 1566
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0788018397790085,
      "learning_rate": 4.943132863579223e-05,
      "loss": 0.1304,
      "step": 1567
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.970151050608334,
      "learning_rate": 4.943027614821569e-05,
      "loss": 0.2368,
      "step": 1568
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3420690778597641,
      "learning_rate": 4.942922269880167e-05,
      "loss": 0.1228,
      "step": 1569
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4159228115676359,
      "learning_rate": 4.9428168287591634e-05,
      "loss": 0.3168,
      "step": 1570
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.157929111737107,
      "learning_rate": 4.942711291462712e-05,
      "loss": 0.3066,
      "step": 1571
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.282823271208563,
      "learning_rate": 4.942605657994965e-05,
      "loss": 0.1552,
      "step": 1572
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9074390950092471,
      "learning_rate": 4.942499928360084e-05,
      "loss": 0.1499,
      "step": 1573
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6049078079380414,
      "learning_rate": 4.942394102562229e-05,
      "loss": 0.304,
      "step": 1574
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3909098225834011,
      "learning_rate": 4.942288180605569e-05,
      "loss": 0.204,
      "step": 1575
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0149706963896687,
      "learning_rate": 4.9421821624942724e-05,
      "loss": 0.224,
      "step": 1576
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1575094436038906,
      "learning_rate": 4.9420760482325134e-05,
      "loss": 0.1738,
      "step": 1577
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.098338398307798,
      "learning_rate": 4.94196983782447e-05,
      "loss": 0.1815,
      "step": 1578
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2115010568775515,
      "learning_rate": 4.941863531274326e-05,
      "loss": 0.2029,
      "step": 1579
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8257536952850172,
      "learning_rate": 4.941757128586263e-05,
      "loss": 0.1196,
      "step": 1580
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2444701052658778,
      "learning_rate": 4.941650629764472e-05,
      "loss": 0.1978,
      "step": 1581
    },
    {
      "epoch": 0.1,
      "grad_norm": 11.75800416598325,
      "learning_rate": 4.941544034813147e-05,
      "loss": 1.0636,
      "step": 1582
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3670048605379561,
      "learning_rate": 4.941437343736482e-05,
      "loss": 0.2141,
      "step": 1583
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.7361726231738484,
      "learning_rate": 4.94133055653868e-05,
      "loss": 0.3287,
      "step": 1584
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1788067652467138,
      "learning_rate": 4.941223673223944e-05,
      "loss": 0.2008,
      "step": 1585
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2536787561082123,
      "learning_rate": 4.941116693796483e-05,
      "loss": 0.204,
      "step": 1586
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.5303959584539766,
      "learning_rate": 4.941009618260507e-05,
      "loss": 0.7107,
      "step": 1587
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5352639532578447,
      "learning_rate": 4.940902446620233e-05,
      "loss": 0.3236,
      "step": 1588
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8529391227476534,
      "learning_rate": 4.940795178879882e-05,
      "loss": 0.1511,
      "step": 1589
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8120053327478307,
      "learning_rate": 4.940687815043675e-05,
      "loss": 0.3296,
      "step": 1590
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1550030780372167,
      "learning_rate": 4.940580355115839e-05,
      "loss": 0.1567,
      "step": 1591
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1776294886245309,
      "learning_rate": 4.940472799100605e-05,
      "loss": 0.2136,
      "step": 1592
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6913357279086016,
      "learning_rate": 4.940365147002208e-05,
      "loss": 0.2709,
      "step": 1593
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.513886216648881,
      "learning_rate": 4.940257398824887e-05,
      "loss": 0.2,
      "step": 1594
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4083764129626433,
      "learning_rate": 4.940149554572883e-05,
      "loss": 0.4522,
      "step": 1595
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6773010307817757,
      "learning_rate": 4.940041614250442e-05,
      "loss": 0.3183,
      "step": 1596
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6914404482088716,
      "learning_rate": 4.9399335778618146e-05,
      "loss": 0.2136,
      "step": 1597
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0460189609844786,
      "learning_rate": 4.939825445411253e-05,
      "loss": 0.1655,
      "step": 1598
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4631686524397411,
      "learning_rate": 4.9397172169030146e-05,
      "loss": 0.2024,
      "step": 1599
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.942386478253508,
      "learning_rate": 4.939608892341362e-05,
      "loss": 0.4684,
      "step": 1600
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1769500453198973,
      "learning_rate": 4.9395004717305584e-05,
      "loss": 0.2593,
      "step": 1601
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.762958217734549,
      "learning_rate": 4.9393919550748724e-05,
      "loss": 0.733,
      "step": 1602
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7961206928158961,
      "learning_rate": 4.939283342378578e-05,
      "loss": 0.2322,
      "step": 1603
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1447991275984677,
      "learning_rate": 4.939174633645949e-05,
      "loss": 0.1649,
      "step": 1604
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6983936168643643,
      "learning_rate": 4.9390658288812675e-05,
      "loss": 0.3121,
      "step": 1605
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.786875748607846,
      "learning_rate": 4.938956928088816e-05,
      "loss": 0.3837,
      "step": 1606
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.341577577922953,
      "learning_rate": 4.938847931272882e-05,
      "loss": 0.7979,
      "step": 1607
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.981556433962132,
      "learning_rate": 4.9387388384377576e-05,
      "loss": 0.1118,
      "step": 1608
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0824549958127963,
      "learning_rate": 4.9386296495877374e-05,
      "loss": 0.1549,
      "step": 1609
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4150134356822355,
      "learning_rate": 4.9385203647271195e-05,
      "loss": 0.2563,
      "step": 1610
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.904116523223936,
      "learning_rate": 4.9384109838602074e-05,
      "loss": 0.2771,
      "step": 1611
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0756511139066656,
      "learning_rate": 4.9383015069913075e-05,
      "loss": 0.463,
      "step": 1612
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1642175833758965,
      "learning_rate": 4.9381919341247295e-05,
      "loss": 0.1171,
      "step": 1613
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.024450144314874,
      "learning_rate": 4.9380822652647885e-05,
      "loss": 0.1476,
      "step": 1614
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0019645615339983,
      "learning_rate": 4.937972500415801e-05,
      "loss": 0.1062,
      "step": 1615
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9988466626336976,
      "learning_rate": 4.937862639582088e-05,
      "loss": 0.1023,
      "step": 1616
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0856745346572612,
      "learning_rate": 4.9377526827679766e-05,
      "loss": 0.0996,
      "step": 1617
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.239589224100096,
      "learning_rate": 4.937642629977795e-05,
      "loss": 0.2066,
      "step": 1618
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4610873460293,
      "learning_rate": 4.9375324812158765e-05,
      "loss": 0.173,
      "step": 1619
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2748624096136691,
      "learning_rate": 4.937422236486556e-05,
      "loss": 0.1187,
      "step": 1620
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2641301618208076,
      "learning_rate": 4.937311895794177e-05,
      "loss": 0.5646,
      "step": 1621
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7884026384945044,
      "learning_rate": 4.937201459143081e-05,
      "loss": 0.3639,
      "step": 1622
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0063378910763066,
      "learning_rate": 4.937090926537616e-05,
      "loss": 0.1776,
      "step": 1623
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9031186405325171,
      "learning_rate": 4.9369802979821355e-05,
      "loss": 0.3358,
      "step": 1624
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.005239268523047,
      "learning_rate": 4.936869573480994e-05,
      "loss": 0.1435,
      "step": 1625
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9873171454001552,
      "learning_rate": 4.9367587530385505e-05,
      "loss": 0.3234,
      "step": 1626
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.10751837677073,
      "learning_rate": 4.9366478366591694e-05,
      "loss": 0.4854,
      "step": 1627
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.764808114439875,
      "learning_rate": 4.9365368243472166e-05,
      "loss": 0.3837,
      "step": 1628
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9986718362602094,
      "learning_rate": 4.936425716107062e-05,
      "loss": 0.1629,
      "step": 1629
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.960392316584048,
      "learning_rate": 4.936314511943082e-05,
      "loss": 0.1536,
      "step": 1630
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4279164107879803,
      "learning_rate": 4.9362032118596526e-05,
      "loss": 0.2602,
      "step": 1631
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2796868326153656,
      "learning_rate": 4.936091815861157e-05,
      "loss": 0.2051,
      "step": 1632
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3548745013679746,
      "learning_rate": 4.9359803239519806e-05,
      "loss": 0.235,
      "step": 1633
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4826860319489226,
      "learning_rate": 4.9358687361365136e-05,
      "loss": 0.4417,
      "step": 1634
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3521524775777491,
      "learning_rate": 4.935757052419149e-05,
      "loss": 0.224,
      "step": 1635
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.6265241096240426,
      "learning_rate": 4.9356452728042826e-05,
      "loss": 0.8915,
      "step": 1636
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9231072568576103,
      "learning_rate": 4.935533397296317e-05,
      "loss": 0.1183,
      "step": 1637
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1317149032123728,
      "learning_rate": 4.9354214258996554e-05,
      "loss": 0.1969,
      "step": 1638
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1217078992832268,
      "learning_rate": 4.935309358618707e-05,
      "loss": 0.151,
      "step": 1639
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3290007704085227,
      "learning_rate": 4.935197195457885e-05,
      "loss": 0.1217,
      "step": 1640
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9336036567885855,
      "learning_rate": 4.9350849364216025e-05,
      "loss": 0.2617,
      "step": 1641
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.4120070019408644,
      "learning_rate": 4.9349725815142815e-05,
      "loss": 0.5133,
      "step": 1642
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8959227557718366,
      "learning_rate": 4.9348601307403456e-05,
      "loss": 0.396,
      "step": 1643
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9699996537742668,
      "learning_rate": 4.93474758410422e-05,
      "loss": 0.1376,
      "step": 1644
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9934753286667865,
      "learning_rate": 4.9346349416103376e-05,
      "loss": 0.1934,
      "step": 1645
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.994941305878683,
      "learning_rate": 4.934522203263133e-05,
      "loss": 0.1092,
      "step": 1646
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.914009136607483,
      "learning_rate": 4.934409369067044e-05,
      "loss": 0.4512,
      "step": 1647
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5560088920006945,
      "learning_rate": 4.9342964390265145e-05,
      "loss": 0.5471,
      "step": 1648
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1165839929167853,
      "learning_rate": 4.934183413145989e-05,
      "loss": 0.2022,
      "step": 1649
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9409463006000899,
      "learning_rate": 4.934070291429918e-05,
      "loss": 0.2159,
      "step": 1650
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1452033881097148,
      "learning_rate": 4.9339570738827553e-05,
      "loss": 0.1212,
      "step": 1651
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0387217642198858,
      "learning_rate": 4.933843760508958e-05,
      "loss": 0.1273,
      "step": 1652
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.784308752432407,
      "learning_rate": 4.933730351312988e-05,
      "loss": 0.5688,
      "step": 1653
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9625284001312705,
      "learning_rate": 4.933616846299309e-05,
      "loss": 0.1787,
      "step": 1654
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1499238109069603,
      "learning_rate": 4.933503245472392e-05,
      "loss": 0.1678,
      "step": 1655
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8167966747547537,
      "learning_rate": 4.933389548836707e-05,
      "loss": 0.3709,
      "step": 1656
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5508809557341412,
      "learning_rate": 4.933275756396732e-05,
      "loss": 0.2239,
      "step": 1657
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.420724556536571,
      "learning_rate": 4.933161868156947e-05,
      "loss": 0.3052,
      "step": 1658
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1471300749084923,
      "learning_rate": 4.933047884121835e-05,
      "loss": 0.1481,
      "step": 1659
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9413474757267073,
      "learning_rate": 4.932933804295884e-05,
      "loss": 0.1265,
      "step": 1660
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2789401335122867,
      "learning_rate": 4.932819628683585e-05,
      "loss": 0.1514,
      "step": 1661
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0619433922538084,
      "learning_rate": 4.932705357289434e-05,
      "loss": 0.1328,
      "step": 1662
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8446606583813403,
      "learning_rate": 4.9325909901179304e-05,
      "loss": 0.3389,
      "step": 1663
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2031038187748548,
      "learning_rate": 4.932476527173576e-05,
      "loss": 0.2261,
      "step": 1664
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.771346494337497,
      "learning_rate": 4.932361968460877e-05,
      "loss": 0.3384,
      "step": 1665
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3017913979782352,
      "learning_rate": 4.932247313984344e-05,
      "loss": 0.2391,
      "step": 1666
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5213296007307937,
      "learning_rate": 4.932132563748492e-05,
      "loss": 0.6995,
      "step": 1667
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8749838055521671,
      "learning_rate": 4.932017717757836e-05,
      "loss": 0.0988,
      "step": 1668
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8469558196906415,
      "learning_rate": 4.9319027760169014e-05,
      "loss": 0.4028,
      "step": 1669
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0393931360151436,
      "learning_rate": 4.931787738530211e-05,
      "loss": 0.1808,
      "step": 1670
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8335320325700573,
      "learning_rate": 4.931672605302294e-05,
      "loss": 0.0715,
      "step": 1671
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1939668136823387,
      "learning_rate": 4.931557376337684e-05,
      "loss": 0.1613,
      "step": 1672
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.616929151738316,
      "learning_rate": 4.9314420516409186e-05,
      "loss": 0.2726,
      "step": 1673
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0130919232520472,
      "learning_rate": 4.931326631216536e-05,
      "loss": 0.1547,
      "step": 1674
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0713548064220153,
      "learning_rate": 4.931211115069082e-05,
      "loss": 0.1579,
      "step": 1675
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.319935814429877,
      "learning_rate": 4.931095503203104e-05,
      "loss": 0.1625,
      "step": 1676
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1682164952402254,
      "learning_rate": 4.930979795623153e-05,
      "loss": 0.1685,
      "step": 1677
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3409995525942187,
      "learning_rate": 4.9308639923337854e-05,
      "loss": 0.1687,
      "step": 1678
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1274429605859841,
      "learning_rate": 4.930748093339561e-05,
      "loss": 0.2149,
      "step": 1679
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.022853262701058,
      "learning_rate": 4.930632098645041e-05,
      "loss": 0.1536,
      "step": 1680
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3503058270178898,
      "learning_rate": 4.930516008254794e-05,
      "loss": 0.1934,
      "step": 1681
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.081236395540778,
      "learning_rate": 4.93039982217339e-05,
      "loss": 0.1249,
      "step": 1682
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.083531937550976,
      "learning_rate": 4.930283540405403e-05,
      "loss": 0.1687,
      "step": 1683
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9361958829968801,
      "learning_rate": 4.930167162955411e-05,
      "loss": 0.1034,
      "step": 1684
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1204989814924247,
      "learning_rate": 4.930050689827996e-05,
      "loss": 0.1515,
      "step": 1685
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.03870837689166,
      "learning_rate": 4.929934121027745e-05,
      "loss": 0.4744,
      "step": 1686
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.264772730784084,
      "learning_rate": 4.929817456559245e-05,
      "loss": 0.785,
      "step": 1687
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.2867170696075,
      "learning_rate": 4.929700696427091e-05,
      "loss": 0.7011,
      "step": 1688
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8805544446199836,
      "learning_rate": 4.929583840635879e-05,
      "loss": 0.3275,
      "step": 1689
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1697073944913159,
      "learning_rate": 4.92946688919021e-05,
      "loss": 0.236,
      "step": 1690
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9522683044929854,
      "learning_rate": 4.929349842094688e-05,
      "loss": 0.1263,
      "step": 1691
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5939365199628475,
      "learning_rate": 4.929232699353923e-05,
      "loss": 0.2092,
      "step": 1692
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1150266041755934,
      "learning_rate": 4.929115460972524e-05,
      "loss": 0.5074,
      "step": 1693
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0983254184127629,
      "learning_rate": 4.92899812695511e-05,
      "loss": 0.2811,
      "step": 1694
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3110919230262508,
      "learning_rate": 4.928880697306297e-05,
      "loss": 0.2606,
      "step": 1695
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.888971940324134,
      "learning_rate": 4.928763172030712e-05,
      "loss": 0.1203,
      "step": 1696
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2968132963897878,
      "learning_rate": 4.9286455511329796e-05,
      "loss": 0.2334,
      "step": 1697
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.421989569118189,
      "learning_rate": 4.928527834617731e-05,
      "loss": 0.2395,
      "step": 1698
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1297340287215976,
      "learning_rate": 4.928410022489602e-05,
      "loss": 0.1578,
      "step": 1699
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.352218751544378,
      "learning_rate": 4.928292114753229e-05,
      "loss": 0.2309,
      "step": 1700
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0613672187387546,
      "learning_rate": 4.9281741114132565e-05,
      "loss": 0.0983,
      "step": 1701
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6308519698719066,
      "learning_rate": 4.9280560124743275e-05,
      "loss": 0.1274,
      "step": 1702
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9115461186281772,
      "learning_rate": 4.927937817941094e-05,
      "loss": 0.417,
      "step": 1703
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1274605596701344,
      "learning_rate": 4.9278195278182084e-05,
      "loss": 0.206,
      "step": 1704
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3667564095404425,
      "learning_rate": 4.927701142110328e-05,
      "loss": 0.2851,
      "step": 1705
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7887990512589613,
      "learning_rate": 4.927582660822113e-05,
      "loss": 0.4849,
      "step": 1706
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.9161674012659353,
      "learning_rate": 4.92746408395823e-05,
      "loss": 0.4272,
      "step": 1707
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7470686188377746,
      "learning_rate": 4.927345411523346e-05,
      "loss": 0.4187,
      "step": 1708
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5591181393580151,
      "learning_rate": 4.927226643522134e-05,
      "loss": 0.3379,
      "step": 1709
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0180190882616693,
      "learning_rate": 4.9271077799592685e-05,
      "loss": 0.1915,
      "step": 1710
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3628959544437682,
      "learning_rate": 4.926988820839431e-05,
      "loss": 0.3427,
      "step": 1711
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.313852498125834,
      "learning_rate": 4.926869766167304e-05,
      "loss": 0.68,
      "step": 1712
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1967426627149818,
      "learning_rate": 4.926750615947575e-05,
      "loss": 0.2829,
      "step": 1713
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.263756182158462,
      "learning_rate": 4.926631370184935e-05,
      "loss": 0.6153,
      "step": 1714
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0684997926830193,
      "learning_rate": 4.926512028884079e-05,
      "loss": 0.1338,
      "step": 1715
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.375956235038261,
      "learning_rate": 4.926392592049705e-05,
      "loss": 0.2646,
      "step": 1716
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8430182639938189,
      "learning_rate": 4.926273059686516e-05,
      "loss": 0.3924,
      "step": 1717
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1072075547257572,
      "learning_rate": 4.926153431799218e-05,
      "loss": 0.1595,
      "step": 1718
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.80030515553084,
      "learning_rate": 4.92603370839252e-05,
      "loss": 0.1285,
      "step": 1719
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9758117740024469,
      "learning_rate": 4.925913889471137e-05,
      "loss": 0.2203,
      "step": 1720
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9188676600157808,
      "learning_rate": 4.925793975039785e-05,
      "loss": 0.1128,
      "step": 1721
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2142200590191377,
      "learning_rate": 4.925673965103187e-05,
      "loss": 0.2298,
      "step": 1722
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.171703392197084,
      "learning_rate": 4.925553859666065e-05,
      "loss": 0.1593,
      "step": 1723
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8274354399095961,
      "learning_rate": 4.925433658733149e-05,
      "loss": 0.1055,
      "step": 1724
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1983342252216296,
      "learning_rate": 4.925313362309173e-05,
      "loss": 0.2788,
      "step": 1725
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.2505791761707887,
      "learning_rate": 4.925192970398871e-05,
      "loss": 0.5022,
      "step": 1726
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0572625745021693,
      "learning_rate": 4.925072483006984e-05,
      "loss": 0.1428,
      "step": 1727
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2164261861893315,
      "learning_rate": 4.924951900138255e-05,
      "loss": 0.1537,
      "step": 1728
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0408810219521625,
      "learning_rate": 4.924831221797433e-05,
      "loss": 0.1999,
      "step": 1729
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.117352729074319,
      "learning_rate": 4.924710447989266e-05,
      "loss": 0.3663,
      "step": 1730
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9963432197631333,
      "learning_rate": 4.924589578718512e-05,
      "loss": 0.1259,
      "step": 1731
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3491096897332224,
      "learning_rate": 4.924468613989929e-05,
      "loss": 0.2245,
      "step": 1732
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1772302947544775,
      "learning_rate": 4.9243475538082784e-05,
      "loss": 0.2332,
      "step": 1733
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.957103018217962,
      "learning_rate": 4.924226398178327e-05,
      "loss": 0.1703,
      "step": 1734
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5885861050931955,
      "learning_rate": 4.924105147104846e-05,
      "loss": 0.2787,
      "step": 1735
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0147153280914347,
      "learning_rate": 4.9239838005926066e-05,
      "loss": 0.4779,
      "step": 1736
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9740944705984098,
      "learning_rate": 4.9238623586463884e-05,
      "loss": 0.1367,
      "step": 1737
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1046732924979972,
      "learning_rate": 4.9237408212709725e-05,
      "loss": 0.187,
      "step": 1738
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3016038843917968,
      "learning_rate": 4.923619188471143e-05,
      "loss": 0.234,
      "step": 1739
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.057235009784906,
      "learning_rate": 4.923497460251689e-05,
      "loss": 0.1765,
      "step": 1740
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4492016555129035,
      "learning_rate": 4.923375636617403e-05,
      "loss": 0.215,
      "step": 1741
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3240609361568125,
      "learning_rate": 4.9232537175730816e-05,
      "loss": 0.3008,
      "step": 1742
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7487918775152973,
      "learning_rate": 4.923131703123526e-05,
      "loss": 0.3739,
      "step": 1743
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.189101327646949,
      "learning_rate": 4.923009593273536e-05,
      "loss": 0.1738,
      "step": 1744
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.175497256687372,
      "learning_rate": 4.922887388027924e-05,
      "loss": 0.1285,
      "step": 1745
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2482085427295613,
      "learning_rate": 4.922765087391499e-05,
      "loss": 0.2811,
      "step": 1746
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.250521257231786,
      "learning_rate": 4.922642691369075e-05,
      "loss": 0.2295,
      "step": 1747
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2144303245571944,
      "learning_rate": 4.9225201999654725e-05,
      "loss": 0.1777,
      "step": 1748
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3113827980774762,
      "learning_rate": 4.9223976131855135e-05,
      "loss": 0.2904,
      "step": 1749
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8511409484277999,
      "learning_rate": 4.922274931034024e-05,
      "loss": 0.3401,
      "step": 1750
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0637731211544765,
      "learning_rate": 4.922152153515836e-05,
      "loss": 0.7469,
      "step": 1751
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5048770146396668,
      "learning_rate": 4.922029280635781e-05,
      "loss": 0.2608,
      "step": 1752
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9643720593830695,
      "learning_rate": 4.9219063123986965e-05,
      "loss": 0.1476,
      "step": 1753
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0409873474838272,
      "learning_rate": 4.921783248809425e-05,
      "loss": 0.2705,
      "step": 1754
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1512470150303504,
      "learning_rate": 4.921660089872812e-05,
      "loss": 0.301,
      "step": 1755
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7634718819442828,
      "learning_rate": 4.921536835593706e-05,
      "loss": 0.1293,
      "step": 1756
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9642512150962009,
      "learning_rate": 4.921413485976959e-05,
      "loss": 0.1321,
      "step": 1757
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.290417385060622,
      "learning_rate": 4.921290041027427e-05,
      "loss": 0.3414,
      "step": 1758
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.184632789342789,
      "learning_rate": 4.921166500749972e-05,
      "loss": 1.7461,
      "step": 1759
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.677189398513952,
      "learning_rate": 4.9210428651494553e-05,
      "loss": 0.3621,
      "step": 1760
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.991051122786479,
      "learning_rate": 4.920919134230747e-05,
      "loss": 0.1128,
      "step": 1761
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.518915318451536,
      "learning_rate": 4.920795307998717e-05,
      "loss": 0.1248,
      "step": 1762
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.876215179276024,
      "learning_rate": 4.920671386458241e-05,
      "loss": 0.1436,
      "step": 1763
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0552738678177078,
      "learning_rate": 4.920547369614197e-05,
      "loss": 0.1288,
      "step": 1764
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.500593629629617,
      "learning_rate": 4.9204232574714694e-05,
      "loss": 0.2352,
      "step": 1765
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.220951681744721,
      "learning_rate": 4.9202990500349425e-05,
      "loss": 0.1643,
      "step": 1766
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3995766267862353,
      "learning_rate": 4.920174747309508e-05,
      "loss": 0.2908,
      "step": 1767
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8443198476072107,
      "learning_rate": 4.920050349300059e-05,
      "loss": 0.3053,
      "step": 1768
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.073999126483501,
      "learning_rate": 4.9199258560114936e-05,
      "loss": 0.1921,
      "step": 1769
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0004522512441711,
      "learning_rate": 4.919801267448712e-05,
      "loss": 0.0793,
      "step": 1770
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6823306750971254,
      "learning_rate": 4.919676583616621e-05,
      "loss": 0.2584,
      "step": 1771
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9140747552747124,
      "learning_rate": 4.919551804520129e-05,
      "loss": 0.0932,
      "step": 1772
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2702283577826303,
      "learning_rate": 4.919426930164149e-05,
      "loss": 0.2288,
      "step": 1773
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.793593729269314,
      "learning_rate": 4.919301960553596e-05,
      "loss": 0.3445,
      "step": 1774
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2214988263961442,
      "learning_rate": 4.919176895693391e-05,
      "loss": 0.2179,
      "step": 1775
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0961549048609163,
      "learning_rate": 4.919051735588458e-05,
      "loss": 0.1276,
      "step": 1776
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.130228622062317,
      "learning_rate": 4.918926480243724e-05,
      "loss": 0.1481,
      "step": 1777
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.441278985486559,
      "learning_rate": 4.918801129664121e-05,
      "loss": 0.3181,
      "step": 1778
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1122569465732441,
      "learning_rate": 4.918675683854585e-05,
      "loss": 0.1651,
      "step": 1779
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2634439852055324,
      "learning_rate": 4.918550142820053e-05,
      "loss": 0.2108,
      "step": 1780
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.175083950521469,
      "learning_rate": 4.918424506565469e-05,
      "loss": 0.4037,
      "step": 1781
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5144002329060167,
      "learning_rate": 4.9182987750957786e-05,
      "loss": 0.2966,
      "step": 1782
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.031053424592771,
      "learning_rate": 4.918172948415932e-05,
      "loss": 0.4726,
      "step": 1783
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2729823299325154,
      "learning_rate": 4.9180470265308845e-05,
      "loss": 0.2846,
      "step": 1784
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1457567268344162,
      "learning_rate": 4.9179210094455916e-05,
      "loss": 0.149,
      "step": 1785
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2081680223420688,
      "learning_rate": 4.917794897165017e-05,
      "loss": 0.551,
      "step": 1786
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4765921830378206,
      "learning_rate": 4.917668689694123e-05,
      "loss": 0.2265,
      "step": 1787
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.581700012335117,
      "learning_rate": 4.91754238703788e-05,
      "loss": 0.196,
      "step": 1788
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.360039692429743,
      "learning_rate": 4.9174159892012615e-05,
      "loss": 0.1476,
      "step": 1789
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7548163968974723,
      "learning_rate": 4.917289496189243e-05,
      "loss": 0.1062,
      "step": 1790
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.412804418657549,
      "learning_rate": 4.917162908006804e-05,
      "loss": 0.1913,
      "step": 1791
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4358480671131777,
      "learning_rate": 4.91703622465893e-05,
      "loss": 0.2731,
      "step": 1792
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5292500031068557,
      "learning_rate": 4.9169094461506064e-05,
      "loss": 0.2598,
      "step": 1793
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1406405559486958,
      "learning_rate": 4.916782572486827e-05,
      "loss": 0.1752,
      "step": 1794
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.318956070548706,
      "learning_rate": 4.916655603672584e-05,
      "loss": 0.2182,
      "step": 1795
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.418303963073898,
      "learning_rate": 4.916528539712879e-05,
      "loss": 0.6965,
      "step": 1796
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.9865695427190215,
      "learning_rate": 4.916401380612714e-05,
      "loss": 0.6383,
      "step": 1797
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.371834521232282,
      "learning_rate": 4.916274126377094e-05,
      "loss": 0.2587,
      "step": 1798
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9092618743748475,
      "learning_rate": 4.9161467770110304e-05,
      "loss": 0.1098,
      "step": 1799
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1626547712696782,
      "learning_rate": 4.916019332519536e-05,
      "loss": 0.128,
      "step": 1800
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0859153197949265,
      "learning_rate": 4.9158917929076295e-05,
      "loss": 0.0994,
      "step": 1801
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.811210927157997,
      "learning_rate": 4.9157641581803313e-05,
      "loss": 0.0836,
      "step": 1802
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2878753760750892,
      "learning_rate": 4.915636428342668e-05,
      "loss": 0.2289,
      "step": 1803
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.146128892559166,
      "learning_rate": 4.9155086033996665e-05,
      "loss": 0.1245,
      "step": 1804
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8430098325854473,
      "learning_rate": 4.91538068335636e-05,
      "loss": 0.1138,
      "step": 1805
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1542476384141827,
      "learning_rate": 4.915252668217785e-05,
      "loss": 0.1438,
      "step": 1806
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7971633595452214,
      "learning_rate": 4.9151245579889816e-05,
      "loss": 0.3404,
      "step": 1807
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4739640870939574,
      "learning_rate": 4.9149963526749936e-05,
      "loss": 0.2194,
      "step": 1808
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5278017890045221,
      "learning_rate": 4.914868052280869e-05,
      "loss": 0.2663,
      "step": 1809
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.8148879361820858,
      "learning_rate": 4.914739656811657e-05,
      "loss": 0.2891,
      "step": 1810
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9640530811593162,
      "learning_rate": 4.9146111662724145e-05,
      "loss": 0.1332,
      "step": 1811
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4313462826853216,
      "learning_rate": 4.914482580668201e-05,
      "loss": 0.2789,
      "step": 1812
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.257553310768324,
      "learning_rate": 4.9143539000040774e-05,
      "loss": 0.2527,
      "step": 1813
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8133598485480102,
      "learning_rate": 4.91422512428511e-05,
      "loss": 0.3941,
      "step": 1814
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.982064407170167,
      "learning_rate": 4.9140962535163695e-05,
      "loss": 1.0851,
      "step": 1815
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1364880699799305,
      "learning_rate": 4.9139672877029294e-05,
      "loss": 0.2042,
      "step": 1816
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.620057332761475,
      "learning_rate": 4.913838226849867e-05,
      "loss": 0.4146,
      "step": 1817
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3973939426678654,
      "learning_rate": 4.913709070962265e-05,
      "loss": 0.282,
      "step": 1818
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3519742740524712,
      "learning_rate": 4.9135798200452054e-05,
      "loss": 0.2586,
      "step": 1819
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0613110133794077,
      "learning_rate": 4.913450474103779e-05,
      "loss": 0.1433,
      "step": 1820
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7938720023817354,
      "learning_rate": 4.913321033143078e-05,
      "loss": 0.1259,
      "step": 1821
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5733541882991844,
      "learning_rate": 4.913191497168198e-05,
      "loss": 0.2925,
      "step": 1822
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0199898445079119,
      "learning_rate": 4.9130618661842396e-05,
      "loss": 0.084,
      "step": 1823
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8541241444216552,
      "learning_rate": 4.912932140196306e-05,
      "loss": 0.4276,
      "step": 1824
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6841052770388545,
      "learning_rate": 4.912802319209505e-05,
      "loss": 0.2412,
      "step": 1825
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9476541719715055,
      "learning_rate": 4.912672403228947e-05,
      "loss": 0.3046,
      "step": 1826
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1004887875081724,
      "learning_rate": 4.912542392259747e-05,
      "loss": 0.3615,
      "step": 1827
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9188714860550259,
      "learning_rate": 4.9124122863070255e-05,
      "loss": 0.1567,
      "step": 1828
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.32185423114952,
      "learning_rate": 4.9122820853759024e-05,
      "loss": 0.4405,
      "step": 1829
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1158263919717277,
      "learning_rate": 4.912151789471505e-05,
      "loss": 0.3696,
      "step": 1830
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.535591145623637,
      "learning_rate": 4.9120213985989624e-05,
      "loss": 0.2743,
      "step": 1831
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.34849319839235,
      "learning_rate": 4.9118909127634095e-05,
      "loss": 0.2919,
      "step": 1832
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0351867210558894,
      "learning_rate": 4.911760331969982e-05,
      "loss": 0.2168,
      "step": 1833
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3922840289439855,
      "learning_rate": 4.911629656223822e-05,
      "loss": 0.3926,
      "step": 1834
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4257694516150996,
      "learning_rate": 4.911498885530075e-05,
      "loss": 0.2814,
      "step": 1835
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6505450638384151,
      "learning_rate": 4.911368019893887e-05,
      "loss": 0.2799,
      "step": 1836
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7533039835443478,
      "learning_rate": 4.9112370593204135e-05,
      "loss": 0.2742,
      "step": 1837
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3177381208961207,
      "learning_rate": 4.911106003814808e-05,
      "loss": 0.3878,
      "step": 1838
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6905642407359736,
      "learning_rate": 4.9109748533822315e-05,
      "loss": 0.2977,
      "step": 1839
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1169320028857967,
      "learning_rate": 4.9108436080278467e-05,
      "loss": 0.1824,
      "step": 1840
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1653410291463067,
      "learning_rate": 4.910712267756822e-05,
      "loss": 0.2538,
      "step": 1841
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.16402828116375,
      "learning_rate": 4.9105808325743266e-05,
      "loss": 0.3399,
      "step": 1842
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5702266873232973,
      "learning_rate": 4.910449302485537e-05,
      "loss": 0.2043,
      "step": 1843
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1004339244400025,
      "learning_rate": 4.910317677495631e-05,
      "loss": 0.5181,
      "step": 1844
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8204712384969937,
      "learning_rate": 4.9101859576097905e-05,
      "loss": 0.4765,
      "step": 1845
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6368786692784834,
      "learning_rate": 4.9100541428332006e-05,
      "loss": 0.4756,
      "step": 1846
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.669564129282918,
      "learning_rate": 4.909922233171053e-05,
      "loss": 0.4007,
      "step": 1847
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3508670874793067,
      "learning_rate": 4.909790228628539e-05,
      "loss": 0.2036,
      "step": 1848
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9752762590875356,
      "learning_rate": 4.9096581292108577e-05,
      "loss": 0.1234,
      "step": 1849
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2698159438099093,
      "learning_rate": 4.909525934923208e-05,
      "loss": 0.3508,
      "step": 1850
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1610586450516396,
      "learning_rate": 4.909393645770796e-05,
      "loss": 0.2175,
      "step": 1851
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9032359845253575,
      "learning_rate": 4.9092612617588294e-05,
      "loss": 0.1342,
      "step": 1852
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.33571686621738,
      "learning_rate": 4.9091287828925205e-05,
      "loss": 0.2635,
      "step": 1853
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1687908752197087,
      "learning_rate": 4.908996209177084e-05,
      "loss": 0.2202,
      "step": 1854
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4823915902197207,
      "learning_rate": 4.908863540617741e-05,
      "loss": 0.3675,
      "step": 1855
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9350213798538802,
      "learning_rate": 4.908730777219713e-05,
      "loss": 0.1301,
      "step": 1856
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3941793868186811,
      "learning_rate": 4.908597918988229e-05,
      "loss": 0.3314,
      "step": 1857
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0046675077975327,
      "learning_rate": 4.908464965928519e-05,
      "loss": 0.0802,
      "step": 1858
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2197224757874834,
      "learning_rate": 4.9083319180458164e-05,
      "loss": 0.2092,
      "step": 1859
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.375617408091814,
      "learning_rate": 4.908198775345361e-05,
      "loss": 0.1648,
      "step": 1860
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3188652542722983,
      "learning_rate": 4.908065537832393e-05,
      "loss": 0.3042,
      "step": 1861
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8584546831730028,
      "learning_rate": 4.9079322055121594e-05,
      "loss": 0.1514,
      "step": 1862
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2663833095143358,
      "learning_rate": 4.9077987783899094e-05,
      "loss": 0.2986,
      "step": 1863
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8644422463657884,
      "learning_rate": 4.907665256470896e-05,
      "loss": 0.104,
      "step": 1864
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9121668553263789,
      "learning_rate": 4.9075316397603755e-05,
      "loss": 0.9855,
      "step": 1865
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9823351872309236,
      "learning_rate": 4.907397928263609e-05,
      "loss": 0.4526,
      "step": 1866
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8503653755922103,
      "learning_rate": 4.90726412198586e-05,
      "loss": 0.4487,
      "step": 1867
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2093190849894542,
      "learning_rate": 4.907130220932399e-05,
      "loss": 0.2879,
      "step": 1868
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0184851996197342,
      "learning_rate": 4.906996225108495e-05,
      "loss": 0.1521,
      "step": 1869
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1674475402530708,
      "learning_rate": 4.906862134519424e-05,
      "loss": 0.1334,
      "step": 1870
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7763555304064915,
      "learning_rate": 4.906727949170468e-05,
      "loss": 0.0867,
      "step": 1871
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9499794918569222,
      "learning_rate": 4.906593669066906e-05,
      "loss": 0.1193,
      "step": 1872
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.6891380574888175,
      "learning_rate": 4.906459294214026e-05,
      "loss": 0.6682,
      "step": 1873
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1878794733193705,
      "learning_rate": 4.90632482461712e-05,
      "loss": 0.2077,
      "step": 1874
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.259050240939776,
      "learning_rate": 4.906190260281481e-05,
      "loss": 0.2745,
      "step": 1875
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4053128790465272,
      "learning_rate": 4.906055601212407e-05,
      "loss": 0.2881,
      "step": 1876
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0776564174979573,
      "learning_rate": 4.9059208474151996e-05,
      "loss": 0.187,
      "step": 1877
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1871575441121114,
      "learning_rate": 4.9057859988951635e-05,
      "loss": 0.2238,
      "step": 1878
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2054526851088243,
      "learning_rate": 4.9056510556576094e-05,
      "loss": 0.2201,
      "step": 1879
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4153506655113803,
      "learning_rate": 4.9055160177078494e-05,
      "loss": 0.2692,
      "step": 1880
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4282553580384585,
      "learning_rate": 4.905380885051199e-05,
      "loss": 0.3967,
      "step": 1881
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9833198187394526,
      "learning_rate": 4.90524565769298e-05,
      "loss": 0.2452,
      "step": 1882
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2160107909046187,
      "learning_rate": 4.9051103356385145e-05,
      "loss": 0.159,
      "step": 1883
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0401234707880656,
      "learning_rate": 4.904974918893133e-05,
      "loss": 0.1762,
      "step": 1884
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.01569579930993,
      "learning_rate": 4.904839407462164e-05,
      "loss": 0.5244,
      "step": 1885
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.367506494109201,
      "learning_rate": 4.904703801350945e-05,
      "loss": 0.6106,
      "step": 1886
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.063123329960304,
      "learning_rate": 4.904568100564812e-05,
      "loss": 0.194,
      "step": 1887
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4057464549897962,
      "learning_rate": 4.904432305109112e-05,
      "loss": 0.2322,
      "step": 1888
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.155965220651428,
      "learning_rate": 4.904296414989188e-05,
      "loss": 0.1673,
      "step": 1889
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.487269323963028,
      "learning_rate": 4.904160430210391e-05,
      "loss": 0.5616,
      "step": 1890
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9266803445535313,
      "learning_rate": 4.904024350778075e-05,
      "loss": 0.135,
      "step": 1891
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3974265481796926,
      "learning_rate": 4.903888176697598e-05,
      "loss": 0.2657,
      "step": 1892
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3805084762345305,
      "learning_rate": 4.903751907974319e-05,
      "loss": 0.3591,
      "step": 1893
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2972655065278749,
      "learning_rate": 4.9036155446136055e-05,
      "loss": 0.2883,
      "step": 1894
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1369317013408493,
      "learning_rate": 4.9034790866208255e-05,
      "loss": 0.1514,
      "step": 1895
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8545432426751194,
      "learning_rate": 4.903342534001352e-05,
      "loss": 0.5323,
      "step": 1896
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0685753211194176,
      "learning_rate": 4.9032058867605595e-05,
      "loss": 0.24,
      "step": 1897
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0167597316362134,
      "learning_rate": 4.9030691449038295e-05,
      "loss": 0.1684,
      "step": 1898
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.097372186639514,
      "learning_rate": 4.9029323084365455e-05,
      "loss": 0.4507,
      "step": 1899
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7108072830947182,
      "learning_rate": 4.9027953773640934e-05,
      "loss": 0.0725,
      "step": 1900
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5041174551716536,
      "learning_rate": 4.902658351691866e-05,
      "loss": 0.2138,
      "step": 1901
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8912848981865793,
      "learning_rate": 4.902521231425257e-05,
      "loss": 0.1583,
      "step": 1902
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6832679494067918,
      "learning_rate": 4.902384016569665e-05,
      "loss": 0.0901,
      "step": 1903
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8718714544146269,
      "learning_rate": 4.902246707130493e-05,
      "loss": 0.1574,
      "step": 1904
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4448886807045522,
      "learning_rate": 4.902109303113146e-05,
      "loss": 0.2242,
      "step": 1905
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.035832957767392,
      "learning_rate": 4.901971804523034e-05,
      "loss": 0.1363,
      "step": 1906
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.93445775727053,
      "learning_rate": 4.901834211365572e-05,
      "loss": 0.1255,
      "step": 1907
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1603279282357397,
      "learning_rate": 4.901696523646175e-05,
      "loss": 0.2333,
      "step": 1908
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0817228884709487,
      "learning_rate": 4.9015587413702645e-05,
      "loss": 0.1198,
      "step": 1909
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8245687180993748,
      "learning_rate": 4.901420864543265e-05,
      "loss": 0.3734,
      "step": 1910
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.413943965641211,
      "learning_rate": 4.9012828931706056e-05,
      "loss": 0.1978,
      "step": 1911
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.8730562233743724,
      "learning_rate": 4.9011448272577175e-05,
      "loss": 0.5389,
      "step": 1912
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.964164869181759,
      "learning_rate": 4.901006666810037e-05,
      "loss": 0.1874,
      "step": 1913
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8596608228836888,
      "learning_rate": 4.900868411833003e-05,
      "loss": 0.1372,
      "step": 1914
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2032410381061025,
      "learning_rate": 4.900730062332058e-05,
      "loss": 0.2728,
      "step": 1915
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4704853532063142,
      "learning_rate": 4.900591618312652e-05,
      "loss": 0.3944,
      "step": 1916
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2028568028334932,
      "learning_rate": 4.900453079780232e-05,
      "loss": 0.2656,
      "step": 1917
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3762778782562963,
      "learning_rate": 4.9003144467402544e-05,
      "loss": 0.306,
      "step": 1918
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1127069769925482,
      "learning_rate": 4.900175719198177e-05,
      "loss": 0.1851,
      "step": 1919
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9900846241048313,
      "learning_rate": 4.900036897159461e-05,
      "loss": 0.1917,
      "step": 1920
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.269646948365611,
      "learning_rate": 4.8998979806295716e-05,
      "loss": 0.1968,
      "step": 1921
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0441326025635262,
      "learning_rate": 4.89975896961398e-05,
      "loss": 0.253,
      "step": 1922
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.823900762692339,
      "learning_rate": 4.899619864118158e-05,
      "loss": 0.2829,
      "step": 1923
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.19022344615097,
      "learning_rate": 4.8994806641475815e-05,
      "loss": 0.1089,
      "step": 1924
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0624588204695962,
      "learning_rate": 4.899341369707732e-05,
      "loss": 0.1432,
      "step": 1925
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0021579218634413,
      "learning_rate": 4.899201980804093e-05,
      "loss": 0.2342,
      "step": 1926
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4311194417106734,
      "learning_rate": 4.899062497442153e-05,
      "loss": 0.2442,
      "step": 1927
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2165742476476062,
      "learning_rate": 4.898922919627402e-05,
      "loss": 0.1917,
      "step": 1928
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.540874237377553,
      "learning_rate": 4.898783247365337e-05,
      "loss": 0.411,
      "step": 1929
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3361445162106982,
      "learning_rate": 4.898643480661457e-05,
      "loss": 0.1256,
      "step": 1930
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6496386693828102,
      "learning_rate": 4.898503619521264e-05,
      "loss": 0.413,
      "step": 1931
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3485573857385547,
      "learning_rate": 4.898363663950264e-05,
      "loss": 0.103,
      "step": 1932
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.525606188197916,
      "learning_rate": 4.898223613953968e-05,
      "loss": 0.6606,
      "step": 1933
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.82691762553622,
      "learning_rate": 4.898083469537889e-05,
      "loss": 0.1854,
      "step": 1934
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7554778782865945,
      "learning_rate": 4.897943230707546e-05,
      "loss": 0.5006,
      "step": 1935
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4595846134550472,
      "learning_rate": 4.897802897468459e-05,
      "loss": 0.1865,
      "step": 1936
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9783322477147879,
      "learning_rate": 4.897662469826154e-05,
      "loss": 0.2034,
      "step": 1937
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.490880214727145,
      "learning_rate": 4.897521947786158e-05,
      "loss": 0.1981,
      "step": 1938
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.139912335527559,
      "learning_rate": 4.8973813313540065e-05,
      "loss": 0.1934,
      "step": 1939
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.687174753632847,
      "learning_rate": 4.8972406205352325e-05,
      "loss": 0.8476,
      "step": 1940
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4434167339966903,
      "learning_rate": 4.8970998153353774e-05,
      "loss": 0.4152,
      "step": 1941
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2164328257809187,
      "learning_rate": 4.896958915759985e-05,
      "loss": 0.2125,
      "step": 1942
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.032563559913352,
      "learning_rate": 4.8968179218146016e-05,
      "loss": 0.1392,
      "step": 1943
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4309587208530279,
      "learning_rate": 4.89667683350478e-05,
      "loss": 0.259,
      "step": 1944
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6897149297964673,
      "learning_rate": 4.8965356508360725e-05,
      "loss": 0.3572,
      "step": 1945
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2760592095824146,
      "learning_rate": 4.8963943738140395e-05,
      "loss": 0.2058,
      "step": 1946
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1815670744038003,
      "learning_rate": 4.896253002444243e-05,
      "loss": 0.2314,
      "step": 1947
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9307233302245065,
      "learning_rate": 4.8961115367322474e-05,
      "loss": 0.1366,
      "step": 1948
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.945414847958867,
      "learning_rate": 4.895969976683624e-05,
      "loss": 0.1098,
      "step": 1949
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.254657200043174,
      "learning_rate": 4.895828322303946e-05,
      "loss": 0.1964,
      "step": 1950
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.239312432577765,
      "learning_rate": 4.895686573598789e-05,
      "loss": 0.2995,
      "step": 1951
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0245907342165808,
      "learning_rate": 4.895544730573736e-05,
      "loss": 0.1235,
      "step": 1952
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.7184751489238517,
      "learning_rate": 4.8954027932343686e-05,
      "loss": 0.6826,
      "step": 1953
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1744137239315053,
      "learning_rate": 4.895260761586277e-05,
      "loss": 0.2094,
      "step": 1954
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8426856596066157,
      "learning_rate": 4.895118635635053e-05,
      "loss": 0.2521,
      "step": 1955
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4501359914463976,
      "learning_rate": 4.894976415386292e-05,
      "loss": 0.2928,
      "step": 1956
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0598005349857162,
      "learning_rate": 4.894834100845592e-05,
      "loss": 0.1299,
      "step": 1957
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8064237489622855,
      "learning_rate": 4.894691692018558e-05,
      "loss": 0.0798,
      "step": 1958
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0809458065984971,
      "learning_rate": 4.894549188910795e-05,
      "loss": 0.2073,
      "step": 1959
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3316130423187784,
      "learning_rate": 4.894406591527916e-05,
      "loss": 0.1922,
      "step": 1960
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2576453025638645,
      "learning_rate": 4.894263899875532e-05,
      "loss": 0.2097,
      "step": 1961
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1891902026979946,
      "learning_rate": 4.894121113959263e-05,
      "loss": 0.1659,
      "step": 1962
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.194020780953523,
      "learning_rate": 4.8939782337847296e-05,
      "loss": 0.2149,
      "step": 1963
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2147190232678615,
      "learning_rate": 4.893835259357558e-05,
      "loss": 0.2242,
      "step": 1964
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0548085533916154,
      "learning_rate": 4.893692190683376e-05,
      "loss": 0.1334,
      "step": 1965
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3752116950395534,
      "learning_rate": 4.893549027767817e-05,
      "loss": 0.184,
      "step": 1966
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0011174315946734,
      "learning_rate": 4.893405770616517e-05,
      "loss": 0.1671,
      "step": 1967
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3456359482098883,
      "learning_rate": 4.893262419235117e-05,
      "loss": 0.296,
      "step": 1968
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8492383143133642,
      "learning_rate": 4.893118973629261e-05,
      "loss": 0.1147,
      "step": 1969
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.693838569256478,
      "learning_rate": 4.892975433804595e-05,
      "loss": 0.4508,
      "step": 1970
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0629710393291332,
      "learning_rate": 4.892831799766772e-05,
      "loss": 0.1762,
      "step": 1971
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.125020997384932,
      "learning_rate": 4.8926880715214454e-05,
      "loss": 0.2486,
      "step": 1972
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.21557211839818,
      "learning_rate": 4.8925442490742754e-05,
      "loss": 0.6276,
      "step": 1973
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.137249852133099,
      "learning_rate": 4.892400332430923e-05,
      "loss": 0.1138,
      "step": 1974
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9360175199932848,
      "learning_rate": 4.892256321597055e-05,
      "loss": 0.1635,
      "step": 1975
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7054547894451093,
      "learning_rate": 4.892112216578342e-05,
      "loss": 0.365,
      "step": 1976
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.13143343892693,
      "learning_rate": 4.891968017380455e-05,
      "loss": 0.228,
      "step": 1977
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9059781912273158,
      "learning_rate": 4.891823724009074e-05,
      "loss": 0.1266,
      "step": 1978
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2695790958032924,
      "learning_rate": 4.891679336469879e-05,
      "loss": 0.2251,
      "step": 1979
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9847141659918748,
      "learning_rate": 4.8915348547685534e-05,
      "loss": 0.4517,
      "step": 1980
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4030549135519552,
      "learning_rate": 4.891390278910788e-05,
      "loss": 0.4549,
      "step": 1981
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.9705070197365377,
      "learning_rate": 4.891245608902273e-05,
      "loss": 0.9263,
      "step": 1982
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.117443635866304,
      "learning_rate": 4.891100844748704e-05,
      "loss": 0.1351,
      "step": 1983
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8218429260680674,
      "learning_rate": 4.890955986455782e-05,
      "loss": 0.1302,
      "step": 1984
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.4834462941701205,
      "learning_rate": 4.8908110340292086e-05,
      "loss": 0.5223,
      "step": 1985
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9858598592131081,
      "learning_rate": 4.890665987474692e-05,
      "loss": 0.1972,
      "step": 1986
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0634242370576477,
      "learning_rate": 4.890520846797941e-05,
      "loss": 0.125,
      "step": 1987
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9861300305654473,
      "learning_rate": 4.890375612004672e-05,
      "loss": 0.1321,
      "step": 1988
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0514716388957484,
      "learning_rate": 4.890230283100602e-05,
      "loss": 0.1613,
      "step": 1989
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3580109190228193,
      "learning_rate": 4.890084860091452e-05,
      "loss": 0.2614,
      "step": 1990
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.098239551150053,
      "learning_rate": 4.889939342982949e-05,
      "loss": 0.1469,
      "step": 1991
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1542885564595524,
      "learning_rate": 4.8897937317808205e-05,
      "loss": 0.1808,
      "step": 1992
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9351094514504111,
      "learning_rate": 4.889648026490801e-05,
      "loss": 0.1182,
      "step": 1993
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3121516569785205,
      "learning_rate": 4.889502227118625e-05,
      "loss": 0.256,
      "step": 1994
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4160599802812777,
      "learning_rate": 4.889356333670035e-05,
      "loss": 0.2558,
      "step": 1995
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.4447258161730656,
      "learning_rate": 4.889210346150773e-05,
      "loss": 0.5613,
      "step": 1996
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1892820969359288,
      "learning_rate": 4.889064264566587e-05,
      "loss": 0.1536,
      "step": 1997
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1612410391819208,
      "learning_rate": 4.88891808892323e-05,
      "loss": 0.1825,
      "step": 1998
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9886528183953531,
      "learning_rate": 4.8887718192264545e-05,
      "loss": 0.1324,
      "step": 1999
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2090815088678204,
      "learning_rate": 4.8886254554820215e-05,
      "loss": 0.1748,
      "step": 2000
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.075248745861815,
      "learning_rate": 4.8884789976956925e-05,
      "loss": 0.3837,
      "step": 2001
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4933986816602773,
      "learning_rate": 4.888332445873233e-05,
      "loss": 0.3965,
      "step": 2002
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5500968280007634,
      "learning_rate": 4.888185800020413e-05,
      "loss": 0.39,
      "step": 2003
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4027914355366928,
      "learning_rate": 4.888039060143007e-05,
      "loss": 0.1837,
      "step": 2004
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1315461894496885,
      "learning_rate": 4.887892226246792e-05,
      "loss": 0.3441,
      "step": 2005
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1806637983822628,
      "learning_rate": 4.887745298337548e-05,
      "loss": 0.1913,
      "step": 2006
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2181428807990302,
      "learning_rate": 4.8875982764210616e-05,
      "loss": 0.3054,
      "step": 2007
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7575707796725613,
      "learning_rate": 4.8874511605031194e-05,
      "loss": 0.3688,
      "step": 2008
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8781830252183183,
      "learning_rate": 4.8873039505895135e-05,
      "loss": 0.0958,
      "step": 2009
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6884985389126113,
      "learning_rate": 4.8871566466860405e-05,
      "loss": 0.3959,
      "step": 2010
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2343844803611492,
      "learning_rate": 4.8870092487984995e-05,
      "loss": 0.2833,
      "step": 2011
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2630357429387498,
      "learning_rate": 4.8868617569326936e-05,
      "loss": 0.1961,
      "step": 2012
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9449817661377293,
      "learning_rate": 4.8867141710944294e-05,
      "loss": 0.1329,
      "step": 2013
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.188389041180782,
      "learning_rate": 4.8865664912895184e-05,
      "loss": 0.3357,
      "step": 2014
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0722889020591946,
      "learning_rate": 4.8864187175237744e-05,
      "loss": 0.1826,
      "step": 2015
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3017168717027536,
      "learning_rate": 4.886270849803014e-05,
      "loss": 0.2436,
      "step": 2016
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0252689238976422,
      "learning_rate": 4.8861228881330615e-05,
      "loss": 0.1462,
      "step": 2017
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.092484854018464,
      "learning_rate": 4.8859748325197396e-05,
      "loss": 0.2132,
      "step": 2018
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.28076318988425,
      "learning_rate": 4.8858266829688795e-05,
      "loss": 0.366,
      "step": 2019
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.761382840841203,
      "learning_rate": 4.8856784394863123e-05,
      "loss": 0.5444,
      "step": 2020
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8008911548384982,
      "learning_rate": 4.885530102077876e-05,
      "loss": 0.1094,
      "step": 2021
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2553074885630093,
      "learning_rate": 4.8853816707494095e-05,
      "loss": 0.2971,
      "step": 2022
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7726089001024132,
      "learning_rate": 4.885233145506757e-05,
      "loss": 0.114,
      "step": 2023
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1263941564002806,
      "learning_rate": 4.885084526355766e-05,
      "loss": 0.2898,
      "step": 2024
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6564729831579412,
      "learning_rate": 4.8849358133022884e-05,
      "loss": 0.2694,
      "step": 2025
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0889332197604213,
      "learning_rate": 4.884787006352178e-05,
      "loss": 0.4645,
      "step": 2026
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2280117459258657,
      "learning_rate": 4.8846381055112946e-05,
      "loss": 0.1551,
      "step": 2027
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8223513249866401,
      "learning_rate": 4.8844891107854995e-05,
      "loss": 0.1187,
      "step": 2028
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8081990964643502,
      "learning_rate": 4.884340022180659e-05,
      "loss": 0.4327,
      "step": 2029
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6209068581836146,
      "learning_rate": 4.884190839702644e-05,
      "loss": 0.2675,
      "step": 2030
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1925739771366415,
      "learning_rate": 4.8840415633573254e-05,
      "loss": 0.1941,
      "step": 2031
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6692031234112367,
      "learning_rate": 4.883892193150583e-05,
      "loss": 0.0832,
      "step": 2032
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7691177573108958,
      "learning_rate": 4.883742729088297e-05,
      "loss": 0.4659,
      "step": 2033
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9081800039440157,
      "learning_rate": 4.88359317117635e-05,
      "loss": 0.1288,
      "step": 2034
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9802950978130155,
      "learning_rate": 4.883443519420633e-05,
      "loss": 0.1009,
      "step": 2035
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.939213291748127,
      "learning_rate": 4.883293773827036e-05,
      "loss": 0.6905,
      "step": 2036
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.378053568994463,
      "learning_rate": 4.883143934401454e-05,
      "loss": 0.2502,
      "step": 2037
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4623302446059292,
      "learning_rate": 4.882994001149789e-05,
      "loss": 0.2519,
      "step": 2038
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0940474274652585,
      "learning_rate": 4.882843974077941e-05,
      "loss": 0.1108,
      "step": 2039
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9315154184463045,
      "learning_rate": 4.882693853191819e-05,
      "loss": 0.1581,
      "step": 2040
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5241013501871965,
      "learning_rate": 4.882543638497332e-05,
      "loss": 0.191,
      "step": 2041
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4706657814345427,
      "learning_rate": 4.882393330000394e-05,
      "loss": 0.3361,
      "step": 2042
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5663143744307704,
      "learning_rate": 4.8822429277069235e-05,
      "loss": 0.2179,
      "step": 2043
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0433088264739403,
      "learning_rate": 4.8820924316228425e-05,
      "loss": 0.1658,
      "step": 2044
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2666715079761506,
      "learning_rate": 4.881941841754074e-05,
      "loss": 0.2849,
      "step": 2045
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0479495765168911,
      "learning_rate": 4.881791158106549e-05,
      "loss": 0.1072,
      "step": 2046
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0667138466814232,
      "learning_rate": 4.881640380686199e-05,
      "loss": 0.1967,
      "step": 2047
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8589280300332426,
      "learning_rate": 4.88148950949896e-05,
      "loss": 0.1122,
      "step": 2048
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8346011259459827,
      "learning_rate": 4.881338544550772e-05,
      "loss": 0.3459,
      "step": 2049
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.261761812701033,
      "learning_rate": 4.8811874858475795e-05,
      "loss": 0.1056,
      "step": 2050
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.186614219608772,
      "learning_rate": 4.881036333395329e-05,
      "loss": 0.1895,
      "step": 2051
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7117893863981312,
      "learning_rate": 4.8808850871999714e-05,
      "loss": 0.1264,
      "step": 2052
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.907567193169549,
      "learning_rate": 4.880733747267461e-05,
      "loss": 0.4018,
      "step": 2053
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.280172343111016,
      "learning_rate": 4.880582313603757e-05,
      "loss": 0.1777,
      "step": 2054
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1402775635657028,
      "learning_rate": 4.8804307862148225e-05,
      "loss": 0.2333,
      "step": 2055
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8164783524739612,
      "learning_rate": 4.880279165106622e-05,
      "loss": 0.1521,
      "step": 2056
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.134736418862461,
      "learning_rate": 4.880127450285124e-05,
      "loss": 0.2341,
      "step": 2057
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1206742895430337,
      "learning_rate": 4.879975641756302e-05,
      "loss": 0.1578,
      "step": 2058
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6815711919472643,
      "learning_rate": 4.8798237395261336e-05,
      "loss": 0.0786,
      "step": 2059
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8954827250558474,
      "learning_rate": 4.879671743600599e-05,
      "loss": 0.199,
      "step": 2060
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.829647479567098,
      "learning_rate": 4.879519653985683e-05,
      "loss": 1.3554,
      "step": 2061
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2735000903476887,
      "learning_rate": 4.879367470687372e-05,
      "loss": 0.1958,
      "step": 2062
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9491967559986151,
      "learning_rate": 4.879215193711659e-05,
      "loss": 0.2151,
      "step": 2063
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0205313173899841,
      "learning_rate": 4.8790628230645394e-05,
      "loss": 0.1744,
      "step": 2064
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9355200344418408,
      "learning_rate": 4.87891035875201e-05,
      "loss": 0.6626,
      "step": 2065
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0514707683818791,
      "learning_rate": 4.878757800780076e-05,
      "loss": 0.1277,
      "step": 2066
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0796568183454864,
      "learning_rate": 4.8786051491547425e-05,
      "loss": 0.2589,
      "step": 2067
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0804516832328146,
      "learning_rate": 4.878452403882019e-05,
      "loss": 0.2246,
      "step": 2068
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8789057586513113,
      "learning_rate": 4.87829956496792e-05,
      "loss": 0.0965,
      "step": 2069
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9939540690700542,
      "learning_rate": 4.878146632418463e-05,
      "loss": 0.099,
      "step": 2070
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.4630906397025636,
      "learning_rate": 4.877993606239668e-05,
      "loss": 0.5248,
      "step": 2071
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8621507468261158,
      "learning_rate": 4.877840486437561e-05,
      "loss": 0.0861,
      "step": 2072
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4070066599432216,
      "learning_rate": 4.87768727301817e-05,
      "loss": 0.2415,
      "step": 2073
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8357678351077127,
      "learning_rate": 4.877533965987527e-05,
      "loss": 0.2251,
      "step": 2074
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3788824331879135,
      "learning_rate": 4.877380565351668e-05,
      "loss": 0.1532,
      "step": 2075
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3592775839155558,
      "learning_rate": 4.877227071116632e-05,
      "loss": 0.3203,
      "step": 2076
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0264103540976406,
      "learning_rate": 4.877073483288462e-05,
      "loss": 0.185,
      "step": 2077
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4334176265727012,
      "learning_rate": 4.876919801873207e-05,
      "loss": 0.1487,
      "step": 2078
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5546251449935897,
      "learning_rate": 4.8767660268769146e-05,
      "loss": 0.318,
      "step": 2079
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1188851244654112,
      "learning_rate": 4.87661215830564e-05,
      "loss": 0.1723,
      "step": 2080
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1021065124938285,
      "learning_rate": 4.876458196165442e-05,
      "loss": 0.1845,
      "step": 2081
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6301322781606273,
      "learning_rate": 4.876304140462382e-05,
      "loss": 0.1736,
      "step": 2082
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.150555559530796,
      "learning_rate": 4.876149991202525e-05,
      "loss": 0.1816,
      "step": 2083
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8875739194398652,
      "learning_rate": 4.875995748391939e-05,
      "loss": 0.235,
      "step": 2084
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6034147178470661,
      "learning_rate": 4.875841412036698e-05,
      "loss": 0.4751,
      "step": 2085
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.3207768739488044,
      "learning_rate": 4.8756869821428784e-05,
      "loss": 0.5715,
      "step": 2086
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2969650160124258,
      "learning_rate": 4.875532458716559e-05,
      "loss": 0.21,
      "step": 2087
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.266409782110788,
      "learning_rate": 4.875377841763825e-05,
      "loss": 0.4116,
      "step": 2088
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.255539596619959,
      "learning_rate": 4.875223131290763e-05,
      "loss": 0.6439,
      "step": 2089
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8685050474138729,
      "learning_rate": 4.8750683273034634e-05,
      "loss": 0.1943,
      "step": 2090
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.140901733735988,
      "learning_rate": 4.874913429808022e-05,
      "loss": 0.2019,
      "step": 2091
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8601226543665768,
      "learning_rate": 4.874758438810537e-05,
      "loss": 0.4458,
      "step": 2092
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.623963870835606,
      "learning_rate": 4.87460335431711e-05,
      "loss": 0.0903,
      "step": 2093
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.573545600980923,
      "learning_rate": 4.874448176333847e-05,
      "loss": 0.1828,
      "step": 2094
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3295211307169599,
      "learning_rate": 4.874292904866858e-05,
      "loss": 0.158,
      "step": 2095
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.32120214177031,
      "learning_rate": 4.874137539922256e-05,
      "loss": 0.18,
      "step": 2096
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3646195123060707,
      "learning_rate": 4.8739820815061565e-05,
      "loss": 0.2417,
      "step": 2097
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9852751541424978,
      "learning_rate": 4.873826529624682e-05,
      "loss": 0.1741,
      "step": 2098
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1756824380611839,
      "learning_rate": 4.873670884283955e-05,
      "loss": 0.1265,
      "step": 2099
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8061531600239311,
      "learning_rate": 4.873515145490104e-05,
      "loss": 0.4595,
      "step": 2100
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.995384434470805,
      "learning_rate": 4.8733593132492614e-05,
      "loss": 0.1434,
      "step": 2101
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0811777548152284,
      "learning_rate": 4.873203387567562e-05,
      "loss": 0.1947,
      "step": 2102
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1456814174389238,
      "learning_rate": 4.873047368451144e-05,
      "loss": 0.2531,
      "step": 2103
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1427934247454905,
      "learning_rate": 4.8728912559061505e-05,
      "loss": 0.2143,
      "step": 2104
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1021672154863649,
      "learning_rate": 4.872735049938727e-05,
      "loss": 0.1012,
      "step": 2105
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9026104692714951,
      "learning_rate": 4.8725787505550244e-05,
      "loss": 0.1179,
      "step": 2106
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0320654923557797,
      "learning_rate": 4.872422357761197e-05,
      "loss": 0.1834,
      "step": 2107
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9788471593287027,
      "learning_rate": 4.8722658715634006e-05,
      "loss": 0.3823,
      "step": 2108
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3471216615335242,
      "learning_rate": 4.872109291967796e-05,
      "loss": 0.1302,
      "step": 2109
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9003605432811969,
      "learning_rate": 4.871952618980549e-05,
      "loss": 0.0967,
      "step": 2110
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0623250911632098,
      "learning_rate": 4.871795852607828e-05,
      "loss": 0.174,
      "step": 2111
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.967833257395101,
      "learning_rate": 4.8716389928558035e-05,
      "loss": 0.1567,
      "step": 2112
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7728926519552897,
      "learning_rate": 4.871482039730653e-05,
      "loss": 0.0671,
      "step": 2113
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.805900101579691,
      "learning_rate": 4.8713249932385544e-05,
      "loss": 0.2145,
      "step": 2114
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8298179616545647,
      "learning_rate": 4.871167853385692e-05,
      "loss": 0.3248,
      "step": 2115
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4383851130381213,
      "learning_rate": 4.8710106201782516e-05,
      "loss": 0.1644,
      "step": 2116
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4535244754911345,
      "learning_rate": 4.870853293622423e-05,
      "loss": 0.2988,
      "step": 2117
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8458420068057155,
      "learning_rate": 4.870695873724402e-05,
      "loss": 0.1489,
      "step": 2118
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.134903996181738,
      "learning_rate": 4.870538360490385e-05,
      "loss": 0.1328,
      "step": 2119
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2967728980899695,
      "learning_rate": 4.870380753926575e-05,
      "loss": 0.2649,
      "step": 2120
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9957922202288174,
      "learning_rate": 4.870223054039174e-05,
      "loss": 0.1424,
      "step": 2121
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5503986214526957,
      "learning_rate": 4.8700652608343936e-05,
      "loss": 0.1362,
      "step": 2122
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5616335918361095,
      "learning_rate": 4.8699073743184454e-05,
      "loss": 0.2649,
      "step": 2123
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1001569968904321,
      "learning_rate": 4.869749394497546e-05,
      "loss": 0.1646,
      "step": 2124
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1902686277024217,
      "learning_rate": 4.869591321377914e-05,
      "loss": 0.3794,
      "step": 2125
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1600347780552132,
      "learning_rate": 4.8694331549657735e-05,
      "loss": 0.2017,
      "step": 2126
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1558478405297514,
      "learning_rate": 4.8692748952673514e-05,
      "loss": 0.182,
      "step": 2127
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1709704281604068,
      "learning_rate": 4.8691165422888785e-05,
      "loss": 0.1817,
      "step": 2128
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1623096045829937,
      "learning_rate": 4.86895809603659e-05,
      "loss": 0.2444,
      "step": 2129
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9838337426704679,
      "learning_rate": 4.8687995565167235e-05,
      "loss": 0.1986,
      "step": 2130
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.9802365642263724,
      "learning_rate": 4.86864092373552e-05,
      "loss": 0.479,
      "step": 2131
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2892574394572538,
      "learning_rate": 4.8684821976992266e-05,
      "loss": 0.2332,
      "step": 2132
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2175338490719987,
      "learning_rate": 4.868323378414091e-05,
      "loss": 0.3127,
      "step": 2133
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2729999871240751,
      "learning_rate": 4.8681644658863675e-05,
      "loss": 0.1099,
      "step": 2134
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2047962853225498,
      "learning_rate": 4.8680054601223114e-05,
      "loss": 0.3637,
      "step": 2135
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3089513050019017,
      "learning_rate": 4.8678463611281835e-05,
      "loss": 0.2692,
      "step": 2136
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9733920820860201,
      "learning_rate": 4.8676871689102464e-05,
      "loss": 0.1562,
      "step": 2137
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0625530378599823,
      "learning_rate": 4.8675278834747694e-05,
      "loss": 0.1979,
      "step": 2138
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9518736708437044,
      "learning_rate": 4.867368504828023e-05,
      "loss": 0.129,
      "step": 2139
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3096799372477583,
      "learning_rate": 4.867209032976281e-05,
      "loss": 0.2372,
      "step": 2140
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4332113374989413,
      "learning_rate": 4.8670494679258236e-05,
      "loss": 0.2758,
      "step": 2141
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8251287481607785,
      "learning_rate": 4.866889809682932e-05,
      "loss": 0.1078,
      "step": 2142
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9290575408137677,
      "learning_rate": 4.8667300582538935e-05,
      "loss": 0.1072,
      "step": 2143
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0662950618197737,
      "learning_rate": 4.866570213644995e-05,
      "loss": 0.1372,
      "step": 2144
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1924140845031803,
      "learning_rate": 4.866410275862531e-05,
      "loss": 0.1592,
      "step": 2145
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.28318361157764,
      "learning_rate": 4.8662502449127995e-05,
      "loss": 0.2584,
      "step": 2146
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.123562138729882,
      "learning_rate": 4.8660901208020995e-05,
      "loss": 0.166,
      "step": 2147
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2284483311145329,
      "learning_rate": 4.865929903536735e-05,
      "loss": 0.2598,
      "step": 2148
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5034187827538568,
      "learning_rate": 4.865769593123016e-05,
      "loss": 0.4393,
      "step": 2149
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3986980141224754,
      "learning_rate": 4.8656091895672525e-05,
      "loss": 0.2454,
      "step": 2150
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0458156634605305,
      "learning_rate": 4.8654486928757595e-05,
      "loss": 0.1772,
      "step": 2151
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.116565019734311,
      "learning_rate": 4.865288103054856e-05,
      "loss": 0.2139,
      "step": 2152
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1518029384252764,
      "learning_rate": 4.865127420110864e-05,
      "loss": 0.152,
      "step": 2153
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0653277216616144,
      "learning_rate": 4.864966644050112e-05,
      "loss": 0.617,
      "step": 2154
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5740698559956,
      "learning_rate": 4.864805774878927e-05,
      "loss": 0.5502,
      "step": 2155
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9185989274278409,
      "learning_rate": 4.864644812603645e-05,
      "loss": 0.1479,
      "step": 2156
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.968602783301261,
      "learning_rate": 4.864483757230601e-05,
      "loss": 0.1226,
      "step": 2157
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7953502875525675,
      "learning_rate": 4.864322608766138e-05,
      "loss": 0.1265,
      "step": 2158
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2402912650234814,
      "learning_rate": 4.864161367216598e-05,
      "loss": 0.416,
      "step": 2159
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0802763326548512,
      "learning_rate": 4.864000032588332e-05,
      "loss": 0.152,
      "step": 2160
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3025001082053937,
      "learning_rate": 4.863838604887689e-05,
      "loss": 0.2521,
      "step": 2161
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7060308444037673,
      "learning_rate": 4.863677084121028e-05,
      "loss": 0.3722,
      "step": 2162
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1242478261805795,
      "learning_rate": 4.8635154702947045e-05,
      "loss": 0.2577,
      "step": 2163
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1768657325600063,
      "learning_rate": 4.8633537634150836e-05,
      "loss": 0.1293,
      "step": 2164
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9838295752818695,
      "learning_rate": 4.8631919634885316e-05,
      "loss": 0.1437,
      "step": 2165
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1555922650724966,
      "learning_rate": 4.863030070521418e-05,
      "loss": 0.1788,
      "step": 2166
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9337627747591563,
      "learning_rate": 4.862868084520118e-05,
      "loss": 0.0925,
      "step": 2167
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9790311785581519,
      "learning_rate": 4.862706005491007e-05,
      "loss": 0.2011,
      "step": 2168
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6914691906235044,
      "learning_rate": 4.862543833440467e-05,
      "loss": 0.2865,
      "step": 2169
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1266682271110737,
      "learning_rate": 4.862381568374884e-05,
      "loss": 0.1708,
      "step": 2170
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3059356925285683,
      "learning_rate": 4.8622192103006445e-05,
      "loss": 0.2536,
      "step": 2171
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8949919636649909,
      "learning_rate": 4.862056759224143e-05,
      "loss": 0.1405,
      "step": 2172
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2756569630317434,
      "learning_rate": 4.861894215151773e-05,
      "loss": 0.113,
      "step": 2173
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8683165638577597,
      "learning_rate": 4.8617315780899366e-05,
      "loss": 0.0874,
      "step": 2174
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2607943477005055,
      "learning_rate": 4.861568848045034e-05,
      "loss": 0.3641,
      "step": 2175
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4499408385160306,
      "learning_rate": 4.861406025023474e-05,
      "loss": 0.3096,
      "step": 2176
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0766365076765778,
      "learning_rate": 4.861243109031666e-05,
      "loss": 0.1885,
      "step": 2177
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0820097925993148,
      "learning_rate": 4.8610801000760244e-05,
      "loss": 0.2108,
      "step": 2178
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1553608528766335,
      "learning_rate": 4.860916998162968e-05,
      "loss": 0.201,
      "step": 2179
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5374018346029406,
      "learning_rate": 4.8607538032989164e-05,
      "loss": 0.1818,
      "step": 2180
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8202659201223187,
      "learning_rate": 4.860590515490296e-05,
      "loss": 0.1347,
      "step": 2181
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7981401454788642,
      "learning_rate": 4.8604271347435356e-05,
      "loss": 0.1212,
      "step": 2182
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.040196578862719,
      "learning_rate": 4.860263661065066e-05,
      "loss": 0.2262,
      "step": 2183
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2339795434242504,
      "learning_rate": 4.860100094461326e-05,
      "loss": 0.1759,
      "step": 2184
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0786187389771913,
      "learning_rate": 4.859936434938754e-05,
      "loss": 0.2049,
      "step": 2185
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1699765990630795,
      "learning_rate": 4.859772682503792e-05,
      "loss": 0.1943,
      "step": 2186
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9079846844945355,
      "learning_rate": 4.859608837162889e-05,
      "loss": 0.1486,
      "step": 2187
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.8568371348604864,
      "learning_rate": 4.859444898922495e-05,
      "loss": 0.6535,
      "step": 2188
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3468837441716697,
      "learning_rate": 4.859280867789065e-05,
      "loss": 0.2367,
      "step": 2189
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0565317850337594,
      "learning_rate": 4.859116743769055e-05,
      "loss": 0.1472,
      "step": 2190
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1990569199439987,
      "learning_rate": 4.858952526868929e-05,
      "loss": 0.2409,
      "step": 2191
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2846995135475274,
      "learning_rate": 4.858788217095152e-05,
      "loss": 0.2165,
      "step": 2192
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8913860088518346,
      "learning_rate": 4.8586238144541907e-05,
      "loss": 0.0954,
      "step": 2193
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0148009978729904,
      "learning_rate": 4.8584593189525204e-05,
      "loss": 0.1516,
      "step": 2194
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0804667414532894,
      "learning_rate": 4.858294730596616e-05,
      "loss": 0.2516,
      "step": 2195
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3577524848484521,
      "learning_rate": 4.8581300493929586e-05,
      "loss": 0.2237,
      "step": 2196
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0255508262840018,
      "learning_rate": 4.857965275348031e-05,
      "loss": 0.1596,
      "step": 2197
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1990128167454974,
      "learning_rate": 4.85780040846832e-05,
      "loss": 0.1981,
      "step": 2198
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8406343052485026,
      "learning_rate": 4.857635448760318e-05,
      "loss": 0.1252,
      "step": 2199
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0159186579075166,
      "learning_rate": 4.857470396230518e-05,
      "loss": 0.1224,
      "step": 2200
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.504249273887803,
      "learning_rate": 4.85730525088542e-05,
      "loss": 0.1766,
      "step": 2201
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.299451738935014,
      "learning_rate": 4.857140012731524e-05,
      "loss": 0.2656,
      "step": 2202
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5728391828703516,
      "learning_rate": 4.856974681775337e-05,
      "loss": 0.3106,
      "step": 2203
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.318296806387356,
      "learning_rate": 4.856809258023367e-05,
      "loss": 0.2886,
      "step": 2204
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.187619026927005,
      "learning_rate": 4.856643741482128e-05,
      "loss": 0.2109,
      "step": 2205
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.088604565976481,
      "learning_rate": 4.856478132158135e-05,
      "loss": 0.2038,
      "step": 2206
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7450038765393351,
      "learning_rate": 4.85631243005791e-05,
      "loss": 0.1122,
      "step": 2207
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0593555946481061,
      "learning_rate": 4.856146635187975e-05,
      "loss": 0.1279,
      "step": 2208
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.049288419328524,
      "learning_rate": 4.855980747554859e-05,
      "loss": 0.1277,
      "step": 2209
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.024803652202222,
      "learning_rate": 4.855814767165093e-05,
      "loss": 0.1349,
      "step": 2210
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.9352757108249157,
      "learning_rate": 4.855648694025211e-05,
      "loss": 0.4228,
      "step": 2211
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4723826012685455,
      "learning_rate": 4.855482528141751e-05,
      "loss": 0.2257,
      "step": 2212
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.5483698448683305,
      "learning_rate": 4.855316269521256e-05,
      "loss": 0.5551,
      "step": 2213
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8482464040170452,
      "learning_rate": 4.855149918170272e-05,
      "loss": 0.1233,
      "step": 2214
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.999426540970038,
      "learning_rate": 4.8549834740953475e-05,
      "loss": 0.1179,
      "step": 2215
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5751677114820803,
      "learning_rate": 4.854816937303036e-05,
      "loss": 0.3147,
      "step": 2216
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3357344470528258,
      "learning_rate": 4.854650307799894e-05,
      "loss": 0.2451,
      "step": 2217
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1402104056927531,
      "learning_rate": 4.854483585592481e-05,
      "loss": 0.2009,
      "step": 2218
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8854959655916674,
      "learning_rate": 4.854316770687363e-05,
      "loss": 0.1918,
      "step": 2219
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.980436406675039,
      "learning_rate": 4.8541498630911065e-05,
      "loss": 0.2483,
      "step": 2220
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8916350128369058,
      "learning_rate": 4.8539828628102824e-05,
      "loss": 0.0966,
      "step": 2221
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9876356319610753,
      "learning_rate": 4.8538157698514656e-05,
      "loss": 0.1824,
      "step": 2222
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4635041923568304,
      "learning_rate": 4.853648584221235e-05,
      "loss": 0.2066,
      "step": 2223
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7236247689416788,
      "learning_rate": 4.853481305926173e-05,
      "loss": 0.0702,
      "step": 2224
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.649638843334975,
      "learning_rate": 4.853313934972866e-05,
      "loss": 0.3079,
      "step": 2225
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2008520159817395,
      "learning_rate": 4.853146471367902e-05,
      "loss": 0.2036,
      "step": 2226
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7362564848534779,
      "learning_rate": 4.852978915117875e-05,
      "loss": 0.123,
      "step": 2227
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2193136831104852,
      "learning_rate": 4.852811266229382e-05,
      "loss": 0.1926,
      "step": 2228
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9608460576420926,
      "learning_rate": 4.852643524709024e-05,
      "loss": 0.172,
      "step": 2229
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1025147435778362,
      "learning_rate": 4.852475690563403e-05,
      "loss": 0.1974,
      "step": 2230
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4300066726916916,
      "learning_rate": 4.852307763799129e-05,
      "loss": 0.3799,
      "step": 2231
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8943960642407137,
      "learning_rate": 4.852139744422812e-05,
      "loss": 0.1824,
      "step": 2232
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1376558818211995,
      "learning_rate": 4.851971632441068e-05,
      "loss": 0.1801,
      "step": 2233
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0440268648968998,
      "learning_rate": 4.8518034278605154e-05,
      "loss": 0.2433,
      "step": 2234
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0688643817740549,
      "learning_rate": 4.851635130687775e-05,
      "loss": 0.1675,
      "step": 2235
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.5110960033880056,
      "learning_rate": 4.851466740929476e-05,
      "loss": 0.31,
      "step": 2236
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5866695388214898,
      "learning_rate": 4.851298258592245e-05,
      "loss": 0.2172,
      "step": 2237
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7426138738945661,
      "learning_rate": 4.851129683682716e-05,
      "loss": 0.1516,
      "step": 2238
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0504351039371922,
      "learning_rate": 4.8509610162075277e-05,
      "loss": 0.163,
      "step": 2239
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.325979191938307,
      "learning_rate": 4.8507922561733186e-05,
      "loss": 0.2464,
      "step": 2240
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9397362612112262,
      "learning_rate": 4.850623403586734e-05,
      "loss": 0.1515,
      "step": 2241
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7870294008216843,
      "learning_rate": 4.850454458454421e-05,
      "loss": 0.0843,
      "step": 2242
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8866175912126308,
      "learning_rate": 4.8502854207830315e-05,
      "loss": 0.1186,
      "step": 2243
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6408319161168834,
      "learning_rate": 4.850116290579221e-05,
      "loss": 0.3819,
      "step": 2244
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8200338207427887,
      "learning_rate": 4.8499470678496474e-05,
      "loss": 0.1362,
      "step": 2245
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0577922915792572,
      "learning_rate": 4.8497777526009736e-05,
      "loss": 0.6461,
      "step": 2246
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0760778223887242,
      "learning_rate": 4.8496083448398654e-05,
      "loss": 0.1766,
      "step": 2247
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2433564439124858,
      "learning_rate": 4.8494388445729935e-05,
      "loss": 0.2262,
      "step": 2248
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.144573263461714,
      "learning_rate": 4.8492692518070296e-05,
      "loss": 0.3041,
      "step": 2249
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9406913657000097,
      "learning_rate": 4.8490995665486526e-05,
      "loss": 0.1007,
      "step": 2250
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2545154897274253,
      "learning_rate": 4.848929788804542e-05,
      "loss": 0.12,
      "step": 2251
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0553550919359844,
      "learning_rate": 4.848759918581382e-05,
      "loss": 0.1125,
      "step": 2252
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.7814407442884397,
      "learning_rate": 4.8485899558858596e-05,
      "loss": 0.1693,
      "step": 2253
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9230568209669906,
      "learning_rate": 4.848419900724669e-05,
      "loss": 0.1582,
      "step": 2254
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1423351896390033,
      "learning_rate": 4.848249753104503e-05,
      "loss": 0.2424,
      "step": 2255
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9996172492149092,
      "learning_rate": 4.848079513032061e-05,
      "loss": 0.1658,
      "step": 2256
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8438614098202502,
      "learning_rate": 4.847909180514046e-05,
      "loss": 0.1104,
      "step": 2257
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9495725994476066,
      "learning_rate": 4.847738755557165e-05,
      "loss": 0.1972,
      "step": 2258
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0661499944007904,
      "learning_rate": 4.847568238168125e-05,
      "loss": 0.1159,
      "step": 2259
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.091613386651492,
      "learning_rate": 4.8473976283536423e-05,
      "loss": 0.2838,
      "step": 2260
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0874226578525323,
      "learning_rate": 4.847226926120433e-05,
      "loss": 0.1223,
      "step": 2261
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8304299422016095,
      "learning_rate": 4.847056131475216e-05,
      "loss": 0.2047,
      "step": 2262
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7378304090665635,
      "learning_rate": 4.846885244424718e-05,
      "loss": 0.2064,
      "step": 2263
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3055493929721504,
      "learning_rate": 4.846714264975665e-05,
      "loss": 0.2259,
      "step": 2264
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8207172635757823,
      "learning_rate": 4.8465431931347904e-05,
      "loss": 0.1124,
      "step": 2265
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8121912676470675,
      "learning_rate": 4.846372028908829e-05,
      "loss": 0.464,
      "step": 2266
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9985942656988371,
      "learning_rate": 4.8462007723045176e-05,
      "loss": 0.1298,
      "step": 2267
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8254603328031229,
      "learning_rate": 4.846029423328602e-05,
      "loss": 0.1226,
      "step": 2268
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.891652986021686,
      "learning_rate": 4.8458579819878257e-05,
      "loss": 0.1136,
      "step": 2269
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.913996909785254,
      "learning_rate": 4.84568644828894e-05,
      "loss": 0.0939,
      "step": 2270
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2550589690750509,
      "learning_rate": 4.845514822238697e-05,
      "loss": 0.404,
      "step": 2271
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9675691613606758,
      "learning_rate": 4.845343103843855e-05,
      "loss": 0.1175,
      "step": 2272
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7204448204761466,
      "learning_rate": 4.845171293111174e-05,
      "loss": 0.0816,
      "step": 2273
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.228779963280907,
      "learning_rate": 4.8449993900474187e-05,
      "loss": 1.2091,
      "step": 2274
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2253824433510574,
      "learning_rate": 4.844827394659357e-05,
      "loss": 0.1486,
      "step": 2275
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.925472185006558,
      "learning_rate": 4.84465530695376e-05,
      "loss": 0.1434,
      "step": 2276
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4796156597845567,
      "learning_rate": 4.844483126937403e-05,
      "loss": 0.2299,
      "step": 2277
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9776816338486283,
      "learning_rate": 4.844310854617065e-05,
      "loss": 0.1413,
      "step": 2278
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.136418977204843,
      "learning_rate": 4.844138489999529e-05,
      "loss": 0.1668,
      "step": 2279
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1060360557630207,
      "learning_rate": 4.84396603309158e-05,
      "loss": 0.141,
      "step": 2280
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.408889215274059,
      "learning_rate": 4.843793483900009e-05,
      "loss": 0.4787,
      "step": 2281
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2205547340895622,
      "learning_rate": 4.843620842431609e-05,
      "loss": 0.2046,
      "step": 2282
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2470865755916118,
      "learning_rate": 4.8434481086931764e-05,
      "loss": 0.1677,
      "step": 2283
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2691207153566162,
      "learning_rate": 4.8432752826915126e-05,
      "loss": 0.1588,
      "step": 2284
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9012007043245924,
      "learning_rate": 4.843102364433421e-05,
      "loss": 0.163,
      "step": 2285
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3403868591370904,
      "learning_rate": 4.8429293539257103e-05,
      "loss": 0.3244,
      "step": 2286
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0255048106812674,
      "learning_rate": 4.842756251175192e-05,
      "loss": 0.1411,
      "step": 2287
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0485926903304081,
      "learning_rate": 4.842583056188681e-05,
      "loss": 0.1742,
      "step": 2288
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.073619065054702,
      "learning_rate": 4.8424097689729965e-05,
      "loss": 0.1718,
      "step": 2289
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1854334279392427,
      "learning_rate": 4.84223638953496e-05,
      "loss": 0.1992,
      "step": 2290
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5527577088471307,
      "learning_rate": 4.842062917881399e-05,
      "loss": 0.3624,
      "step": 2291
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3487498765620574,
      "learning_rate": 4.841889354019142e-05,
      "loss": 0.191,
      "step": 2292
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0604230253355993,
      "learning_rate": 4.841715697955023e-05,
      "loss": 0.1901,
      "step": 2293
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1530667721248065,
      "learning_rate": 4.841541949695878e-05,
      "loss": 0.1193,
      "step": 2294
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0293644945576979,
      "learning_rate": 4.841368109248548e-05,
      "loss": 0.2335,
      "step": 2295
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3269353814836524,
      "learning_rate": 4.8411941766198786e-05,
      "loss": 0.2027,
      "step": 2296
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0670632794101298,
      "learning_rate": 4.841020151816717e-05,
      "loss": 0.1433,
      "step": 2297
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7340097212468448,
      "learning_rate": 4.8408460348459136e-05,
      "loss": 0.3801,
      "step": 2298
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6786020180624812,
      "learning_rate": 4.840671825714324e-05,
      "loss": 0.09,
      "step": 2299
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.116694001837011,
      "learning_rate": 4.8404975244288076e-05,
      "loss": 0.4934,
      "step": 2300
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9356344181330498,
      "learning_rate": 4.8403231309962263e-05,
      "loss": 0.1294,
      "step": 2301
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7423929571714678,
      "learning_rate": 4.840148645423445e-05,
      "loss": 0.4981,
      "step": 2302
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5695515247438632,
      "learning_rate": 4.839974067717336e-05,
      "loss": 0.2278,
      "step": 2303
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2786336529201683,
      "learning_rate": 4.839799397884769e-05,
      "loss": 0.2493,
      "step": 2304
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7274052682256251,
      "learning_rate": 4.8396246359326245e-05,
      "loss": 0.3004,
      "step": 2305
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6793370814539341,
      "learning_rate": 4.839449781867781e-05,
      "loss": 0.3305,
      "step": 2306
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.903634997714657,
      "learning_rate": 4.839274835697123e-05,
      "loss": 0.1457,
      "step": 2307
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.303267896923215,
      "learning_rate": 4.839099797427538e-05,
      "loss": 1.0714,
      "step": 2308
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.11369777064729,
      "learning_rate": 4.838924667065917e-05,
      "loss": 0.6627,
      "step": 2309
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9767543336998574,
      "learning_rate": 4.838749444619156e-05,
      "loss": 0.1625,
      "step": 2310
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9718656190099415,
      "learning_rate": 4.8385741300941535e-05,
      "loss": 0.1598,
      "step": 2311
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8329073310929569,
      "learning_rate": 4.8383987234978114e-05,
      "loss": 0.1109,
      "step": 2312
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.6999867679296967,
      "learning_rate": 4.838223224837036e-05,
      "loss": 0.2568,
      "step": 2313
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.357338087945475,
      "learning_rate": 4.838047634118736e-05,
      "loss": 0.7718,
      "step": 2314
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.057811975000381,
      "learning_rate": 4.837871951349826e-05,
      "loss": 0.1662,
      "step": 2315
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.739390751539372,
      "learning_rate": 4.837696176537221e-05,
      "loss": 0.7055,
      "step": 2316
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.755911242461867,
      "learning_rate": 4.8375203096878416e-05,
      "loss": 0.6529,
      "step": 2317
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0543730415947734,
      "learning_rate": 4.837344350808613e-05,
      "loss": 0.1524,
      "step": 2318
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.8636805598082042,
      "learning_rate": 4.837168299906462e-05,
      "loss": 0.408,
      "step": 2319
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9522253889713805,
      "learning_rate": 4.83699215698832e-05,
      "loss": 0.7627,
      "step": 2320
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.115617618609018,
      "learning_rate": 4.8368159220611225e-05,
      "loss": 0.3257,
      "step": 2321
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2254293529234146,
      "learning_rate": 4.836639595131807e-05,
      "loss": 0.1731,
      "step": 2322
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9206602121113399,
      "learning_rate": 4.836463176207316e-05,
      "loss": 0.4594,
      "step": 2323
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.410375900341364,
      "learning_rate": 4.836286665294596e-05,
      "loss": 0.286,
      "step": 2324
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4733805834982083,
      "learning_rate": 4.836110062400595e-05,
      "loss": 0.4105,
      "step": 2325
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3430510858059417,
      "learning_rate": 4.835933367532266e-05,
      "loss": 0.2509,
      "step": 2326
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8524773439555875,
      "learning_rate": 4.835756580696566e-05,
      "loss": 0.3626,
      "step": 2327
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.049703390140765,
      "learning_rate": 4.835579701900457e-05,
      "loss": 0.1923,
      "step": 2328
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.239187936188134,
      "learning_rate": 4.8354027311509e-05,
      "loss": 0.6549,
      "step": 2329
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.383987019947468,
      "learning_rate": 4.8352256684548644e-05,
      "loss": 0.5968,
      "step": 2330
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4775129033813197,
      "learning_rate": 4.8350485138193205e-05,
      "loss": 0.1799,
      "step": 2331
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9682668508772005,
      "learning_rate": 4.834871267251243e-05,
      "loss": 0.3646,
      "step": 2332
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.225499591889593,
      "learning_rate": 4.834693928757611e-05,
      "loss": 0.3551,
      "step": 2333
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1866604290416352,
      "learning_rate": 4.834516498345405e-05,
      "loss": 0.1572,
      "step": 2334
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.011383803443606,
      "learning_rate": 4.834338976021612e-05,
      "loss": 0.166,
      "step": 2335
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.274683445551539,
      "learning_rate": 4.83416136179322e-05,
      "loss": 0.1967,
      "step": 2336
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1968367284033687,
      "learning_rate": 4.833983655667223e-05,
      "loss": 0.2967,
      "step": 2337
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1759688415454925,
      "learning_rate": 4.8338058576506164e-05,
      "loss": 0.4,
      "step": 2338
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0782596090170296,
      "learning_rate": 4.8336279677504e-05,
      "loss": 0.1315,
      "step": 2339
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2057560517448838,
      "learning_rate": 4.8334499859735785e-05,
      "loss": 0.166,
      "step": 2340
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.4504804558084903,
      "learning_rate": 4.833271912327159e-05,
      "loss": 0.484,
      "step": 2341
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5776314055049951,
      "learning_rate": 4.833093746818153e-05,
      "loss": 0.3298,
      "step": 2342
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8960184632215562,
      "learning_rate": 4.832915489453572e-05,
      "loss": 0.0907,
      "step": 2343
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.493919508388506,
      "learning_rate": 4.8327371402404375e-05,
      "loss": 0.2112,
      "step": 2344
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2664255633429824,
      "learning_rate": 4.83255869918577e-05,
      "loss": 0.2466,
      "step": 2345
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3979653504813256,
      "learning_rate": 4.832380166296595e-05,
      "loss": 0.1313,
      "step": 2346
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1008198933358002,
      "learning_rate": 4.8322015415799404e-05,
      "loss": 0.1535,
      "step": 2347
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3380007396251843,
      "learning_rate": 4.83202282504284e-05,
      "loss": 0.1524,
      "step": 2348
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3064724253273705,
      "learning_rate": 4.83184401669233e-05,
      "loss": 0.2009,
      "step": 2349
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2130477535391266,
      "learning_rate": 4.8316651165354496e-05,
      "loss": 0.1663,
      "step": 2350
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6652115138840993,
      "learning_rate": 4.831486124579243e-05,
      "loss": 0.2965,
      "step": 2351
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.3972252418486213,
      "learning_rate": 4.831307040830756e-05,
      "loss": 0.3572,
      "step": 2352
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3195683404316654,
      "learning_rate": 4.831127865297041e-05,
      "loss": 0.3144,
      "step": 2353
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.069712756410624,
      "learning_rate": 4.83094859798515e-05,
      "loss": 0.2363,
      "step": 2354
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.068838850512623,
      "learning_rate": 4.830769238902143e-05,
      "loss": 0.3265,
      "step": 2355
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5205054672933354,
      "learning_rate": 4.830589788055081e-05,
      "loss": 0.4355,
      "step": 2356
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.312559219457824,
      "learning_rate": 4.830410245451028e-05,
      "loss": 0.2564,
      "step": 2357
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4634471191295517,
      "learning_rate": 4.830230611097054e-05,
      "loss": 0.3176,
      "step": 2358
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.1117069394619374,
      "learning_rate": 4.8300508850002304e-05,
      "loss": 0.7775,
      "step": 2359
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.946908851757376,
      "learning_rate": 4.829871067167634e-05,
      "loss": 0.1953,
      "step": 2360
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9707244225534016,
      "learning_rate": 4.829691157606344e-05,
      "loss": 0.1242,
      "step": 2361
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8909450025316494,
      "learning_rate": 4.829511156323443e-05,
      "loss": 0.1326,
      "step": 2362
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.78972464137139,
      "learning_rate": 4.829331063326018e-05,
      "loss": 0.1053,
      "step": 2363
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0356801711977959,
      "learning_rate": 4.82915087862116e-05,
      "loss": 0.2535,
      "step": 2364
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1336120699902372,
      "learning_rate": 4.828970602215964e-05,
      "loss": 0.1973,
      "step": 2365
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4152918775724581,
      "learning_rate": 4.828790234117525e-05,
      "loss": 0.3455,
      "step": 2366
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9585208717846574,
      "learning_rate": 4.8286097743329453e-05,
      "loss": 0.3166,
      "step": 2367
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.486943167111476,
      "learning_rate": 4.82842922286933e-05,
      "loss": 0.5143,
      "step": 2368
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.203150616286725,
      "learning_rate": 4.828248579733788e-05,
      "loss": 0.3634,
      "step": 2369
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.231029059022259,
      "learning_rate": 4.82806784493343e-05,
      "loss": 0.1398,
      "step": 2370
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9110279306157129,
      "learning_rate": 4.827887018475373e-05,
      "loss": 0.0746,
      "step": 2371
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9682275171190458,
      "learning_rate": 4.827706100366736e-05,
      "loss": 0.1361,
      "step": 2372
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0009462668690536,
      "learning_rate": 4.827525090614642e-05,
      "loss": 0.1491,
      "step": 2373
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0090792837447051,
      "learning_rate": 4.827343989226216e-05,
      "loss": 0.1262,
      "step": 2374
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.3534318334820434,
      "learning_rate": 4.8271627962085894e-05,
      "loss": 0.4216,
      "step": 2375
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4257078396090916,
      "learning_rate": 4.8269815115688964e-05,
      "loss": 0.3545,
      "step": 2376
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4517653392594936,
      "learning_rate": 4.8268001353142736e-05,
      "loss": 0.2503,
      "step": 2377
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8595521791743324,
      "learning_rate": 4.826618667451861e-05,
      "loss": 0.0873,
      "step": 2378
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.304617784833954,
      "learning_rate": 4.826437107988804e-05,
      "loss": 0.1947,
      "step": 2379
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0711040505948513,
      "learning_rate": 4.826255456932252e-05,
      "loss": 0.1641,
      "step": 2380
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0575808797223887,
      "learning_rate": 4.826073714289354e-05,
      "loss": 0.2443,
      "step": 2381
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.275336342127802,
      "learning_rate": 4.825891880067268e-05,
      "loss": 0.184,
      "step": 2382
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1395952028198406,
      "learning_rate": 4.8257099542731516e-05,
      "loss": 0.1336,
      "step": 2383
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.067311359739354,
      "learning_rate": 4.825527936914167e-05,
      "loss": 0.1544,
      "step": 2384
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3180637396613935,
      "learning_rate": 4.8253458279974814e-05,
      "loss": 0.2578,
      "step": 2385
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2275337274613778,
      "learning_rate": 4.825163627530264e-05,
      "loss": 0.2414,
      "step": 2386
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.242187226424314,
      "learning_rate": 4.824981335519688e-05,
      "loss": 0.5403,
      "step": 2387
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9737679439313267,
      "learning_rate": 4.8247989519729306e-05,
      "loss": 0.1024,
      "step": 2388
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.019424525541356,
      "learning_rate": 4.824616476897172e-05,
      "loss": 0.1414,
      "step": 2389
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1516375308578197,
      "learning_rate": 4.8244339102995976e-05,
      "loss": 0.1841,
      "step": 2390
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9920568631217891,
      "learning_rate": 4.824251252187393e-05,
      "loss": 0.1389,
      "step": 2391
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2369968176376716,
      "learning_rate": 4.824068502567752e-05,
      "loss": 0.3037,
      "step": 2392
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.127300845861164,
      "learning_rate": 4.823885661447868e-05,
      "loss": 0.2114,
      "step": 2393
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8886471370757577,
      "learning_rate": 4.82370272883494e-05,
      "loss": 0.0954,
      "step": 2394
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8479241750496196,
      "learning_rate": 4.8235197047361705e-05,
      "loss": 0.116,
      "step": 2395
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0144268622632364,
      "learning_rate": 4.823336589158765e-05,
      "loss": 0.1458,
      "step": 2396
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9654112411115894,
      "learning_rate": 4.823153382109932e-05,
      "loss": 0.1532,
      "step": 2397
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0220725471464258,
      "learning_rate": 4.8229700835968864e-05,
      "loss": 0.1172,
      "step": 2398
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5079459432237725,
      "learning_rate": 4.822786693626844e-05,
      "loss": 0.2553,
      "step": 2399
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.920257085224616,
      "learning_rate": 4.822603212207024e-05,
      "loss": 0.1305,
      "step": 2400
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0290242119463742,
      "learning_rate": 4.8224196393446506e-05,
      "loss": 0.1652,
      "step": 2401
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2516693080232586,
      "learning_rate": 4.8222359750469526e-05,
      "loss": 0.4969,
      "step": 2402
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1019202119023477,
      "learning_rate": 4.82205221932116e-05,
      "loss": 0.1341,
      "step": 2403
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6649723341750476,
      "learning_rate": 4.8218683721745065e-05,
      "loss": 0.1363,
      "step": 2404
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8703052237009872,
      "learning_rate": 4.821684433614232e-05,
      "loss": 0.1014,
      "step": 2405
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0843975680751325,
      "learning_rate": 4.821500403647577e-05,
      "loss": 0.2376,
      "step": 2406
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4434979258059204,
      "learning_rate": 4.821316282281788e-05,
      "loss": 0.6397,
      "step": 2407
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8219366522373355,
      "learning_rate": 4.8211320695241125e-05,
      "loss": 0.5436,
      "step": 2408
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4591418433508732,
      "learning_rate": 4.820947765381804e-05,
      "loss": 0.1789,
      "step": 2409
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1248972656187786,
      "learning_rate": 4.8207633698621194e-05,
      "loss": 0.2135,
      "step": 2410
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.000270445486383,
      "learning_rate": 4.8205788829723164e-05,
      "loss": 0.5062,
      "step": 2411
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.916629471664839,
      "learning_rate": 4.820394304719661e-05,
      "loss": 0.138,
      "step": 2412
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9765807207375293,
      "learning_rate": 4.820209635111418e-05,
      "loss": 0.1438,
      "step": 2413
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.640667645382271,
      "learning_rate": 4.820024874154859e-05,
      "loss": 0.4685,
      "step": 2414
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0734854435161165,
      "learning_rate": 4.819840021857258e-05,
      "loss": 0.6913,
      "step": 2415
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3204394193803162,
      "learning_rate": 4.819655078225893e-05,
      "loss": 0.1257,
      "step": 2416
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6981665049077711,
      "learning_rate": 4.819470043268045e-05,
      "loss": 0.0822,
      "step": 2417
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8630979231704798,
      "learning_rate": 4.8192849169909996e-05,
      "loss": 0.4351,
      "step": 2418
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1440734738624017,
      "learning_rate": 4.819099699402044e-05,
      "loss": 0.1451,
      "step": 2419
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0577157869895764,
      "learning_rate": 4.8189143905084716e-05,
      "loss": 0.1277,
      "step": 2420
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9294562660061381,
      "learning_rate": 4.8187289903175784e-05,
      "loss": 0.1489,
      "step": 2421
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9929786780770813,
      "learning_rate": 4.8185434988366616e-05,
      "loss": 0.5815,
      "step": 2422
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.043764263164325,
      "learning_rate": 4.818357916073026e-05,
      "loss": 0.4899,
      "step": 2423
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7363069357827448,
      "learning_rate": 4.818172242033979e-05,
      "loss": 0.1512,
      "step": 2424
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2677858340061028,
      "learning_rate": 4.817986476726828e-05,
      "loss": 0.3418,
      "step": 2425
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8345409939113002,
      "learning_rate": 4.817800620158889e-05,
      "loss": 0.1099,
      "step": 2426
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.081773213230627,
      "learning_rate": 4.8176146723374784e-05,
      "loss": 0.1568,
      "step": 2427
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7810577670270287,
      "learning_rate": 4.817428633269917e-05,
      "loss": 0.1834,
      "step": 2428
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1210839863031727,
      "learning_rate": 4.817242502963529e-05,
      "loss": 0.2061,
      "step": 2429
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0804962607037545,
      "learning_rate": 4.817056281425644e-05,
      "loss": 0.1442,
      "step": 2430
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.210085101702654,
      "learning_rate": 4.816869968663592e-05,
      "loss": 0.2542,
      "step": 2431
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9817311120378549,
      "learning_rate": 4.816683564684709e-05,
      "loss": 0.1621,
      "step": 2432
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8338159143192986,
      "learning_rate": 4.816497069496333e-05,
      "loss": 0.1317,
      "step": 2433
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.226376473953481,
      "learning_rate": 4.816310483105808e-05,
      "loss": 0.1291,
      "step": 2434
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8703469068766079,
      "learning_rate": 4.8161238055204796e-05,
      "loss": 0.1336,
      "step": 2435
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1921745609748544,
      "learning_rate": 4.815937036747697e-05,
      "loss": 0.1574,
      "step": 2436
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1549244830401093,
      "learning_rate": 4.815750176794812e-05,
      "loss": 0.1596,
      "step": 2437
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.089544621802173,
      "learning_rate": 4.815563225669184e-05,
      "loss": 0.1255,
      "step": 2438
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0938913904788845,
      "learning_rate": 4.8153761833781726e-05,
      "loss": 0.1946,
      "step": 2439
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0142786590716293,
      "learning_rate": 4.8151890499291406e-05,
      "loss": 0.2733,
      "step": 2440
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0424933824053535,
      "learning_rate": 4.815001825329457e-05,
      "loss": 0.1475,
      "step": 2441
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9063347563998592,
      "learning_rate": 4.814814509586493e-05,
      "loss": 0.1295,
      "step": 2442
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1341445066962388,
      "learning_rate": 4.814627102707622e-05,
      "loss": 0.2173,
      "step": 2443
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9230311960862688,
      "learning_rate": 4.814439604700224e-05,
      "loss": 0.1808,
      "step": 2444
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1276340357548225,
      "learning_rate": 4.8142520155716785e-05,
      "loss": 0.142,
      "step": 2445
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0591912513773976,
      "learning_rate": 4.814064335329374e-05,
      "loss": 0.1798,
      "step": 2446
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.839381096826647,
      "learning_rate": 4.813876563980698e-05,
      "loss": 0.1183,
      "step": 2447
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9070221011383643,
      "learning_rate": 4.813688701533043e-05,
      "loss": 0.117,
      "step": 2448
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8738054741521245,
      "learning_rate": 4.8135007479938055e-05,
      "loss": 0.0823,
      "step": 2449
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.098142105149676,
      "learning_rate": 4.813312703370386e-05,
      "loss": 0.1394,
      "step": 2450
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.728595172802493,
      "learning_rate": 4.8131245676701866e-05,
      "loss": 0.1242,
      "step": 2451
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0067724935881097,
      "learning_rate": 4.812936340900617e-05,
      "loss": 0.216,
      "step": 2452
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1193627050498476,
      "learning_rate": 4.812748023069084e-05,
      "loss": 0.1985,
      "step": 2453
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1484347601258416,
      "learning_rate": 4.812559614183005e-05,
      "loss": 0.147,
      "step": 2454
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.25513051528431,
      "learning_rate": 4.8123711142497964e-05,
      "loss": 0.2673,
      "step": 2455
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.155974695652438,
      "learning_rate": 4.81218252327688e-05,
      "loss": 0.1519,
      "step": 2456
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2198176594920904,
      "learning_rate": 4.81199384127168e-05,
      "loss": 0.2034,
      "step": 2457
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1863721871442618,
      "learning_rate": 4.811805068241626e-05,
      "loss": 0.2557,
      "step": 2458
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3828205833997804,
      "learning_rate": 4.8116162041941495e-05,
      "loss": 0.2692,
      "step": 2459
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1212153022540496,
      "learning_rate": 4.811427249136686e-05,
      "loss": 0.2737,
      "step": 2460
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8152662450499271,
      "learning_rate": 4.811238203076676e-05,
      "loss": 0.1755,
      "step": 2461
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0071332182948893,
      "learning_rate": 4.8110490660215615e-05,
      "loss": 0.2302,
      "step": 2462
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8834228327737765,
      "learning_rate": 4.8108598379787886e-05,
      "loss": 0.1392,
      "step": 2463
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8332660078810055,
      "learning_rate": 4.810670518955808e-05,
      "loss": 0.1032,
      "step": 2464
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7930025529788532,
      "learning_rate": 4.810481108960072e-05,
      "loss": 0.0967,
      "step": 2465
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.218627909288332,
      "learning_rate": 4.8102916079990404e-05,
      "loss": 0.2383,
      "step": 2466
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8369457578447255,
      "learning_rate": 4.810102016080171e-05,
      "loss": 0.1381,
      "step": 2467
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0847998350938561,
      "learning_rate": 4.809912333210931e-05,
      "loss": 0.2859,
      "step": 2468
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1143859570355246,
      "learning_rate": 4.809722559398786e-05,
      "loss": 0.1952,
      "step": 2469
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.886183199873812,
      "learning_rate": 4.8095326946512086e-05,
      "loss": 0.1914,
      "step": 2470
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1891300586043883,
      "learning_rate": 4.8093427389756745e-05,
      "loss": 0.2895,
      "step": 2471
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8364906770293413,
      "learning_rate": 4.809152692379661e-05,
      "loss": 0.1197,
      "step": 2472
    },
    {
      "epoch": 0.15,
      "grad_norm": 14.596702711771206,
      "learning_rate": 4.8089625548706516e-05,
      "loss": 1.0676,
      "step": 2473
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3439138109929927,
      "learning_rate": 4.8087723264561315e-05,
      "loss": 0.3308,
      "step": 2474
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0319764508227338,
      "learning_rate": 4.808582007143591e-05,
      "loss": 0.1352,
      "step": 2475
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1228898815254247,
      "learning_rate": 4.808391596940521e-05,
      "loss": 0.2506,
      "step": 2476
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9518336360716051,
      "learning_rate": 4.8082010958544197e-05,
      "loss": 0.1333,
      "step": 2477
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9914179254484229,
      "learning_rate": 4.808010503892788e-05,
      "loss": 0.1966,
      "step": 2478
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0203504737921683,
      "learning_rate": 4.8078198210631274e-05,
      "loss": 0.1581,
      "step": 2479
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.082241665562355,
      "learning_rate": 4.807629047372948e-05,
      "loss": 0.1662,
      "step": 2480
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3631105311258165,
      "learning_rate": 4.8074381828297584e-05,
      "loss": 0.2642,
      "step": 2481
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9529320395231324,
      "learning_rate": 4.807247227441074e-05,
      "loss": 0.3073,
      "step": 2482
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.252276020537615,
      "learning_rate": 4.807056181214412e-05,
      "loss": 0.1871,
      "step": 2483
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.977991246362876,
      "learning_rate": 4.806865044157295e-05,
      "loss": 1.0993,
      "step": 2484
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0681141612126317,
      "learning_rate": 4.806673816277249e-05,
      "loss": 0.1912,
      "step": 2485
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9393219735412167,
      "learning_rate": 4.8064824975818e-05,
      "loss": 0.1458,
      "step": 2486
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9613135239665471,
      "learning_rate": 4.8062910880784836e-05,
      "loss": 0.1165,
      "step": 2487
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9575376382297303,
      "learning_rate": 4.806099587774834e-05,
      "loss": 0.2756,
      "step": 2488
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2192437349638696,
      "learning_rate": 4.80590799667839e-05,
      "loss": 0.1341,
      "step": 2489
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9819415316495224,
      "learning_rate": 4.805716314796696e-05,
      "loss": 0.5246,
      "step": 2490
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.9512250030535125,
      "learning_rate": 4.805524542137298e-05,
      "loss": 0.7416,
      "step": 2491
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.453995046289058,
      "learning_rate": 4.805332678707747e-05,
      "loss": 0.2964,
      "step": 2492
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9973700815844817,
      "learning_rate": 4.8051407245155965e-05,
      "loss": 0.6124,
      "step": 2493
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.224400308029225,
      "learning_rate": 4.804948679568403e-05,
      "loss": 0.1819,
      "step": 2494
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2234324553685378,
      "learning_rate": 4.8047565438737284e-05,
      "loss": 0.1977,
      "step": 2495
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.013601903046746,
      "learning_rate": 4.804564317439136e-05,
      "loss": 0.0965,
      "step": 2496
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2234311183526594,
      "learning_rate": 4.804372000272196e-05,
      "loss": 0.5689,
      "step": 2497
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7949218512962709,
      "learning_rate": 4.8041795923804786e-05,
      "loss": 0.1073,
      "step": 2498
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9810904400057584,
      "learning_rate": 4.8039870937715595e-05,
      "loss": 0.1415,
      "step": 2499
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9708973701803821,
      "learning_rate": 4.803794504453016e-05,
      "loss": 0.098,
      "step": 2500
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.670144244770401,
      "learning_rate": 4.803601824432433e-05,
      "loss": 0.7087,
      "step": 2501
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0114733473636925,
      "learning_rate": 4.803409053717395e-05,
      "loss": 0.2242,
      "step": 2502
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.343048745906834,
      "learning_rate": 4.803216192315492e-05,
      "loss": 0.2539,
      "step": 2503
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.175880697578412,
      "learning_rate": 4.8030232402343164e-05,
      "loss": 0.2483,
      "step": 2504
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0141588020937362,
      "learning_rate": 4.802830197481466e-05,
      "loss": 0.1328,
      "step": 2505
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3962782552568078,
      "learning_rate": 4.8026370640645404e-05,
      "loss": 0.3338,
      "step": 2506
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8597996079687971,
      "learning_rate": 4.802443839991142e-05,
      "loss": 0.0949,
      "step": 2507
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9822458276327841,
      "learning_rate": 4.8022505252688804e-05,
      "loss": 0.1134,
      "step": 2508
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6783080942937543,
      "learning_rate": 4.802057119905366e-05,
      "loss": 0.3867,
      "step": 2509
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5627580135615704,
      "learning_rate": 4.8018636239082124e-05,
      "loss": 0.3062,
      "step": 2510
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7231804997019442,
      "learning_rate": 4.801670037285039e-05,
      "loss": 0.499,
      "step": 2511
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7358760831185861,
      "learning_rate": 4.801476360043465e-05,
      "loss": 0.0862,
      "step": 2512
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1086473659017608,
      "learning_rate": 4.8012825921911185e-05,
      "loss": 0.1924,
      "step": 2513
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9613386612936556,
      "learning_rate": 4.801088733735627e-05,
      "loss": 0.1231,
      "step": 2514
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9252150542536497,
      "learning_rate": 4.800894784684623e-05,
      "loss": 0.1165,
      "step": 2515
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1403939063122226,
      "learning_rate": 4.800700745045742e-05,
      "loss": 0.1829,
      "step": 2516
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0678570225520414,
      "learning_rate": 4.8005066148266245e-05,
      "loss": 0.1795,
      "step": 2517
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1243878484979146,
      "learning_rate": 4.800312394034912e-05,
      "loss": 0.3164,
      "step": 2518
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1144311175369195,
      "learning_rate": 4.800118082678252e-05,
      "loss": 0.1838,
      "step": 2519
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2807384569522182,
      "learning_rate": 4.799923680764296e-05,
      "loss": 0.2201,
      "step": 2520
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0196346438915358,
      "learning_rate": 4.799729188300694e-05,
      "loss": 0.1637,
      "step": 2521
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9072093616058075,
      "learning_rate": 4.7995346052951076e-05,
      "loss": 0.5489,
      "step": 2522
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.135184738653344,
      "learning_rate": 4.799339931755196e-05,
      "loss": 0.2103,
      "step": 2523
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2500314453862178,
      "learning_rate": 4.799145167688622e-05,
      "loss": 0.2531,
      "step": 2524
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9440103104068013,
      "learning_rate": 4.7989503131030556e-05,
      "loss": 0.2714,
      "step": 2525
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3214076668445194,
      "learning_rate": 4.798755368006168e-05,
      "loss": 0.1844,
      "step": 2526
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8837324571848402,
      "learning_rate": 4.798560332405634e-05,
      "loss": 0.4939,
      "step": 2527
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.452777188771596,
      "learning_rate": 4.7983652063091325e-05,
      "loss": 0.4847,
      "step": 2528
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2712399463450181,
      "learning_rate": 4.798169989724345e-05,
      "loss": 0.3335,
      "step": 2529
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.539329180797662,
      "learning_rate": 4.797974682658959e-05,
      "loss": 0.119,
      "step": 2530
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.936326281396551,
      "learning_rate": 4.797779285120662e-05,
      "loss": 0.1196,
      "step": 2531
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6050049938597584,
      "learning_rate": 4.7975837971171484e-05,
      "loss": 0.17,
      "step": 2532
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9525359336333639,
      "learning_rate": 4.797388218656114e-05,
      "loss": 0.2128,
      "step": 2533
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.849288675412868,
      "learning_rate": 4.7971925497452586e-05,
      "loss": 0.4821,
      "step": 2534
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5569514304272016,
      "learning_rate": 4.7969967903922864e-05,
      "loss": 0.2444,
      "step": 2535
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9782053454910203,
      "learning_rate": 4.796800940604904e-05,
      "loss": 0.0935,
      "step": 2536
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0070183907001238,
      "learning_rate": 4.796605000390823e-05,
      "loss": 0.4746,
      "step": 2537
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6903808647911396,
      "learning_rate": 4.796408969757756e-05,
      "loss": 0.0704,
      "step": 2538
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1024609234204739,
      "learning_rate": 4.796212848713424e-05,
      "loss": 0.2541,
      "step": 2539
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1888668420825386,
      "learning_rate": 4.7960166372655454e-05,
      "loss": 0.1488,
      "step": 2540
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9876071258097637,
      "learning_rate": 4.7958203354218464e-05,
      "loss": 0.1031,
      "step": 2541
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9135785310253368,
      "learning_rate": 4.795623943190055e-05,
      "loss": 0.1147,
      "step": 2542
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.512715195663901,
      "learning_rate": 4.795427460577905e-05,
      "loss": 0.211,
      "step": 2543
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2449289869299305,
      "learning_rate": 4.795230887593129e-05,
      "loss": 0.2032,
      "step": 2544
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.000834277339612,
      "learning_rate": 4.7950342242434696e-05,
      "loss": 0.1169,
      "step": 2545
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3031867995101352,
      "learning_rate": 4.794837470536668e-05,
      "loss": 0.1356,
      "step": 2546
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1236020753158922,
      "learning_rate": 4.79464062648047e-05,
      "loss": 0.1554,
      "step": 2547
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0228495816272094,
      "learning_rate": 4.794443692082626e-05,
      "loss": 0.1589,
      "step": 2548
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.608745357310565,
      "learning_rate": 4.794246667350889e-05,
      "loss": 0.3932,
      "step": 2549
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3275666071071224,
      "learning_rate": 4.7940495522930175e-05,
      "loss": 0.1898,
      "step": 2550
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1970106781494927,
      "learning_rate": 4.7938523469167705e-05,
      "loss": 0.2164,
      "step": 2551
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.116095143975448,
      "learning_rate": 4.793655051229913e-05,
      "loss": 0.1341,
      "step": 2552
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2805047816616184,
      "learning_rate": 4.793457665240211e-05,
      "loss": 0.2182,
      "step": 2553
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9480714084928181,
      "learning_rate": 4.7932601889554386e-05,
      "loss": 0.1144,
      "step": 2554
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1303820409775165,
      "learning_rate": 4.7930626223833683e-05,
      "loss": 0.1765,
      "step": 2555
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4338613973865175,
      "learning_rate": 4.7928649655317795e-05,
      "loss": 0.3555,
      "step": 2556
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1283631669035987,
      "learning_rate": 4.792667218408454e-05,
      "loss": 0.4971,
      "step": 2557
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.761106561123657,
      "learning_rate": 4.792469381021176e-05,
      "loss": 0.2865,
      "step": 2558
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4339628789909071,
      "learning_rate": 4.792271453377737e-05,
      "loss": 0.5253,
      "step": 2559
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.747466019229733,
      "learning_rate": 4.7920734354859266e-05,
      "loss": 0.5783,
      "step": 2560
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0141332518437605,
      "learning_rate": 4.791875327353543e-05,
      "loss": 0.1384,
      "step": 2561
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0498456981076272,
      "learning_rate": 4.7916771289883856e-05,
      "loss": 0.1804,
      "step": 2562
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0191654030215551,
      "learning_rate": 4.791478840398257e-05,
      "loss": 0.1394,
      "step": 2563
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0564537069876234,
      "learning_rate": 4.791280461590964e-05,
      "loss": 0.307,
      "step": 2564
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0997756281282822,
      "learning_rate": 4.7910819925743166e-05,
      "loss": 0.1642,
      "step": 2565
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0159605739001898,
      "learning_rate": 4.79088343335613e-05,
      "loss": 0.124,
      "step": 2566
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9198008184340165,
      "learning_rate": 4.7906847839442206e-05,
      "loss": 0.1918,
      "step": 2567
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0345220725447672,
      "learning_rate": 4.79048604434641e-05,
      "loss": 0.1412,
      "step": 2568
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1582461859629078,
      "learning_rate": 4.790287214570522e-05,
      "loss": 0.1891,
      "step": 2569
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8035660633141334,
      "learning_rate": 4.790088294624385e-05,
      "loss": 0.133,
      "step": 2570
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.0220419541105112,
      "learning_rate": 4.78988928451583e-05,
      "loss": 0.7278,
      "step": 2571
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1725872424449952,
      "learning_rate": 4.789690184252693e-05,
      "loss": 0.2022,
      "step": 2572
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.480092463730138,
      "learning_rate": 4.789490993842813e-05,
      "loss": 0.3236,
      "step": 2573
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0836912829588254,
      "learning_rate": 4.789291713294032e-05,
      "loss": 0.1516,
      "step": 2574
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1063940248456197,
      "learning_rate": 4.7890923426141946e-05,
      "loss": 0.1626,
      "step": 2575
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2268318795157065,
      "learning_rate": 4.788892881811152e-05,
      "loss": 0.1974,
      "step": 2576
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5713683807813006,
      "learning_rate": 4.7886933308927553e-05,
      "loss": 0.3039,
      "step": 2577
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5287271089171584,
      "learning_rate": 4.788493689866863e-05,
      "loss": 0.3961,
      "step": 2578
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9552416247575497,
      "learning_rate": 4.788293958741333e-05,
      "loss": 0.4718,
      "step": 2579
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3276224250760642,
      "learning_rate": 4.788094137524031e-05,
      "loss": 0.2862,
      "step": 2580
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.952112321933526,
      "learning_rate": 4.787894226222822e-05,
      "loss": 0.1242,
      "step": 2581
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0309053275646562,
      "learning_rate": 4.787694224845578e-05,
      "loss": 0.1856,
      "step": 2582
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.144330978651856,
      "learning_rate": 4.787494133400173e-05,
      "loss": 0.5318,
      "step": 2583
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.99108859944128,
      "learning_rate": 4.787293951894484e-05,
      "loss": 0.153,
      "step": 2584
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9339147571313386,
      "learning_rate": 4.787093680336393e-05,
      "loss": 0.1559,
      "step": 2585
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1743656393148583,
      "learning_rate": 4.786893318733785e-05,
      "loss": 0.2224,
      "step": 2586
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0472930372169904,
      "learning_rate": 4.786692867094548e-05,
      "loss": 0.2022,
      "step": 2587
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1898961816405,
      "learning_rate": 4.786492325426575e-05,
      "loss": 0.1914,
      "step": 2588
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4970541906992214,
      "learning_rate": 4.7862916937377586e-05,
      "loss": 0.2948,
      "step": 2589
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.41740819717493,
      "learning_rate": 4.786090972036001e-05,
      "loss": 0.4754,
      "step": 2590
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8924980148289051,
      "learning_rate": 4.7858901603292027e-05,
      "loss": 0.1471,
      "step": 2591
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6500347202907666,
      "learning_rate": 4.785689258625271e-05,
      "loss": 0.4808,
      "step": 2592
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1482280436487418,
      "learning_rate": 4.785488266932116e-05,
      "loss": 0.2004,
      "step": 2593
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2438464162546417,
      "learning_rate": 4.7852871852576484e-05,
      "loss": 0.1856,
      "step": 2594
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.443704839840469,
      "learning_rate": 4.785086013609787e-05,
      "loss": 0.5445,
      "step": 2595
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1267281275920684,
      "learning_rate": 4.784884751996452e-05,
      "loss": 0.2368,
      "step": 2596
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3143304051601115,
      "learning_rate": 4.784683400425566e-05,
      "loss": 0.2825,
      "step": 2597
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6494206374681442,
      "learning_rate": 4.7844819589050576e-05,
      "loss": 0.2376,
      "step": 2598
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2999810785428307,
      "learning_rate": 4.784280427442858e-05,
      "loss": 0.2486,
      "step": 2599
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8930283686776538,
      "learning_rate": 4.7840788060469e-05,
      "loss": 0.159,
      "step": 2600
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2016613693624176,
      "learning_rate": 4.783877094725122e-05,
      "loss": 0.2887,
      "step": 2601
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1540569199875303,
      "learning_rate": 4.783675293485467e-05,
      "loss": 0.2029,
      "step": 2602
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9271682106487206,
      "learning_rate": 4.783473402335879e-05,
      "loss": 0.1455,
      "step": 2603
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8647492139936422,
      "learning_rate": 4.7832714212843054e-05,
      "loss": 0.1067,
      "step": 2604
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8528961001162931,
      "learning_rate": 4.783069350338701e-05,
      "loss": 0.5429,
      "step": 2605
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1017268662325492,
      "learning_rate": 4.78286718950702e-05,
      "loss": 0.1957,
      "step": 2606
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0003116168380246,
      "learning_rate": 4.782664938797221e-05,
      "loss": 0.1353,
      "step": 2607
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9461794434050331,
      "learning_rate": 4.7824625982172674e-05,
      "loss": 0.1372,
      "step": 2608
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7912684514517996,
      "learning_rate": 4.782260167775125e-05,
      "loss": 0.0829,
      "step": 2609
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.353091316997063,
      "learning_rate": 4.782057647478766e-05,
      "loss": 0.1423,
      "step": 2610
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.434838231402527,
      "learning_rate": 4.7818550373361606e-05,
      "loss": 0.6909,
      "step": 2611
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9072516316331298,
      "learning_rate": 4.781652337355287e-05,
      "loss": 0.1211,
      "step": 2612
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.101275888853495,
      "learning_rate": 4.781449547544126e-05,
      "loss": 0.1798,
      "step": 2613
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0201396090727572,
      "learning_rate": 4.7812466679106616e-05,
      "loss": 0.1268,
      "step": 2614
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3085442240674463,
      "learning_rate": 4.7810436984628805e-05,
      "loss": 0.1464,
      "step": 2615
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4541896046608163,
      "learning_rate": 4.780840639208775e-05,
      "loss": 0.3221,
      "step": 2616
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1301141939341814,
      "learning_rate": 4.7806374901563376e-05,
      "loss": 0.2912,
      "step": 2617
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.740420106881376,
      "learning_rate": 4.780434251313568e-05,
      "loss": 0.3668,
      "step": 2618
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9373878316188949,
      "learning_rate": 4.780230922688468e-05,
      "loss": 0.1663,
      "step": 2619
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7310131136051458,
      "learning_rate": 4.780027504289042e-05,
      "loss": 0.0844,
      "step": 2620
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.267441610712804,
      "learning_rate": 4.7798239961233e-05,
      "loss": 0.1909,
      "step": 2621
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8778430934012809,
      "learning_rate": 4.7796203981992526e-05,
      "loss": 0.1039,
      "step": 2622
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9412297608077126,
      "learning_rate": 4.779416710524916e-05,
      "loss": 0.1654,
      "step": 2623
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.351690540354306,
      "learning_rate": 4.7792129331083114e-05,
      "loss": 0.2382,
      "step": 2624
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.905674673115208,
      "learning_rate": 4.7790090659574596e-05,
      "loss": 0.4533,
      "step": 2625
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9711038859900797,
      "learning_rate": 4.778805109080387e-05,
      "loss": 0.2216,
      "step": 2626
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5752716510378066,
      "learning_rate": 4.778601062485124e-05,
      "loss": 0.5663,
      "step": 2627
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5808315381864597,
      "learning_rate": 4.7783969261797044e-05,
      "loss": 0.2513,
      "step": 2628
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4022249318354363,
      "learning_rate": 4.778192700172166e-05,
      "loss": 0.2572,
      "step": 2629
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.0608534480631224,
      "learning_rate": 4.777988384470547e-05,
      "loss": 0.7187,
      "step": 2630
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0307743336962272,
      "learning_rate": 4.7777839790828925e-05,
      "loss": 0.216,
      "step": 2631
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2035449945075327,
      "learning_rate": 4.777579484017252e-05,
      "loss": 0.1197,
      "step": 2632
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4319927448913163,
      "learning_rate": 4.777374899281673e-05,
      "loss": 0.1952,
      "step": 2633
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.169938571920905,
      "learning_rate": 4.777170224884213e-05,
      "loss": 0.48,
      "step": 2634
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8688709627864932,
      "learning_rate": 4.77696546083293e-05,
      "loss": 0.1301,
      "step": 2635
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8138278802221621,
      "learning_rate": 4.7767606071358836e-05,
      "loss": 0.104,
      "step": 2636
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2045626326810719,
      "learning_rate": 4.776555663801141e-05,
      "loss": 0.1782,
      "step": 2637
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.95795607287711,
      "learning_rate": 4.77635063083677e-05,
      "loss": 0.3585,
      "step": 2638
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.180555921078603,
      "learning_rate": 4.776145508250843e-05,
      "loss": 0.5427,
      "step": 2639
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2467593638007153,
      "learning_rate": 4.775940296051437e-05,
      "loss": 0.2197,
      "step": 2640
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.371262493494068,
      "learning_rate": 4.77573499424663e-05,
      "loss": 0.3494,
      "step": 2641
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3383019821069815,
      "learning_rate": 4.775529602844505e-05,
      "loss": 0.2481,
      "step": 2642
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.747248479946469,
      "learning_rate": 4.77532412185315e-05,
      "loss": 0.4695,
      "step": 2643
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1258352561380975,
      "learning_rate": 4.775118551280652e-05,
      "loss": 0.1942,
      "step": 2644
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.5793243574818066,
      "learning_rate": 4.774912891135107e-05,
      "loss": 0.253,
      "step": 2645
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7973030374243057,
      "learning_rate": 4.7747071414246114e-05,
      "loss": 0.1058,
      "step": 2646
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2169628411493019,
      "learning_rate": 4.774501302157265e-05,
      "loss": 0.1833,
      "step": 2647
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0052299016585784,
      "learning_rate": 4.774295373341172e-05,
      "loss": 0.1088,
      "step": 2648
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9127110947930006,
      "learning_rate": 4.77408935498444e-05,
      "loss": 0.2305,
      "step": 2649
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3164915750545105,
      "learning_rate": 4.773883247095181e-05,
      "loss": 0.2922,
      "step": 2650
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3217894558255054,
      "learning_rate": 4.7736770496815084e-05,
      "loss": 0.2476,
      "step": 2651
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.353919508679148,
      "learning_rate": 4.7734707627515416e-05,
      "loss": 0.2091,
      "step": 2652
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.882197858718005,
      "learning_rate": 4.7732643863134016e-05,
      "loss": 0.185,
      "step": 2653
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.273905185211497,
      "learning_rate": 4.7730579203752125e-05,
      "loss": 0.3356,
      "step": 2654
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3428763074903296,
      "learning_rate": 4.7728513649451045e-05,
      "loss": 0.3663,
      "step": 2655
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5966137356720618,
      "learning_rate": 4.77264472003121e-05,
      "loss": 0.2434,
      "step": 2656
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.010328794081537,
      "learning_rate": 4.772437985641663e-05,
      "loss": 0.1859,
      "step": 2657
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0529630689034133,
      "learning_rate": 4.772231161784605e-05,
      "loss": 0.1595,
      "step": 2658
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8316988048938364,
      "learning_rate": 4.7720242484681765e-05,
      "loss": 0.1707,
      "step": 2659
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0451615028325647,
      "learning_rate": 4.771817245700526e-05,
      "loss": 0.4905,
      "step": 2660
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1030086831456507,
      "learning_rate": 4.7716101534898025e-05,
      "loss": 0.2064,
      "step": 2661
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.004726399874203,
      "learning_rate": 4.7714029718441586e-05,
      "loss": 0.3845,
      "step": 2662
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0424911484464128,
      "learning_rate": 4.771195700771752e-05,
      "loss": 0.138,
      "step": 2663
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.249918883803028,
      "learning_rate": 4.770988340280743e-05,
      "loss": 0.278,
      "step": 2664
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.019701125811718,
      "learning_rate": 4.7707808903792956e-05,
      "loss": 0.1865,
      "step": 2665
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1468977839215813,
      "learning_rate": 4.770573351075577e-05,
      "loss": 0.3608,
      "step": 2666
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7492818084684729,
      "learning_rate": 4.7703657223777585e-05,
      "loss": 0.1391,
      "step": 2667
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1428340695103283,
      "learning_rate": 4.770158004294014e-05,
      "loss": 0.1482,
      "step": 2668
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9440936708013165,
      "learning_rate": 4.769950196832522e-05,
      "loss": 0.1459,
      "step": 2669
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9738292648786793,
      "learning_rate": 4.769742300001464e-05,
      "loss": 0.1273,
      "step": 2670
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0892381762096666,
      "learning_rate": 4.769534313809025e-05,
      "loss": 0.2214,
      "step": 2671
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2018706471530294,
      "learning_rate": 4.769326238263394e-05,
      "loss": 0.2351,
      "step": 2672
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0750371555382394,
      "learning_rate": 4.769118073372763e-05,
      "loss": 0.1444,
      "step": 2673
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8743914673224242,
      "learning_rate": 4.7689098191453256e-05,
      "loss": 0.1768,
      "step": 2674
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2652353437378576,
      "learning_rate": 4.768701475589284e-05,
      "loss": 0.2915,
      "step": 2675
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0368490768940184,
      "learning_rate": 4.768493042712838e-05,
      "loss": 0.1918,
      "step": 2676
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0925772195725212,
      "learning_rate": 4.7682845205241965e-05,
      "loss": 0.1508,
      "step": 2677
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.222050032765253,
      "learning_rate": 4.7680759090315675e-05,
      "loss": 0.18,
      "step": 2678
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.381445309895366,
      "learning_rate": 4.767867208243164e-05,
      "loss": 0.3351,
      "step": 2679
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9998454252804758,
      "learning_rate": 4.767658418167203e-05,
      "loss": 0.3619,
      "step": 2680
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9267198579426894,
      "learning_rate": 4.7674495388119053e-05,
      "loss": 0.134,
      "step": 2681
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0412320102356385,
      "learning_rate": 4.767240570185495e-05,
      "loss": 0.149,
      "step": 2682
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3599130960005212,
      "learning_rate": 4.7670315122961975e-05,
      "loss": 0.2802,
      "step": 2683
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8449013091094293,
      "learning_rate": 4.766822365152245e-05,
      "loss": 0.202,
      "step": 2684
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1783552572436773,
      "learning_rate": 4.766613128761872e-05,
      "loss": 0.2009,
      "step": 2685
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9494985190020789,
      "learning_rate": 4.7664038031333145e-05,
      "loss": 0.1612,
      "step": 2686
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2365163848735101,
      "learning_rate": 4.7661943882748155e-05,
      "loss": 0.1313,
      "step": 2687
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0550965143895248,
      "learning_rate": 4.7659848841946195e-05,
      "loss": 0.279,
      "step": 2688
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8530803756126442,
      "learning_rate": 4.765775290900974e-05,
      "loss": 0.2118,
      "step": 2689
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.025749381728045,
      "learning_rate": 4.765565608402133e-05,
      "loss": 0.1604,
      "step": 2690
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1908381495215974,
      "learning_rate": 4.765355836706349e-05,
      "loss": 0.1236,
      "step": 2691
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1827864642037036,
      "learning_rate": 4.765145975821882e-05,
      "loss": 0.2225,
      "step": 2692
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9494707925342367,
      "learning_rate": 4.7649360257569955e-05,
      "loss": 0.3204,
      "step": 2693
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.966437242530297,
      "learning_rate": 4.764725986519954e-05,
      "loss": 0.1464,
      "step": 2694
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0579767387214702,
      "learning_rate": 4.7645158581190275e-05,
      "loss": 0.2686,
      "step": 2695
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9028134626455603,
      "learning_rate": 4.7643056405624886e-05,
      "loss": 0.165,
      "step": 2696
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9098193790788149,
      "learning_rate": 4.764095333858614e-05,
      "loss": 0.1467,
      "step": 2697
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9839315024626225,
      "learning_rate": 4.763884938015683e-05,
      "loss": 0.1613,
      "step": 2698
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1471125681573846,
      "learning_rate": 4.7636744530419795e-05,
      "loss": 0.2242,
      "step": 2699
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8465251863790705,
      "learning_rate": 4.763463878945791e-05,
      "loss": 0.1164,
      "step": 2700
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.376554164487828,
      "learning_rate": 4.763253215735408e-05,
      "loss": 0.8985,
      "step": 2701
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8694183195365406,
      "learning_rate": 4.763042463419123e-05,
      "loss": 0.1459,
      "step": 2702
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6595029975686046,
      "learning_rate": 4.7628316220052346e-05,
      "loss": 0.4039,
      "step": 2703
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8963334906079046,
      "learning_rate": 4.762620691502043e-05,
      "loss": 0.1016,
      "step": 2704
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0519715114049928,
      "learning_rate": 4.7624096719178534e-05,
      "loss": 0.2112,
      "step": 2705
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.23627226481902,
      "learning_rate": 4.7621985632609746e-05,
      "loss": 0.2562,
      "step": 2706
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.071805274291091,
      "learning_rate": 4.7619873655397165e-05,
      "loss": 0.3428,
      "step": 2707
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1447715659379325,
      "learning_rate": 4.761776078762394e-05,
      "loss": 0.2236,
      "step": 2708
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.745478894927679,
      "learning_rate": 4.761564702937327e-05,
      "loss": 0.5246,
      "step": 2709
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8003311715542716,
      "learning_rate": 4.761353238072836e-05,
      "loss": 0.2961,
      "step": 2710
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0627126468365204,
      "learning_rate": 4.761141684177248e-05,
      "loss": 0.2373,
      "step": 2711
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.060814156872276,
      "learning_rate": 4.760930041258892e-05,
      "loss": 0.1438,
      "step": 2712
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8422498631423695,
      "learning_rate": 4.7607183093261e-05,
      "loss": 0.1308,
      "step": 2713
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7687854650578567,
      "learning_rate": 4.760506488387207e-05,
      "loss": 0.5662,
      "step": 2714
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7210439341077284,
      "learning_rate": 4.760294578450554e-05,
      "loss": 0.3885,
      "step": 2715
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4359460554801566,
      "learning_rate": 4.760082579524483e-05,
      "loss": 0.182,
      "step": 2716
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9486100987109635,
      "learning_rate": 4.759870491617342e-05,
      "loss": 0.1923,
      "step": 2717
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4901827225268804,
      "learning_rate": 4.759658314737479e-05,
      "loss": 0.1806,
      "step": 2718
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2570598482414828,
      "learning_rate": 4.759446048893249e-05,
      "loss": 0.2506,
      "step": 2719
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9523588755927477,
      "learning_rate": 4.75923369409301e-05,
      "loss": 0.1567,
      "step": 2720
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2447635971839437,
      "learning_rate": 4.759021250345121e-05,
      "loss": 0.2615,
      "step": 2721
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0087283688821402,
      "learning_rate": 4.758808717657945e-05,
      "loss": 0.1959,
      "step": 2722
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8554203982454026,
      "learning_rate": 4.758596096039852e-05,
      "loss": 0.127,
      "step": 2723
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2024966481066346,
      "learning_rate": 4.758383385499212e-05,
      "loss": 0.2562,
      "step": 2724
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9351723089304345,
      "learning_rate": 4.758170586044399e-05,
      "loss": 0.1524,
      "step": 2725
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7299852356073983,
      "learning_rate": 4.757957697683792e-05,
      "loss": 0.0857,
      "step": 2726
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8999470464048716,
      "learning_rate": 4.757744720425773e-05,
      "loss": 0.1356,
      "step": 2727
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9267897587210965,
      "learning_rate": 4.757531654278725e-05,
      "loss": 0.1473,
      "step": 2728
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8076464592751684,
      "learning_rate": 4.757318499251039e-05,
      "loss": 0.1063,
      "step": 2729
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7657945929779507,
      "learning_rate": 4.757105255351105e-05,
      "loss": 0.0937,
      "step": 2730
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7724301398331959,
      "learning_rate": 4.75689192258732e-05,
      "loss": 0.352,
      "step": 2731
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8287865350424338,
      "learning_rate": 4.7566785009680827e-05,
      "loss": 0.1352,
      "step": 2732
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4872938010696384,
      "learning_rate": 4.756464990501795e-05,
      "loss": 0.3773,
      "step": 2733
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3793169943162813,
      "learning_rate": 4.7562513911968634e-05,
      "loss": 0.1715,
      "step": 2734
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8367620673276737,
      "learning_rate": 4.756037703061698e-05,
      "loss": 0.0957,
      "step": 2735
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.426372175966948,
      "learning_rate": 4.755823926104711e-05,
      "loss": 0.3326,
      "step": 2736
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9847504339923538,
      "learning_rate": 4.75561006033432e-05,
      "loss": 0.1377,
      "step": 2737
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4936275042017115,
      "learning_rate": 4.755396105758944e-05,
      "loss": 0.4459,
      "step": 2738
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.838671979779754,
      "learning_rate": 4.7551820623870065e-05,
      "loss": 0.1528,
      "step": 2739
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9462234252757351,
      "learning_rate": 4.754967930226936e-05,
      "loss": 0.1614,
      "step": 2740
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0434957694778886,
      "learning_rate": 4.7547537092871606e-05,
      "loss": 0.184,
      "step": 2741
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.378499765298319,
      "learning_rate": 4.754539399576117e-05,
      "loss": 0.3254,
      "step": 2742
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4430788824291096,
      "learning_rate": 4.7543250011022405e-05,
      "loss": 0.3409,
      "step": 2743
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0460398655825505,
      "learning_rate": 4.754110513873974e-05,
      "loss": 0.2345,
      "step": 2744
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7121719029604466,
      "learning_rate": 4.75389593789976e-05,
      "loss": 0.0494,
      "step": 2745
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.931221105237122,
      "learning_rate": 4.7536812731880487e-05,
      "loss": 0.1621,
      "step": 2746
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.951664864522014,
      "learning_rate": 4.753466519747289e-05,
      "loss": 0.1817,
      "step": 2747
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8552233449142579,
      "learning_rate": 4.7532516775859394e-05,
      "loss": 0.0758,
      "step": 2748
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4279367318772704,
      "learning_rate": 4.753036746712455e-05,
      "loss": 0.2364,
      "step": 2749
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.72841074220806,
      "learning_rate": 4.7528217271353005e-05,
      "loss": 0.3461,
      "step": 2750
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1417744200914057,
      "learning_rate": 4.752606618862939e-05,
      "loss": 0.2455,
      "step": 2751
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5624701947303037,
      "learning_rate": 4.752391421903841e-05,
      "loss": 0.3185,
      "step": 2752
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1843222929467585,
      "learning_rate": 4.752176136266478e-05,
      "loss": 0.2342,
      "step": 2753
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3888345348097104,
      "learning_rate": 4.751960761959326e-05,
      "loss": 0.1751,
      "step": 2754
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1913230965572355,
      "learning_rate": 4.7517452989908665e-05,
      "loss": 0.2164,
      "step": 2755
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.095501475917447,
      "learning_rate": 4.75152974736958e-05,
      "loss": 0.4473,
      "step": 2756
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.716289341460436,
      "learning_rate": 4.751314107103954e-05,
      "loss": 0.4618,
      "step": 2757
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.469402005865431,
      "learning_rate": 4.751098378202477e-05,
      "loss": 0.4754,
      "step": 2758
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9170696428115753,
      "learning_rate": 4.750882560673644e-05,
      "loss": 0.1277,
      "step": 2759
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1828704251352804,
      "learning_rate": 4.750666654525952e-05,
      "loss": 0.208,
      "step": 2760
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8555178310547185,
      "learning_rate": 4.7504506597679006e-05,
      "loss": 0.148,
      "step": 2761
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3677901147335234,
      "learning_rate": 4.750234576407994e-05,
      "loss": 0.3279,
      "step": 2762
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2666575170352277,
      "learning_rate": 4.750018404454739e-05,
      "loss": 0.291,
      "step": 2763
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2467447465148829,
      "learning_rate": 4.7498021439166474e-05,
      "loss": 0.1574,
      "step": 2764
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2456787297918541,
      "learning_rate": 4.7495857948022323e-05,
      "loss": 0.159,
      "step": 2765
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7704196390570156,
      "learning_rate": 4.7493693571200134e-05,
      "loss": 0.1077,
      "step": 2766
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9534084190683816,
      "learning_rate": 4.7491528308785096e-05,
      "loss": 0.2251,
      "step": 2767
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.478615591905268,
      "learning_rate": 4.7489362160862476e-05,
      "loss": 0.3417,
      "step": 2768
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.829379430679548,
      "learning_rate": 4.7487195127517545e-05,
      "loss": 0.3713,
      "step": 2769
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9621292435006404,
      "learning_rate": 4.748502720883564e-05,
      "loss": 0.1701,
      "step": 2770
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.2553334030595193,
      "learning_rate": 4.748285840490209e-05,
      "loss": 0.4935,
      "step": 2771
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4256546955132672,
      "learning_rate": 4.748068871580228e-05,
      "loss": 0.2648,
      "step": 2772
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0034031132202221,
      "learning_rate": 4.7478518141621665e-05,
      "loss": 0.1119,
      "step": 2773
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0441103968263314,
      "learning_rate": 4.747634668244567e-05,
      "loss": 0.1136,
      "step": 2774
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8228451051396253,
      "learning_rate": 4.74741743383598e-05,
      "loss": 0.216,
      "step": 2775
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.9163392480296113,
      "learning_rate": 4.747200110944958e-05,
      "loss": 0.6854,
      "step": 2776
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.936603762553329,
      "learning_rate": 4.7469826995800564e-05,
      "loss": 0.123,
      "step": 2777
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.889377930717304,
      "learning_rate": 4.746765199749836e-05,
      "loss": 0.1173,
      "step": 2778
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9367566111547371,
      "learning_rate": 4.74654761146286e-05,
      "loss": 0.2364,
      "step": 2779
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9375424881904373,
      "learning_rate": 4.746329934727694e-05,
      "loss": 0.1484,
      "step": 2780
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3219236277195838,
      "learning_rate": 4.7461121695529086e-05,
      "loss": 0.1686,
      "step": 2781
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3558966827061,
      "learning_rate": 4.7458943159470776e-05,
      "loss": 0.2649,
      "step": 2782
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.6928722892204116,
      "learning_rate": 4.745676373918778e-05,
      "loss": 1.0563,
      "step": 2783
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9501750880808996,
      "learning_rate": 4.7454583434765896e-05,
      "loss": 0.1078,
      "step": 2784
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9713671999380794,
      "learning_rate": 4.745240224629097e-05,
      "loss": 0.1475,
      "step": 2785
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0562012998385022,
      "learning_rate": 4.745022017384888e-05,
      "loss": 0.1709,
      "step": 2786
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1733578971150245,
      "learning_rate": 4.744803721752553e-05,
      "loss": 0.2592,
      "step": 2787
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.019673355013092,
      "learning_rate": 4.744585337740687e-05,
      "loss": 0.2256,
      "step": 2788
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.068729356141102,
      "learning_rate": 4.744366865357886e-05,
      "loss": 0.3698,
      "step": 2789
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9731026220787469,
      "learning_rate": 4.7441483046127554e-05,
      "loss": 0.2036,
      "step": 2790
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9478189099140635,
      "learning_rate": 4.743929655513897e-05,
      "loss": 0.4768,
      "step": 2791
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0935280712985818,
      "learning_rate": 4.743710918069919e-05,
      "loss": 0.2018,
      "step": 2792
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1951611744400226,
      "learning_rate": 4.7434920922894346e-05,
      "loss": 0.4288,
      "step": 2793
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.209541584656268,
      "learning_rate": 4.743273178181059e-05,
      "loss": 0.105,
      "step": 2794
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9715430706289827,
      "learning_rate": 4.74305417575341e-05,
      "loss": 1.1023,
      "step": 2795
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.843754960280237,
      "learning_rate": 4.742835085015112e-05,
      "loss": 0.3839,
      "step": 2796
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2224655007487373,
      "learning_rate": 4.742615905974788e-05,
      "loss": 0.1881,
      "step": 2797
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5795533413267409,
      "learning_rate": 4.742396638641069e-05,
      "loss": 0.2603,
      "step": 2798
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.185617862872508,
      "learning_rate": 4.7421772830225866e-05,
      "loss": 0.2218,
      "step": 2799
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8217812004366736,
      "learning_rate": 4.7419578391279784e-05,
      "loss": 0.4391,
      "step": 2800
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.375440791278468,
      "learning_rate": 4.7417383069658835e-05,
      "loss": 0.2789,
      "step": 2801
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.914389225423815,
      "learning_rate": 4.7415186865449444e-05,
      "loss": 0.1694,
      "step": 2802
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.39328804600653,
      "learning_rate": 4.741298977873808e-05,
      "loss": 0.6868,
      "step": 2803
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5018606223081872,
      "learning_rate": 4.741079180961125e-05,
      "loss": 0.2071,
      "step": 2804
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.921900634091636,
      "learning_rate": 4.740859295815549e-05,
      "loss": 0.4725,
      "step": 2805
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4505964297575034,
      "learning_rate": 4.740639322445735e-05,
      "loss": 0.3311,
      "step": 2806
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5510096509938134,
      "learning_rate": 4.740419260860346e-05,
      "loss": 0.2548,
      "step": 2807
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6663212451712857,
      "learning_rate": 4.7401991110680453e-05,
      "loss": 0.4657,
      "step": 2808
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1850295567696192,
      "learning_rate": 4.7399788730775e-05,
      "loss": 0.1357,
      "step": 2809
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9093777959704124,
      "learning_rate": 4.7397585468973805e-05,
      "loss": 0.0874,
      "step": 2810
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.765474948131659,
      "learning_rate": 4.739538132536362e-05,
      "loss": 0.708,
      "step": 2811
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.885122197043401,
      "learning_rate": 4.7393176300031226e-05,
      "loss": 0.843,
      "step": 2812
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0367629359661015,
      "learning_rate": 4.7390970393063436e-05,
      "loss": 0.1787,
      "step": 2813
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8567236048717017,
      "learning_rate": 4.7388763604547094e-05,
      "loss": 0.1056,
      "step": 2814
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0571541254147339,
      "learning_rate": 4.7386555934569074e-05,
      "loss": 0.1344,
      "step": 2815
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1206025173181857,
      "learning_rate": 4.738434738321631e-05,
      "loss": 0.24,
      "step": 2816
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2369666938670467,
      "learning_rate": 4.738213795057574e-05,
      "loss": 0.41,
      "step": 2817
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.308385634587871,
      "learning_rate": 4.737992763673437e-05,
      "loss": 0.2764,
      "step": 2818
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.307772554726186,
      "learning_rate": 4.7377716441779204e-05,
      "loss": 0.2412,
      "step": 2819
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5689749312636037,
      "learning_rate": 4.73755043657973e-05,
      "loss": 0.2861,
      "step": 2820
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.739850619461903,
      "learning_rate": 4.737329140887576e-05,
      "loss": 0.3452,
      "step": 2821
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4213689965819902,
      "learning_rate": 4.7371077571101693e-05,
      "loss": 0.255,
      "step": 2822
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0095485639272381,
      "learning_rate": 4.736886285256228e-05,
      "loss": 0.1844,
      "step": 2823
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8220849909461868,
      "learning_rate": 4.7366647253344705e-05,
      "loss": 0.0866,
      "step": 2824
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7711118513341213,
      "learning_rate": 4.7364430773536195e-05,
      "loss": 0.2568,
      "step": 2825
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.3778519357001184,
      "learning_rate": 4.736221341322401e-05,
      "loss": 0.2863,
      "step": 2826
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2293235214585696,
      "learning_rate": 4.735999517249546e-05,
      "loss": 0.2964,
      "step": 2827
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4307801466616061,
      "learning_rate": 4.735777605143789e-05,
      "loss": 0.2369,
      "step": 2828
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1720852611062111,
      "learning_rate": 4.7355556050138635e-05,
      "loss": 0.1598,
      "step": 2829
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.050377731462632,
      "learning_rate": 4.735333516868512e-05,
      "loss": 0.1838,
      "step": 2830
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2681070488447779,
      "learning_rate": 4.735111340716479e-05,
      "loss": 0.1413,
      "step": 2831
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.5607777344552582,
      "learning_rate": 4.73488907656651e-05,
      "loss": 0.4181,
      "step": 2832
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6011847214853903,
      "learning_rate": 4.734666724427356e-05,
      "loss": 0.303,
      "step": 2833
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.324745904229068,
      "learning_rate": 4.734444284307772e-05,
      "loss": 0.2163,
      "step": 2834
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4842007251005913,
      "learning_rate": 4.734221756216515e-05,
      "loss": 0.3012,
      "step": 2835
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2609152582854661,
      "learning_rate": 4.733999140162346e-05,
      "loss": 0.1708,
      "step": 2836
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.793830692764268,
      "learning_rate": 4.733776436154031e-05,
      "loss": 0.1073,
      "step": 2837
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3807534250150886,
      "learning_rate": 4.7335536442003353e-05,
      "loss": 0.2232,
      "step": 2838
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.682031915518108,
      "learning_rate": 4.733330764310033e-05,
      "loss": 0.4345,
      "step": 2839
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8037071948860829,
      "learning_rate": 4.733107796491897e-05,
      "loss": 0.0991,
      "step": 2840
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.901625968579164,
      "learning_rate": 4.732884740754707e-05,
      "loss": 0.1492,
      "step": 2841
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3857531867798072,
      "learning_rate": 4.732661597107245e-05,
      "loss": 0.1971,
      "step": 2842
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7133380085296852,
      "learning_rate": 4.7324383655582946e-05,
      "loss": 0.2765,
      "step": 2843
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.179559694735889,
      "learning_rate": 4.732215046116647e-05,
      "loss": 0.2821,
      "step": 2844
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7248839949108374,
      "learning_rate": 4.731991638791093e-05,
      "loss": 0.4523,
      "step": 2845
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1290235285517585,
      "learning_rate": 4.731768143590428e-05,
      "loss": 0.2024,
      "step": 2846
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9044236287304251,
      "learning_rate": 4.7315445605234525e-05,
      "loss": 0.1366,
      "step": 2847
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3957081310933313,
      "learning_rate": 4.731320889598968e-05,
      "loss": 0.2929,
      "step": 2848
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2206424310759778,
      "learning_rate": 4.731097130825781e-05,
      "loss": 0.1733,
      "step": 2849
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.093898170056122,
      "learning_rate": 4.730873284212701e-05,
      "loss": 0.1853,
      "step": 2850
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8865588978922223,
      "learning_rate": 4.730649349768541e-05,
      "loss": 0.1342,
      "step": 2851
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.853395718273984,
      "learning_rate": 4.7304253275021175e-05,
      "loss": 0.7259,
      "step": 2852
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5709819975694101,
      "learning_rate": 4.73020121742225e-05,
      "loss": 0.1972,
      "step": 2853
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5159279290376388,
      "learning_rate": 4.729977019537763e-05,
      "loss": 0.3575,
      "step": 2854
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6664471495771557,
      "learning_rate": 4.7297527338574823e-05,
      "loss": 0.1161,
      "step": 2855
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2946989212823334,
      "learning_rate": 4.729528360390238e-05,
      "loss": 0.2785,
      "step": 2856
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.667300064901507,
      "learning_rate": 4.729303899144864e-05,
      "loss": 0.6057,
      "step": 2857
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.979431225271062,
      "learning_rate": 4.729079350130199e-05,
      "loss": 0.1323,
      "step": 2858
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8728304412411625,
      "learning_rate": 4.728854713355082e-05,
      "loss": 0.3151,
      "step": 2859
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2421550750992552,
      "learning_rate": 4.728629988828357e-05,
      "loss": 0.1834,
      "step": 2860
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1794044605400313,
      "learning_rate": 4.7284051765588726e-05,
      "loss": 0.189,
      "step": 2861
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9088507617783637,
      "learning_rate": 4.728180276555479e-05,
      "loss": 0.1135,
      "step": 2862
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2768167526115894,
      "learning_rate": 4.7279552888270315e-05,
      "loss": 0.2742,
      "step": 2863
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9714352220046002,
      "learning_rate": 4.7277302133823875e-05,
      "loss": 0.1597,
      "step": 2864
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5148522209151398,
      "learning_rate": 4.7275050502304084e-05,
      "loss": 0.3048,
      "step": 2865
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9161857914246689,
      "learning_rate": 4.727279799379959e-05,
      "loss": 0.1617,
      "step": 2866
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.382610539476269,
      "learning_rate": 4.727054460839908e-05,
      "loss": 0.2749,
      "step": 2867
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8420841667564254,
      "learning_rate": 4.7268290346191265e-05,
      "loss": 0.1098,
      "step": 2868
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.459126374970266,
      "learning_rate": 4.72660352072649e-05,
      "loss": 0.633,
      "step": 2869
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3088127913635164,
      "learning_rate": 4.7263779191708766e-05,
      "loss": 0.2525,
      "step": 2870
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2226258918161927,
      "learning_rate": 4.7261522299611696e-05,
      "loss": 0.2373,
      "step": 2871
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.833700007006543,
      "learning_rate": 4.7259264531062544e-05,
      "loss": 0.3451,
      "step": 2872
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4873847474834656,
      "learning_rate": 4.725700588615019e-05,
      "loss": 0.3084,
      "step": 2873
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1798667708199935,
      "learning_rate": 4.725474636496355e-05,
      "loss": 0.2075,
      "step": 2874
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3585450896792541,
      "learning_rate": 4.725248596759161e-05,
      "loss": 0.2382,
      "step": 2875
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.316701749860798,
      "learning_rate": 4.7250224694123344e-05,
      "loss": 0.3025,
      "step": 2876
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.743613910970847,
      "learning_rate": 4.72479625446478e-05,
      "loss": 0.1107,
      "step": 2877
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2712876967681899,
      "learning_rate": 4.7245699519254003e-05,
      "loss": 0.2405,
      "step": 2878
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.218390768804121,
      "learning_rate": 4.724343561803109e-05,
      "loss": 0.3729,
      "step": 2879
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7603316425902318,
      "learning_rate": 4.724117084106817e-05,
      "loss": 0.1042,
      "step": 2880
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.030878798969395,
      "learning_rate": 4.723890518845442e-05,
      "loss": 0.2909,
      "step": 2881
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1102170043586792,
      "learning_rate": 4.7236638660279026e-05,
      "loss": 0.1755,
      "step": 2882
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8769939485276816,
      "learning_rate": 4.7234371256631234e-05,
      "loss": 0.1157,
      "step": 2883
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3497427974748182,
      "learning_rate": 4.7232102977600315e-05,
      "loss": 0.2272,
      "step": 2884
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.335202556612719,
      "learning_rate": 4.722983382327556e-05,
      "loss": 0.546,
      "step": 2885
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.89016354209023,
      "learning_rate": 4.722756379374632e-05,
      "loss": 0.1496,
      "step": 2886
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9691821696578371,
      "learning_rate": 4.722529288910197e-05,
      "loss": 0.1827,
      "step": 2887
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9937592016999147,
      "learning_rate": 4.7223021109431896e-05,
      "loss": 0.169,
      "step": 2888
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4672299004043845,
      "learning_rate": 4.722074845482557e-05,
      "loss": 0.2598,
      "step": 2889
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7336840735484139,
      "learning_rate": 4.721847492537245e-05,
      "loss": 0.0884,
      "step": 2890
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.149335799208993,
      "learning_rate": 4.721620052116204e-05,
      "loss": 0.132,
      "step": 2891
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7515705913358255,
      "learning_rate": 4.7213925242283895e-05,
      "loss": 0.1578,
      "step": 2892
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8261974083536339,
      "learning_rate": 4.7211649088827594e-05,
      "loss": 0.1903,
      "step": 2893
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7658640606256394,
      "learning_rate": 4.7209372060882755e-05,
      "loss": 0.1514,
      "step": 2894
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7462181129005407,
      "learning_rate": 4.720709415853902e-05,
      "loss": 0.1318,
      "step": 2895
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9632394603400618,
      "learning_rate": 4.720481538188607e-05,
      "loss": 0.1694,
      "step": 2896
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9639592214905492,
      "learning_rate": 4.720253573101363e-05,
      "loss": 0.1564,
      "step": 2897
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0713934298166914,
      "learning_rate": 4.7200255206011436e-05,
      "loss": 0.2584,
      "step": 2898
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1777926131035672,
      "learning_rate": 4.719797380696929e-05,
      "loss": 0.1965,
      "step": 2899
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8221878674141913,
      "learning_rate": 4.719569153397701e-05,
      "loss": 0.1285,
      "step": 2900
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.6416790552206764,
      "learning_rate": 4.7193408387124445e-05,
      "loss": 0.4373,
      "step": 2901
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9505210707678027,
      "learning_rate": 4.719112436650148e-05,
      "loss": 0.1405,
      "step": 2902
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.2807575845816657,
      "learning_rate": 4.718883947219806e-05,
      "loss": 0.5518,
      "step": 2903
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1953385162182604,
      "learning_rate": 4.718655370430411e-05,
      "loss": 0.2019,
      "step": 2904
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9423190654177085,
      "learning_rate": 4.718426706290965e-05,
      "loss": 0.1194,
      "step": 2905
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.256821085004945,
      "learning_rate": 4.718197954810469e-05,
      "loss": 0.6621,
      "step": 2906
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3366199636929115,
      "learning_rate": 4.71796911599793e-05,
      "loss": 0.2262,
      "step": 2907
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8741580890053252,
      "learning_rate": 4.7177401898623586e-05,
      "loss": 0.137,
      "step": 2908
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1451257701783881,
      "learning_rate": 4.717511176412766e-05,
      "loss": 0.2169,
      "step": 2909
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.086302613125262,
      "learning_rate": 4.717282075658168e-05,
      "loss": 0.1887,
      "step": 2910
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4626249800222966,
      "learning_rate": 4.717052887607586e-05,
      "loss": 0.2058,
      "step": 2911
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0195586912120558,
      "learning_rate": 4.716823612270043e-05,
      "loss": 0.1647,
      "step": 2912
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.96901833631213,
      "learning_rate": 4.716594249654566e-05,
      "loss": 0.1343,
      "step": 2913
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4896796269172925,
      "learning_rate": 4.7163647997701845e-05,
      "loss": 0.4598,
      "step": 2914
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.75039789567677,
      "learning_rate": 4.716135262625933e-05,
      "loss": 0.3198,
      "step": 2915
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4460889320574615,
      "learning_rate": 4.715905638230848e-05,
      "loss": 0.1361,
      "step": 2916
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9822201851138084,
      "learning_rate": 4.7156759265939685e-05,
      "loss": 0.389,
      "step": 2917
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9801189160419712,
      "learning_rate": 4.715446127724341e-05,
      "loss": 0.1183,
      "step": 2918
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.17561613558927,
      "learning_rate": 4.715216241631012e-05,
      "loss": 0.242,
      "step": 2919
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.128417029612515,
      "learning_rate": 4.714986268323032e-05,
      "loss": 0.1896,
      "step": 2920
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1593466281503928,
      "learning_rate": 4.7147562078094546e-05,
      "loss": 0.155,
      "step": 2921
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5849719698572842,
      "learning_rate": 4.714526060099338e-05,
      "loss": 0.2406,
      "step": 2922
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3573569264823055,
      "learning_rate": 4.714295825201743e-05,
      "loss": 0.2984,
      "step": 2923
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.910118436632627,
      "learning_rate": 4.714065503125735e-05,
      "loss": 0.1961,
      "step": 2924
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2713531894401973,
      "learning_rate": 4.713835093880381e-05,
      "loss": 0.3145,
      "step": 2925
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.854914216884508,
      "learning_rate": 4.713604597474753e-05,
      "loss": 0.1066,
      "step": 2926
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0577995226979626,
      "learning_rate": 4.713374013917926e-05,
      "loss": 0.1379,
      "step": 2927
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0880364720448534,
      "learning_rate": 4.713143343218978e-05,
      "loss": 0.1432,
      "step": 2928
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3923045081658876,
      "learning_rate": 4.7129125853869897e-05,
      "loss": 0.2125,
      "step": 2929
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1242746406407553,
      "learning_rate": 4.7126817404310466e-05,
      "loss": 0.1544,
      "step": 2930
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8405018201356741,
      "learning_rate": 4.712450808360238e-05,
      "loss": 0.1292,
      "step": 2931
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1175149537588194,
      "learning_rate": 4.7122197891836565e-05,
      "loss": 0.1857,
      "step": 2932
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0436425923033787,
      "learning_rate": 4.711988682910395e-05,
      "loss": 0.2345,
      "step": 2933
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0712566947247915,
      "learning_rate": 4.7117574895495544e-05,
      "loss": 0.1367,
      "step": 2934
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1598565279349662,
      "learning_rate": 4.7115262091102366e-05,
      "loss": 0.1598,
      "step": 2935
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2885934201274165,
      "learning_rate": 4.711294841601546e-05,
      "loss": 0.3304,
      "step": 2936
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9181516243464897,
      "learning_rate": 4.711063387032594e-05,
      "loss": 0.1578,
      "step": 2937
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8499342084145424,
      "learning_rate": 4.710831845412491e-05,
      "loss": 0.1499,
      "step": 2938
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.761264526114105,
      "learning_rate": 4.7106002167503535e-05,
      "loss": 0.1886,
      "step": 2939
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0211178605154942,
      "learning_rate": 4.710368501055302e-05,
      "loss": 0.1594,
      "step": 2940
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.51442119480348,
      "learning_rate": 4.7101366983364584e-05,
      "loss": 0.4338,
      "step": 2941
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.629812147873258,
      "learning_rate": 4.7099048086029486e-05,
      "loss": 0.49,
      "step": 2942
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0971421741062828,
      "learning_rate": 4.7096728318639025e-05,
      "loss": 0.1764,
      "step": 2943
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0424006347364188,
      "learning_rate": 4.7094407681284544e-05,
      "loss": 0.2388,
      "step": 2944
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9646403899675094,
      "learning_rate": 4.7092086174057394e-05,
      "loss": 0.1275,
      "step": 2945
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.154400432491111,
      "learning_rate": 4.7089763797048974e-05,
      "loss": 0.1423,
      "step": 2946
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9603656155826177,
      "learning_rate": 4.708744055035073e-05,
      "loss": 0.1093,
      "step": 2947
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8139176194100475,
      "learning_rate": 4.708511643405411e-05,
      "loss": 0.1603,
      "step": 2948
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.189788807911476,
      "learning_rate": 4.708279144825064e-05,
      "loss": 0.2353,
      "step": 2949
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6972951972267443,
      "learning_rate": 4.7080465593031844e-05,
      "loss": 0.0972,
      "step": 2950
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1972076210124363,
      "learning_rate": 4.7078138868489284e-05,
      "loss": 0.1858,
      "step": 2951
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8713259465599941,
      "learning_rate": 4.7075811274714575e-05,
      "loss": 0.1127,
      "step": 2952
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8601434359844583,
      "learning_rate": 4.7073482811799365e-05,
      "loss": 0.165,
      "step": 2953
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.014264368851175,
      "learning_rate": 4.7071153479835304e-05,
      "loss": 0.6485,
      "step": 2954
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2018820657010216,
      "learning_rate": 4.706882327891412e-05,
      "loss": 0.1564,
      "step": 2955
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6857852236589501,
      "learning_rate": 4.706649220912755e-05,
      "loss": 0.1054,
      "step": 2956
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4648664140458616,
      "learning_rate": 4.706416027056736e-05,
      "loss": 0.2636,
      "step": 2957
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6425970258871064,
      "learning_rate": 4.706182746332537e-05,
      "loss": 0.1112,
      "step": 2958
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.830524292868232,
      "learning_rate": 4.705949378749343e-05,
      "loss": 0.3216,
      "step": 2959
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7927503327340782,
      "learning_rate": 4.705715924316341e-05,
      "loss": 0.0823,
      "step": 2960
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.729392201012853,
      "learning_rate": 4.705482383042721e-05,
      "loss": 0.1079,
      "step": 2961
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.458777826645148,
      "learning_rate": 4.7052487549376796e-05,
      "loss": 0.182,
      "step": 2962
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9634286290835157,
      "learning_rate": 4.705015040010414e-05,
      "loss": 0.2012,
      "step": 2963
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7516212440749948,
      "learning_rate": 4.704781238270126e-05,
      "loss": 0.0826,
      "step": 2964
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9656795779856795,
      "learning_rate": 4.7045473497260214e-05,
      "loss": 0.1328,
      "step": 2965
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.123897149848455,
      "learning_rate": 4.704313374387307e-05,
      "loss": 0.7042,
      "step": 2966
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8708970075597656,
      "learning_rate": 4.704079312263195e-05,
      "loss": 0.0876,
      "step": 2967
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2492246275134526,
      "learning_rate": 4.703845163362901e-05,
      "loss": 0.2645,
      "step": 2968
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.222401735264598,
      "learning_rate": 4.703610927695644e-05,
      "loss": 0.451,
      "step": 2969
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0966682917055788,
      "learning_rate": 4.703376605270645e-05,
      "loss": 0.1801,
      "step": 2970
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6246474791356442,
      "learning_rate": 4.703142196097131e-05,
      "loss": 0.2861,
      "step": 2971
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1118447259803226,
      "learning_rate": 4.702907700184328e-05,
      "loss": 0.143,
      "step": 2972
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1579220581424055,
      "learning_rate": 4.7026731175414716e-05,
      "loss": 0.2703,
      "step": 2973
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.404478032865518,
      "learning_rate": 4.702438448177795e-05,
      "loss": 0.3261,
      "step": 2974
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1050431646439487,
      "learning_rate": 4.702203692102539e-05,
      "loss": 0.1639,
      "step": 2975
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.208480968400657,
      "learning_rate": 4.701968849324945e-05,
      "loss": 0.3946,
      "step": 2976
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0596577112619867,
      "learning_rate": 4.701733919854259e-05,
      "loss": 0.1522,
      "step": 2977
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7618844685868058,
      "learning_rate": 4.7014989036997314e-05,
      "loss": 0.3853,
      "step": 2978
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.3627981075762765,
      "learning_rate": 4.701263800870613e-05,
      "loss": 0.402,
      "step": 2979
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3608714882154236,
      "learning_rate": 4.701028611376163e-05,
      "loss": 0.2249,
      "step": 2980
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.019059017480566,
      "learning_rate": 4.7007933352256376e-05,
      "loss": 0.1624,
      "step": 2981
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9162724180698998,
      "learning_rate": 4.7005579724283025e-05,
      "loss": 0.1152,
      "step": 2982
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1063408160748205,
      "learning_rate": 4.7003225229934215e-05,
      "loss": 0.1879,
      "step": 2983
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3741268740698918,
      "learning_rate": 4.7000869869302664e-05,
      "loss": 0.1957,
      "step": 2984
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.571606341934553,
      "learning_rate": 4.699851364248111e-05,
      "loss": 0.7398,
      "step": 2985
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.022011172182063,
      "learning_rate": 4.69961565495623e-05,
      "loss": 0.1332,
      "step": 2986
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1118434887044175,
      "learning_rate": 4.699379859063904e-05,
      "loss": 0.1772,
      "step": 2987
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3011527573215533,
      "learning_rate": 4.699143976580417e-05,
      "loss": 0.2469,
      "step": 2988
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0146127965139862,
      "learning_rate": 4.698908007515055e-05,
      "loss": 0.1841,
      "step": 2989
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2455736269131412,
      "learning_rate": 4.6986719518771096e-05,
      "loss": 0.2608,
      "step": 2990
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8342507236338245,
      "learning_rate": 4.698435809675873e-05,
      "loss": 0.1665,
      "step": 2991
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1275591801270137,
      "learning_rate": 4.698199580920644e-05,
      "loss": 0.2633,
      "step": 2992
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.585120665744948,
      "learning_rate": 4.697963265620722e-05,
      "loss": 0.2324,
      "step": 2993
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8589075328565623,
      "learning_rate": 4.69772686378541e-05,
      "loss": 0.1584,
      "step": 2994
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0202452050412498,
      "learning_rate": 4.697490375424017e-05,
      "loss": 0.1647,
      "step": 2995
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.980688763194602,
      "learning_rate": 4.6972538005458526e-05,
      "loss": 0.1384,
      "step": 2996
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4934977843731205,
      "learning_rate": 4.697017139160232e-05,
      "loss": 0.2988,
      "step": 2997
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.2248641004758185,
      "learning_rate": 4.696780391276472e-05,
      "loss": 0.5468,
      "step": 2998
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.164748604532203,
      "learning_rate": 4.696543556903893e-05,
      "loss": 0.3068,
      "step": 2999
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8852346697193055,
      "learning_rate": 4.6963066360518206e-05,
      "loss": 0.1426,
      "step": 3000
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7973947416298135,
      "learning_rate": 4.696069628729582e-05,
      "loss": 0.1222,
      "step": 3001
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.754851096651054,
      "learning_rate": 4.695832534946507e-05,
      "loss": 0.1071,
      "step": 3002
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9540619190563213,
      "learning_rate": 4.695595354711932e-05,
      "loss": 0.1779,
      "step": 3003
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.268392941246154,
      "learning_rate": 4.695358088035195e-05,
      "loss": 0.6059,
      "step": 3004
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.729272253554572,
      "learning_rate": 4.695120734925637e-05,
      "loss": 0.0914,
      "step": 3005
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.168946359155228,
      "learning_rate": 4.694883295392601e-05,
      "loss": 0.3171,
      "step": 3006
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4799231064177505,
      "learning_rate": 4.694645769445438e-05,
      "loss": 0.2883,
      "step": 3007
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9978413988852277,
      "learning_rate": 4.6944081570934973e-05,
      "loss": 0.1739,
      "step": 3008
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.626636463452345,
      "learning_rate": 4.6941704583461355e-05,
      "loss": 0.3372,
      "step": 3009
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1614186348055413,
      "learning_rate": 4.69393267321271e-05,
      "loss": 0.2869,
      "step": 3010
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1606879777802255,
      "learning_rate": 4.693694801702583e-05,
      "loss": 0.2345,
      "step": 3011
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1830490515104277,
      "learning_rate": 4.693456843825118e-05,
      "loss": 0.2284,
      "step": 3012
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7579564228208566,
      "learning_rate": 4.693218799589687e-05,
      "loss": 0.0874,
      "step": 3013
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8799665771869576,
      "learning_rate": 4.6929806690056596e-05,
      "loss": 0.116,
      "step": 3014
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1028621185879892,
      "learning_rate": 4.6927424520824106e-05,
      "loss": 0.3006,
      "step": 3015
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1155884525430313,
      "learning_rate": 4.692504148829321e-05,
      "loss": 0.157,
      "step": 3016
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4681251537554862,
      "learning_rate": 4.692265759255772e-05,
      "loss": 0.2816,
      "step": 3017
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7057182473562559,
      "learning_rate": 4.692027283371148e-05,
      "loss": 0.1015,
      "step": 3018
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.167131797647168,
      "learning_rate": 4.6917887211848386e-05,
      "loss": 0.1969,
      "step": 3019
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7852039775672959,
      "learning_rate": 4.691550072706237e-05,
      "loss": 0.0994,
      "step": 3020
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0525694979135782,
      "learning_rate": 4.691311337944738e-05,
      "loss": 0.1865,
      "step": 3021
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.165197329624012,
      "learning_rate": 4.6910725169097415e-05,
      "loss": 0.1807,
      "step": 3022
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8749187692952701,
      "learning_rate": 4.69083360961065e-05,
      "loss": 0.1399,
      "step": 3023
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.724820700055452,
      "learning_rate": 4.690594616056869e-05,
      "loss": 0.1004,
      "step": 3024
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.323648471337604,
      "learning_rate": 4.690355536257808e-05,
      "loss": 0.4003,
      "step": 3025
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.917124036148339,
      "learning_rate": 4.690116370222879e-05,
      "loss": 0.1486,
      "step": 3026
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8471861914478649,
      "learning_rate": 4.6898771179614996e-05,
      "loss": 0.135,
      "step": 3027
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7510118622076235,
      "learning_rate": 4.6896377794830894e-05,
      "loss": 0.0832,
      "step": 3028
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9479665065589514,
      "learning_rate": 4.6893983547970695e-05,
      "loss": 0.1888,
      "step": 3029
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9532192891427875,
      "learning_rate": 4.689158843912868e-05,
      "loss": 0.1889,
      "step": 3030
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2589475774394105,
      "learning_rate": 4.6889192468399126e-05,
      "loss": 0.1116,
      "step": 3031
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2142646865076048,
      "learning_rate": 4.688679563587639e-05,
      "loss": 0.2249,
      "step": 3032
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2340589280329568,
      "learning_rate": 4.6884397941654815e-05,
      "loss": 0.3262,
      "step": 3033
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7406504710633595,
      "learning_rate": 4.688199938582882e-05,
      "loss": 0.2021,
      "step": 3034
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8773004564722385,
      "learning_rate": 4.687959996849282e-05,
      "loss": 0.1542,
      "step": 3035
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9448255816708534,
      "learning_rate": 4.687719968974129e-05,
      "loss": 0.1773,
      "step": 3036
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5524380406901999,
      "learning_rate": 4.687479854966873e-05,
      "loss": 0.2187,
      "step": 3037
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7858543476549766,
      "learning_rate": 4.687239654836967e-05,
      "loss": 0.1049,
      "step": 3038
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.779642603563697,
      "learning_rate": 4.686999368593868e-05,
      "loss": 0.1288,
      "step": 3039
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9550326154673041,
      "learning_rate": 4.686758996247037e-05,
      "loss": 0.1866,
      "step": 3040
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0921088166291824,
      "learning_rate": 4.686518537805937e-05,
      "loss": 0.2092,
      "step": 3041
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.248604507614488,
      "learning_rate": 4.686277993280036e-05,
      "loss": 0.2952,
      "step": 3042
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6991764446419586,
      "learning_rate": 4.6860373626788025e-05,
      "loss": 0.1117,
      "step": 3043
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0036861065258191,
      "learning_rate": 4.685796646011712e-05,
      "loss": 0.1433,
      "step": 3044
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9505735495298994,
      "learning_rate": 4.6855558432882406e-05,
      "loss": 0.126,
      "step": 3045
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6907395770111506,
      "learning_rate": 4.68531495451787e-05,
      "loss": 0.2798,
      "step": 3046
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6190581993317148,
      "learning_rate": 4.685073979710083e-05,
      "loss": 0.4363,
      "step": 3047
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0551571690764165,
      "learning_rate": 4.684832918874367e-05,
      "loss": 0.1834,
      "step": 3048
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7619272942416742,
      "learning_rate": 4.6845917720202135e-05,
      "loss": 0.1049,
      "step": 3049
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.970724668509815,
      "learning_rate": 4.684350539157116e-05,
      "loss": 0.1392,
      "step": 3050
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0029586171593285,
      "learning_rate": 4.684109220294573e-05,
      "loss": 0.1475,
      "step": 3051
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9651944343653809,
      "learning_rate": 4.6838678154420844e-05,
      "loss": 0.1697,
      "step": 3052
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6174859904369708,
      "learning_rate": 4.683626324609155e-05,
      "loss": 0.082,
      "step": 3053
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.36091857885567,
      "learning_rate": 4.6833847478052914e-05,
      "loss": 0.2346,
      "step": 3054
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8147101723585704,
      "learning_rate": 4.683143085040006e-05,
      "loss": 0.1038,
      "step": 3055
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7668626213021229,
      "learning_rate": 4.682901336322813e-05,
      "loss": 0.1104,
      "step": 3056
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8909413355216962,
      "learning_rate": 4.68265950166323e-05,
      "loss": 0.1203,
      "step": 3057
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1583771635439957,
      "learning_rate": 4.682417581070778e-05,
      "loss": 0.1448,
      "step": 3058
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7776727225633522,
      "learning_rate": 4.682175574554981e-05,
      "loss": 0.4377,
      "step": 3059
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9897642483645213,
      "learning_rate": 4.681933482125369e-05,
      "loss": 0.9578,
      "step": 3060
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8487938515721485,
      "learning_rate": 4.681691303791471e-05,
      "loss": 0.1366,
      "step": 3061
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7841921791745533,
      "learning_rate": 4.6814490395628235e-05,
      "loss": 0.1216,
      "step": 3062
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8182041763055087,
      "learning_rate": 4.6812066894489634e-05,
      "loss": 0.1446,
      "step": 3063
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3895759460478367,
      "learning_rate": 4.6809642534594336e-05,
      "loss": 0.2688,
      "step": 3064
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2031095435963648,
      "learning_rate": 4.680721731603778e-05,
      "loss": 0.1568,
      "step": 3065
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1105550969744877,
      "learning_rate": 4.680479123891543e-05,
      "loss": 0.2954,
      "step": 3066
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8847262972338188,
      "learning_rate": 4.680236430332284e-05,
      "loss": 0.2668,
      "step": 3067
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5590337719035618,
      "learning_rate": 4.679993650935555e-05,
      "loss": 0.1867,
      "step": 3068
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.4367073735410045,
      "learning_rate": 4.679750785710912e-05,
      "loss": 0.4356,
      "step": 3069
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8982675258228031,
      "learning_rate": 4.679507834667919e-05,
      "loss": 0.2155,
      "step": 3070
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.6290904569385862,
      "learning_rate": 4.6792647978161405e-05,
      "loss": 0.8876,
      "step": 3071
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3115480866245133,
      "learning_rate": 4.679021675165145e-05,
      "loss": 0.2464,
      "step": 3072
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3652939766618182,
      "learning_rate": 4.678778466724505e-05,
      "loss": 0.3039,
      "step": 3073
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2810810069587213,
      "learning_rate": 4.678535172503794e-05,
      "loss": 0.2536,
      "step": 3074
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.219223081829493,
      "learning_rate": 4.678291792512594e-05,
      "loss": 0.6432,
      "step": 3075
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8834050996266353,
      "learning_rate": 4.678048326760483e-05,
      "loss": 0.0885,
      "step": 3076
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2850627713617127,
      "learning_rate": 4.677804775257049e-05,
      "loss": 0.3323,
      "step": 3077
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.283918861056975,
      "learning_rate": 4.677561138011881e-05,
      "loss": 0.2359,
      "step": 3078
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9308197555735036,
      "learning_rate": 4.6773174150345695e-05,
      "loss": 0.103,
      "step": 3079
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.301688726457502,
      "learning_rate": 4.677073606334711e-05,
      "loss": 0.3977,
      "step": 3080
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8735741549870905,
      "learning_rate": 4.676829711921905e-05,
      "loss": 0.091,
      "step": 3081
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8514496757501646,
      "learning_rate": 4.6765857318057526e-05,
      "loss": 0.0796,
      "step": 3082
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.868434668515441,
      "learning_rate": 4.67634166599586e-05,
      "loss": 0.2816,
      "step": 3083
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0947646547885803,
      "learning_rate": 4.676097514501837e-05,
      "loss": 0.2768,
      "step": 3084
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2310558342211173,
      "learning_rate": 4.6758532773332946e-05,
      "loss": 0.2005,
      "step": 3085
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.100692009802669,
      "learning_rate": 4.6756089544998496e-05,
      "loss": 0.1743,
      "step": 3086
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8008583286128126,
      "learning_rate": 4.6753645460111206e-05,
      "loss": 0.1341,
      "step": 3087
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7084905281222558,
      "learning_rate": 4.675120051876731e-05,
      "loss": 0.3999,
      "step": 3088
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8016112223627349,
      "learning_rate": 4.674875472106305e-05,
      "loss": 0.1475,
      "step": 3089
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0969766330739406,
      "learning_rate": 4.674630806709474e-05,
      "loss": 0.1846,
      "step": 3090
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.992665155658614,
      "learning_rate": 4.67438605569587e-05,
      "loss": 0.4201,
      "step": 3091
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6568128135307212,
      "learning_rate": 4.6741412190751277e-05,
      "loss": 0.1099,
      "step": 3092
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9478104128972964,
      "learning_rate": 4.673896296856888e-05,
      "loss": 0.1589,
      "step": 3093
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9645240365358021,
      "learning_rate": 4.6736512890507925e-05,
      "loss": 0.1309,
      "step": 3094
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.817134192775966,
      "learning_rate": 4.673406195666488e-05,
      "loss": 0.1164,
      "step": 3095
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3772218163723065,
      "learning_rate": 4.673161016713624e-05,
      "loss": 0.2766,
      "step": 3096
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3552888569224135,
      "learning_rate": 4.672915752201854e-05,
      "loss": 0.142,
      "step": 3097
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2004046618056154,
      "learning_rate": 4.672670402140833e-05,
      "loss": 0.2219,
      "step": 3098
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2164580823870232,
      "learning_rate": 4.672424966540221e-05,
      "loss": 0.1991,
      "step": 3099
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8411414949659851,
      "learning_rate": 4.672179445409682e-05,
      "loss": 0.1315,
      "step": 3100
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9369198707150717,
      "learning_rate": 4.671933838758881e-05,
      "loss": 0.3455,
      "step": 3101
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3187378231706237,
      "learning_rate": 4.6716881465974876e-05,
      "loss": 0.2126,
      "step": 3102
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9916323106025235,
      "learning_rate": 4.671442368935176e-05,
      "loss": 0.7771,
      "step": 3103
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6952096331206561,
      "learning_rate": 4.671196505781622e-05,
      "loss": 0.1014,
      "step": 3104
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6078557716043191,
      "learning_rate": 4.670950557146506e-05,
      "loss": 0.3516,
      "step": 3105
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9236968773366521,
      "learning_rate": 4.6707045230395105e-05,
      "loss": 0.12,
      "step": 3106
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3644896153751869,
      "learning_rate": 4.670458403470322e-05,
      "loss": 0.387,
      "step": 3107
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8989751299304911,
      "learning_rate": 4.67021219844863e-05,
      "loss": 0.6457,
      "step": 3108
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9299629319458841,
      "learning_rate": 4.6699659079841296e-05,
      "loss": 0.2055,
      "step": 3109
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8770594314195567,
      "learning_rate": 4.6697195320865164e-05,
      "loss": 0.1188,
      "step": 3110
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4398100569672263,
      "learning_rate": 4.669473070765489e-05,
      "loss": 0.3847,
      "step": 3111
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8272329303685343,
      "learning_rate": 4.669226524030753e-05,
      "loss": 0.1053,
      "step": 3112
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.233615277688985,
      "learning_rate": 4.6689798918920135e-05,
      "loss": 0.2205,
      "step": 3113
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2655097600566234,
      "learning_rate": 4.6687331743589815e-05,
      "loss": 0.2252,
      "step": 3114
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8690975968007695,
      "learning_rate": 4.66848637144137e-05,
      "loss": 0.1221,
      "step": 3115
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5422249759428823,
      "learning_rate": 4.668239483148896e-05,
      "loss": 0.2596,
      "step": 3116
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7338577235315324,
      "learning_rate": 4.6679925094912795e-05,
      "loss": 0.0958,
      "step": 3117
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.607535619637202,
      "learning_rate": 4.667745450478245e-05,
      "loss": 0.4741,
      "step": 3118
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.808549422720486,
      "learning_rate": 4.667498306119518e-05,
      "loss": 0.3594,
      "step": 3119
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6186118568937617,
      "learning_rate": 4.667251076424829e-05,
      "loss": 0.3343,
      "step": 3120
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9277889113677489,
      "learning_rate": 4.667003761403912e-05,
      "loss": 0.1398,
      "step": 3121
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9651541992728636,
      "learning_rate": 4.6667563610665035e-05,
      "loss": 0.6053,
      "step": 3122
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8686504881270342,
      "learning_rate": 4.666508875422345e-05,
      "loss": 0.1682,
      "step": 3123
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2936011748391358,
      "learning_rate": 4.666261304481179e-05,
      "loss": 0.2151,
      "step": 3124
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6768431941639639,
      "learning_rate": 4.666013648252753e-05,
      "loss": 0.071,
      "step": 3125
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1052653286073433,
      "learning_rate": 4.665765906746817e-05,
      "loss": 0.2166,
      "step": 3126
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9812138170102965,
      "learning_rate": 4.665518079973126e-05,
      "loss": 0.1844,
      "step": 3127
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.872769475936063,
      "learning_rate": 4.6652701679414354e-05,
      "loss": 0.1332,
      "step": 3128
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3879466751039022,
      "learning_rate": 4.665022170661506e-05,
      "loss": 0.1769,
      "step": 3129
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3060699736645216,
      "learning_rate": 4.6647740881431037e-05,
      "loss": 0.3125,
      "step": 3130
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.245782771902087,
      "learning_rate": 4.664525920395993e-05,
      "loss": 0.4221,
      "step": 3131
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.50692282734712,
      "learning_rate": 4.664277667429946e-05,
      "loss": 0.5718,
      "step": 3132
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.365276504152244,
      "learning_rate": 4.6640293292547354e-05,
      "loss": 0.332,
      "step": 3133
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5286873210424885,
      "learning_rate": 4.66378090588014e-05,
      "loss": 0.2783,
      "step": 3134
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3504130091257769,
      "learning_rate": 4.663532397315939e-05,
      "loss": 0.2361,
      "step": 3135
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.266834924230692,
      "learning_rate": 4.663283803571918e-05,
      "loss": 0.1921,
      "step": 3136
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7608384838559219,
      "learning_rate": 4.663035124657863e-05,
      "loss": 0.0732,
      "step": 3137
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0525744873432585,
      "learning_rate": 4.662786360583564e-05,
      "loss": 0.4015,
      "step": 3138
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9148482369532323,
      "learning_rate": 4.662537511358816e-05,
      "loss": 0.3575,
      "step": 3139
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2160322787070954,
      "learning_rate": 4.6622885769934175e-05,
      "loss": 0.3403,
      "step": 3140
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8059476610138201,
      "learning_rate": 4.6620395574971675e-05,
      "loss": 0.1563,
      "step": 3141
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8994349479394509,
      "learning_rate": 4.661790452879871e-05,
      "loss": 0.1483,
      "step": 3142
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0564492577771813,
      "learning_rate": 4.661541263151334e-05,
      "loss": 0.1918,
      "step": 3143
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4055608408583,
      "learning_rate": 4.661291988321369e-05,
      "loss": 0.2597,
      "step": 3144
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9232062727539824,
      "learning_rate": 4.66104262839979e-05,
      "loss": 0.1203,
      "step": 3145
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1052432184139043,
      "learning_rate": 4.660793183396413e-05,
      "loss": 0.2413,
      "step": 3146
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1543119180157113,
      "learning_rate": 4.6605436533210606e-05,
      "loss": 0.5809,
      "step": 3147
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1012816283738647,
      "learning_rate": 4.660294038183556e-05,
      "loss": 0.1785,
      "step": 3148
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0005440828573486,
      "learning_rate": 4.6600443379937265e-05,
      "loss": 0.2668,
      "step": 3149
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8133251608537518,
      "learning_rate": 4.659794552761404e-05,
      "loss": 0.1563,
      "step": 3150
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1847431876228933,
      "learning_rate": 4.659544682496422e-05,
      "loss": 0.2348,
      "step": 3151
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.586341458339089,
      "learning_rate": 4.6592947272086174e-05,
      "loss": 0.385,
      "step": 3152
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.043652493580807,
      "learning_rate": 4.659044686907832e-05,
      "loss": 0.285,
      "step": 3153
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7759588078947411,
      "learning_rate": 4.6587945616039105e-05,
      "loss": 0.0992,
      "step": 3154
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.070099387743307,
      "learning_rate": 4.658544351306699e-05,
      "loss": 0.1399,
      "step": 3155
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7124043240827523,
      "learning_rate": 4.658294056026051e-05,
      "loss": 0.0962,
      "step": 3156
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9039392275371122,
      "learning_rate": 4.6580436757718186e-05,
      "loss": 0.1091,
      "step": 3157
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9838443811341983,
      "learning_rate": 4.657793210553859e-05,
      "loss": 0.1229,
      "step": 3158
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.155687966448821,
      "learning_rate": 4.657542660382035e-05,
      "loss": 0.1392,
      "step": 3159
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8856265732608561,
      "learning_rate": 4.6572920252662106e-05,
      "loss": 0.139,
      "step": 3160
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7615277226672543,
      "learning_rate": 4.6570413052162524e-05,
      "loss": 0.4293,
      "step": 3161
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8405371566584174,
      "learning_rate": 4.6567905002420324e-05,
      "loss": 0.183,
      "step": 3162
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.3457495622493836,
      "learning_rate": 4.6565396103534256e-05,
      "loss": 0.5307,
      "step": 3163
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7421179677488765,
      "learning_rate": 4.6562886355603076e-05,
      "loss": 0.2147,
      "step": 3164
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.6659325686919346,
      "learning_rate": 4.65603757587256e-05,
      "loss": 0.7859,
      "step": 3165
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0655541976039717,
      "learning_rate": 4.6557864313000695e-05,
      "loss": 0.1358,
      "step": 3166
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0773455939019627,
      "learning_rate": 4.655535201852722e-05,
      "loss": 0.1635,
      "step": 3167
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0161434415080517,
      "learning_rate": 4.655283887540408e-05,
      "loss": 0.1191,
      "step": 3168
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0576407051197332,
      "learning_rate": 4.6550324883730236e-05,
      "loss": 0.1394,
      "step": 3169
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0430808330979648,
      "learning_rate": 4.654781004360464e-05,
      "loss": 0.2102,
      "step": 3170
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9821375890443387,
      "learning_rate": 4.654529435512634e-05,
      "loss": 0.147,
      "step": 3171
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9389283799353386,
      "learning_rate": 4.654277781839435e-05,
      "loss": 0.1825,
      "step": 3172
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9316611043755486,
      "learning_rate": 4.6540260433507765e-05,
      "loss": 0.157,
      "step": 3173
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8740364104632404,
      "learning_rate": 4.653774220056568e-05,
      "loss": 0.1125,
      "step": 3174
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.407205070610484,
      "learning_rate": 4.6535223119667254e-05,
      "loss": 0.2963,
      "step": 3175
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2625902833365856,
      "learning_rate": 4.6532703190911656e-05,
      "loss": 0.6116,
      "step": 3176
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6991485716290333,
      "learning_rate": 4.6530182414398104e-05,
      "loss": 0.2441,
      "step": 3177
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.124964678129625,
      "learning_rate": 4.6527660790225845e-05,
      "loss": 0.1797,
      "step": 3178
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8343314765196942,
      "learning_rate": 4.652513831849414e-05,
      "loss": 0.1829,
      "step": 3179
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1574691246723792,
      "learning_rate": 4.6522614999302324e-05,
      "loss": 0.2538,
      "step": 3180
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8907181642877843,
      "learning_rate": 4.652009083274973e-05,
      "loss": 0.156,
      "step": 3181
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1098754081653819,
      "learning_rate": 4.651756581893574e-05,
      "loss": 0.1457,
      "step": 3182
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.389332694052459,
      "learning_rate": 4.6515039957959756e-05,
      "loss": 0.2137,
      "step": 3183
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7076998698954642,
      "learning_rate": 4.651251324992123e-05,
      "loss": 0.1173,
      "step": 3184
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1923334974482132,
      "learning_rate": 4.6509985694919646e-05,
      "loss": 0.1459,
      "step": 3185
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8940863000228324,
      "learning_rate": 4.6507457293054505e-05,
      "loss": 0.0893,
      "step": 3186
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.307296468798955,
      "learning_rate": 4.650492804442535e-05,
      "loss": 0.1802,
      "step": 3187
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4483207137510836,
      "learning_rate": 4.6502397949131774e-05,
      "loss": 0.5154,
      "step": 3188
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.911647251816741,
      "learning_rate": 4.649986700727338e-05,
      "loss": 0.1415,
      "step": 3189
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9254649328911329,
      "learning_rate": 4.649733521894981e-05,
      "loss": 0.203,
      "step": 3190
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.086557567653099,
      "learning_rate": 4.649480258426074e-05,
      "loss": 0.2156,
      "step": 3191
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0566615213223531,
      "learning_rate": 4.6492269103305904e-05,
      "loss": 0.1896,
      "step": 3192
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.655004650873325,
      "learning_rate": 4.648973477618502e-05,
      "loss": 0.8778,
      "step": 3193
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6050860288516826,
      "learning_rate": 4.648719960299788e-05,
      "loss": 0.4975,
      "step": 3194
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.785690717648514,
      "learning_rate": 4.648466358384429e-05,
      "loss": 0.15,
      "step": 3195
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.737393649039845,
      "learning_rate": 4.648212671882409e-05,
      "loss": 0.1317,
      "step": 3196
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.847067040357843,
      "learning_rate": 4.6479589008037175e-05,
      "loss": 0.3025,
      "step": 3197
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8053892598592457,
      "learning_rate": 4.6477050451583445e-05,
      "loss": 0.0824,
      "step": 3198
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0735071164149432,
      "learning_rate": 4.647451104956285e-05,
      "loss": 0.1644,
      "step": 3199
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6906845628791708,
      "learning_rate": 4.647197080207536e-05,
      "loss": 0.0975,
      "step": 3200
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.168372105647454,
      "learning_rate": 4.6469429709221e-05,
      "loss": 0.6225,
      "step": 3201
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0230577667479388,
      "learning_rate": 4.6466887771099796e-05,
      "loss": 0.1947,
      "step": 3202
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7686759163031579,
      "learning_rate": 4.646434498781184e-05,
      "loss": 0.122,
      "step": 3203
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8808035427128771,
      "learning_rate": 4.6461801359457234e-05,
      "loss": 0.1441,
      "step": 3204
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9221039701615754,
      "learning_rate": 4.6459256886136126e-05,
      "loss": 0.1797,
      "step": 3205
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.47736218898142,
      "learning_rate": 4.6456711567948705e-05,
      "loss": 0.761,
      "step": 3206
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2086878403614982,
      "learning_rate": 4.645416540499517e-05,
      "loss": 0.3357,
      "step": 3207
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.6047944259511295,
      "learning_rate": 4.645161839737576e-05,
      "loss": 0.4835,
      "step": 3208
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9928282149225642,
      "learning_rate": 4.6449070545190764e-05,
      "loss": 0.1191,
      "step": 3209
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7472513119569392,
      "learning_rate": 4.6446521848540484e-05,
      "loss": 0.105,
      "step": 3210
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2851972909446985,
      "learning_rate": 4.6443972307525274e-05,
      "loss": 0.2257,
      "step": 3211
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4808643451256562,
      "learning_rate": 4.644142192224551e-05,
      "loss": 0.4337,
      "step": 3212
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9345452318711733,
      "learning_rate": 4.643887069280159e-05,
      "loss": 0.1367,
      "step": 3213
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5740514547871758,
      "learning_rate": 4.643631861929397e-05,
      "loss": 0.1997,
      "step": 3214
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7873264946169076,
      "learning_rate": 4.6433765701823114e-05,
      "loss": 0.0914,
      "step": 3215
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8687924583191011,
      "learning_rate": 4.643121194048956e-05,
      "loss": 0.1219,
      "step": 3216
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6298687841460939,
      "learning_rate": 4.642865733539382e-05,
      "loss": 0.0812,
      "step": 3217
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3878694063786046,
      "learning_rate": 4.642610188663648e-05,
      "loss": 0.221,
      "step": 3218
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.499181850064262,
      "learning_rate": 4.6423545594318164e-05,
      "loss": 0.2879,
      "step": 3219
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.498498672618418,
      "learning_rate": 4.64209884585395e-05,
      "loss": 0.5292,
      "step": 3220
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1824291994147762,
      "learning_rate": 4.6418430479401164e-05,
      "loss": 0.2936,
      "step": 3221
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7303228826118875,
      "learning_rate": 4.6415871657003875e-05,
      "loss": 0.4152,
      "step": 3222
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3685914100971703,
      "learning_rate": 4.6413311991448373e-05,
      "loss": 0.1829,
      "step": 3223
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0140092306758486,
      "learning_rate": 4.641075148283542e-05,
      "loss": 0.1725,
      "step": 3224
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8352884626208574,
      "learning_rate": 4.640819013126585e-05,
      "loss": 0.4311,
      "step": 3225
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9703945276694713,
      "learning_rate": 4.6405627936840485e-05,
      "loss": 0.1983,
      "step": 3226
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1086416168491808,
      "learning_rate": 4.6403064899660207e-05,
      "loss": 0.2115,
      "step": 3227
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1426034237480833,
      "learning_rate": 4.6400501019825924e-05,
      "loss": 0.2764,
      "step": 3228
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1452295561782633,
      "learning_rate": 4.639793629743859e-05,
      "loss": 0.2113,
      "step": 3229
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.4665965213875496,
      "learning_rate": 4.639537073259915e-05,
      "loss": 0.7131,
      "step": 3230
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.5895989105702006,
      "learning_rate": 4.6392804325408644e-05,
      "loss": 0.4716,
      "step": 3231
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.754161661694711,
      "learning_rate": 4.6390237075968104e-05,
      "loss": 0.1076,
      "step": 3232
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0921029496441232,
      "learning_rate": 4.638766898437858e-05,
      "loss": 0.1187,
      "step": 3233
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0160389264817324,
      "learning_rate": 4.638510005074122e-05,
      "loss": 0.1229,
      "step": 3234
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8179925081190744,
      "learning_rate": 4.6382530275157136e-05,
      "loss": 0.4252,
      "step": 3235
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1645900875195534,
      "learning_rate": 4.6379959657727515e-05,
      "loss": 0.1851,
      "step": 3236
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9824577145951418,
      "learning_rate": 4.637738819855356e-05,
      "loss": 0.1342,
      "step": 3237
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3771498277436296,
      "learning_rate": 4.6374815897736505e-05,
      "loss": 0.2218,
      "step": 3238
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.768233604408166,
      "learning_rate": 4.637224275537763e-05,
      "loss": 0.2954,
      "step": 3239
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.69139607553345,
      "learning_rate": 4.6369668771578246e-05,
      "loss": 0.8788,
      "step": 3240
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3481708316636278,
      "learning_rate": 4.636709394643969e-05,
      "loss": 0.1466,
      "step": 3241
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.264624487400413,
      "learning_rate": 4.636451828006332e-05,
      "loss": 0.155,
      "step": 3242
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7114562998077016,
      "learning_rate": 4.636194177255057e-05,
      "loss": 0.2911,
      "step": 3243
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7568196054825623,
      "learning_rate": 4.635936442400285e-05,
      "loss": 0.093,
      "step": 3244
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.884580345717764,
      "learning_rate": 4.635678623452164e-05,
      "loss": 0.853,
      "step": 3245
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9337514357783746,
      "learning_rate": 4.6354207204208466e-05,
      "loss": 0.1614,
      "step": 3246
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5975681908761414,
      "learning_rate": 4.6351627333164844e-05,
      "loss": 0.4135,
      "step": 3247
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3111839350139243,
      "learning_rate": 4.6349046621492345e-05,
      "loss": 0.2251,
      "step": 3248
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1984008306239087,
      "learning_rate": 4.6346465069292586e-05,
      "loss": 0.3041,
      "step": 3249
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8874657693446334,
      "learning_rate": 4.6343882676667204e-05,
      "loss": 0.1248,
      "step": 3250
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2154434600488428,
      "learning_rate": 4.634129944371786e-05,
      "loss": 0.2525,
      "step": 3251
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8325888968645773,
      "learning_rate": 4.633871537054626e-05,
      "loss": 0.1597,
      "step": 3252
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9908320378663807,
      "learning_rate": 4.633613045725415e-05,
      "loss": 0.179,
      "step": 3253
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0571650785821518,
      "learning_rate": 4.633354470394328e-05,
      "loss": 0.1615,
      "step": 3254
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7667194479984237,
      "learning_rate": 4.633095811071547e-05,
      "loss": 0.1189,
      "step": 3255
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8239450967081005,
      "learning_rate": 4.632837067767256e-05,
      "loss": 0.099,
      "step": 3256
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8494122233993717,
      "learning_rate": 4.63257824049164e-05,
      "loss": 0.1405,
      "step": 3257
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.625540672918284,
      "learning_rate": 4.632319329254891e-05,
      "loss": 0.2889,
      "step": 3258
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2362771419715954,
      "learning_rate": 4.632060334067202e-05,
      "loss": 0.3104,
      "step": 3259
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.135704063016671,
      "learning_rate": 4.631801254938769e-05,
      "loss": 0.1863,
      "step": 3260
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8359494889415977,
      "learning_rate": 4.631542091879794e-05,
      "loss": 0.1722,
      "step": 3261
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.986712265957759,
      "learning_rate": 4.631282844900478e-05,
      "loss": 0.5654,
      "step": 3262
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3943199255240086,
      "learning_rate": 4.631023514011029e-05,
      "loss": 0.4026,
      "step": 3263
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.122830956865703,
      "learning_rate": 4.630764099221657e-05,
      "loss": 0.2092,
      "step": 3264
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1963170890212182,
      "learning_rate": 4.630504600542576e-05,
      "loss": 0.1695,
      "step": 3265
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8759642972605162,
      "learning_rate": 4.630245017984002e-05,
      "loss": 0.6968,
      "step": 3266
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.386337185431698,
      "learning_rate": 4.6299853515561543e-05,
      "loss": 0.3117,
      "step": 3267
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0182669584155137,
      "learning_rate": 4.6297256012692566e-05,
      "loss": 1.0874,
      "step": 3268
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.673643082782213,
      "learning_rate": 4.629465767133536e-05,
      "loss": 0.1264,
      "step": 3269
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9197309360979631,
      "learning_rate": 4.629205849159223e-05,
      "loss": 0.1422,
      "step": 3270
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.795320736889757,
      "learning_rate": 4.6289458473565485e-05,
      "loss": 0.1217,
      "step": 3271
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0161098750772772,
      "learning_rate": 4.628685761735751e-05,
      "loss": 0.1434,
      "step": 3272
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.920800332004365,
      "learning_rate": 4.628425592307069e-05,
      "loss": 0.1751,
      "step": 3273
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9617300144626909,
      "learning_rate": 4.628165339080747e-05,
      "loss": 0.1385,
      "step": 3274
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8570602169036474,
      "learning_rate": 4.6279050020670286e-05,
      "loss": 0.1159,
      "step": 3275
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8671955671801549,
      "learning_rate": 4.627644581276167e-05,
      "loss": 0.0894,
      "step": 3276
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4836393675166435,
      "learning_rate": 4.627384076718412e-05,
      "loss": 0.3892,
      "step": 3277
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8332661475353587,
      "learning_rate": 4.627123488404022e-05,
      "loss": 0.1139,
      "step": 3278
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7217373267182235,
      "learning_rate": 4.626862816343256e-05,
      "loss": 0.0697,
      "step": 3279
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8387147775526198,
      "learning_rate": 4.626602060546377e-05,
      "loss": 0.4229,
      "step": 3280
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.505996710303731,
      "learning_rate": 4.626341221023649e-05,
      "loss": 0.3673,
      "step": 3281
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0235864086065396,
      "learning_rate": 4.626080297785345e-05,
      "loss": 0.5197,
      "step": 3282
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2971015153940109,
      "learning_rate": 4.625819290841735e-05,
      "loss": 0.2241,
      "step": 3283
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.023607637507246,
      "learning_rate": 4.625558200203097e-05,
      "loss": 0.7171,
      "step": 3284
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3179315260001303,
      "learning_rate": 4.6252970258797084e-05,
      "loss": 0.2077,
      "step": 3285
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7833273773285616,
      "learning_rate": 4.625035767881853e-05,
      "loss": 0.1466,
      "step": 3286
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8062730590909183,
      "learning_rate": 4.624774426219817e-05,
      "loss": 0.2275,
      "step": 3287
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.797167681041771,
      "learning_rate": 4.624513000903888e-05,
      "loss": 0.1168,
      "step": 3288
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1435597377661828,
      "learning_rate": 4.62425149194436e-05,
      "loss": 0.1225,
      "step": 3289
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7983970280180931,
      "learning_rate": 4.623989899351529e-05,
      "loss": 0.1062,
      "step": 3290
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9578701005700266,
      "learning_rate": 4.623728223135693e-05,
      "loss": 0.3565,
      "step": 3291
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8528149393942988,
      "learning_rate": 4.6234664633071545e-05,
      "loss": 0.1546,
      "step": 3292
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.783808096807133,
      "learning_rate": 4.6232046198762196e-05,
      "loss": 0.4335,
      "step": 3293
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8910308781350521,
      "learning_rate": 4.622942692853197e-05,
      "loss": 0.245,
      "step": 3294
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9850714755901501,
      "learning_rate": 4.6226806822484e-05,
      "loss": 0.1405,
      "step": 3295
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6114977808289093,
      "learning_rate": 4.622418588072143e-05,
      "loss": 0.3993,
      "step": 3296
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8980713143924949,
      "learning_rate": 4.6221564103347445e-05,
      "loss": 0.1424,
      "step": 3297
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2445995660757887,
      "learning_rate": 4.621894149046528e-05,
      "loss": 0.32,
      "step": 3298
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0251571700675903,
      "learning_rate": 4.621631804217818e-05,
      "loss": 0.2898,
      "step": 3299
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9723832058183517,
      "learning_rate": 4.621369375858943e-05,
      "loss": 0.1688,
      "step": 3300
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3479853322088717,
      "learning_rate": 4.6211068639802354e-05,
      "loss": 0.2529,
      "step": 3301
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6437991630573323,
      "learning_rate": 4.6208442685920306e-05,
      "loss": 0.5022,
      "step": 3302
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.994802383672562,
      "learning_rate": 4.620581589704667e-05,
      "loss": 0.2033,
      "step": 3303
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9342535108039687,
      "learning_rate": 4.620318827328486e-05,
      "loss": 0.1744,
      "step": 3304
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8990927374506988,
      "learning_rate": 4.620055981473833e-05,
      "loss": 0.1744,
      "step": 3305
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6038552357295166,
      "learning_rate": 4.619793052151057e-05,
      "loss": 0.4653,
      "step": 3306
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2101989015664547,
      "learning_rate": 4.6195300393705096e-05,
      "loss": 0.2323,
      "step": 3307
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8646346128421499,
      "learning_rate": 4.619266943142545e-05,
      "loss": 0.1128,
      "step": 3308
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1310356996315865,
      "learning_rate": 4.619003763477523e-05,
      "loss": 0.1818,
      "step": 3309
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8351305725346538,
      "learning_rate": 4.618740500385803e-05,
      "loss": 0.0897,
      "step": 3310
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2273231439147916,
      "learning_rate": 4.618477153877752e-05,
      "loss": 0.1855,
      "step": 3311
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2970594312745949,
      "learning_rate": 4.618213723963737e-05,
      "loss": 0.3264,
      "step": 3312
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9485532194324118,
      "learning_rate": 4.6179502106541294e-05,
      "loss": 0.2224,
      "step": 3313
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1230893740869632,
      "learning_rate": 4.6176866139593044e-05,
      "loss": 0.1615,
      "step": 3314
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4887267055170617,
      "learning_rate": 4.617422933889639e-05,
      "loss": 0.3753,
      "step": 3315
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9389024069291676,
      "learning_rate": 4.617159170455516e-05,
      "loss": 0.1738,
      "step": 3316
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3270787208780785,
      "learning_rate": 4.6168953236673185e-05,
      "loss": 0.2034,
      "step": 3317
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9171529021804887,
      "learning_rate": 4.6166313935354354e-05,
      "loss": 0.1313,
      "step": 3318
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.69965246993716,
      "learning_rate": 4.6163673800702575e-05,
      "loss": 0.405,
      "step": 3319
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9515539098223701,
      "learning_rate": 4.616103283282179e-05,
      "loss": 0.1502,
      "step": 3320
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.204092109753703,
      "learning_rate": 4.615839103181597e-05,
      "loss": 0.2388,
      "step": 3321
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7287192749662312,
      "learning_rate": 4.615574839778914e-05,
      "loss": 0.0894,
      "step": 3322
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6862332701918549,
      "learning_rate": 4.6153104930845334e-05,
      "loss": 0.092,
      "step": 3323
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9956833449375608,
      "learning_rate": 4.615046063108862e-05,
      "loss": 0.4468,
      "step": 3324
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0515967296291666,
      "learning_rate": 4.614781549862312e-05,
      "loss": 0.2329,
      "step": 3325
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.047557027317358,
      "learning_rate": 4.614516953355297e-05,
      "loss": 0.2695,
      "step": 3326
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0603447562823092,
      "learning_rate": 4.614252273598233e-05,
      "loss": 0.2094,
      "step": 3327
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9286734854305939,
      "learning_rate": 4.613987510601542e-05,
      "loss": 0.1382,
      "step": 3328
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0425683365868788,
      "learning_rate": 4.613722664375648e-05,
      "loss": 0.1478,
      "step": 3329
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9872183891903564,
      "learning_rate": 4.613457734930978e-05,
      "loss": 0.1437,
      "step": 3330
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8853379684657089,
      "learning_rate": 4.6131927222779625e-05,
      "loss": 0.1642,
      "step": 3331
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.214015930656575,
      "learning_rate": 4.612927626427035e-05,
      "loss": 0.3402,
      "step": 3332
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6980111290251523,
      "learning_rate": 4.612662447388632e-05,
      "loss": 0.2477,
      "step": 3333
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.920270760052367,
      "learning_rate": 4.612397185173195e-05,
      "loss": 0.1063,
      "step": 3334
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0310462272245817,
      "learning_rate": 4.612131839791166e-05,
      "loss": 0.2196,
      "step": 3335
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2078119299531174,
      "learning_rate": 4.611866411252994e-05,
      "loss": 0.2553,
      "step": 3336
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0195600789603856,
      "learning_rate": 4.611600899569127e-05,
      "loss": 0.1287,
      "step": 3337
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9714616690753559,
      "learning_rate": 4.6113353047500194e-05,
      "loss": 0.1362,
      "step": 3338
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3150393396874953,
      "learning_rate": 4.611069626806128e-05,
      "loss": 0.3185,
      "step": 3339
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0444376806658902,
      "learning_rate": 4.610803865747912e-05,
      "loss": 0.2208,
      "step": 3340
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4724353659501834,
      "learning_rate": 4.610538021585835e-05,
      "loss": 0.27,
      "step": 3341
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7683938807036015,
      "learning_rate": 4.610272094330364e-05,
      "loss": 0.1231,
      "step": 3342
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9892189588133714,
      "learning_rate": 4.610006083991968e-05,
      "loss": 0.185,
      "step": 3343
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8103170505217456,
      "learning_rate": 4.60973999058112e-05,
      "loss": 0.0825,
      "step": 3344
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8469287190860388,
      "learning_rate": 4.6094738141082975e-05,
      "loss": 0.7586,
      "step": 3345
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.971275479698773,
      "learning_rate": 4.6092075545839785e-05,
      "loss": 0.1186,
      "step": 3346
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3282556339089464,
      "learning_rate": 4.608941212018646e-05,
      "loss": 0.3093,
      "step": 3347
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9048460816145619,
      "learning_rate": 4.6086747864227874e-05,
      "loss": 0.1665,
      "step": 3348
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.985591215807669,
      "learning_rate": 4.608408277806891e-05,
      "loss": 0.138,
      "step": 3349
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9202754079472747,
      "learning_rate": 4.60814168618145e-05,
      "loss": 0.1428,
      "step": 3350
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3843864959889125,
      "learning_rate": 4.607875011556959e-05,
      "loss": 0.3097,
      "step": 3351
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2593157902389083,
      "learning_rate": 4.607608253943918e-05,
      "loss": 0.1527,
      "step": 3352
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1093103863592875,
      "learning_rate": 4.60734141335283e-05,
      "loss": 0.2114,
      "step": 3353
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.267238671635974,
      "learning_rate": 4.6070744897942005e-05,
      "loss": 0.1744,
      "step": 3354
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3981128935530756,
      "learning_rate": 4.606807483278539e-05,
      "loss": 0.1848,
      "step": 3355
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.470351824306898,
      "learning_rate": 4.606540393816356e-05,
      "loss": 0.3245,
      "step": 3356
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2508459944421544,
      "learning_rate": 4.606273221418168e-05,
      "loss": 0.2089,
      "step": 3357
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7419885688647455,
      "learning_rate": 4.606005966094494e-05,
      "loss": 0.0811,
      "step": 3358
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9971352461282927,
      "learning_rate": 4.6057386278558554e-05,
      "loss": 0.2261,
      "step": 3359
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1467600779996623,
      "learning_rate": 4.605471206712779e-05,
      "loss": 0.3454,
      "step": 3360
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2427869937379095,
      "learning_rate": 4.605203702675791e-05,
      "loss": 0.4,
      "step": 3361
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.861094801266783,
      "learning_rate": 4.6049361157554254e-05,
      "loss": 0.0999,
      "step": 3362
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4348444478949405,
      "learning_rate": 4.604668445962216e-05,
      "loss": 0.2993,
      "step": 3363
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0550751268950624,
      "learning_rate": 4.604400693306702e-05,
      "loss": 0.2308,
      "step": 3364
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7514262122123876,
      "learning_rate": 4.6041328577994235e-05,
      "loss": 0.272,
      "step": 3365
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2760900110851798,
      "learning_rate": 4.603864939450927e-05,
      "loss": 0.4805,
      "step": 3366
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9491417605820389,
      "learning_rate": 4.60359693827176e-05,
      "loss": 0.2785,
      "step": 3367
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9963243087843482,
      "learning_rate": 4.603328854272474e-05,
      "loss": 0.4969,
      "step": 3368
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9449283330130849,
      "learning_rate": 4.6030606874636236e-05,
      "loss": 0.14,
      "step": 3369
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9445427039263002,
      "learning_rate": 4.6027924378557665e-05,
      "loss": 0.1492,
      "step": 3370
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8049628817908148,
      "learning_rate": 4.602524105459464e-05,
      "loss": 0.1368,
      "step": 3371
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.104982869082257,
      "learning_rate": 4.6022556902852807e-05,
      "loss": 0.1574,
      "step": 3372
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9384056938473173,
      "learning_rate": 4.601987192343785e-05,
      "loss": 0.108,
      "step": 3373
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2583569720829852,
      "learning_rate": 4.601718611645546e-05,
      "loss": 0.3137,
      "step": 3374
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4062667553630606,
      "learning_rate": 4.60144994820114e-05,
      "loss": 0.316,
      "step": 3375
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.978863388404365,
      "learning_rate": 4.601181202021143e-05,
      "loss": 0.1669,
      "step": 3376
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.030944664798841,
      "learning_rate": 4.6009123731161363e-05,
      "loss": 0.1263,
      "step": 3377
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.889682087054565,
      "learning_rate": 4.6006434614967035e-05,
      "loss": 0.4536,
      "step": 3378
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7390560652305452,
      "learning_rate": 4.600374467173433e-05,
      "loss": 0.1653,
      "step": 3379
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.441760387148752,
      "learning_rate": 4.600105390156914e-05,
      "loss": 0.3069,
      "step": 3380
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3089002414872986,
      "learning_rate": 4.59983623045774e-05,
      "loss": 0.2772,
      "step": 3381
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1005489349457196,
      "learning_rate": 4.599566988086509e-05,
      "loss": 0.2128,
      "step": 3382
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8580039800585221,
      "learning_rate": 4.599297663053822e-05,
      "loss": 0.1399,
      "step": 3383
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.221937488043072,
      "learning_rate": 4.59902825537028e-05,
      "loss": 0.2959,
      "step": 3384
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6552505828634845,
      "learning_rate": 4.598758765046492e-05,
      "loss": 0.0997,
      "step": 3385
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9780217359169219,
      "learning_rate": 4.598489192093067e-05,
      "loss": 0.1153,
      "step": 3386
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.245358632983039,
      "learning_rate": 4.598219536520619e-05,
      "loss": 0.1397,
      "step": 3387
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.597814684067119,
      "learning_rate": 4.597949798339764e-05,
      "loss": 0.3655,
      "step": 3388
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9137664558621312,
      "learning_rate": 4.597679977561122e-05,
      "loss": 0.4952,
      "step": 3389
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0695277375263748,
      "learning_rate": 4.5974100741953154e-05,
      "loss": 0.1599,
      "step": 3390
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.115120155500918,
      "learning_rate": 4.597140088252971e-05,
      "loss": 0.133,
      "step": 3391
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.857868240117267,
      "learning_rate": 4.596870019744719e-05,
      "loss": 0.126,
      "step": 3392
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.961806337533026,
      "learning_rate": 4.5965998686811916e-05,
      "loss": 0.1973,
      "step": 3393
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8097541389217274,
      "learning_rate": 4.5963296350730245e-05,
      "loss": 0.125,
      "step": 3394
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1912570531060311,
      "learning_rate": 4.596059318930857e-05,
      "loss": 0.3678,
      "step": 3395
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.665152640217248,
      "learning_rate": 4.595788920265332e-05,
      "loss": 0.4939,
      "step": 3396
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3477336066546386,
      "learning_rate": 4.5955184390870956e-05,
      "loss": 0.2964,
      "step": 3397
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9560735705892537,
      "learning_rate": 4.595247875406796e-05,
      "loss": 0.1555,
      "step": 3398
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1887891180656636,
      "learning_rate": 4.594977229235087e-05,
      "loss": 0.1971,
      "step": 3399
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5643492063249897,
      "learning_rate": 4.5947065005826215e-05,
      "loss": 0.0783,
      "step": 3400
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1013275442333383,
      "learning_rate": 4.594435689460061e-05,
      "loss": 0.1835,
      "step": 3401
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3386966132234726,
      "learning_rate": 4.594164795878066e-05,
      "loss": 0.3343,
      "step": 3402
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9184378933869148,
      "learning_rate": 4.593893819847302e-05,
      "loss": 0.1613,
      "step": 3403
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0947946598057465,
      "learning_rate": 4.5936227613784376e-05,
      "loss": 0.3252,
      "step": 3404
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0629135347176062,
      "learning_rate": 4.593351620482145e-05,
      "loss": 0.1358,
      "step": 3405
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.964911279987277,
      "learning_rate": 4.593080397169099e-05,
      "loss": 0.1428,
      "step": 3406
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4759057521365055,
      "learning_rate": 4.592809091449977e-05,
      "loss": 0.448,
      "step": 3407
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2530866461796157,
      "learning_rate": 4.5925377033354615e-05,
      "loss": 0.3194,
      "step": 3408
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6846043851672838,
      "learning_rate": 4.5922662328362374e-05,
      "loss": 0.0796,
      "step": 3409
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8848761229263219,
      "learning_rate": 4.591994679962992e-05,
      "loss": 0.1095,
      "step": 3410
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1340066517692338,
      "learning_rate": 4.5917230447264164e-05,
      "loss": 0.212,
      "step": 3411
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.261620311462931,
      "learning_rate": 4.591451327137206e-05,
      "loss": 0.1661,
      "step": 3412
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8732927019988725,
      "learning_rate": 4.5911795272060575e-05,
      "loss": 0.1378,
      "step": 3413
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.584246711764388,
      "learning_rate": 4.590907644943673e-05,
      "loss": 0.2534,
      "step": 3414
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.360105514085889,
      "learning_rate": 4.5906356803607544e-05,
      "loss": 0.4024,
      "step": 3415
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.289508126568836,
      "learning_rate": 4.590363633468012e-05,
      "loss": 0.3864,
      "step": 3416
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3008298109076737,
      "learning_rate": 4.590091504276156e-05,
      "loss": 0.3698,
      "step": 3417
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0751760324979753,
      "learning_rate": 4.589819292795898e-05,
      "loss": 0.1456,
      "step": 3418
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2556000200695814,
      "learning_rate": 4.5895469990379575e-05,
      "loss": 0.2579,
      "step": 3419
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1732938258674122,
      "learning_rate": 4.589274623013054e-05,
      "loss": 0.2418,
      "step": 3420
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9041340403631016,
      "learning_rate": 4.589002164731911e-05,
      "loss": 0.1936,
      "step": 3421
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7890295591186188,
      "learning_rate": 4.588729624205256e-05,
      "loss": 0.1203,
      "step": 3422
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8198939171492943,
      "learning_rate": 4.588457001443819e-05,
      "loss": 0.1454,
      "step": 3423
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1777301493856611,
      "learning_rate": 4.588184296458332e-05,
      "loss": 0.236,
      "step": 3424
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8899651513400335,
      "learning_rate": 4.587911509259534e-05,
      "loss": 0.3125,
      "step": 3425
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.136078833903871,
      "learning_rate": 4.587638639858163e-05,
      "loss": 0.1661,
      "step": 3426
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1384799511063683,
      "learning_rate": 4.5873656882649626e-05,
      "loss": 0.2602,
      "step": 3427
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9407631789278379,
      "learning_rate": 4.587092654490679e-05,
      "loss": 0.1488,
      "step": 3428
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0384683490929236,
      "learning_rate": 4.586819538546062e-05,
      "loss": 0.3016,
      "step": 3429
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.833035976248515,
      "learning_rate": 4.586546340441864e-05,
      "loss": 0.0965,
      "step": 3430
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4946318079587615,
      "learning_rate": 4.586273060188842e-05,
      "loss": 0.2194,
      "step": 3431
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1175875629499472,
      "learning_rate": 4.585999697797753e-05,
      "loss": 0.2352,
      "step": 3432
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6818514079505323,
      "learning_rate": 4.585726253279362e-05,
      "loss": 0.0918,
      "step": 3433
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7786865803599082,
      "learning_rate": 4.585452726644433e-05,
      "loss": 0.1186,
      "step": 3434
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.29032995517202,
      "learning_rate": 4.585179117903736e-05,
      "loss": 0.1586,
      "step": 3435
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.768890800195211,
      "learning_rate": 4.5849054270680426e-05,
      "loss": 0.1425,
      "step": 3436
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1493791079817948,
      "learning_rate": 4.584631654148128e-05,
      "loss": 0.3732,
      "step": 3437
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8403680120892576,
      "learning_rate": 4.584357799154771e-05,
      "loss": 0.4065,
      "step": 3438
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0716799187582773,
      "learning_rate": 4.584083862098755e-05,
      "loss": 0.2083,
      "step": 3439
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.250903355422917,
      "learning_rate": 4.583809842990863e-05,
      "loss": 0.3601,
      "step": 3440
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1021303322016789,
      "learning_rate": 4.5835357418418834e-05,
      "loss": 0.2488,
      "step": 3441
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7563182114021599,
      "learning_rate": 4.583261558662609e-05,
      "loss": 0.2222,
      "step": 3442
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7517729427086943,
      "learning_rate": 4.582987293463834e-05,
      "loss": 0.0659,
      "step": 3443
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7356138086490441,
      "learning_rate": 4.582712946256357e-05,
      "loss": 0.1153,
      "step": 3444
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3343148994756147,
      "learning_rate": 4.582438517050978e-05,
      "loss": 0.2673,
      "step": 3445
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8334422563834211,
      "learning_rate": 4.582164005858502e-05,
      "loss": 0.1184,
      "step": 3446
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7929029439948511,
      "learning_rate": 4.581889412689737e-05,
      "loss": 0.1139,
      "step": 3447
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.06900009127907,
      "learning_rate": 4.581614737555494e-05,
      "loss": 0.2366,
      "step": 3448
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8375894645681251,
      "learning_rate": 4.581339980466587e-05,
      "loss": 0.1924,
      "step": 3449
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8882732618104039,
      "learning_rate": 4.581065141433832e-05,
      "loss": 0.0834,
      "step": 3450
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1871343560402041,
      "learning_rate": 4.580790220468053e-05,
      "loss": 0.2339,
      "step": 3451
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8851805968658663,
      "learning_rate": 4.58051521758007e-05,
      "loss": 0.167,
      "step": 3452
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.1701932472712375,
      "learning_rate": 4.5802401327807134e-05,
      "loss": 0.5435,
      "step": 3453
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1625895301944091,
      "learning_rate": 4.579964966080811e-05,
      "loss": 0.2701,
      "step": 3454
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9015692406685732,
      "learning_rate": 4.579689717491197e-05,
      "loss": 0.1588,
      "step": 3455
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.42733178511113,
      "learning_rate": 4.579414387022709e-05,
      "loss": 0.3006,
      "step": 3456
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8926788598953639,
      "learning_rate": 4.579138974686186e-05,
      "loss": 0.1091,
      "step": 3457
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9958624299928467,
      "learning_rate": 4.578863480492472e-05,
      "loss": 0.2055,
      "step": 3458
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.780202937928995,
      "learning_rate": 4.5785879044524125e-05,
      "loss": 0.4116,
      "step": 3459
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2131857599395222,
      "learning_rate": 4.578312246576858e-05,
      "loss": 0.2983,
      "step": 3460
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3673152221179412,
      "learning_rate": 4.57803650687666e-05,
      "loss": 0.5582,
      "step": 3461
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.330733519118605,
      "learning_rate": 4.5777606853626757e-05,
      "loss": 0.2424,
      "step": 3462
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2996467916991112,
      "learning_rate": 4.5774847820457644e-05,
      "loss": 0.2784,
      "step": 3463
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6570949163974953,
      "learning_rate": 4.577208796936788e-05,
      "loss": 0.1206,
      "step": 3464
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0641397409737918,
      "learning_rate": 4.576932730046613e-05,
      "loss": 0.2385,
      "step": 3465
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8857522235421643,
      "learning_rate": 4.576656581386108e-05,
      "loss": 0.1944,
      "step": 3466
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7597935783837612,
      "learning_rate": 4.576380350966145e-05,
      "loss": 0.0986,
      "step": 3467
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.950878782425257,
      "learning_rate": 4.5761040387975995e-05,
      "loss": 0.1793,
      "step": 3468
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9289128388341569,
      "learning_rate": 4.5758276448913496e-05,
      "loss": 0.1818,
      "step": 3469
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8750495176648221,
      "learning_rate": 4.575551169258278e-05,
      "loss": 0.1862,
      "step": 3470
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2292594992181936,
      "learning_rate": 4.5752746119092696e-05,
      "loss": 0.3779,
      "step": 3471
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0645670366335298,
      "learning_rate": 4.5749979728552116e-05,
      "loss": 0.2504,
      "step": 3472
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2249291451595319,
      "learning_rate": 4.574721252106997e-05,
      "loss": 0.1607,
      "step": 3473
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1211260283255995,
      "learning_rate": 4.57444444967552e-05,
      "loss": 0.28,
      "step": 3474
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1275270735000869,
      "learning_rate": 4.5741675655716774e-05,
      "loss": 0.2501,
      "step": 3475
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0409010981735924,
      "learning_rate": 4.573890599806372e-05,
      "loss": 0.2797,
      "step": 3476
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8175557413931198,
      "learning_rate": 4.573613552390508e-05,
      "loss": 0.1203,
      "step": 3477
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2605722199250167,
      "learning_rate": 4.573336423334991e-05,
      "loss": 0.1932,
      "step": 3478
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4503065530015506,
      "learning_rate": 4.573059212650734e-05,
      "loss": 0.3157,
      "step": 3479
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.244595793981776,
      "learning_rate": 4.57278192034865e-05,
      "loss": 0.2836,
      "step": 3480
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1591380229043335,
      "learning_rate": 4.572504546439656e-05,
      "loss": 0.2805,
      "step": 3481
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7919077443972106,
      "learning_rate": 4.572227090934673e-05,
      "loss": 0.0993,
      "step": 3482
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1066385548524456,
      "learning_rate": 4.571949553844624e-05,
      "loss": 0.2247,
      "step": 3483
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7742080655728363,
      "learning_rate": 4.571671935180436e-05,
      "loss": 0.1047,
      "step": 3484
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2639462944354802,
      "learning_rate": 4.571394234953039e-05,
      "loss": 0.2369,
      "step": 3485
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7269885878932533,
      "learning_rate": 4.571116453173367e-05,
      "loss": 0.1333,
      "step": 3486
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.618344528083889,
      "learning_rate": 4.570838589852356e-05,
      "loss": 0.476,
      "step": 3487
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8044415767773819,
      "learning_rate": 4.570560645000945e-05,
      "loss": 0.1491,
      "step": 3488
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.216613155193758,
      "learning_rate": 4.570282618630079e-05,
      "loss": 0.2066,
      "step": 3489
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9125232904466954,
      "learning_rate": 4.5700045107507e-05,
      "loss": 0.1351,
      "step": 3490
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5990228410395595,
      "learning_rate": 4.5697263213737616e-05,
      "loss": 0.7925,
      "step": 3491
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2210909719218581,
      "learning_rate": 4.5694480505102145e-05,
      "loss": 0.2783,
      "step": 3492
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7032021295315822,
      "learning_rate": 4.5691696981710146e-05,
      "loss": 0.084,
      "step": 3493
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.126376915764212,
      "learning_rate": 4.5688912643671214e-05,
      "loss": 0.191,
      "step": 3494
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.735510058643272,
      "learning_rate": 4.568612749109495e-05,
      "loss": 0.4774,
      "step": 3495
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8691936875707643,
      "learning_rate": 4.568334152409103e-05,
      "loss": 0.1913,
      "step": 3496
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.851851317048531,
      "learning_rate": 4.568055474276912e-05,
      "loss": 0.1482,
      "step": 3497
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7921819968986792,
      "learning_rate": 4.567776714723896e-05,
      "loss": 0.2063,
      "step": 3498
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3442631635375941,
      "learning_rate": 4.567497873761028e-05,
      "loss": 0.2336,
      "step": 3499
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8812442795225989,
      "learning_rate": 4.5672189513992883e-05,
      "loss": 0.2229,
      "step": 3500
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8675756093736169,
      "learning_rate": 4.566939947649655e-05,
      "loss": 0.0899,
      "step": 3501
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4992216531581228,
      "learning_rate": 4.5666608625231166e-05,
      "loss": 0.4633,
      "step": 3502
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8880211350934661,
      "learning_rate": 4.566381696030658e-05,
      "loss": 0.4702,
      "step": 3503
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5045407053763145,
      "learning_rate": 4.5661024481832714e-05,
      "loss": 0.2796,
      "step": 3504
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9550905670983637,
      "learning_rate": 4.5658231189919506e-05,
      "loss": 0.1668,
      "step": 3505
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8345542141420104,
      "learning_rate": 4.5655437084676926e-05,
      "loss": 0.1093,
      "step": 3506
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.652149626365335,
      "learning_rate": 4.565264216621499e-05,
      "loss": 0.3248,
      "step": 3507
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.41570064390511,
      "learning_rate": 4.5649846434643726e-05,
      "loss": 0.2879,
      "step": 3508
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7862166659864972,
      "learning_rate": 4.564704989007321e-05,
      "loss": 0.1378,
      "step": 3509
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.94677483641482,
      "learning_rate": 4.564425253261355e-05,
      "loss": 0.1572,
      "step": 3510
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.977239653327181,
      "learning_rate": 4.564145436237487e-05,
      "loss": 0.2473,
      "step": 3511
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8309432385286684,
      "learning_rate": 4.563865537946733e-05,
      "loss": 0.1209,
      "step": 3512
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.549926376583289,
      "learning_rate": 4.563585558400114e-05,
      "loss": 0.4074,
      "step": 3513
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3620731697581865,
      "learning_rate": 4.563305497608653e-05,
      "loss": 0.2042,
      "step": 3514
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9156715331734778,
      "learning_rate": 4.5630253555833754e-05,
      "loss": 0.1464,
      "step": 3515
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8995248457895183,
      "learning_rate": 4.562745132335312e-05,
      "loss": 0.1035,
      "step": 3516
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8858074362084025,
      "learning_rate": 4.562464827875493e-05,
      "loss": 0.1268,
      "step": 3517
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0828383667659471,
      "learning_rate": 4.562184442214957e-05,
      "loss": 0.1439,
      "step": 3518
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9721889009938286,
      "learning_rate": 4.561903975364741e-05,
      "loss": 0.2492,
      "step": 3519
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2389724248413483,
      "learning_rate": 4.5616234273358876e-05,
      "loss": 0.1549,
      "step": 3520
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.181473584435597,
      "learning_rate": 4.561342798139442e-05,
      "loss": 0.2717,
      "step": 3521
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5225898560692896,
      "learning_rate": 4.561062087786454e-05,
      "loss": 0.4555,
      "step": 3522
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7911853613628209,
      "learning_rate": 4.560781296287974e-05,
      "loss": 0.1297,
      "step": 3523
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9387924053959411,
      "learning_rate": 4.560500423655057e-05,
      "loss": 0.1521,
      "step": 3524
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3208559327985372,
      "learning_rate": 4.560219469898763e-05,
      "loss": 0.3409,
      "step": 3525
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.170425462915318,
      "learning_rate": 4.559938435030151e-05,
      "loss": 0.1991,
      "step": 3526
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0668180372760707,
      "learning_rate": 4.5596573190602867e-05,
      "loss": 0.0834,
      "step": 3527
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.1987386623834664,
      "learning_rate": 4.559376122000237e-05,
      "loss": 0.4228,
      "step": 3528
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4749770504118407,
      "learning_rate": 4.5590948438610745e-05,
      "loss": 0.1658,
      "step": 3529
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7334797071949276,
      "learning_rate": 4.558813484653872e-05,
      "loss": 0.1276,
      "step": 3530
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1898402506623171,
      "learning_rate": 4.5585320443897076e-05,
      "loss": 0.3024,
      "step": 3531
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0859863360839492,
      "learning_rate": 4.55825052307966e-05,
      "loss": 0.1632,
      "step": 3532
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1492268269664332,
      "learning_rate": 4.557968920734815e-05,
      "loss": 0.2608,
      "step": 3533
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0114533378240504,
      "learning_rate": 4.5576872373662585e-05,
      "loss": 0.1313,
      "step": 3534
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8604816612217524,
      "learning_rate": 4.557405472985081e-05,
      "loss": 0.1511,
      "step": 3535
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0743316810351915,
      "learning_rate": 4.557123627602375e-05,
      "loss": 0.1478,
      "step": 3536
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0811947311116026,
      "learning_rate": 4.556841701229238e-05,
      "loss": 0.2145,
      "step": 3537
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7606563709474505,
      "learning_rate": 4.556559693876769e-05,
      "loss": 0.0961,
      "step": 3538
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8205929820983349,
      "learning_rate": 4.556277605556071e-05,
      "loss": 0.1537,
      "step": 3539
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6286121409600902,
      "learning_rate": 4.5559954362782495e-05,
      "loss": 0.1005,
      "step": 3540
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9122220990365679,
      "learning_rate": 4.5557131860544156e-05,
      "loss": 0.167,
      "step": 3541
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2434961402580011,
      "learning_rate": 4.5554308548956795e-05,
      "loss": 0.4536,
      "step": 3542
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9253583119406239,
      "learning_rate": 4.555148442813157e-05,
      "loss": 0.2238,
      "step": 3543
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9788602130327818,
      "learning_rate": 4.554865949817968e-05,
      "loss": 0.1644,
      "step": 3544
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8615485038979397,
      "learning_rate": 4.554583375921234e-05,
      "loss": 0.2219,
      "step": 3545
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7412355905734465,
      "learning_rate": 4.554300721134079e-05,
      "loss": 0.1222,
      "step": 3546
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2267412492598104,
      "learning_rate": 4.554017985467633e-05,
      "loss": 0.2209,
      "step": 3547
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6596605489472712,
      "learning_rate": 4.5537351689330274e-05,
      "loss": 0.1121,
      "step": 3548
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7942609117814922,
      "learning_rate": 4.5534522715413964e-05,
      "loss": 0.1768,
      "step": 3549
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8368278385006828,
      "learning_rate": 4.553169293303877e-05,
      "loss": 0.1225,
      "step": 3550
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9355158223454277,
      "learning_rate": 4.5528862342316114e-05,
      "loss": 0.1695,
      "step": 3551
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9153123637242966,
      "learning_rate": 4.552603094335743e-05,
      "loss": 0.1406,
      "step": 3552
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7310004341268711,
      "learning_rate": 4.552319873627421e-05,
      "loss": 0.1612,
      "step": 3553
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8839395330607979,
      "learning_rate": 4.5520365721177935e-05,
      "loss": 0.1022,
      "step": 3554
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.46996502934790724,
      "learning_rate": 4.551753189818016e-05,
      "loss": 0.0704,
      "step": 3555
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9834923181066842,
      "learning_rate": 4.551469726739245e-05,
      "loss": 0.1711,
      "step": 3556
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2029093652915694,
      "learning_rate": 4.5511861828926405e-05,
      "loss": 0.2654,
      "step": 3557
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6940694142852998,
      "learning_rate": 4.550902558289366e-05,
      "loss": 0.0832,
      "step": 3558
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9014237926696365,
      "learning_rate": 4.550618852940588e-05,
      "loss": 0.1739,
      "step": 3559
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9137598608749139,
      "learning_rate": 4.550335066857476e-05,
      "loss": 0.196,
      "step": 3560
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7961103609073629,
      "learning_rate": 4.550051200051203e-05,
      "loss": 0.0992,
      "step": 3561
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7462136245916111,
      "learning_rate": 4.549767252532945e-05,
      "loss": 0.146,
      "step": 3562
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8218823774751624,
      "learning_rate": 4.549483224313882e-05,
      "loss": 0.1308,
      "step": 3563
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2439559270104763,
      "learning_rate": 4.549199115405195e-05,
      "loss": 0.3628,
      "step": 3564
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.314911288324003,
      "learning_rate": 4.5489149258180696e-05,
      "loss": 0.2803,
      "step": 3565
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6744054081104535,
      "learning_rate": 4.548630655563696e-05,
      "loss": 0.1081,
      "step": 3566
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2720712527097644,
      "learning_rate": 4.548346304653265e-05,
      "loss": 0.3114,
      "step": 3567
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9460672076957016,
      "learning_rate": 4.548061873097972e-05,
      "loss": 0.1301,
      "step": 3568
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.645610817115761,
      "learning_rate": 4.547777360909016e-05,
      "loss": 0.0717,
      "step": 3569
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9239193764227558,
      "learning_rate": 4.547492768097597e-05,
      "loss": 0.1546,
      "step": 3570
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.948605993424982,
      "learning_rate": 4.5472080946749196e-05,
      "loss": 0.5276,
      "step": 3571
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8149163461796406,
      "learning_rate": 4.5469233406521935e-05,
      "loss": 0.1416,
      "step": 3572
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8045789779524943,
      "learning_rate": 4.546638506040628e-05,
      "loss": 0.177,
      "step": 3573
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.8229479931719648,
      "learning_rate": 4.546353590851438e-05,
      "loss": 0.1643,
      "step": 3574
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.741778927781217,
      "learning_rate": 4.5460685950958403e-05,
      "loss": 0.1335,
      "step": 3575
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5949540767146297,
      "learning_rate": 4.5457835187850555e-05,
      "loss": 0.0721,
      "step": 3576
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.160339022644702,
      "learning_rate": 4.545498361930308e-05,
      "loss": 0.1785,
      "step": 3577
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8199432357444022,
      "learning_rate": 4.545213124542824e-05,
      "loss": 0.2254,
      "step": 3578
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.3926312257982145,
      "learning_rate": 4.544927806633833e-05,
      "loss": 0.3302,
      "step": 3579
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.642997795463776,
      "learning_rate": 4.544642408214569e-05,
      "loss": 0.1417,
      "step": 3580
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1580032944596264,
      "learning_rate": 4.544356929296268e-05,
      "loss": 0.3518,
      "step": 3581
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3231376555028371,
      "learning_rate": 4.544071369890169e-05,
      "loss": 0.462,
      "step": 3582
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5463727692846843,
      "learning_rate": 4.543785730007516e-05,
      "loss": 0.15,
      "step": 3583
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2971593336110394,
      "learning_rate": 4.543500009659554e-05,
      "loss": 0.3809,
      "step": 3584
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.081214903937814,
      "learning_rate": 4.5432142088575316e-05,
      "loss": 0.1991,
      "step": 3585
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1390514694442362,
      "learning_rate": 4.5429283276127024e-05,
      "loss": 0.1899,
      "step": 3586
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8221281554966847,
      "learning_rate": 4.54264236593632e-05,
      "loss": 0.0923,
      "step": 3587
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8223052736237242,
      "learning_rate": 4.5423563238396436e-05,
      "loss": 0.163,
      "step": 3588
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.912387462434139,
      "learning_rate": 4.5420702013339355e-05,
      "loss": 0.2451,
      "step": 3589
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.890403016344995,
      "learning_rate": 4.54178399843046e-05,
      "loss": 0.1327,
      "step": 3590
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3784500508866893,
      "learning_rate": 4.541497715140485e-05,
      "loss": 0.2882,
      "step": 3591
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8635962689775866,
      "learning_rate": 4.5412113514752826e-05,
      "loss": 0.2244,
      "step": 3592
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9840139693946567,
      "learning_rate": 4.5409249074461256e-05,
      "loss": 0.1707,
      "step": 3593
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0118657637262016,
      "learning_rate": 4.540638383064293e-05,
      "loss": 0.206,
      "step": 3594
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0745977203025918,
      "learning_rate": 4.540351778341064e-05,
      "loss": 0.2202,
      "step": 3595
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8834370587471599,
      "learning_rate": 4.540065093287724e-05,
      "loss": 0.1234,
      "step": 3596
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0027336595919931,
      "learning_rate": 4.539778327915558e-05,
      "loss": 0.1545,
      "step": 3597
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9213406374549867,
      "learning_rate": 4.5394914822358586e-05,
      "loss": 0.0742,
      "step": 3598
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9611827904659438,
      "learning_rate": 4.539204556259917e-05,
      "loss": 0.6405,
      "step": 3599
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6566196735802353,
      "learning_rate": 4.538917549999031e-05,
      "loss": 0.101,
      "step": 3600
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6114999063361382,
      "learning_rate": 4.5386304634645006e-05,
      "loss": 0.3307,
      "step": 3601
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1186702594704705,
      "learning_rate": 4.538343296667627e-05,
      "loss": 0.3227,
      "step": 3602
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7953360984554194,
      "learning_rate": 4.538056049619717e-05,
      "loss": 0.274,
      "step": 3603
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0401656480889914,
      "learning_rate": 4.5377687223320794e-05,
      "loss": 0.1501,
      "step": 3604
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2418775206721708,
      "learning_rate": 4.537481314816027e-05,
      "loss": 0.1747,
      "step": 3605
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.913033440119189,
      "learning_rate": 4.537193827082875e-05,
      "loss": 0.1963,
      "step": 3606
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7218538386388459,
      "learning_rate": 4.536906259143943e-05,
      "loss": 0.1542,
      "step": 3607
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9506320369796594,
      "learning_rate": 4.53661861101055e-05,
      "loss": 0.1428,
      "step": 3608
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6153773839152874,
      "learning_rate": 4.536330882694024e-05,
      "loss": 0.0874,
      "step": 3609
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0361908144908871,
      "learning_rate": 4.536043074205692e-05,
      "loss": 0.1197,
      "step": 3610
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7360643780016531,
      "learning_rate": 4.5357551855568844e-05,
      "loss": 0.5105,
      "step": 3611
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7864556297880607,
      "learning_rate": 4.535467216758936e-05,
      "loss": 0.1373,
      "step": 3612
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0827170141549347,
      "learning_rate": 4.535179167823185e-05,
      "loss": 0.4293,
      "step": 3613
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5099612683594801,
      "learning_rate": 4.534891038760971e-05,
      "loss": 0.0633,
      "step": 3614
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.57002868802137,
      "learning_rate": 4.5346028295836386e-05,
      "loss": 0.3088,
      "step": 3615
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0928377476262066,
      "learning_rate": 4.534314540302534e-05,
      "loss": 0.2134,
      "step": 3616
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9831362507545841,
      "learning_rate": 4.534026170929009e-05,
      "loss": 0.2821,
      "step": 3617
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0381879507621319,
      "learning_rate": 4.533737721474416e-05,
      "loss": 0.1769,
      "step": 3618
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.912481421965437,
      "learning_rate": 4.533449191950111e-05,
      "loss": 0.6505,
      "step": 3619
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0508852135101545,
      "learning_rate": 4.533160582367454e-05,
      "loss": 0.352,
      "step": 3620
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7388022957584565,
      "learning_rate": 4.5328718927378076e-05,
      "loss": 0.103,
      "step": 3621
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.250141191239249,
      "learning_rate": 4.532583123072538e-05,
      "loss": 0.8062,
      "step": 3622
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6708470810673387,
      "learning_rate": 4.5322942733830146e-05,
      "loss": 0.1195,
      "step": 3623
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9493252046890456,
      "learning_rate": 4.5320053436806087e-05,
      "loss": 0.2472,
      "step": 3624
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.317044939673463,
      "learning_rate": 4.5317163339766964e-05,
      "loss": 0.1436,
      "step": 3625
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.895647939250875,
      "learning_rate": 4.531427244282655e-05,
      "loss": 0.0931,
      "step": 3626
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0805218061048285,
      "learning_rate": 4.5311380746098685e-05,
      "loss": 0.2673,
      "step": 3627
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6625305640111843,
      "learning_rate": 4.5308488249697204e-05,
      "loss": 0.2178,
      "step": 3628
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7516068766391192,
      "learning_rate": 4.530559495373598e-05,
      "loss": 0.1086,
      "step": 3629
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8064099032645671,
      "learning_rate": 4.530270085832893e-05,
      "loss": 0.1383,
      "step": 3630
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7413612199544677,
      "learning_rate": 4.529980596358999e-05,
      "loss": 0.427,
      "step": 3631
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3662655098477843,
      "learning_rate": 4.529691026963315e-05,
      "loss": 0.2172,
      "step": 3632
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7481874478983686,
      "learning_rate": 4.52940137765724e-05,
      "loss": 0.1039,
      "step": 3633
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9822633178384097,
      "learning_rate": 4.529111648452179e-05,
      "loss": 0.1828,
      "step": 3634
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7951950086825363,
      "learning_rate": 4.528821839359538e-05,
      "loss": 0.1281,
      "step": 3635
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6372008594622236,
      "learning_rate": 4.528531950390727e-05,
      "loss": 0.3193,
      "step": 3636
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1297898757784126,
      "learning_rate": 4.52824198155716e-05,
      "loss": 0.1407,
      "step": 3637
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1042196210002542,
      "learning_rate": 4.527951932870251e-05,
      "loss": 0.1572,
      "step": 3638
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.109851157043204,
      "learning_rate": 4.527661804341422e-05,
      "loss": 0.1614,
      "step": 3639
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9554888899667222,
      "learning_rate": 4.527371595982095e-05,
      "loss": 0.3848,
      "step": 3640
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7525757484976291,
      "learning_rate": 4.527081307803695e-05,
      "loss": 0.0913,
      "step": 3641
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7358154001544794,
      "learning_rate": 4.526790939817651e-05,
      "loss": 0.3688,
      "step": 3642
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2233526732461235,
      "learning_rate": 4.5265004920353954e-05,
      "loss": 0.1551,
      "step": 3643
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2058681128820612,
      "learning_rate": 4.526209964468363e-05,
      "loss": 0.1223,
      "step": 3644
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2469162477503342,
      "learning_rate": 4.525919357127992e-05,
      "loss": 0.2242,
      "step": 3645
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6693459032919,
      "learning_rate": 4.525628670025724e-05,
      "loss": 0.1323,
      "step": 3646
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9566824071248365,
      "learning_rate": 4.5253379031730036e-05,
      "loss": 0.2606,
      "step": 3647
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8030168900662744,
      "learning_rate": 4.5250470565812785e-05,
      "loss": 0.3925,
      "step": 3648
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7395982358579656,
      "learning_rate": 4.5247561302619986e-05,
      "loss": 0.1126,
      "step": 3649
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0768817796927597,
      "learning_rate": 4.52446512422662e-05,
      "loss": 0.1667,
      "step": 3650
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5481842223346067,
      "learning_rate": 4.524174038486598e-05,
      "loss": 0.6028,
      "step": 3651
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0779290993608612,
      "learning_rate": 4.523882873053394e-05,
      "loss": 0.2536,
      "step": 3652
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2652102258766005,
      "learning_rate": 4.5235916279384706e-05,
      "loss": 0.2692,
      "step": 3653
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.945708246042362,
      "learning_rate": 4.5233003031532936e-05,
      "loss": 0.1126,
      "step": 3654
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1508146555350527,
      "learning_rate": 4.523008898709334e-05,
      "loss": 0.1709,
      "step": 3655
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.813047983611079,
      "learning_rate": 4.522717414618065e-05,
      "loss": 0.1027,
      "step": 3656
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2732862500606736,
      "learning_rate": 4.522425850890962e-05,
      "loss": 0.3125,
      "step": 3657
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8835623731497703,
      "learning_rate": 4.5221342075395035e-05,
      "loss": 0.1907,
      "step": 3658
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7301345684175311,
      "learning_rate": 4.521842484575172e-05,
      "loss": 0.1302,
      "step": 3659
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9445386196420262,
      "learning_rate": 4.521550682009452e-05,
      "loss": 0.1556,
      "step": 3660
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9676194934704446,
      "learning_rate": 4.5212587998538335e-05,
      "loss": 0.1079,
      "step": 3661
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7667443243249816,
      "learning_rate": 4.520966838119807e-05,
      "loss": 0.0982,
      "step": 3662
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2445597734431904,
      "learning_rate": 4.520674796818868e-05,
      "loss": 0.5446,
      "step": 3663
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8933528846980184,
      "learning_rate": 4.5203826759625144e-05,
      "loss": 0.1543,
      "step": 3664
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7031033188191352,
      "learning_rate": 4.520090475562247e-05,
      "loss": 0.1053,
      "step": 3665
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5126262471583545,
      "learning_rate": 4.5197981956295696e-05,
      "loss": 0.045,
      "step": 3666
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.046406463340658,
      "learning_rate": 4.519505836175989e-05,
      "loss": 0.1766,
      "step": 3667
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0545094351995832,
      "learning_rate": 4.5192133972130173e-05,
      "loss": 0.3035,
      "step": 3668
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.013846661315229,
      "learning_rate": 4.5189208787521665e-05,
      "loss": 0.2654,
      "step": 3669
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4012454998867294,
      "learning_rate": 4.518628280804954e-05,
      "loss": 0.5604,
      "step": 3670
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1351226473004288,
      "learning_rate": 4.5183356033828996e-05,
      "loss": 0.1265,
      "step": 3671
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2331251472350762,
      "learning_rate": 4.5180428464975255e-05,
      "loss": 0.17,
      "step": 3672
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7371960156053666,
      "learning_rate": 4.5177500101603586e-05,
      "loss": 0.1941,
      "step": 3673
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8354698123495881,
      "learning_rate": 4.5174570943829274e-05,
      "loss": 0.1918,
      "step": 3674
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6852933774471248,
      "learning_rate": 4.517164099176765e-05,
      "loss": 0.3932,
      "step": 3675
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8904499084060206,
      "learning_rate": 4.516871024553406e-05,
      "loss": 0.1193,
      "step": 3676
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.405649369916691,
      "learning_rate": 4.51657787052439e-05,
      "loss": 0.2371,
      "step": 3677
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4304728856234101,
      "learning_rate": 4.516284637101257e-05,
      "loss": 0.4028,
      "step": 3678
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5919680569186185,
      "learning_rate": 4.5159913242955534e-05,
      "loss": 0.4518,
      "step": 3679
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7287177248046701,
      "learning_rate": 4.5156979321188264e-05,
      "loss": 0.0972,
      "step": 3680
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.580984717280125,
      "learning_rate": 4.515404460582627e-05,
      "loss": 0.3646,
      "step": 3681
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0100573929062293,
      "learning_rate": 4.51511090969851e-05,
      "loss": 0.2383,
      "step": 3682
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0027406199850857,
      "learning_rate": 4.514817279478032e-05,
      "loss": 0.216,
      "step": 3683
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1613151456483164,
      "learning_rate": 4.514523569932754e-05,
      "loss": 0.1624,
      "step": 3684
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.733115353716856,
      "learning_rate": 4.51422978107424e-05,
      "loss": 0.4238,
      "step": 3685
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4056159800434205,
      "learning_rate": 4.5139359129140545e-05,
      "loss": 0.3183,
      "step": 3686
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0061984293635062,
      "learning_rate": 4.5136419654637696e-05,
      "loss": 0.2653,
      "step": 3687
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8972722503804238,
      "learning_rate": 4.513347938734957e-05,
      "loss": 0.2588,
      "step": 3688
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.63222432944165,
      "learning_rate": 4.513053832739194e-05,
      "loss": 0.0908,
      "step": 3689
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7211053841027487,
      "learning_rate": 4.512759647488057e-05,
      "loss": 0.105,
      "step": 3690
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0601017224022793,
      "learning_rate": 4.512465382993132e-05,
      "loss": 0.2048,
      "step": 3691
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6994032218318307,
      "learning_rate": 4.512171039266002e-05,
      "loss": 0.1257,
      "step": 3692
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7982827065358407,
      "learning_rate": 4.5118766163182564e-05,
      "loss": 0.2022,
      "step": 3693
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0098901946135896,
      "learning_rate": 4.511582114161486e-05,
      "loss": 0.1837,
      "step": 3694
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1231489339213008,
      "learning_rate": 4.511287532807287e-05,
      "loss": 0.3068,
      "step": 3695
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6956558704379532,
      "learning_rate": 4.5109928722672556e-05,
      "loss": 0.1027,
      "step": 3696
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6152487455864495,
      "learning_rate": 4.5106981325529935e-05,
      "loss": 0.4739,
      "step": 3697
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7902703603061944,
      "learning_rate": 4.510403313676105e-05,
      "loss": 0.099,
      "step": 3698
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.912345421169405,
      "learning_rate": 4.5101084156481975e-05,
      "loss": 0.2187,
      "step": 3699
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.268291592610643,
      "learning_rate": 4.509813438480881e-05,
      "loss": 0.2377,
      "step": 3700
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2865315218239692,
      "learning_rate": 4.5095183821857686e-05,
      "loss": 0.3672,
      "step": 3701
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4960711944273124,
      "learning_rate": 4.509223246774478e-05,
      "loss": 0.4553,
      "step": 3702
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2241750706511896,
      "learning_rate": 4.508928032258628e-05,
      "loss": 0.3421,
      "step": 3703
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2059312031942218,
      "learning_rate": 4.508632738649842e-05,
      "loss": 0.2084,
      "step": 3704
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2681606097513967,
      "learning_rate": 4.508337365959745e-05,
      "loss": 0.2201,
      "step": 3705
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.782273654955275,
      "learning_rate": 4.508041914199967e-05,
      "loss": 0.1172,
      "step": 3706
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1598096906625301,
      "learning_rate": 4.507746383382139e-05,
      "loss": 0.3388,
      "step": 3707
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5412576118445338,
      "learning_rate": 4.5074507735178976e-05,
      "loss": 0.0486,
      "step": 3708
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4647277816437663,
      "learning_rate": 4.50715508461888e-05,
      "loss": 0.3159,
      "step": 3709
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.014825918554259,
      "learning_rate": 4.5068593166967295e-05,
      "loss": 0.2493,
      "step": 3710
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8856268018460656,
      "learning_rate": 4.506563469763089e-05,
      "loss": 0.1745,
      "step": 3711
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7838233195691937,
      "learning_rate": 4.5062675438296056e-05,
      "loss": 0.1101,
      "step": 3712
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8569209232964142,
      "learning_rate": 4.505971538907932e-05,
      "loss": 0.1579,
      "step": 3713
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.233807228653152,
      "learning_rate": 4.5056754550097205e-05,
      "loss": 0.2716,
      "step": 3714
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9115677511030541,
      "learning_rate": 4.5053792921466296e-05,
      "loss": 0.1595,
      "step": 3715
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.8374581438132696,
      "learning_rate": 4.505083050330319e-05,
      "loss": 1.1461,
      "step": 3716
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9311071982130718,
      "learning_rate": 4.5047867295724514e-05,
      "loss": 0.1488,
      "step": 3717
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8432874196530625,
      "learning_rate": 4.5044903298846927e-05,
      "loss": 0.165,
      "step": 3718
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7013813975178893,
      "learning_rate": 4.504193851278714e-05,
      "loss": 0.0779,
      "step": 3719
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.867187475655485,
      "learning_rate": 4.503897293766187e-05,
      "loss": 0.1222,
      "step": 3720
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7427303676323581,
      "learning_rate": 4.503600657358787e-05,
      "loss": 0.1207,
      "step": 3721
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.035719175688726,
      "learning_rate": 4.503303942068194e-05,
      "loss": 0.1829,
      "step": 3722
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9377937434007438,
      "learning_rate": 4.5030071479060885e-05,
      "loss": 0.1574,
      "step": 3723
    },
    {
      "epoch": 0.23,
      "grad_norm": 13.240199748472913,
      "learning_rate": 4.502710274884157e-05,
      "loss": 0.6998,
      "step": 3724
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.848202004430582,
      "learning_rate": 4.502413323014085e-05,
      "loss": 0.3101,
      "step": 3725
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8663398335781671,
      "learning_rate": 4.502116292307567e-05,
      "loss": 0.2144,
      "step": 3726
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3533135751451277,
      "learning_rate": 4.5018191827762945e-05,
      "loss": 0.1644,
      "step": 3727
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7706482441523095,
      "learning_rate": 4.501521994431966e-05,
      "loss": 0.2778,
      "step": 3728
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9920645678409723,
      "learning_rate": 4.501224727286283e-05,
      "loss": 0.7204,
      "step": 3729
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.114289496866365,
      "learning_rate": 4.5009273813509486e-05,
      "loss": 0.4199,
      "step": 3730
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2792984856590015,
      "learning_rate": 4.5006299566376685e-05,
      "loss": 0.2035,
      "step": 3731
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8757964652931084,
      "learning_rate": 4.500332453158153e-05,
      "loss": 0.1218,
      "step": 3732
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.360627510275651,
      "learning_rate": 4.500034870924116e-05,
      "loss": 0.5321,
      "step": 3733
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9076810980763855,
      "learning_rate": 4.499737209947272e-05,
      "loss": 0.2168,
      "step": 3734
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3258944087690996,
      "learning_rate": 4.499439470239341e-05,
      "loss": 0.1587,
      "step": 3735
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.803227148813003,
      "learning_rate": 4.499141651812045e-05,
      "loss": 0.243,
      "step": 3736
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.280392445982938,
      "learning_rate": 4.49884375467711e-05,
      "loss": 0.1935,
      "step": 3737
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.405804341541834,
      "learning_rate": 4.498545778846264e-05,
      "loss": 0.2641,
      "step": 3738
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3457984247875368,
      "learning_rate": 4.498247724331237e-05,
      "loss": 0.2689,
      "step": 3739
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2091525169271347,
      "learning_rate": 4.4979495911437665e-05,
      "loss": 0.279,
      "step": 3740
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7140840162540512,
      "learning_rate": 4.497651379295588e-05,
      "loss": 0.0981,
      "step": 3741
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7598629447915408,
      "learning_rate": 4.497353088798443e-05,
      "loss": 0.3716,
      "step": 3742
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8967569351904255,
      "learning_rate": 4.497054719664075e-05,
      "loss": 0.1805,
      "step": 3743
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1284105750757634,
      "learning_rate": 4.496756271904233e-05,
      "loss": 0.3204,
      "step": 3744
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1105850144980494,
      "learning_rate": 4.496457745530665e-05,
      "loss": 0.1552,
      "step": 3745
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4969648914904268,
      "learning_rate": 4.496159140555124e-05,
      "loss": 0.412,
      "step": 3746
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4253321472863694,
      "learning_rate": 4.495860456989367e-05,
      "loss": 0.3017,
      "step": 3747
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0668785891030976,
      "learning_rate": 4.495561694845154e-05,
      "loss": 0.148,
      "step": 3748
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8785171927569925,
      "learning_rate": 4.495262854134247e-05,
      "loss": 0.1638,
      "step": 3749
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1364969642000853,
      "learning_rate": 4.494963934868411e-05,
      "loss": 0.2547,
      "step": 3750
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3124618355200066,
      "learning_rate": 4.4946649370594155e-05,
      "loss": 0.2333,
      "step": 3751
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9140461196039148,
      "learning_rate": 4.4943658607190316e-05,
      "loss": 0.1129,
      "step": 3752
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.511221760538784,
      "learning_rate": 4.4940667058590344e-05,
      "loss": 0.4114,
      "step": 3753
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.024021770089967,
      "learning_rate": 4.4937674724912024e-05,
      "loss": 0.1423,
      "step": 3754
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.403280213957324,
      "learning_rate": 4.493468160627315e-05,
      "loss": 0.2956,
      "step": 3755
    },
    {
      "epoch": 0.23,
      "grad_norm": 10.23536063866866,
      "learning_rate": 4.493168770279159e-05,
      "loss": 1.1253,
      "step": 3756
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.17552809913365,
      "learning_rate": 4.492869301458519e-05,
      "loss": 0.1913,
      "step": 3757
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5421736719972408,
      "learning_rate": 4.492569754177187e-05,
      "loss": 0.2847,
      "step": 3758
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.9797438889828896,
      "learning_rate": 4.4922701284469556e-05,
      "loss": 0.4881,
      "step": 3759
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9516350857198749,
      "learning_rate": 4.4919704242796215e-05,
      "loss": 0.1761,
      "step": 3760
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.168758522002726,
      "learning_rate": 4.491670641686983e-05,
      "loss": 0.2833,
      "step": 3761
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8324505370113664,
      "learning_rate": 4.491370780680846e-05,
      "loss": 0.1174,
      "step": 3762
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7859192810657281,
      "learning_rate": 4.491070841273013e-05,
      "loss": 0.1507,
      "step": 3763
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3256688576055728,
      "learning_rate": 4.490770823475294e-05,
      "loss": 0.251,
      "step": 3764
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1084904328547682,
      "learning_rate": 4.490470727299501e-05,
      "loss": 0.2418,
      "step": 3765
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9685049875326323,
      "learning_rate": 4.490170552757449e-05,
      "loss": 0.1535,
      "step": 3766
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2617350991850158,
      "learning_rate": 4.489870299860956e-05,
      "loss": 0.2211,
      "step": 3767
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2905952828471903,
      "learning_rate": 4.489569968621843e-05,
      "loss": 0.1284,
      "step": 3768
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0762973655762784,
      "learning_rate": 4.489269559051935e-05,
      "loss": 0.2231,
      "step": 3769
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8063517291478507,
      "learning_rate": 4.488969071163058e-05,
      "loss": 0.1112,
      "step": 3770
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8026042376635956,
      "learning_rate": 4.488668504967044e-05,
      "loss": 0.1131,
      "step": 3771
    },
    {
      "epoch": 0.23,
      "grad_norm": 10.638793489108343,
      "learning_rate": 4.4883678604757264e-05,
      "loss": 1.6065,
      "step": 3772
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.334774588994899,
      "learning_rate": 4.48806713770094e-05,
      "loss": 0.1944,
      "step": 3773
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9061301635770161,
      "learning_rate": 4.487766336654526e-05,
      "loss": 0.1573,
      "step": 3774
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1564925669726165,
      "learning_rate": 4.4874654573483264e-05,
      "loss": 0.2141,
      "step": 3775
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0820540798715044,
      "learning_rate": 4.4871644997941875e-05,
      "loss": 0.2125,
      "step": 3776
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7117865870138387,
      "learning_rate": 4.486863464003958e-05,
      "loss": 0.378,
      "step": 3777
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1370698081482753,
      "learning_rate": 4.486562349989491e-05,
      "loss": 0.1404,
      "step": 3778
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9400976075866434,
      "learning_rate": 4.486261157762639e-05,
      "loss": 0.1651,
      "step": 3779
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7914058315978206,
      "learning_rate": 4.485959887335263e-05,
      "loss": 0.1558,
      "step": 3780
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4257364359319102,
      "learning_rate": 4.485658538719222e-05,
      "loss": 0.3785,
      "step": 3781
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0403334298791085,
      "learning_rate": 4.485357111926381e-05,
      "loss": 0.0827,
      "step": 3782
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2243123939442262,
      "learning_rate": 4.485055606968609e-05,
      "loss": 0.1756,
      "step": 3783
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7878234243746648,
      "learning_rate": 4.484754023857774e-05,
      "loss": 0.1331,
      "step": 3784
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.007833205729439,
      "learning_rate": 4.484452362605751e-05,
      "loss": 0.1816,
      "step": 3785
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.138532548262207,
      "learning_rate": 4.484150623224416e-05,
      "loss": 0.2664,
      "step": 3786
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1149414042727148,
      "learning_rate": 4.4838488057256495e-05,
      "loss": 0.2498,
      "step": 3787
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1269379657026322,
      "learning_rate": 4.483546910121333e-05,
      "loss": 0.2101,
      "step": 3788
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.910222258000282,
      "learning_rate": 4.483244936423354e-05,
      "loss": 0.0804,
      "step": 3789
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1011332998452608,
      "learning_rate": 4.4829428846436004e-05,
      "loss": 0.1612,
      "step": 3790
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.035448490527162,
      "learning_rate": 4.4826407547939644e-05,
      "loss": 0.1487,
      "step": 3791
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7620337133208266,
      "learning_rate": 4.4823385468863406e-05,
      "loss": 0.135,
      "step": 3792
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8388542292006866,
      "learning_rate": 4.4820362609326274e-05,
      "loss": 0.1471,
      "step": 3793
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8420090524725199,
      "learning_rate": 4.4817338969447266e-05,
      "loss": 0.1373,
      "step": 3794
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9974600911368304,
      "learning_rate": 4.4814314549345414e-05,
      "loss": 0.3275,
      "step": 3795
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8404201370448383,
      "learning_rate": 4.48112893491398e-05,
      "loss": 0.4137,
      "step": 3796
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7415983798886124,
      "learning_rate": 4.480826336894953e-05,
      "loss": 0.0788,
      "step": 3797
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8603313464999239,
      "learning_rate": 4.480523660889373e-05,
      "loss": 0.1366,
      "step": 3798
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3757221571685727,
      "learning_rate": 4.480220906909157e-05,
      "loss": 0.3043,
      "step": 3799
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7462548854524056,
      "learning_rate": 4.479918074966225e-05,
      "loss": 0.1215,
      "step": 3800
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2885003338204863,
      "learning_rate": 4.4796151650724994e-05,
      "loss": 0.2266,
      "step": 3801
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7881394849066706,
      "learning_rate": 4.479312177239906e-05,
      "loss": 0.1135,
      "step": 3802
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9549085777684071,
      "learning_rate": 4.4790091114803734e-05,
      "loss": 0.1398,
      "step": 3803
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2326904353786001,
      "learning_rate": 4.478705967805834e-05,
      "loss": 0.2602,
      "step": 3804
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9845132149795789,
      "learning_rate": 4.478402746228223e-05,
      "loss": 0.12,
      "step": 3805
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1770375600300522,
      "learning_rate": 4.4780994467594775e-05,
      "loss": 0.2427,
      "step": 3806
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3495385596487686,
      "learning_rate": 4.4777960694115393e-05,
      "loss": 0.2512,
      "step": 3807
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6981241093062793,
      "learning_rate": 4.477492614196352e-05,
      "loss": 0.1008,
      "step": 3808
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9384992775655207,
      "learning_rate": 4.477189081125864e-05,
      "loss": 0.1981,
      "step": 3809
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2632207923698366,
      "learning_rate": 4.4768854702120245e-05,
      "loss": 0.4266,
      "step": 3810
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5972265786416409,
      "learning_rate": 4.476581781466788e-05,
      "loss": 0.0684,
      "step": 3811
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9553911813180392,
      "learning_rate": 4.476278014902109e-05,
      "loss": 0.2388,
      "step": 3812
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1852871632500934,
      "learning_rate": 4.475974170529948e-05,
      "loss": 0.2187,
      "step": 3813
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5710132039026863,
      "learning_rate": 4.4756702483622694e-05,
      "loss": 0.4321,
      "step": 3814
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3629316605605282,
      "learning_rate": 4.475366248411036e-05,
      "loss": 0.3764,
      "step": 3815
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0451032070679183,
      "learning_rate": 4.475062170688218e-05,
      "loss": 0.1737,
      "step": 3816
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2076154755484663,
      "learning_rate": 4.474758015205787e-05,
      "loss": 0.1994,
      "step": 3817
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6915937058627983,
      "learning_rate": 4.474453781975717e-05,
      "loss": 0.0928,
      "step": 3818
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.858826044533993,
      "learning_rate": 4.4741494710099875e-05,
      "loss": 0.1255,
      "step": 3819
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7409585840026852,
      "learning_rate": 4.473845082320578e-05,
      "loss": 0.1671,
      "step": 3820
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.548468253985336,
      "learning_rate": 4.473540615919473e-05,
      "loss": 0.3881,
      "step": 3821
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1229033367882697,
      "learning_rate": 4.4732360718186594e-05,
      "loss": 0.1923,
      "step": 3822
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4883417000085222,
      "learning_rate": 4.472931450030128e-05,
      "loss": 0.2224,
      "step": 3823
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2296981907621938,
      "learning_rate": 4.472626750565871e-05,
      "loss": 0.2647,
      "step": 3824
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7580855436156928,
      "learning_rate": 4.472321973437885e-05,
      "loss": 0.1124,
      "step": 3825
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8781380525008032,
      "learning_rate": 4.47201711865817e-05,
      "loss": 0.1603,
      "step": 3826
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0525856365858768,
      "learning_rate": 4.471712186238727e-05,
      "loss": 0.1025,
      "step": 3827
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.031861240378026,
      "learning_rate": 4.471407176191563e-05,
      "loss": 0.145,
      "step": 3828
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6567945211902038,
      "learning_rate": 4.4711020885286846e-05,
      "loss": 0.2876,
      "step": 3829
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7994727060471548,
      "learning_rate": 4.470796923262105e-05,
      "loss": 0.1644,
      "step": 3830
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6827479923950334,
      "learning_rate": 4.4704916804038385e-05,
      "loss": 0.1103,
      "step": 3831
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2554449654159516,
      "learning_rate": 4.470186359965901e-05,
      "loss": 0.2583,
      "step": 3832
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9736756705503848,
      "learning_rate": 4.469880961960316e-05,
      "loss": 0.196,
      "step": 3833
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4944388686728434,
      "learning_rate": 4.469575486399105e-05,
      "loss": 0.3244,
      "step": 3834
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8733303872698683,
      "learning_rate": 4.469269933294296e-05,
      "loss": 0.1389,
      "step": 3835
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8519226016349845,
      "learning_rate": 4.468964302657918e-05,
      "loss": 0.1615,
      "step": 3836
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9925766897838113,
      "learning_rate": 4.468658594502004e-05,
      "loss": 0.1677,
      "step": 3837
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9686731209992029,
      "learning_rate": 4.468352808838591e-05,
      "loss": 0.1998,
      "step": 3838
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1010121981755518,
      "learning_rate": 4.4680469456797166e-05,
      "loss": 0.1675,
      "step": 3839
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8829221990850742,
      "learning_rate": 4.467741005037425e-05,
      "loss": 0.3342,
      "step": 3840
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.913049774783547,
      "learning_rate": 4.467434986923758e-05,
      "loss": 0.1595,
      "step": 3841
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8560811255457202,
      "learning_rate": 4.467128891350766e-05,
      "loss": 0.1105,
      "step": 3842
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0197735803168357,
      "learning_rate": 4.4668227183305e-05,
      "loss": 0.1969,
      "step": 3843
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3419379616570684,
      "learning_rate": 4.4665164678750135e-05,
      "loss": 0.1532,
      "step": 3844
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7202042796794819,
      "learning_rate": 4.466210139996365e-05,
      "loss": 0.2752,
      "step": 3845
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3626702681059646,
      "learning_rate": 4.465903734706614e-05,
      "loss": 0.2645,
      "step": 3846
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8718068059388894,
      "learning_rate": 4.465597252017824e-05,
      "loss": 0.1303,
      "step": 3847
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1765715305742581,
      "learning_rate": 4.465290691942061e-05,
      "loss": 0.228,
      "step": 3848
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2031671359402167,
      "learning_rate": 4.464984054491396e-05,
      "loss": 0.1913,
      "step": 3849
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7394239041454039,
      "learning_rate": 4.4646773396779e-05,
      "loss": 0.1165,
      "step": 3850
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8722121412248567,
      "learning_rate": 4.4643705475136485e-05,
      "loss": 0.1071,
      "step": 3851
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0171362151478225,
      "learning_rate": 4.464063678010721e-05,
      "loss": 0.1644,
      "step": 3852
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0386057522568506,
      "learning_rate": 4.4637567311812e-05,
      "loss": 0.3037,
      "step": 3853
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.746923621103738,
      "learning_rate": 4.4634497070371686e-05,
      "loss": 0.1253,
      "step": 3854
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8424700245083862,
      "learning_rate": 4.463142605590714e-05,
      "loss": 0.1413,
      "step": 3855
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7750997590982913,
      "learning_rate": 4.462835426853929e-05,
      "loss": 0.1045,
      "step": 3856
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.920036721443994,
      "learning_rate": 4.4625281708389066e-05,
      "loss": 0.1675,
      "step": 3857
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1718112363230762,
      "learning_rate": 4.4622208375577434e-05,
      "loss": 0.2322,
      "step": 3858
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2333613555326082,
      "learning_rate": 4.461913427022539e-05,
      "loss": 0.2074,
      "step": 3859
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0142898710846224,
      "learning_rate": 4.461605939245398e-05,
      "loss": 0.2367,
      "step": 3860
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.948936905448284,
      "learning_rate": 4.461298374238425e-05,
      "loss": 0.1677,
      "step": 3861
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8396501869445407,
      "learning_rate": 4.460990732013729e-05,
      "loss": 0.1675,
      "step": 3862
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.5382218663007556,
      "learning_rate": 4.460683012583423e-05,
      "loss": 0.4581,
      "step": 3863
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.697994880154048,
      "learning_rate": 4.460375215959621e-05,
      "loss": 0.1298,
      "step": 3864
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9724007727730202,
      "learning_rate": 4.4600673421544426e-05,
      "loss": 0.1228,
      "step": 3865
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8263459531350532,
      "learning_rate": 4.459759391180007e-05,
      "loss": 0.2641,
      "step": 3866
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8253216232890066,
      "learning_rate": 4.459451363048441e-05,
      "loss": 0.1392,
      "step": 3867
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7989503377369083,
      "learning_rate": 4.4591432577718695e-05,
      "loss": 0.1398,
      "step": 3868
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.052557672067105,
      "learning_rate": 4.458835075362424e-05,
      "loss": 0.092,
      "step": 3869
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7901872913745029,
      "learning_rate": 4.4585268158322385e-05,
      "loss": 0.1001,
      "step": 3870
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7779838255089543,
      "learning_rate": 4.4582184791934475e-05,
      "loss": 0.1199,
      "step": 3871
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0200859805437,
      "learning_rate": 4.457910065458193e-05,
      "loss": 0.2561,
      "step": 3872
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8237173489280012,
      "learning_rate": 4.457601574638615e-05,
      "loss": 0.1308,
      "step": 3873
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4043164559323726,
      "learning_rate": 4.457293006746861e-05,
      "loss": 0.2838,
      "step": 3874
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3167661298782132,
      "learning_rate": 4.456984361795078e-05,
      "loss": 0.209,
      "step": 3875
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9688542668112335,
      "learning_rate": 4.456675639795418e-05,
      "loss": 0.1398,
      "step": 3876
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.035226744265297,
      "learning_rate": 4.456366840760036e-05,
      "loss": 0.1195,
      "step": 3877
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7576539689040714,
      "learning_rate": 4.4560579647010885e-05,
      "loss": 0.1148,
      "step": 3878
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8313083704779554,
      "learning_rate": 4.4557490116307376e-05,
      "loss": 0.1052,
      "step": 3879
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8636132000007022,
      "learning_rate": 4.455439981561147e-05,
      "loss": 0.1036,
      "step": 3880
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6784833123214514,
      "learning_rate": 4.455130874504483e-05,
      "loss": 0.0803,
      "step": 3881
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1656490482856139,
      "learning_rate": 4.4548216904729144e-05,
      "loss": 0.1561,
      "step": 3882
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.5959326255085773,
      "learning_rate": 4.4545124294786155e-05,
      "loss": 0.3667,
      "step": 3883
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1400070652108303,
      "learning_rate": 4.4542030915337606e-05,
      "loss": 0.6424,
      "step": 3884
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2999196859060822,
      "learning_rate": 4.453893676650531e-05,
      "loss": 0.2187,
      "step": 3885
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5615332992459694,
      "learning_rate": 4.4535841848411057e-05,
      "loss": 0.1537,
      "step": 3886
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9480023589805879,
      "learning_rate": 4.4532746161176716e-05,
      "loss": 0.1162,
      "step": 3887
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8528722738652809,
      "learning_rate": 4.452964970492416e-05,
      "loss": 0.1387,
      "step": 3888
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0663848610282416,
      "learning_rate": 4.4526552479775294e-05,
      "loss": 0.2361,
      "step": 3889
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5474182144338524,
      "learning_rate": 4.452345448585207e-05,
      "loss": 0.3889,
      "step": 3890
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7212078373409538,
      "learning_rate": 4.452035572327645e-05,
      "loss": 0.0743,
      "step": 3891
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8343293214413294,
      "learning_rate": 4.451725619217043e-05,
      "loss": 0.1538,
      "step": 3892
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.243032856114318,
      "learning_rate": 4.451415589265605e-05,
      "loss": 0.9508,
      "step": 3893
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0066498046465548,
      "learning_rate": 4.451105482485537e-05,
      "loss": 0.2096,
      "step": 3894
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.089294963416566,
      "learning_rate": 4.450795298889047e-05,
      "loss": 0.1567,
      "step": 3895
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4625516051820693,
      "learning_rate": 4.450485038488349e-05,
      "loss": 0.5134,
      "step": 3896
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.22902774977674,
      "learning_rate": 4.4501747012956565e-05,
      "loss": 0.4042,
      "step": 3897
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9095047314501582,
      "learning_rate": 4.449864287323188e-05,
      "loss": 0.088,
      "step": 3898
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9931242180075477,
      "learning_rate": 4.449553796583166e-05,
      "loss": 0.1346,
      "step": 3899
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2797526067120595,
      "learning_rate": 4.4492432290878136e-05,
      "loss": 0.1934,
      "step": 3900
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6587901486377729,
      "learning_rate": 4.448932584849358e-05,
      "loss": 0.1166,
      "step": 3901
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0031877419493902,
      "learning_rate": 4.4486218638800296e-05,
      "loss": 0.2197,
      "step": 3902
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7903068056031806,
      "learning_rate": 4.448311066192063e-05,
      "loss": 0.1346,
      "step": 3903
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0193239860239318,
      "learning_rate": 4.448000191797692e-05,
      "loss": 0.122,
      "step": 3904
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4109224993327303,
      "learning_rate": 4.447689240709158e-05,
      "loss": 0.4698,
      "step": 3905
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7350039269390584,
      "learning_rate": 4.447378212938703e-05,
      "loss": 0.0929,
      "step": 3906
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7138514937427662,
      "learning_rate": 4.447067108498572e-05,
      "loss": 0.1033,
      "step": 3907
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0378636862866815,
      "learning_rate": 4.4467559274010126e-05,
      "loss": 0.4095,
      "step": 3908
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6654523060874777,
      "learning_rate": 4.4464446696582786e-05,
      "loss": 0.383,
      "step": 3909
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8252952293738162,
      "learning_rate": 4.446133335282622e-05,
      "loss": 0.126,
      "step": 3910
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9860473743905518,
      "learning_rate": 4.445821924286302e-05,
      "loss": 0.1298,
      "step": 3911
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.92978677469106,
      "learning_rate": 4.4455104366815775e-05,
      "loss": 0.134,
      "step": 3912
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1505966805838093,
      "learning_rate": 4.445198872480713e-05,
      "loss": 0.2071,
      "step": 3913
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8706897042455929,
      "learning_rate": 4.4448872316959746e-05,
      "loss": 0.1245,
      "step": 3914
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.098236891164167,
      "learning_rate": 4.4445755143396326e-05,
      "loss": 0.1388,
      "step": 3915
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9396502452941651,
      "learning_rate": 4.444263720423959e-05,
      "loss": 0.1954,
      "step": 3916
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7632642544119991,
      "learning_rate": 4.443951849961229e-05,
      "loss": 0.1246,
      "step": 3917
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9438128064889374,
      "learning_rate": 4.443639902963722e-05,
      "loss": 0.1944,
      "step": 3918
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1106962845131512,
      "learning_rate": 4.443327879443718e-05,
      "loss": 0.2472,
      "step": 3919
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7409656256418602,
      "learning_rate": 4.4430157794135044e-05,
      "loss": 0.1298,
      "step": 3920
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7956869457760736,
      "learning_rate": 4.4427036028853654e-05,
      "loss": 0.1318,
      "step": 3921
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0591668682936461,
      "learning_rate": 4.442391349871594e-05,
      "loss": 0.2874,
      "step": 3922
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8562252582839917,
      "learning_rate": 4.4420790203844834e-05,
      "loss": 0.1476,
      "step": 3923
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6489013925625496,
      "learning_rate": 4.441766614436329e-05,
      "loss": 0.0753,
      "step": 3924
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8600839100003176,
      "learning_rate": 4.441454132039432e-05,
      "loss": 0.1385,
      "step": 3925
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.8820364904183844,
      "learning_rate": 4.4411415732060944e-05,
      "loss": 0.5046,
      "step": 3926
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9877737416933182,
      "learning_rate": 4.440828937948621e-05,
      "loss": 0.2201,
      "step": 3927
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6875339879526414,
      "learning_rate": 4.440516226279323e-05,
      "loss": 0.3104,
      "step": 3928
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7531988128882009,
      "learning_rate": 4.440203438210509e-05,
      "loss": 0.0905,
      "step": 3929
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7122289353117652,
      "learning_rate": 4.439890573754496e-05,
      "loss": 0.0983,
      "step": 3930
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8520492219727082,
      "learning_rate": 4.439577632923601e-05,
      "loss": 0.1374,
      "step": 3931
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.94486339550688,
      "learning_rate": 4.4392646157301437e-05,
      "loss": 0.2078,
      "step": 3932
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8434652757080605,
      "learning_rate": 4.438951522186449e-05,
      "loss": 0.1191,
      "step": 3933
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5523473908036896,
      "learning_rate": 4.4386383523048435e-05,
      "loss": 0.3106,
      "step": 3934
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1960957630806814,
      "learning_rate": 4.4383251060976565e-05,
      "loss": 0.2194,
      "step": 3935
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8120225076303071,
      "learning_rate": 4.4380117835772205e-05,
      "loss": 0.3508,
      "step": 3936
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5855509629062903,
      "learning_rate": 4.437698384755873e-05,
      "loss": 0.0841,
      "step": 3937
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8866673873748654,
      "learning_rate": 4.43738490964595e-05,
      "loss": 0.1203,
      "step": 3938
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8988145890677339,
      "learning_rate": 4.4370713582597956e-05,
      "loss": 0.2705,
      "step": 3939
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.519154691000744,
      "learning_rate": 4.436757730609753e-05,
      "loss": 0.1106,
      "step": 3940
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7265527985459951,
      "learning_rate": 4.436444026708171e-05,
      "loss": 0.1121,
      "step": 3941
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8432942221468483,
      "learning_rate": 4.436130246567399e-05,
      "loss": 0.1537,
      "step": 3942
    },
    {
      "epoch": 0.24,
      "grad_norm": 23.090365789320465,
      "learning_rate": 4.435816390199792e-05,
      "loss": 0.231,
      "step": 3943
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.157970512582314,
      "learning_rate": 4.435502457617707e-05,
      "loss": 0.3177,
      "step": 3944
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5683248156258549,
      "learning_rate": 4.435188448833503e-05,
      "loss": 0.1052,
      "step": 3945
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7478645229415527,
      "learning_rate": 4.434874363859542e-05,
      "loss": 0.1146,
      "step": 3946
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.043660073123926,
      "learning_rate": 4.434560202708191e-05,
      "loss": 0.6334,
      "step": 3947
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.093509110300905,
      "learning_rate": 4.434245965391819e-05,
      "loss": 0.2836,
      "step": 3948
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0833868839763185,
      "learning_rate": 4.433931651922796e-05,
      "loss": 0.1994,
      "step": 3949
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.0648064264814723,
      "learning_rate": 4.433617262313499e-05,
      "loss": 0.4534,
      "step": 3950
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.201961580903252,
      "learning_rate": 4.433302796576304e-05,
      "loss": 0.3982,
      "step": 3951
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7164254645540284,
      "learning_rate": 4.432988254723592e-05,
      "loss": 0.121,
      "step": 3952
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7288867429300174,
      "learning_rate": 4.432673636767748e-05,
      "loss": 0.5352,
      "step": 3953
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2821438947663886,
      "learning_rate": 4.432358942721157e-05,
      "loss": 0.2312,
      "step": 3954
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.232298937429374,
      "learning_rate": 4.4320441725962105e-05,
      "loss": 0.3819,
      "step": 3955
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8341033500428374,
      "learning_rate": 4.431729326405299e-05,
      "loss": 0.112,
      "step": 3956
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4200543095709663,
      "learning_rate": 4.43141440416082e-05,
      "loss": 0.2583,
      "step": 3957
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2917208640833704,
      "learning_rate": 4.431099405875172e-05,
      "loss": 0.2678,
      "step": 3958
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3861104048865758,
      "learning_rate": 4.430784331560757e-05,
      "loss": 0.2425,
      "step": 3959
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0915183158603303,
      "learning_rate": 4.4304691812299783e-05,
      "loss": 0.2706,
      "step": 3960
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0886290040729112,
      "learning_rate": 4.430153954895244e-05,
      "loss": 0.1759,
      "step": 3961
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7379811347944794,
      "learning_rate": 4.4298386525689665e-05,
      "loss": 0.4336,
      "step": 3962
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6833766375236192,
      "learning_rate": 4.429523274263557e-05,
      "loss": 0.3348,
      "step": 3963
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9959645689659056,
      "learning_rate": 4.429207819991433e-05,
      "loss": 0.1559,
      "step": 3964
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6965541939170417,
      "learning_rate": 4.428892289765016e-05,
      "loss": 0.1023,
      "step": 3965
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7573649723242738,
      "learning_rate": 4.4285766835967264e-05,
      "loss": 0.1516,
      "step": 3966
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.248319144672722,
      "learning_rate": 4.4282610014989904e-05,
      "loss": 0.1216,
      "step": 3967
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9596276737588696,
      "learning_rate": 4.427945243484237e-05,
      "loss": 0.1762,
      "step": 3968
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1046476856881176,
      "learning_rate": 4.427629409564898e-05,
      "loss": 0.1733,
      "step": 3969
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.238431632989865,
      "learning_rate": 4.4273134997534074e-05,
      "loss": 0.2455,
      "step": 3970
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6198263238568292,
      "learning_rate": 4.426997514062203e-05,
      "loss": 0.0704,
      "step": 3971
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1625601465541249,
      "learning_rate": 4.426681452503726e-05,
      "loss": 0.187,
      "step": 3972
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.098637134073433,
      "learning_rate": 4.4263653150904184e-05,
      "loss": 0.2041,
      "step": 3973
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.215916217879783,
      "learning_rate": 4.4260491018347285e-05,
      "loss": 0.4008,
      "step": 3974
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5395746074801158,
      "learning_rate": 4.4257328127491046e-05,
      "loss": 0.3341,
      "step": 3975
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.9151526072607887,
      "learning_rate": 4.425416447846e-05,
      "loss": 0.4414,
      "step": 3976
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6930894049898042,
      "learning_rate": 4.42510000713787e-05,
      "loss": 0.2705,
      "step": 3977
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9624398641477447,
      "learning_rate": 4.424783490637173e-05,
      "loss": 0.2579,
      "step": 3978
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8864296616167475,
      "learning_rate": 4.424466898356371e-05,
      "loss": 0.0943,
      "step": 3979
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3772234756796577,
      "learning_rate": 4.424150230307928e-05,
      "loss": 0.2569,
      "step": 3980
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9518078754409139,
      "learning_rate": 4.423833486504311e-05,
      "loss": 0.1664,
      "step": 3981
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.207755643648768,
      "learning_rate": 4.4235166669579904e-05,
      "loss": 0.1746,
      "step": 3982
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1272016875223447,
      "learning_rate": 4.42319977168144e-05,
      "loss": 0.2666,
      "step": 3983
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9462467463815154,
      "learning_rate": 4.422882800687137e-05,
      "loss": 0.246,
      "step": 3984
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1482704068858518,
      "learning_rate": 4.42256575398756e-05,
      "loss": 0.1293,
      "step": 3985
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2396905197662709,
      "learning_rate": 4.4222486315951916e-05,
      "loss": 0.1799,
      "step": 3986
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4942116470885662,
      "learning_rate": 4.4219314335225165e-05,
      "loss": 0.2046,
      "step": 3987
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9466996797902406,
      "learning_rate": 4.421614159782024e-05,
      "loss": 0.1897,
      "step": 3988
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9753275593136006,
      "learning_rate": 4.421296810386204e-05,
      "loss": 0.1506,
      "step": 3989
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.255804670023226,
      "learning_rate": 4.4209793853475526e-05,
      "loss": 0.1749,
      "step": 3990
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0242423584173679,
      "learning_rate": 4.4206618846785655e-05,
      "loss": 0.1531,
      "step": 3991
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.485928411126198,
      "learning_rate": 4.420344308391743e-05,
      "loss": 0.4707,
      "step": 3992
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4813932770957317,
      "learning_rate": 4.42002665649959e-05,
      "loss": 0.3658,
      "step": 3993
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1540645473002675,
      "learning_rate": 4.419708929014611e-05,
      "loss": 0.3295,
      "step": 3994
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1407677206732185,
      "learning_rate": 4.419391125949315e-05,
      "loss": 0.2338,
      "step": 3995
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1004700772387004,
      "learning_rate": 4.419073247316216e-05,
      "loss": 0.4945,
      "step": 3996
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2250453195591717,
      "learning_rate": 4.418755293127827e-05,
      "loss": 0.1326,
      "step": 3997
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.148291715716454,
      "learning_rate": 4.4184372633966675e-05,
      "loss": 0.1525,
      "step": 3998
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3297195586107649,
      "learning_rate": 4.418119158135258e-05,
      "loss": 0.2321,
      "step": 3999
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1265076669555314,
      "learning_rate": 4.4178009773561215e-05,
      "loss": 0.2158,
      "step": 4000
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7742135863621635,
      "learning_rate": 4.4174827210717873e-05,
      "loss": 0.5716,
      "step": 4001
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1210542222698976,
      "learning_rate": 4.417164389294784e-05,
      "loss": 0.3178,
      "step": 4002
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.033999156877901,
      "learning_rate": 4.416845982037644e-05,
      "loss": 0.2207,
      "step": 4003
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.7995823796068118,
      "learning_rate": 4.4165274993129045e-05,
      "loss": 0.3349,
      "step": 4004
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9996462630707867,
      "learning_rate": 4.416208941133104e-05,
      "loss": 0.1414,
      "step": 4005
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7658706226664991,
      "learning_rate": 4.4158903075107844e-05,
      "loss": 0.0817,
      "step": 4006
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1387792882883272,
      "learning_rate": 4.41557159845849e-05,
      "loss": 0.1638,
      "step": 4007
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.093431924223016,
      "learning_rate": 4.4152528139887685e-05,
      "loss": 0.2191,
      "step": 4008
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3895347203961024,
      "learning_rate": 4.414933954114172e-05,
      "loss": 0.3917,
      "step": 4009
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1653320618999325,
      "learning_rate": 4.4146150188472526e-05,
      "loss": 0.2313,
      "step": 4010
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9228455174709908,
      "learning_rate": 4.414296008200568e-05,
      "loss": 0.1499,
      "step": 4011
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9544397729910772,
      "learning_rate": 4.413976922186677e-05,
      "loss": 0.1311,
      "step": 4012
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7186581630333297,
      "learning_rate": 4.413657760818144e-05,
      "loss": 0.0697,
      "step": 4013
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0571242423195273,
      "learning_rate": 4.413338524107533e-05,
      "loss": 0.1635,
      "step": 4014
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8673613101574364,
      "learning_rate": 4.4130192120674126e-05,
      "loss": 0.3152,
      "step": 4015
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6629295999516002,
      "learning_rate": 4.412699824710355e-05,
      "loss": 0.1097,
      "step": 4016
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7204946200534578,
      "learning_rate": 4.4123803620489345e-05,
      "loss": 0.0939,
      "step": 4017
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8329905254551968,
      "learning_rate": 4.4120608240957286e-05,
      "loss": 0.1578,
      "step": 4018
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0678693797430123,
      "learning_rate": 4.411741210863318e-05,
      "loss": 0.2194,
      "step": 4019
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4072442365786462,
      "learning_rate": 4.411421522364286e-05,
      "loss": 0.3039,
      "step": 4020
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9991713300064975,
      "learning_rate": 4.4111017586112184e-05,
      "loss": 0.1527,
      "step": 4021
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4951613664608963,
      "learning_rate": 4.4107819196167054e-05,
      "loss": 0.3686,
      "step": 4022
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.175938966855507,
      "learning_rate": 4.4104620053933386e-05,
      "loss": 1.0278,
      "step": 4023
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8887386304021371,
      "learning_rate": 4.410142015953713e-05,
      "loss": 0.1091,
      "step": 4024
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.475235413526628,
      "learning_rate": 4.409821951310428e-05,
      "loss": 0.4693,
      "step": 4025
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5359265308586907,
      "learning_rate": 4.4095018114760835e-05,
      "loss": 0.3318,
      "step": 4026
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2813475357818285,
      "learning_rate": 4.4091815964632845e-05,
      "loss": 0.2322,
      "step": 4027
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6089026387119008,
      "learning_rate": 4.4088613062846374e-05,
      "loss": 0.3371,
      "step": 4028
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8280880557758538,
      "learning_rate": 4.4085409409527536e-05,
      "loss": 0.1712,
      "step": 4029
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1794532121296994,
      "learning_rate": 4.408220500480244e-05,
      "loss": 0.1753,
      "step": 4030
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0438353770217972,
      "learning_rate": 4.407899984879726e-05,
      "loss": 0.1353,
      "step": 4031
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6972969539092867,
      "learning_rate": 4.407579394163819e-05,
      "loss": 0.1457,
      "step": 4032
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3583017857141915,
      "learning_rate": 4.407258728345143e-05,
      "loss": 0.2468,
      "step": 4033
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9208835911288884,
      "learning_rate": 4.406937987436325e-05,
      "loss": 0.1976,
      "step": 4034
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7343199984067423,
      "learning_rate": 4.406617171449991e-05,
      "loss": 0.1192,
      "step": 4035
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3442883291485122,
      "learning_rate": 4.4062962803987726e-05,
      "loss": 0.3118,
      "step": 4036
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1132729883422068,
      "learning_rate": 4.4059753142953045e-05,
      "loss": 0.2369,
      "step": 4037
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6483955761416236,
      "learning_rate": 4.405654273152221e-05,
      "loss": 0.0881,
      "step": 4038
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8324703131760641,
      "learning_rate": 4.405333156982164e-05,
      "loss": 0.1648,
      "step": 4039
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8043692992450053,
      "learning_rate": 4.405011965797775e-05,
      "loss": 0.1191,
      "step": 4040
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2448790790372248,
      "learning_rate": 4.404690699611699e-05,
      "loss": 0.1198,
      "step": 4041
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1115214754587086,
      "learning_rate": 4.4043693584365856e-05,
      "loss": 0.2423,
      "step": 4042
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0100027418688338,
      "learning_rate": 4.404047942285086e-05,
      "loss": 0.1583,
      "step": 4043
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8139262326014164,
      "learning_rate": 4.403726451169853e-05,
      "loss": 0.102,
      "step": 4044
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7655783109681576,
      "learning_rate": 4.403404885103546e-05,
      "loss": 0.1381,
      "step": 4045
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6800232524144771,
      "learning_rate": 4.403083244098825e-05,
      "loss": 0.1068,
      "step": 4046
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.698013047393882,
      "learning_rate": 4.402761528168352e-05,
      "loss": 0.11,
      "step": 4047
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.413185154074475,
      "learning_rate": 4.402439737324795e-05,
      "loss": 0.3246,
      "step": 4048
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9254444018709128,
      "learning_rate": 4.4021178715808216e-05,
      "loss": 0.2055,
      "step": 4049
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0160193944358344,
      "learning_rate": 4.401795930949104e-05,
      "loss": 0.2287,
      "step": 4050
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7952114154314606,
      "learning_rate": 4.401473915442318e-05,
      "loss": 0.0756,
      "step": 4051
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7345834265464437,
      "learning_rate": 4.4011518250731415e-05,
      "loss": 0.1529,
      "step": 4052
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1918634493117075,
      "learning_rate": 4.400829659854256e-05,
      "loss": 0.1774,
      "step": 4053
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0739602191287538,
      "learning_rate": 4.400507419798342e-05,
      "loss": 0.259,
      "step": 4054
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3112217072117467,
      "learning_rate": 4.400185104918091e-05,
      "loss": 0.1577,
      "step": 4055
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.790036074274514,
      "learning_rate": 4.3998627152261894e-05,
      "loss": 0.223,
      "step": 4056
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1975736520628741,
      "learning_rate": 4.399540250735331e-05,
      "loss": 0.1804,
      "step": 4057
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6835873243338264,
      "learning_rate": 4.3992177114582124e-05,
      "loss": 0.3117,
      "step": 4058
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1454970363665526,
      "learning_rate": 4.398895097407532e-05,
      "loss": 0.2709,
      "step": 4059
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1922913777978907,
      "learning_rate": 4.398572408595989e-05,
      "loss": 0.2035,
      "step": 4060
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3616270215489865,
      "learning_rate": 4.398249645036291e-05,
      "loss": 0.2533,
      "step": 4061
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7052581693555646,
      "learning_rate": 4.397926806741143e-05,
      "loss": 0.108,
      "step": 4062
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1222447499440222,
      "learning_rate": 4.397603893723257e-05,
      "loss": 0.2536,
      "step": 4063
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8508836832772112,
      "learning_rate": 4.397280905995346e-05,
      "loss": 0.1333,
      "step": 4064
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4907636834889249,
      "learning_rate": 4.396957843570125e-05,
      "loss": 0.4445,
      "step": 4065
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7456759477268872,
      "learning_rate": 4.396634706460315e-05,
      "loss": 0.1968,
      "step": 4066
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7667416291026273,
      "learning_rate": 4.396311494678637e-05,
      "loss": 0.0955,
      "step": 4067
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9195650921242788,
      "learning_rate": 4.3959882082378165e-05,
      "loss": 0.1451,
      "step": 4068
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0380291989571753,
      "learning_rate": 4.395664847150582e-05,
      "loss": 0.2162,
      "step": 4069
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8365698420745301,
      "learning_rate": 4.395341411429663e-05,
      "loss": 0.1885,
      "step": 4070
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8048187085914426,
      "learning_rate": 4.395017901087795e-05,
      "loss": 0.0994,
      "step": 4071
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6990607909476126,
      "learning_rate": 4.394694316137714e-05,
      "loss": 0.1666,
      "step": 4072
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.757326694940819,
      "learning_rate": 4.394370656592159e-05,
      "loss": 0.1649,
      "step": 4073
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.260114569144189,
      "learning_rate": 4.394046922463875e-05,
      "loss": 0.2151,
      "step": 4074
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5886517304646326,
      "learning_rate": 4.3937231137656056e-05,
      "loss": 0.3968,
      "step": 4075
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6175414626518683,
      "learning_rate": 4.3933992305101005e-05,
      "loss": 0.3613,
      "step": 4076
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7509301074753579,
      "learning_rate": 4.3930752727101097e-05,
      "loss": 0.0909,
      "step": 4077
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.3737402767451323,
      "learning_rate": 4.3927512403783894e-05,
      "loss": 0.3064,
      "step": 4078
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2112699825117026,
      "learning_rate": 4.3924271335276955e-05,
      "loss": 0.3837,
      "step": 4079
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5730764176493486,
      "learning_rate": 4.39210295217079e-05,
      "loss": 0.393,
      "step": 4080
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.107675971052673,
      "learning_rate": 4.391778696320435e-05,
      "loss": 0.3749,
      "step": 4081
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9962373430335059,
      "learning_rate": 4.391454365989397e-05,
      "loss": 0.1238,
      "step": 4082
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.225007232639733,
      "learning_rate": 4.391129961190445e-05,
      "loss": 0.2975,
      "step": 4083
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2155926651073403,
      "learning_rate": 4.3908054819363506e-05,
      "loss": 0.3693,
      "step": 4084
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2642349353074027,
      "learning_rate": 4.39048092823989e-05,
      "loss": 0.2825,
      "step": 4085
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.12665413503069,
      "learning_rate": 4.3901563001138404e-05,
      "loss": 0.1093,
      "step": 4086
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6605558750662804,
      "learning_rate": 4.389831597570982e-05,
      "loss": 0.0793,
      "step": 4087
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.891310219462331,
      "learning_rate": 4.3895068206241e-05,
      "loss": 0.1596,
      "step": 4088
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9467463990566215,
      "learning_rate": 4.38918196928598e-05,
      "loss": 0.1811,
      "step": 4089
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6793816739537699,
      "learning_rate": 4.388857043569412e-05,
      "loss": 0.1361,
      "step": 4090
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8866388048020718,
      "learning_rate": 4.388532043487188e-05,
      "loss": 0.1574,
      "step": 4091
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9486549336988315,
      "learning_rate": 4.3882069690521045e-05,
      "loss": 0.2082,
      "step": 4092
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7950483807541312,
      "learning_rate": 4.387881820276959e-05,
      "loss": 0.1533,
      "step": 4093
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0202492493593667,
      "learning_rate": 4.387556597174553e-05,
      "loss": 0.1496,
      "step": 4094
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.910168911364218,
      "learning_rate": 4.387231299757692e-05,
      "loss": 0.1385,
      "step": 4095
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9232984392014704,
      "learning_rate": 4.3869059280391817e-05,
      "loss": 0.147,
      "step": 4096
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8514889696933305,
      "learning_rate": 4.386580482031832e-05,
      "loss": 0.2615,
      "step": 4097
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3337525993569253,
      "learning_rate": 4.386254961748458e-05,
      "loss": 0.2294,
      "step": 4098
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6630920236986032,
      "learning_rate": 4.385929367201874e-05,
      "loss": 0.0656,
      "step": 4099
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9094299266208075,
      "learning_rate": 4.385603698404899e-05,
      "loss": 0.1474,
      "step": 4100
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6913207734925122,
      "learning_rate": 4.385277955370355e-05,
      "loss": 0.0994,
      "step": 4101
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2098385040678956,
      "learning_rate": 4.384952138111066e-05,
      "loss": 0.1679,
      "step": 4102
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9047487171512321,
      "learning_rate": 4.384626246639861e-05,
      "loss": 0.1157,
      "step": 4103
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7668789881331194,
      "learning_rate": 4.384300280969571e-05,
      "loss": 0.1051,
      "step": 4104
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9700580373437746,
      "learning_rate": 4.383974241113027e-05,
      "loss": 0.5252,
      "step": 4105
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.781457267687125,
      "learning_rate": 4.3836481270830675e-05,
      "loss": 0.1461,
      "step": 4106
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.9875456973483097,
      "learning_rate": 4.383321938892532e-05,
      "loss": 0.5159,
      "step": 4107
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9209933781017965,
      "learning_rate": 4.382995676554261e-05,
      "loss": 0.146,
      "step": 4108
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2223880981109174,
      "learning_rate": 4.382669340081101e-05,
      "loss": 0.134,
      "step": 4109
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0128929792881411,
      "learning_rate": 4.3823429294858997e-05,
      "loss": 0.1486,
      "step": 4110
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2075405098304608,
      "learning_rate": 4.382016444781508e-05,
      "loss": 0.2845,
      "step": 4111
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.20290324015259,
      "learning_rate": 4.381689885980781e-05,
      "loss": 0.4412,
      "step": 4112
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9871602603121769,
      "learning_rate": 4.3813632530965745e-05,
      "loss": 0.1372,
      "step": 4113
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.091594019459463,
      "learning_rate": 4.381036546141748e-05,
      "loss": 0.1934,
      "step": 4114
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1261818612578365,
      "learning_rate": 4.380709765129164e-05,
      "loss": 0.1525,
      "step": 4115
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9726370438964796,
      "learning_rate": 4.3803829100716895e-05,
      "loss": 0.1132,
      "step": 4116
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9620258418929337,
      "learning_rate": 4.3800559809821926e-05,
      "loss": 0.2061,
      "step": 4117
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9758875838073844,
      "learning_rate": 4.379728977873543e-05,
      "loss": 0.1281,
      "step": 4118
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7693895783857877,
      "learning_rate": 4.3794019007586176e-05,
      "loss": 0.1107,
      "step": 4119
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1403374344256632,
      "learning_rate": 4.379074749650291e-05,
      "loss": 0.3194,
      "step": 4120
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8851134959326051,
      "learning_rate": 4.378747524561447e-05,
      "loss": 0.1197,
      "step": 4121
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0045382441982025,
      "learning_rate": 4.3784202255049656e-05,
      "loss": 0.2791,
      "step": 4122
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.5952878256436245,
      "learning_rate": 4.378092852493733e-05,
      "loss": 0.3817,
      "step": 4123
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8513546631808098,
      "learning_rate": 4.377765405540639e-05,
      "loss": 0.3149,
      "step": 4124
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7253860432562449,
      "learning_rate": 4.377437884658575e-05,
      "loss": 0.1089,
      "step": 4125
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6876364907129559,
      "learning_rate": 4.377110289860437e-05,
      "loss": 0.0883,
      "step": 4126
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8987908323391562,
      "learning_rate": 4.376782621159121e-05,
      "loss": 0.1493,
      "step": 4127
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8836972022040129,
      "learning_rate": 4.3764548785675285e-05,
      "loss": 0.1212,
      "step": 4128
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2180747989936322,
      "learning_rate": 4.3761270620985626e-05,
      "loss": 0.2379,
      "step": 4129
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0530331680113407,
      "learning_rate": 4.375799171765129e-05,
      "loss": 0.1939,
      "step": 4130
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8571370002116905,
      "learning_rate": 4.375471207580139e-05,
      "loss": 0.1368,
      "step": 4131
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2126515325899583,
      "learning_rate": 4.375143169556501e-05,
      "loss": 0.2012,
      "step": 4132
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8967539639787157,
      "learning_rate": 4.374815057707135e-05,
      "loss": 0.1827,
      "step": 4133
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.161785534599862,
      "learning_rate": 4.374486872044955e-05,
      "loss": 0.1854,
      "step": 4134
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8677334520143821,
      "learning_rate": 4.374158612582885e-05,
      "loss": 0.1706,
      "step": 4135
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9501537416872788,
      "learning_rate": 4.373830279333846e-05,
      "loss": 0.1791,
      "step": 4136
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.691592416409324,
      "learning_rate": 4.3735018723107666e-05,
      "loss": 0.0977,
      "step": 4137
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9825408880053856,
      "learning_rate": 4.373173391526576e-05,
      "loss": 0.2734,
      "step": 4138
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8973165913305439,
      "learning_rate": 4.372844836994206e-05,
      "loss": 0.2364,
      "step": 4139
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0453143602427095,
      "learning_rate": 4.3725162087265924e-05,
      "loss": 0.1905,
      "step": 4140
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7739835496698266,
      "learning_rate": 4.3721875067366737e-05,
      "loss": 0.1152,
      "step": 4141
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0305416970549974,
      "learning_rate": 4.3718587310373915e-05,
      "loss": 0.296,
      "step": 4142
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1635378896099295,
      "learning_rate": 4.371529881641689e-05,
      "loss": 0.1859,
      "step": 4143
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.746116648654604,
      "learning_rate": 4.3712009585625145e-05,
      "loss": 0.1188,
      "step": 4144
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.909880596898115,
      "learning_rate": 4.370871961812817e-05,
      "loss": 0.1399,
      "step": 4145
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.909792323224018,
      "learning_rate": 4.370542891405549e-05,
      "loss": 0.1412,
      "step": 4146
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.135472251035223,
      "learning_rate": 4.3702137473536674e-05,
      "loss": 0.2506,
      "step": 4147
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7163486874129064,
      "learning_rate": 4.3698845296701295e-05,
      "loss": 0.3364,
      "step": 4148
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7351813597012693,
      "learning_rate": 4.369555238367899e-05,
      "loss": 0.694,
      "step": 4149
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.535325587085577,
      "learning_rate": 4.369225873459937e-05,
      "loss": 0.3127,
      "step": 4150
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.942063270169803,
      "learning_rate": 4.3688964349592144e-05,
      "loss": 0.2572,
      "step": 4151
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6473147647266878,
      "learning_rate": 4.368566922878698e-05,
      "loss": 0.0879,
      "step": 4152
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.087486477769975,
      "learning_rate": 4.3682373372313644e-05,
      "loss": 0.1797,
      "step": 4153
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2687299794565954,
      "learning_rate": 4.3679076780301877e-05,
      "loss": 0.2029,
      "step": 4154
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.118525431582334,
      "learning_rate": 4.367577945288147e-05,
      "loss": 0.2824,
      "step": 4155
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9766499733221123,
      "learning_rate": 4.367248139018223e-05,
      "loss": 0.1973,
      "step": 4156
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1933332520038988,
      "learning_rate": 4.3669182592334026e-05,
      "loss": 0.2306,
      "step": 4157
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6204077278042217,
      "learning_rate": 4.366588305946672e-05,
      "loss": 0.1043,
      "step": 4158
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.239384550097092,
      "learning_rate": 4.366258279171022e-05,
      "loss": 0.4978,
      "step": 4159
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0147448314282097,
      "learning_rate": 4.365928178919446e-05,
      "loss": 0.16,
      "step": 4160
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1382846213741824,
      "learning_rate": 4.365598005204941e-05,
      "loss": 0.2826,
      "step": 4161
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2005148251708748,
      "learning_rate": 4.365267758040505e-05,
      "loss": 0.2671,
      "step": 4162
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8956968889170409,
      "learning_rate": 4.364937437439141e-05,
      "loss": 0.1705,
      "step": 4163
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0858258140694141,
      "learning_rate": 4.364607043413853e-05,
      "loss": 0.2605,
      "step": 4164
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0052255524386016,
      "learning_rate": 4.364276575977649e-05,
      "loss": 0.2632,
      "step": 4165
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7728568298237761,
      "learning_rate": 4.363946035143541e-05,
      "loss": 0.1107,
      "step": 4166
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9031731961958469,
      "learning_rate": 4.363615420924542e-05,
      "loss": 0.1186,
      "step": 4167
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7607845016422519,
      "learning_rate": 4.363284733333667e-05,
      "loss": 0.1448,
      "step": 4168
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2220019632049048,
      "learning_rate": 4.362953972383938e-05,
      "loss": 0.2633,
      "step": 4169
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.476842359772352,
      "learning_rate": 4.362623138088375e-05,
      "loss": 0.1577,
      "step": 4170
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5492380631159333,
      "learning_rate": 4.3622922304600047e-05,
      "loss": 0.5297,
      "step": 4171
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9275183550136505,
      "learning_rate": 4.3619612495118545e-05,
      "loss": 0.1601,
      "step": 4172
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.811504095527347,
      "learning_rate": 4.3616301952569546e-05,
      "loss": 0.1747,
      "step": 4173
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8455508516332966,
      "learning_rate": 4.3612990677083406e-05,
      "loss": 0.3766,
      "step": 4174
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.067674848847126,
      "learning_rate": 4.3609678668790486e-05,
      "loss": 0.1845,
      "step": 4175
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.657352578918363,
      "learning_rate": 4.360636592782118e-05,
      "loss": 0.521,
      "step": 4176
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0650016460336496,
      "learning_rate": 4.360305245430591e-05,
      "loss": 0.36,
      "step": 4177
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8442897010054579,
      "learning_rate": 4.359973824837513e-05,
      "loss": 0.1286,
      "step": 4178
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2645823477254214,
      "learning_rate": 4.3596423310159326e-05,
      "loss": 0.3096,
      "step": 4179
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5961686348419774,
      "learning_rate": 4.359310763978901e-05,
      "loss": 0.1563,
      "step": 4180
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7107993888186069,
      "learning_rate": 4.358979123739473e-05,
      "loss": 0.1217,
      "step": 4181
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5063430189093117,
      "learning_rate": 4.358647410310703e-05,
      "loss": 0.2328,
      "step": 4182
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1400153887164377,
      "learning_rate": 4.3583156237056535e-05,
      "loss": 0.1561,
      "step": 4183
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8442189896144117,
      "learning_rate": 4.3579837639373856e-05,
      "loss": 0.1403,
      "step": 4184
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6004886460069979,
      "learning_rate": 4.357651831018965e-05,
      "loss": 0.1277,
      "step": 4185
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2076224374654354,
      "learning_rate": 4.3573198249634615e-05,
      "loss": 0.2738,
      "step": 4186
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8635775617285592,
      "learning_rate": 4.356987745783945e-05,
      "loss": 0.1136,
      "step": 4187
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9548192752866095,
      "learning_rate": 4.35665559349349e-05,
      "loss": 0.1496,
      "step": 4188
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.121573770229429,
      "learning_rate": 4.356323368105173e-05,
      "loss": 0.2605,
      "step": 4189
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4355917787462504,
      "learning_rate": 4.355991069632076e-05,
      "loss": 0.3412,
      "step": 4190
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7558907991619772,
      "learning_rate": 4.355658698087279e-05,
      "loss": 0.1172,
      "step": 4191
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9476717643600572,
      "learning_rate": 4.35532625348387e-05,
      "loss": 0.1764,
      "step": 4192
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1877098269734618,
      "learning_rate": 4.3549937358349366e-05,
      "loss": 0.2592,
      "step": 4193
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6849598400278135,
      "learning_rate": 4.3546611451535704e-05,
      "loss": 0.1224,
      "step": 4194
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7803073165784637,
      "learning_rate": 4.354328481452866e-05,
      "loss": 0.0956,
      "step": 4195
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8398801841131402,
      "learning_rate": 4.35399574474592e-05,
      "loss": 0.1149,
      "step": 4196
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0507092232691975,
      "learning_rate": 4.353662935045832e-05,
      "loss": 0.1526,
      "step": 4197
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.116892431371791,
      "learning_rate": 4.353330052365707e-05,
      "loss": 0.1651,
      "step": 4198
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8795071164076559,
      "learning_rate": 4.35299709671865e-05,
      "loss": 0.1239,
      "step": 4199
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9229798934897859,
      "learning_rate": 4.3526640681177676e-05,
      "loss": 0.1114,
      "step": 4200
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6793237628632568,
      "learning_rate": 4.3523309665761746e-05,
      "loss": 0.1161,
      "step": 4201
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6115345800743912,
      "learning_rate": 4.3519977921069834e-05,
      "loss": 0.3893,
      "step": 4202
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5283892095967626,
      "learning_rate": 4.351664544723312e-05,
      "loss": 0.3906,
      "step": 4203
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.727916768573653,
      "learning_rate": 4.35133122443828e-05,
      "loss": 0.4781,
      "step": 4204
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8238698794178658,
      "learning_rate": 4.350997831265011e-05,
      "loss": 0.1105,
      "step": 4205
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7629951608972497,
      "learning_rate": 4.350664365216631e-05,
      "loss": 0.1191,
      "step": 4206
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8893042028637631,
      "learning_rate": 4.350330826306268e-05,
      "loss": 0.264,
      "step": 4207
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.060275069767596,
      "learning_rate": 4.3499972145470547e-05,
      "loss": 0.2799,
      "step": 4208
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7366924535124968,
      "learning_rate": 4.3496635299521254e-05,
      "loss": 0.6309,
      "step": 4209
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.275777955102707,
      "learning_rate": 4.349329772534617e-05,
      "loss": 0.1996,
      "step": 4210
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8812313399430978,
      "learning_rate": 4.348995942307671e-05,
      "loss": 0.1992,
      "step": 4211
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.264569881208058,
      "learning_rate": 4.3486620392844285e-05,
      "loss": 0.223,
      "step": 4212
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6530795623007974,
      "learning_rate": 4.348328063478037e-05,
      "loss": 0.0991,
      "step": 4213
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.051937284137326,
      "learning_rate": 4.347994014901645e-05,
      "loss": 0.2877,
      "step": 4214
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3738371037918466,
      "learning_rate": 4.347659893568404e-05,
      "loss": 0.2232,
      "step": 4215
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.11297567983785,
      "learning_rate": 4.34732569949147e-05,
      "loss": 0.1287,
      "step": 4216
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.031619631242828,
      "learning_rate": 4.3469914326839975e-05,
      "loss": 0.1788,
      "step": 4217
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5186077917363732,
      "learning_rate": 4.34665709315915e-05,
      "loss": 0.3636,
      "step": 4218
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.60450304986524,
      "learning_rate": 4.346322680930089e-05,
      "loss": 0.3759,
      "step": 4219
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2542849671875178,
      "learning_rate": 4.3459881960099806e-05,
      "loss": 0.2515,
      "step": 4220
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8062165218125195,
      "learning_rate": 4.345653638411994e-05,
      "loss": 0.1415,
      "step": 4221
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7552449403033352,
      "learning_rate": 4.345319008149301e-05,
      "loss": 0.2055,
      "step": 4222
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9084777258150448,
      "learning_rate": 4.344984305235076e-05,
      "loss": 0.207,
      "step": 4223
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1020277795049014,
      "learning_rate": 4.344649529682496e-05,
      "loss": 0.27,
      "step": 4224
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0051474125165343,
      "learning_rate": 4.344314681504743e-05,
      "loss": 0.2263,
      "step": 4225
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9249326982783904,
      "learning_rate": 4.343979760714999e-05,
      "loss": 0.1774,
      "step": 4226
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4470418330530075,
      "learning_rate": 4.34364476732645e-05,
      "loss": 0.2547,
      "step": 4227
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8113890065213566,
      "learning_rate": 4.343309701352286e-05,
      "loss": 0.171,
      "step": 4228
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6770134443412552,
      "learning_rate": 4.342974562805697e-05,
      "loss": 0.0742,
      "step": 4229
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8173133570325685,
      "learning_rate": 4.34263935169988e-05,
      "loss": 0.1392,
      "step": 4230
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6542457223023027,
      "learning_rate": 4.34230406804803e-05,
      "loss": 0.1072,
      "step": 4231
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.010216401385533,
      "learning_rate": 4.341968711863348e-05,
      "loss": 0.2091,
      "step": 4232
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9780735124380066,
      "learning_rate": 4.341633283159039e-05,
      "loss": 0.1842,
      "step": 4233
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7792839766698734,
      "learning_rate": 4.341297781948307e-05,
      "loss": 0.108,
      "step": 4234
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4634990239537973,
      "learning_rate": 4.3409622082443626e-05,
      "loss": 0.2385,
      "step": 4235
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.110307075637469,
      "learning_rate": 4.3406265620604156e-05,
      "loss": 0.2334,
      "step": 4236
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0169340999089214,
      "learning_rate": 4.340290843409682e-05,
      "loss": 0.1999,
      "step": 4237
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7659235915091755,
      "learning_rate": 4.3399550523053796e-05,
      "loss": 0.1121,
      "step": 4238
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9138219421977555,
      "learning_rate": 4.3396191887607274e-05,
      "loss": 0.1585,
      "step": 4239
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0997805846701283,
      "learning_rate": 4.3392832527889496e-05,
      "loss": 0.2145,
      "step": 4240
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.392961509527569,
      "learning_rate": 4.338947244403272e-05,
      "loss": 0.2382,
      "step": 4241
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.518133331593881,
      "learning_rate": 4.3386111636169236e-05,
      "loss": 0.0769,
      "step": 4242
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5505639127717717,
      "learning_rate": 4.338275010443136e-05,
      "loss": 0.0645,
      "step": 4243
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8508227239524281,
      "learning_rate": 4.3379387848951434e-05,
      "loss": 0.1399,
      "step": 4244
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7521621950532211,
      "learning_rate": 4.337602486986184e-05,
      "loss": 0.0981,
      "step": 4245
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7140248264219227,
      "learning_rate": 4.337266116729498e-05,
      "loss": 0.3031,
      "step": 4246
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.766731855726656,
      "learning_rate": 4.336929674138328e-05,
      "loss": 0.1439,
      "step": 4247
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6077265667073684,
      "learning_rate": 4.33659315922592e-05,
      "loss": 0.6159,
      "step": 4248
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8691138903376046,
      "learning_rate": 4.336256572005523e-05,
      "loss": 0.4892,
      "step": 4249
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3930715522367882,
      "learning_rate": 4.335919912490389e-05,
      "loss": 0.3443,
      "step": 4250
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.651993253565543,
      "learning_rate": 4.335583180693772e-05,
      "loss": 0.0704,
      "step": 4251
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7252223947720314,
      "learning_rate": 4.335246376628931e-05,
      "loss": 0.1367,
      "step": 4252
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.567884658197308,
      "learning_rate": 4.334909500309123e-05,
      "loss": 0.3032,
      "step": 4253
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6927339149093391,
      "learning_rate": 4.334572551747614e-05,
      "loss": 0.1174,
      "step": 4254
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6884225537958235,
      "learning_rate": 4.334235530957669e-05,
      "loss": 0.1015,
      "step": 4255
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4182368788962145,
      "learning_rate": 4.333898437952556e-05,
      "loss": 0.3392,
      "step": 4256
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9732958003602268,
      "learning_rate": 4.333561272745547e-05,
      "loss": 0.1517,
      "step": 4257
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1904404476012898,
      "learning_rate": 4.333224035349918e-05,
      "loss": 0.1731,
      "step": 4258
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8771143996162064,
      "learning_rate": 4.332886725778943e-05,
      "loss": 0.2582,
      "step": 4259
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.236893790939283,
      "learning_rate": 4.3325493440459055e-05,
      "loss": 0.2185,
      "step": 4260
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9768461957898767,
      "learning_rate": 4.332211890164086e-05,
      "loss": 0.2128,
      "step": 4261
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0970456673602413,
      "learning_rate": 4.331874364146772e-05,
      "loss": 0.2512,
      "step": 4262
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9856544854011364,
      "learning_rate": 4.3315367660072506e-05,
      "loss": 0.1846,
      "step": 4263
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5610695035847595,
      "learning_rate": 4.331199095758815e-05,
      "loss": 0.3028,
      "step": 4264
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.125094215571004,
      "learning_rate": 4.330861353414758e-05,
      "loss": 0.2334,
      "step": 4265
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8557073391211385,
      "learning_rate": 4.330523538988378e-05,
      "loss": 0.1102,
      "step": 4266
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.579492403074713,
      "learning_rate": 4.330185652492974e-05,
      "loss": 0.058,
      "step": 4267
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7522573648104844,
      "learning_rate": 4.3298476939418485e-05,
      "loss": 0.1419,
      "step": 4268
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0428666490219582,
      "learning_rate": 4.329509663348309e-05,
      "loss": 0.2471,
      "step": 4269
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5354044341683464,
      "learning_rate": 4.329171560725663e-05,
      "loss": 0.2858,
      "step": 4270
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5524730347375337,
      "learning_rate": 4.328833386087221e-05,
      "loss": 0.2601,
      "step": 4271
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6499818349628722,
      "learning_rate": 4.328495139446298e-05,
      "loss": 0.0887,
      "step": 4272
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4913569836613764,
      "learning_rate": 4.3281568208162116e-05,
      "loss": 0.0647,
      "step": 4273
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9915841935161767,
      "learning_rate": 4.32781843021028e-05,
      "loss": 0.2348,
      "step": 4274
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5283562800905514,
      "learning_rate": 4.327479967641828e-05,
      "loss": 0.0975,
      "step": 4275
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.755265837299418,
      "learning_rate": 4.327141433124179e-05,
      "loss": 0.0857,
      "step": 4276
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8797327066779451,
      "learning_rate": 4.326802826670662e-05,
      "loss": 0.1536,
      "step": 4277
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8710463735425735,
      "learning_rate": 4.326464148294609e-05,
      "loss": 0.1361,
      "step": 4278
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8365193848762235,
      "learning_rate": 4.326125398009354e-05,
      "loss": 0.1181,
      "step": 4279
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8863690249420753,
      "learning_rate": 4.325786575828233e-05,
      "loss": 0.4333,
      "step": 4280
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6774415368096204,
      "learning_rate": 4.325447681764586e-05,
      "loss": 0.155,
      "step": 4281
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7548217619634631,
      "learning_rate": 4.3251087158317554e-05,
      "loss": 0.1712,
      "step": 4282
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7789013028506063,
      "learning_rate": 4.324769678043087e-05,
      "loss": 0.1322,
      "step": 4283
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0967948718465148,
      "learning_rate": 4.324430568411928e-05,
      "loss": 0.4037,
      "step": 4284
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8963787179918631,
      "learning_rate": 4.32409138695163e-05,
      "loss": 0.1192,
      "step": 4285
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3327068713676826,
      "learning_rate": 4.323752133675547e-05,
      "loss": 0.1436,
      "step": 4286
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6981641864540175,
      "learning_rate": 4.323412808597036e-05,
      "loss": 0.097,
      "step": 4287
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7658110845095596,
      "learning_rate": 4.323073411729456e-05,
      "loss": 0.1441,
      "step": 4288
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8331735716526599,
      "learning_rate": 4.322733943086168e-05,
      "loss": 0.1588,
      "step": 4289
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0411373114426596,
      "learning_rate": 4.32239440268054e-05,
      "loss": 0.2284,
      "step": 4290
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7569262132559242,
      "learning_rate": 4.322054790525937e-05,
      "loss": 0.1473,
      "step": 4291
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4791336882899728,
      "learning_rate": 4.321715106635731e-05,
      "loss": 0.3554,
      "step": 4292
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5263483456486432,
      "learning_rate": 4.321375351023296e-05,
      "loss": 0.0652,
      "step": 4293
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.700978238234923,
      "learning_rate": 4.3210355237020086e-05,
      "loss": 0.0839,
      "step": 4294
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7480540326580117,
      "learning_rate": 4.320695624685247e-05,
      "loss": 0.1056,
      "step": 4295
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.476852470215936,
      "learning_rate": 4.3203556539863944e-05,
      "loss": 0.1581,
      "step": 4296
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9627702785201547,
      "learning_rate": 4.320015611618834e-05,
      "loss": 0.2491,
      "step": 4297
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6307603302486345,
      "learning_rate": 4.3196754975959555e-05,
      "loss": 0.1143,
      "step": 4298
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8533015568505312,
      "learning_rate": 4.3193353119311486e-05,
      "loss": 0.1326,
      "step": 4299
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2485707321686594,
      "learning_rate": 4.318995054637807e-05,
      "loss": 0.2704,
      "step": 4300
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9142714835065522,
      "learning_rate": 4.318654725729325e-05,
      "loss": 0.295,
      "step": 4301
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.20279000756446,
      "learning_rate": 4.3183143252191044e-05,
      "loss": 0.2064,
      "step": 4302
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1405900698024805,
      "learning_rate": 4.317973853120545e-05,
      "loss": 0.2172,
      "step": 4303
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7512376030723652,
      "learning_rate": 4.3176333094470525e-05,
      "loss": 0.1351,
      "step": 4304
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.731958663172371,
      "learning_rate": 4.3172926942120336e-05,
      "loss": 0.6359,
      "step": 4305
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0318699209885978,
      "learning_rate": 4.316952007428899e-05,
      "loss": 0.1638,
      "step": 4306
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9914073783976479,
      "learning_rate": 4.3166112491110624e-05,
      "loss": 0.197,
      "step": 4307
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1284592662988695,
      "learning_rate": 4.316270419271938e-05,
      "loss": 0.2458,
      "step": 4308
    },
    {
      "epoch": 0.26,
      "grad_norm": 17.507345201127443,
      "learning_rate": 4.3159295179249456e-05,
      "loss": 1.3613,
      "step": 4309
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.835313930749939,
      "learning_rate": 4.315588545083507e-05,
      "loss": 0.6649,
      "step": 4310
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2478033423218577,
      "learning_rate": 4.315247500761046e-05,
      "loss": 0.3552,
      "step": 4311
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.51378075578403,
      "learning_rate": 4.3149063849709905e-05,
      "loss": 0.3898,
      "step": 4312
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1067567694492997,
      "learning_rate": 4.3145651977267694e-05,
      "loss": 0.1799,
      "step": 4313
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0252671075049467,
      "learning_rate": 4.314223939041816e-05,
      "loss": 0.4038,
      "step": 4314
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7529346793249359,
      "learning_rate": 4.3138826089295655e-05,
      "loss": 0.4144,
      "step": 4315
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8120579471997423,
      "learning_rate": 4.3135412074034576e-05,
      "loss": 0.1264,
      "step": 4316
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6262376230021782,
      "learning_rate": 4.313199734476932e-05,
      "loss": 0.094,
      "step": 4317
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6974702225462159,
      "learning_rate": 4.3128581901634345e-05,
      "loss": 0.1135,
      "step": 4318
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2241729647904966,
      "learning_rate": 4.3125165744764096e-05,
      "loss": 0.1642,
      "step": 4319
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.362176028730288,
      "learning_rate": 4.3121748874293085e-05,
      "loss": 0.3079,
      "step": 4320
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9153308992573403,
      "learning_rate": 4.311833129035584e-05,
      "loss": 0.1349,
      "step": 4321
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7701493561815074,
      "learning_rate": 4.3114912993086895e-05,
      "loss": 0.1472,
      "step": 4322
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0369551049178867,
      "learning_rate": 4.3111493982620855e-05,
      "loss": 0.1332,
      "step": 4323
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9637987294691345,
      "learning_rate": 4.3108074259092305e-05,
      "loss": 0.234,
      "step": 4324
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.22893592969173,
      "learning_rate": 4.3104653822635904e-05,
      "loss": 0.3155,
      "step": 4325
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8604959255579996,
      "learning_rate": 4.31012326733863e-05,
      "loss": 0.3791,
      "step": 4326
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.131011524095199,
      "learning_rate": 4.3097810811478204e-05,
      "loss": 0.2525,
      "step": 4327
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.8243716219139627,
      "learning_rate": 4.309438823704631e-05,
      "loss": 0.9161,
      "step": 4328
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.79095192264086,
      "learning_rate": 4.309096495022539e-05,
      "loss": 0.1897,
      "step": 4329
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1843809944698591,
      "learning_rate": 4.3087540951150217e-05,
      "loss": 0.2302,
      "step": 4330
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7155157748000853,
      "learning_rate": 4.30841162399556e-05,
      "loss": 0.1191,
      "step": 4331
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7536505268652226,
      "learning_rate": 4.3080690816776356e-05,
      "loss": 0.077,
      "step": 4332
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.47220975823822303,
      "learning_rate": 4.307726468174736e-05,
      "loss": 0.063,
      "step": 4333
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9067900996446171,
      "learning_rate": 4.3073837835003496e-05,
      "loss": 0.1948,
      "step": 4334
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6323846332414174,
      "learning_rate": 4.307041027667969e-05,
      "loss": 0.0649,
      "step": 4335
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9099493855185865,
      "learning_rate": 4.306698200691087e-05,
      "loss": 0.2994,
      "step": 4336
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9948136954340756,
      "learning_rate": 4.306355302583203e-05,
      "loss": 0.1285,
      "step": 4337
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9281296287213378,
      "learning_rate": 4.306012333357815e-05,
      "loss": 0.2051,
      "step": 4338
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.606520413143464,
      "learning_rate": 4.305669293028428e-05,
      "loss": 0.2924,
      "step": 4339
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.161477069023336,
      "learning_rate": 4.3053261816085463e-05,
      "loss": 0.2191,
      "step": 4340
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6282888706098197,
      "learning_rate": 4.30498299911168e-05,
      "loss": 0.0809,
      "step": 4341
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4850003946296124,
      "learning_rate": 4.304639745551339e-05,
      "loss": 0.5068,
      "step": 4342
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6153310455013968,
      "learning_rate": 4.304296420941037e-05,
      "loss": 0.0891,
      "step": 4343
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0842155218211564,
      "learning_rate": 4.3039530252942926e-05,
      "loss": 0.212,
      "step": 4344
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6669288203413712,
      "learning_rate": 4.303609558624625e-05,
      "loss": 0.3102,
      "step": 4345
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9221498071942343,
      "learning_rate": 4.3032660209455564e-05,
      "loss": 0.2464,
      "step": 4346
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8678111187093247,
      "learning_rate": 4.3029224122706116e-05,
      "loss": 0.4513,
      "step": 4347
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8537753611782322,
      "learning_rate": 4.3025787326133206e-05,
      "loss": 0.1944,
      "step": 4348
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1964057625926057,
      "learning_rate": 4.3022349819872115e-05,
      "loss": 0.2763,
      "step": 4349
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7470169753109795,
      "learning_rate": 4.3018911604058205e-05,
      "loss": 0.1114,
      "step": 4350
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8751788170949413,
      "learning_rate": 4.301547267882683e-05,
      "loss": 0.1571,
      "step": 4351
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9979903022059031,
      "learning_rate": 4.301203304431338e-05,
      "loss": 0.2153,
      "step": 4352
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7185642124208904,
      "learning_rate": 4.3008592700653286e-05,
      "loss": 0.127,
      "step": 4353
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.1079615206351017,
      "learning_rate": 4.3005151647981986e-05,
      "loss": 0.427,
      "step": 4354
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7569863426176308,
      "learning_rate": 4.300170988643496e-05,
      "loss": 0.1067,
      "step": 4355
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0465646204699781,
      "learning_rate": 4.299826741614772e-05,
      "loss": 0.1102,
      "step": 4356
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0697860376065338,
      "learning_rate": 4.2994824237255795e-05,
      "loss": 0.2418,
      "step": 4357
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9717680142927346,
      "learning_rate": 4.2991380349894736e-05,
      "loss": 0.1629,
      "step": 4358
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8505735717549987,
      "learning_rate": 4.298793575420014e-05,
      "loss": 0.0921,
      "step": 4359
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9394274007444178,
      "learning_rate": 4.2984490450307625e-05,
      "loss": 0.1207,
      "step": 4360
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3025807644656457,
      "learning_rate": 4.298104443835283e-05,
      "loss": 0.317,
      "step": 4361
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.895688183447213,
      "learning_rate": 4.297759771847143e-05,
      "loss": 0.2032,
      "step": 4362
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8531766208090037,
      "learning_rate": 4.2974150290799126e-05,
      "loss": 0.0821,
      "step": 4363
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7819211551089706,
      "learning_rate": 4.297070215547164e-05,
      "loss": 0.1679,
      "step": 4364
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7756959571144891,
      "learning_rate": 4.296725331262473e-05,
      "loss": 0.1478,
      "step": 4365
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8466737620637749,
      "learning_rate": 4.296380376239418e-05,
      "loss": 0.148,
      "step": 4366
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9113221142125745,
      "learning_rate": 4.2960353504915806e-05,
      "loss": 0.1271,
      "step": 4367
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6399664954841578,
      "learning_rate": 4.295690254032544e-05,
      "loss": 0.1233,
      "step": 4368
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9656974862492914,
      "learning_rate": 4.295345086875896e-05,
      "loss": 0.2139,
      "step": 4369
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1773248861132184,
      "learning_rate": 4.294999849035224e-05,
      "loss": 0.3196,
      "step": 4370
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9421217169248911,
      "learning_rate": 4.2946545405241225e-05,
      "loss": 0.1615,
      "step": 4371
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6987560886932277,
      "learning_rate": 4.294309161356185e-05,
      "loss": 0.1105,
      "step": 4372
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0669742308985672,
      "learning_rate": 4.29396371154501e-05,
      "loss": 0.1524,
      "step": 4373
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.785365961202868,
      "learning_rate": 4.293618191104197e-05,
      "loss": 0.1512,
      "step": 4374
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8507824421090718,
      "learning_rate": 4.293272600047352e-05,
      "loss": 0.1743,
      "step": 4375
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7550275545251707,
      "learning_rate": 4.292926938388079e-05,
      "loss": 0.1134,
      "step": 4376
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6630393766982775,
      "learning_rate": 4.2925812061399874e-05,
      "loss": 0.0871,
      "step": 4377
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7835174716593646,
      "learning_rate": 4.292235403316689e-05,
      "loss": 0.1453,
      "step": 4378
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2490597998777497,
      "learning_rate": 4.291889529931798e-05,
      "loss": 0.2829,
      "step": 4379
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.979211570256319,
      "learning_rate": 4.291543585998932e-05,
      "loss": 0.3225,
      "step": 4380
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.146706246358423,
      "learning_rate": 4.291197571531712e-05,
      "loss": 0.1505,
      "step": 4381
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3218113886569431,
      "learning_rate": 4.290851486543759e-05,
      "loss": 0.2507,
      "step": 4382
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8250553415447528,
      "learning_rate": 4.2905053310486996e-05,
      "loss": 0.4576,
      "step": 4383
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3650065070783817,
      "learning_rate": 4.2901591050601624e-05,
      "loss": 0.2133,
      "step": 4384
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7753132155650436,
      "learning_rate": 4.289812808591778e-05,
      "loss": 0.1475,
      "step": 4385
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7037040378388706,
      "learning_rate": 4.289466441657181e-05,
      "loss": 0.1034,
      "step": 4386
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2535312730309707,
      "learning_rate": 4.2891200042700076e-05,
      "loss": 0.2528,
      "step": 4387
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8595185791818227,
      "learning_rate": 4.288773496443898e-05,
      "loss": 0.1445,
      "step": 4388
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7750085437644122,
      "learning_rate": 4.2884269181924924e-05,
      "loss": 0.1374,
      "step": 4389
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7000934823613615,
      "learning_rate": 4.2880802695294384e-05,
      "loss": 0.11,
      "step": 4390
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1164730437548878,
      "learning_rate": 4.2877335504683825e-05,
      "loss": 0.1913,
      "step": 4391
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1690810007734762,
      "learning_rate": 4.287386761022975e-05,
      "loss": 0.2965,
      "step": 4392
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7577560047282744,
      "learning_rate": 4.28703990120687e-05,
      "loss": 0.1482,
      "step": 4393
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7337289674971564,
      "learning_rate": 4.286692971033724e-05,
      "loss": 0.1289,
      "step": 4394
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7338758866271995,
      "learning_rate": 4.2863459705171945e-05,
      "loss": 0.1767,
      "step": 4395
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.441620861287545,
      "learning_rate": 4.285998899670944e-05,
      "loss": 0.3198,
      "step": 4396
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9759815693468457,
      "learning_rate": 4.2856517585086374e-05,
      "loss": 0.1764,
      "step": 4397
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.926968959683607,
      "learning_rate": 4.285304547043941e-05,
      "loss": 0.7012,
      "step": 4398
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.368915801177635,
      "learning_rate": 4.284957265290525e-05,
      "loss": 0.3522,
      "step": 4399
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7752875976230268,
      "learning_rate": 4.284609913262062e-05,
      "loss": 0.1142,
      "step": 4400
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3089805102814325,
      "learning_rate": 4.2842624909722284e-05,
      "loss": 0.2795,
      "step": 4401
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9584070023228728,
      "learning_rate": 4.283914998434702e-05,
      "loss": 0.2507,
      "step": 4402
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.982348797753341,
      "learning_rate": 4.2835674356631624e-05,
      "loss": 0.1291,
      "step": 4403
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8288438258334536,
      "learning_rate": 4.2832198026712953e-05,
      "loss": 0.1663,
      "step": 4404
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8897934960883991,
      "learning_rate": 4.2828720994727867e-05,
      "loss": 0.1855,
      "step": 4405
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8544895494185365,
      "learning_rate": 4.282524326081325e-05,
      "loss": 0.1213,
      "step": 4406
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.757826002698615,
      "learning_rate": 4.282176482510604e-05,
      "loss": 0.1216,
      "step": 4407
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3581492871295278,
      "learning_rate": 4.281828568774318e-05,
      "loss": 0.1609,
      "step": 4408
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0124759271673127,
      "learning_rate": 4.281480584886163e-05,
      "loss": 0.2138,
      "step": 4409
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.0756265240814638,
      "learning_rate": 4.281132530859842e-05,
      "loss": 0.527,
      "step": 4410
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.90979374858803,
      "learning_rate": 4.280784406709055e-05,
      "loss": 0.1292,
      "step": 4411
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7453893065896686,
      "learning_rate": 4.280436212447511e-05,
      "loss": 0.1715,
      "step": 4412
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1192816310280722,
      "learning_rate": 4.2800879480889174e-05,
      "loss": 0.2416,
      "step": 4413
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.85909466340174,
      "learning_rate": 4.279739613646985e-05,
      "loss": 0.2273,
      "step": 4414
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7078433289036989,
      "learning_rate": 4.279391209135429e-05,
      "loss": 0.1299,
      "step": 4415
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8788823932377752,
      "learning_rate": 4.279042734567965e-05,
      "loss": 0.1582,
      "step": 4416
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6896754965723277,
      "learning_rate": 4.278694189958314e-05,
      "loss": 0.129,
      "step": 4417
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2498055505584589,
      "learning_rate": 4.278345575320198e-05,
      "loss": 0.222,
      "step": 4418
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0525323488017597,
      "learning_rate": 4.277996890667343e-05,
      "loss": 0.3288,
      "step": 4419
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6847469054840604,
      "learning_rate": 4.2776481360134754e-05,
      "loss": 0.142,
      "step": 4420
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.419607748418274,
      "learning_rate": 4.2772993113723266e-05,
      "loss": 0.3228,
      "step": 4421
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6705424343521971,
      "learning_rate": 4.27695041675763e-05,
      "loss": 0.1106,
      "step": 4422
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.545807854103704,
      "learning_rate": 4.2766014521831224e-05,
      "loss": 0.9781,
      "step": 4423
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7776313124241186,
      "learning_rate": 4.2762524176625426e-05,
      "loss": 0.0851,
      "step": 4424
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8601944067986187,
      "learning_rate": 4.275903313209632e-05,
      "loss": 0.155,
      "step": 4425
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8545990473509114,
      "learning_rate": 4.275554138838135e-05,
      "loss": 0.1252,
      "step": 4426
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0373210970260611,
      "learning_rate": 4.2752048945617985e-05,
      "loss": 0.2299,
      "step": 4427
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7738434715699348,
      "learning_rate": 4.2748555803943734e-05,
      "loss": 0.1198,
      "step": 4428
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8400508407390767,
      "learning_rate": 4.274506196349613e-05,
      "loss": 0.1599,
      "step": 4429
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9402940992159453,
      "learning_rate": 4.274156742441271e-05,
      "loss": 0.1587,
      "step": 4430
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6747766237133483,
      "learning_rate": 4.273807218683107e-05,
      "loss": 0.3172,
      "step": 4431
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7866794848910753,
      "learning_rate": 4.273457625088881e-05,
      "loss": 0.108,
      "step": 4432
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.11353515840245,
      "learning_rate": 4.273107961672358e-05,
      "loss": 0.2734,
      "step": 4433
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.498117668071743,
      "learning_rate": 4.272758228447303e-05,
      "loss": 0.2852,
      "step": 4434
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8174397761731199,
      "learning_rate": 4.272408425427487e-05,
      "loss": 0.1325,
      "step": 4435
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8400735032328416,
      "learning_rate": 4.2720585526266806e-05,
      "loss": 0.1673,
      "step": 4436
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9234209856496387,
      "learning_rate": 4.271708610058659e-05,
      "loss": 0.1619,
      "step": 4437
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.93364595734243,
      "learning_rate": 4.2713585977371995e-05,
      "loss": 0.1447,
      "step": 4438
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.489948266805711,
      "learning_rate": 4.271008515676082e-05,
      "loss": 0.3905,
      "step": 4439
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7668978465102352,
      "learning_rate": 4.270658363889092e-05,
      "loss": 0.61,
      "step": 4440
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0493873719324558,
      "learning_rate": 4.2703081423900114e-05,
      "loss": 0.1349,
      "step": 4441
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.756067494473576,
      "learning_rate": 4.269957851192631e-05,
      "loss": 0.1074,
      "step": 4442
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5454536812032746,
      "learning_rate": 4.269607490310741e-05,
      "loss": 0.093,
      "step": 4443
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7578301504348262,
      "learning_rate": 4.2692570597581375e-05,
      "loss": 0.1277,
      "step": 4444
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4244210441352636,
      "learning_rate": 4.268906559548614e-05,
      "loss": 0.0902,
      "step": 4445
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7957359510145328,
      "learning_rate": 4.2685559896959717e-05,
      "loss": 0.1249,
      "step": 4446
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5507686286104234,
      "learning_rate": 4.268205350214013e-05,
      "loss": 0.1082,
      "step": 4447
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0089955273103421,
      "learning_rate": 4.267854641116542e-05,
      "loss": 0.1312,
      "step": 4448
    },
    {
      "epoch": 0.27,
      "grad_norm": 18.898982911994924,
      "learning_rate": 4.267503862417368e-05,
      "loss": 1.821,
      "step": 4449
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9718035495326184,
      "learning_rate": 4.267153014130299e-05,
      "loss": 0.1987,
      "step": 4450
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.637447278144773,
      "learning_rate": 4.26680209626915e-05,
      "loss": 0.1063,
      "step": 4451
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.046752766462287,
      "learning_rate": 4.2664511088477356e-05,
      "loss": 0.2922,
      "step": 4452
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9202039506500853,
      "learning_rate": 4.266100051879875e-05,
      "loss": 0.1007,
      "step": 4453
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2120766072989189,
      "learning_rate": 4.2657489253793906e-05,
      "loss": 0.1774,
      "step": 4454
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9744346966113612,
      "learning_rate": 4.265397729360105e-05,
      "loss": 0.1694,
      "step": 4455
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8157281109939615,
      "learning_rate": 4.2650464638358455e-05,
      "loss": 0.1871,
      "step": 4456
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7125344487005002,
      "learning_rate": 4.2646951288204417e-05,
      "loss": 0.1706,
      "step": 4457
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2602409627965325,
      "learning_rate": 4.264343724327726e-05,
      "loss": 0.3312,
      "step": 4458
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2753648271540854,
      "learning_rate": 4.263992250371534e-05,
      "loss": 0.4056,
      "step": 4459
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8437362437559465,
      "learning_rate": 4.2636407069657025e-05,
      "loss": 0.1395,
      "step": 4460
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3634634595410395,
      "learning_rate": 4.263289094124072e-05,
      "loss": 0.115,
      "step": 4461
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8441602104802413,
      "learning_rate": 4.2629374118604865e-05,
      "loss": 0.4628,
      "step": 4462
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.330871415107855,
      "learning_rate": 4.262585660188791e-05,
      "loss": 0.341,
      "step": 4463
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9047667312568571,
      "learning_rate": 4.262233839122836e-05,
      "loss": 0.1511,
      "step": 4464
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5777769376231638,
      "learning_rate": 4.261881948676471e-05,
      "loss": 0.1177,
      "step": 4465
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7638686973976914,
      "learning_rate": 4.2615299888635515e-05,
      "loss": 0.1581,
      "step": 4466
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2378349545286602,
      "learning_rate": 4.261177959697934e-05,
      "loss": 0.21,
      "step": 4467
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.041789081954717,
      "learning_rate": 4.2608258611934774e-05,
      "loss": 0.2269,
      "step": 4468
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9238479150960511,
      "learning_rate": 4.2604736933640454e-05,
      "loss": 0.1003,
      "step": 4469
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9084240579858329,
      "learning_rate": 4.260121456223502e-05,
      "loss": 0.1907,
      "step": 4470
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.155637375846967,
      "learning_rate": 4.2597691497857156e-05,
      "loss": 0.183,
      "step": 4471
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3357665135380794,
      "learning_rate": 4.2594167740645564e-05,
      "loss": 0.1098,
      "step": 4472
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6767269609376751,
      "learning_rate": 4.259064329073898e-05,
      "loss": 0.5411,
      "step": 4473
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6805256127466202,
      "learning_rate": 4.258711814827616e-05,
      "loss": 0.346,
      "step": 4474
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3516350347301087,
      "learning_rate": 4.25835923133959e-05,
      "loss": 0.357,
      "step": 4475
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7903358667522268,
      "learning_rate": 4.2580065786237e-05,
      "loss": 0.1286,
      "step": 4476
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.115689545539756,
      "learning_rate": 4.257653856693832e-05,
      "loss": 0.4142,
      "step": 4477
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.130658956952041,
      "learning_rate": 4.257301065563871e-05,
      "loss": 0.2069,
      "step": 4478
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2462298543286143,
      "learning_rate": 4.256948205247708e-05,
      "loss": 0.2667,
      "step": 4479
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6990674056057519,
      "learning_rate": 4.2565952757592354e-05,
      "loss": 0.0923,
      "step": 4480
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9835557516238124,
      "learning_rate": 4.2562422771123485e-05,
      "loss": 0.187,
      "step": 4481
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.348692882751644,
      "learning_rate": 4.255889209320943e-05,
      "loss": 0.2929,
      "step": 4482
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8098578191159507,
      "learning_rate": 4.255536072398922e-05,
      "loss": 0.4566,
      "step": 4483
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.435264715637776,
      "learning_rate": 4.2551828663601875e-05,
      "loss": 0.3243,
      "step": 4484
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6603306607570281,
      "learning_rate": 4.254829591218645e-05,
      "loss": 0.095,
      "step": 4485
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6686138976172683,
      "learning_rate": 4.254476246988205e-05,
      "loss": 0.0842,
      "step": 4486
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6251973110670463,
      "learning_rate": 4.254122833682776e-05,
      "loss": 0.1046,
      "step": 4487
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8237988973947366,
      "learning_rate": 4.253769351316276e-05,
      "loss": 0.4202,
      "step": 4488
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0086152693172,
      "learning_rate": 4.253415799902618e-05,
      "loss": 0.1894,
      "step": 4489
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5179325641169001,
      "learning_rate": 4.2530621794557243e-05,
      "loss": 0.343,
      "step": 4490
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7409453262101092,
      "learning_rate": 4.252708489989517e-05,
      "loss": 0.1009,
      "step": 4491
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5986459264864237,
      "learning_rate": 4.252354731517919e-05,
      "loss": 0.0897,
      "step": 4492
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7191083214464774,
      "learning_rate": 4.252000904054859e-05,
      "loss": 0.1316,
      "step": 4493
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.670500333610858,
      "learning_rate": 4.251647007614269e-05,
      "loss": 0.3825,
      "step": 4494
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6675412751695092,
      "learning_rate": 4.25129304221008e-05,
      "loss": 0.0843,
      "step": 4495
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0504962337185264,
      "learning_rate": 4.2509390078562295e-05,
      "loss": 0.2317,
      "step": 4496
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1963462661875242,
      "learning_rate": 4.250584904566655e-05,
      "loss": 0.3076,
      "step": 4497
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8420042286976914,
      "learning_rate": 4.2502307323552986e-05,
      "loss": 0.1665,
      "step": 4498
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6296071236332325,
      "learning_rate": 4.2498764912361036e-05,
      "loss": 0.097,
      "step": 4499
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9887079645160964,
      "learning_rate": 4.2495221812230176e-05,
      "loss": 0.1144,
      "step": 4500
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9218583579374641,
      "learning_rate": 4.249167802329988e-05,
      "loss": 0.126,
      "step": 4501
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.483537784984841,
      "learning_rate": 4.2488133545709705e-05,
      "loss": 0.2432,
      "step": 4502
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0074728861946558,
      "learning_rate": 4.248458837959917e-05,
      "loss": 0.3168,
      "step": 4503
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8722298913462326,
      "learning_rate": 4.2481042525107854e-05,
      "loss": 0.1678,
      "step": 4504
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2094662109943743,
      "learning_rate": 4.247749598237537e-05,
      "loss": 0.253,
      "step": 4505
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1342065813974296,
      "learning_rate": 4.247394875154135e-05,
      "loss": 0.2308,
      "step": 4506
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1596914640482474,
      "learning_rate": 4.2470400832745425e-05,
      "loss": 0.2719,
      "step": 4507
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6276280222433245,
      "learning_rate": 4.246685222612732e-05,
      "loss": 0.0726,
      "step": 4508
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3610630781213127,
      "learning_rate": 4.246330293182671e-05,
      "loss": 0.4067,
      "step": 4509
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.18821845989286,
      "learning_rate": 4.2459752949983356e-05,
      "loss": 0.184,
      "step": 4510
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7952210023365807,
      "learning_rate": 4.245620228073701e-05,
      "loss": 0.094,
      "step": 4511
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8166939338010323,
      "learning_rate": 4.2452650924227475e-05,
      "loss": 0.1563,
      "step": 4512
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8788571134527928,
      "learning_rate": 4.2449098880594576e-05,
      "loss": 0.0942,
      "step": 4513
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0723497306169623,
      "learning_rate": 4.244554614997813e-05,
      "loss": 0.2277,
      "step": 4514
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8720434495470761,
      "learning_rate": 4.244199273251803e-05,
      "loss": 0.2227,
      "step": 4515
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1476048160465897,
      "learning_rate": 4.2438438628354195e-05,
      "loss": 0.2458,
      "step": 4516
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.341188608891246,
      "learning_rate": 4.2434883837626514e-05,
      "loss": 0.1515,
      "step": 4517
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3618119403488202,
      "learning_rate": 4.243132836047497e-05,
      "loss": 0.2342,
      "step": 4518
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6927335540754294,
      "learning_rate": 4.2427772197039526e-05,
      "loss": 0.1316,
      "step": 4519
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0235987986131416,
      "learning_rate": 4.2424215347460214e-05,
      "loss": 0.1975,
      "step": 4520
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0868632113445975,
      "learning_rate": 4.242065781187705e-05,
      "loss": 0.1592,
      "step": 4521
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.039708470573785,
      "learning_rate": 4.2417099590430104e-05,
      "loss": 0.1956,
      "step": 4522
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8561340369654209,
      "learning_rate": 4.2413540683259465e-05,
      "loss": 0.1345,
      "step": 4523
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9534966187937484,
      "learning_rate": 4.2409981090505244e-05,
      "loss": 0.2396,
      "step": 4524
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9642108424228532,
      "learning_rate": 4.2406420812307594e-05,
      "loss": 0.1678,
      "step": 4525
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7609219540003038,
      "learning_rate": 4.240285984880668e-05,
      "loss": 0.1381,
      "step": 4526
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9750321431567326,
      "learning_rate": 4.2399298200142705e-05,
      "loss": 0.1713,
      "step": 4527
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0320347880713534,
      "learning_rate": 4.2395735866455894e-05,
      "loss": 0.1516,
      "step": 4528
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.617083007467392,
      "learning_rate": 4.239217284788649e-05,
      "loss": 0.0921,
      "step": 4529
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6765849021806171,
      "learning_rate": 4.2388609144574776e-05,
      "loss": 0.0917,
      "step": 4530
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9622208715995427,
      "learning_rate": 4.238504475666106e-05,
      "loss": 0.3596,
      "step": 4531
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9569703993518569,
      "learning_rate": 4.238147968428567e-05,
      "loss": 0.2078,
      "step": 4532
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0389233970114984,
      "learning_rate": 4.237791392758897e-05,
      "loss": 0.2297,
      "step": 4533
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0953526878175857,
      "learning_rate": 4.237434748671134e-05,
      "loss": 0.2237,
      "step": 4534
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6713645315922917,
      "learning_rate": 4.2370780361793206e-05,
      "loss": 0.0954,
      "step": 4535
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9759405267996801,
      "learning_rate": 4.236721255297499e-05,
      "loss": 0.2357,
      "step": 4536
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9389855939161423,
      "learning_rate": 4.236364406039718e-05,
      "loss": 0.2103,
      "step": 4537
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6490965103966858,
      "learning_rate": 4.236007488420025e-05,
      "loss": 0.0863,
      "step": 4538
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7658193770482827,
      "learning_rate": 4.235650502452473e-05,
      "loss": 0.0872,
      "step": 4539
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6112942824051406,
      "learning_rate": 4.235293448151118e-05,
      "loss": 0.0838,
      "step": 4540
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0425375222803006,
      "learning_rate": 4.2349363255300145e-05,
      "loss": 0.1679,
      "step": 4541
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5768659896956533,
      "learning_rate": 4.234579134603226e-05,
      "loss": 0.074,
      "step": 4542
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5036361263218736,
      "learning_rate": 4.234221875384813e-05,
      "loss": 0.2236,
      "step": 4543
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1860262703738433,
      "learning_rate": 4.2338645478888425e-05,
      "loss": 0.18,
      "step": 4544
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.3974092877603126,
      "learning_rate": 4.233507152129382e-05,
      "loss": 0.6616,
      "step": 4545
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8772418553481811,
      "learning_rate": 4.233149688120502e-05,
      "loss": 0.1191,
      "step": 4546
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.798319536315143,
      "learning_rate": 4.2327921558762776e-05,
      "loss": 1.117,
      "step": 4547
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0338631991104417,
      "learning_rate": 4.232434555410783e-05,
      "loss": 0.1855,
      "step": 4548
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7486336257219002,
      "learning_rate": 4.232076886738099e-05,
      "loss": 0.4836,
      "step": 4549
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9658895419273399,
      "learning_rate": 4.2317191498723076e-05,
      "loss": 0.1569,
      "step": 4550
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5764904076918609,
      "learning_rate": 4.231361344827491e-05,
      "loss": 0.5646,
      "step": 4551
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3651145684964243,
      "learning_rate": 4.2310034716177375e-05,
      "loss": 0.2217,
      "step": 4552
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4145853813534401,
      "learning_rate": 4.2306455302571366e-05,
      "loss": 0.5088,
      "step": 4553
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3764281061274695,
      "learning_rate": 4.230287520759782e-05,
      "loss": 0.188,
      "step": 4554
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.23301657152409,
      "learning_rate": 4.229929443139766e-05,
      "loss": 0.4737,
      "step": 4555
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0953234178217586,
      "learning_rate": 4.229571297411189e-05,
      "loss": 0.2249,
      "step": 4556
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8805666239370877,
      "learning_rate": 4.2292130835881504e-05,
      "loss": 0.5159,
      "step": 4557
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.5958809118632766,
      "learning_rate": 4.228854801684754e-05,
      "loss": 0.6131,
      "step": 4558
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.265298502661028,
      "learning_rate": 4.228496451715104e-05,
      "loss": 0.3973,
      "step": 4559
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6064682667949766,
      "learning_rate": 4.228138033693311e-05,
      "loss": 0.1119,
      "step": 4560
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7954143696774888,
      "learning_rate": 4.2277795476334836e-05,
      "loss": 0.125,
      "step": 4561
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6865438138172004,
      "learning_rate": 4.227420993549738e-05,
      "loss": 0.0961,
      "step": 4562
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4451567063380328,
      "learning_rate": 4.2270623714561905e-05,
      "loss": 0.331,
      "step": 4563
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.319082837864759,
      "learning_rate": 4.2267036813669585e-05,
      "loss": 0.2847,
      "step": 4564
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9560472759775804,
      "learning_rate": 4.226344923296166e-05,
      "loss": 0.2806,
      "step": 4565
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.022542081274776,
      "learning_rate": 4.225986097257936e-05,
      "loss": 0.2229,
      "step": 4566
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.928084307618574,
      "learning_rate": 4.225627203266397e-05,
      "loss": 0.127,
      "step": 4567
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9533710397070876,
      "learning_rate": 4.225268241335678e-05,
      "loss": 0.2776,
      "step": 4568
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2092039317579446,
      "learning_rate": 4.2249092114799114e-05,
      "loss": 0.5781,
      "step": 4569
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8021305827045223,
      "learning_rate": 4.224550113713234e-05,
      "loss": 0.1312,
      "step": 4570
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8061523001857634,
      "learning_rate": 4.2241909480497814e-05,
      "loss": 0.126,
      "step": 4571
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0682895088755489,
      "learning_rate": 4.223831714503696e-05,
      "loss": 0.234,
      "step": 4572
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4968635993226493,
      "learning_rate": 4.223472413089121e-05,
      "loss": 0.188,
      "step": 4573
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1674418371423194,
      "learning_rate": 4.223113043820202e-05,
      "loss": 0.1942,
      "step": 4574
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1869000985575742,
      "learning_rate": 4.2227536067110875e-05,
      "loss": 0.2113,
      "step": 4575
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0610633953351019,
      "learning_rate": 4.2223941017759285e-05,
      "loss": 0.2316,
      "step": 4576
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9070612710245127,
      "learning_rate": 4.2220345290288785e-05,
      "loss": 0.2152,
      "step": 4577
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8273946224186691,
      "learning_rate": 4.2216748884840964e-05,
      "loss": 0.1488,
      "step": 4578
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.603237081828925,
      "learning_rate": 4.221315180155739e-05,
      "loss": 0.2455,
      "step": 4579
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9744009916601316,
      "learning_rate": 4.22095540405797e-05,
      "loss": 0.1498,
      "step": 4580
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7578750340588093,
      "learning_rate": 4.220595560204954e-05,
      "loss": 0.1062,
      "step": 4581
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0643438144085646,
      "learning_rate": 4.2202356486108563e-05,
      "loss": 0.1286,
      "step": 4582
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.2155276916731004,
      "learning_rate": 4.219875669289849e-05,
      "loss": 0.2864,
      "step": 4583
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6903719835959711,
      "learning_rate": 4.2195156222561036e-05,
      "loss": 0.1126,
      "step": 4584
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1484354691149206,
      "learning_rate": 4.219155507523797e-05,
      "loss": 0.3735,
      "step": 4585
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.60155671792173,
      "learning_rate": 4.218795325107104e-05,
      "loss": 0.2291,
      "step": 4586
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2226802043011442,
      "learning_rate": 4.2184350750202084e-05,
      "loss": 0.2649,
      "step": 4587
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2100409438803252,
      "learning_rate": 4.2180747572772925e-05,
      "loss": 0.262,
      "step": 4588
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6338493730583812,
      "learning_rate": 4.217714371892542e-05,
      "loss": 0.1872,
      "step": 4589
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4564483050093306,
      "learning_rate": 4.217353918880146e-05,
      "loss": 0.3584,
      "step": 4590
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7451485373580203,
      "learning_rate": 4.216993398254294e-05,
      "loss": 0.1078,
      "step": 4591
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5994267321308797,
      "learning_rate": 4.216632810029183e-05,
      "loss": 0.1103,
      "step": 4592
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1753320199995234,
      "learning_rate": 4.216272154219007e-05,
      "loss": 0.2634,
      "step": 4593
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.212895747849766,
      "learning_rate": 4.2159114308379664e-05,
      "loss": 0.2071,
      "step": 4594
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8368669681157734,
      "learning_rate": 4.215550639900263e-05,
      "loss": 0.1289,
      "step": 4595
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7954557537351964,
      "learning_rate": 4.2151897814201026e-05,
      "loss": 0.0856,
      "step": 4596
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6832086947646066,
      "learning_rate": 4.21482885541169e-05,
      "loss": 0.0848,
      "step": 4597
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.6542094547646147,
      "learning_rate": 4.2144678618892364e-05,
      "loss": 0.5498,
      "step": 4598
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.195704901058212,
      "learning_rate": 4.2141068008669546e-05,
      "loss": 0.9878,
      "step": 4599
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8257744320936289,
      "learning_rate": 4.2137456723590595e-05,
      "loss": 0.1937,
      "step": 4600
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8079730650271976,
      "learning_rate": 4.213384476379769e-05,
      "loss": 0.2034,
      "step": 4601
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2591777912233604,
      "learning_rate": 4.2130232129433044e-05,
      "loss": 0.6218,
      "step": 4602
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9808299714441673,
      "learning_rate": 4.212661882063887e-05,
      "loss": 0.153,
      "step": 4603
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.100996780231445,
      "learning_rate": 4.2123004837557455e-05,
      "loss": 0.4313,
      "step": 4604
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8934679350846328,
      "learning_rate": 4.211939018033106e-05,
      "loss": 0.1501,
      "step": 4605
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9648543322876464,
      "learning_rate": 4.211577484910201e-05,
      "loss": 0.1473,
      "step": 4606
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1855730361088295,
      "learning_rate": 4.211215884401263e-05,
      "loss": 0.1421,
      "step": 4607
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1222455675782312,
      "learning_rate": 4.210854216520529e-05,
      "loss": 0.235,
      "step": 4608
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4766858738945963,
      "learning_rate": 4.210492481282239e-05,
      "loss": 0.3799,
      "step": 4609
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2473369903251,
      "learning_rate": 4.210130678700634e-05,
      "loss": 0.164,
      "step": 4610
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9775889282815567,
      "learning_rate": 4.209768808789958e-05,
      "loss": 0.2217,
      "step": 4611
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8386422954261017,
      "learning_rate": 4.20940687156446e-05,
      "loss": 0.106,
      "step": 4612
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2508074845783042,
      "learning_rate": 4.209044867038387e-05,
      "loss": 0.2308,
      "step": 4613
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.225530954838974,
      "learning_rate": 4.208682795225993e-05,
      "loss": 0.2528,
      "step": 4614
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1681186210627927,
      "learning_rate": 4.2083206561415335e-05,
      "loss": 0.3472,
      "step": 4615
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.095186235863427,
      "learning_rate": 4.2079584497992644e-05,
      "loss": 0.5911,
      "step": 4616
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3861413171163892,
      "learning_rate": 4.2075961762134484e-05,
      "loss": 0.2471,
      "step": 4617
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.9199558626201045,
      "learning_rate": 4.207233835398346e-05,
      "loss": 0.1269,
      "step": 4618
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5764620004826653,
      "learning_rate": 4.2068714273682245e-05,
      "loss": 0.1186,
      "step": 4619
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5706392511110978,
      "learning_rate": 4.2065089521373514e-05,
      "loss": 0.3615,
      "step": 4620
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3138757777422,
      "learning_rate": 4.206146409719999e-05,
      "loss": 0.1784,
      "step": 4621
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0538672177732964,
      "learning_rate": 4.2057838001304376e-05,
      "loss": 0.242,
      "step": 4622
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5872025128036836,
      "learning_rate": 4.2054211233829474e-05,
      "loss": 0.0716,
      "step": 4623
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9606287968585523,
      "learning_rate": 4.205058379491804e-05,
      "loss": 0.231,
      "step": 4624
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8048997028055659,
      "learning_rate": 4.204695568471291e-05,
      "loss": 0.1194,
      "step": 4625
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9650649259597053,
      "learning_rate": 4.2043326903356914e-05,
      "loss": 0.1735,
      "step": 4626
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4217730488197287,
      "learning_rate": 4.203969745099292e-05,
      "loss": 0.2693,
      "step": 4627
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9456205051187248,
      "learning_rate": 4.203606732776384e-05,
      "loss": 0.1842,
      "step": 4628
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6010582770918438,
      "learning_rate": 4.203243653381257e-05,
      "loss": 0.3666,
      "step": 4629
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2191490401777298,
      "learning_rate": 4.2028805069282064e-05,
      "loss": 0.2231,
      "step": 4630
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3074803909773562,
      "learning_rate": 4.2025172934315304e-05,
      "loss": 0.4981,
      "step": 4631
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7820493921454144,
      "learning_rate": 4.202154012905528e-05,
      "loss": 0.1429,
      "step": 4632
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.463534393286948,
      "learning_rate": 4.2017906653645025e-05,
      "loss": 0.3119,
      "step": 4633
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.627868434461568,
      "learning_rate": 4.201427250822758e-05,
      "loss": 0.291,
      "step": 4634
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5310285657779967,
      "learning_rate": 4.201063769294604e-05,
      "loss": 0.346,
      "step": 4635
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9724243884804191,
      "learning_rate": 4.200700220794351e-05,
      "loss": 0.2229,
      "step": 4636
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9704236075378293,
      "learning_rate": 4.20033660533631e-05,
      "loss": 0.1506,
      "step": 4637
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8516165930829934,
      "learning_rate": 4.1999729229348e-05,
      "loss": 0.1524,
      "step": 4638
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.927902839026612,
      "learning_rate": 4.1996091736041366e-05,
      "loss": 0.1471,
      "step": 4639
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9355868187611504,
      "learning_rate": 4.199245357358642e-05,
      "loss": 0.1512,
      "step": 4640
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8278701278863864,
      "learning_rate": 4.19888147421264e-05,
      "loss": 0.111,
      "step": 4641
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9176455995714234,
      "learning_rate": 4.1985175241804566e-05,
      "loss": 0.218,
      "step": 4642
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5095385809028234,
      "learning_rate": 4.198153507276421e-05,
      "loss": 0.2965,
      "step": 4643
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0858826134979402,
      "learning_rate": 4.197789423514865e-05,
      "loss": 0.1689,
      "step": 4644
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8921470125000843,
      "learning_rate": 4.1974252729101225e-05,
      "loss": 0.2739,
      "step": 4645
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0082164090698105,
      "learning_rate": 4.197061055476531e-05,
      "loss": 0.1867,
      "step": 4646
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7858812112852881,
      "learning_rate": 4.1966967712284286e-05,
      "loss": 0.1124,
      "step": 4647
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5570933319607586,
      "learning_rate": 4.196332420180159e-05,
      "loss": 0.1497,
      "step": 4648
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9878288088254908,
      "learning_rate": 4.195968002346066e-05,
      "loss": 0.1343,
      "step": 4649
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7635378502255016,
      "learning_rate": 4.195603517740497e-05,
      "loss": 0.0943,
      "step": 4650
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2216694810984376,
      "learning_rate": 4.195238966377804e-05,
      "loss": 0.1772,
      "step": 4651
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0203908482494706,
      "learning_rate": 4.194874348272336e-05,
      "loss": 0.1311,
      "step": 4652
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6249377395350253,
      "learning_rate": 4.194509663438451e-05,
      "loss": 0.086,
      "step": 4653
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7108202819506646,
      "learning_rate": 4.194144911890506e-05,
      "loss": 0.12,
      "step": 4654
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6428672395063995,
      "learning_rate": 4.193780093642862e-05,
      "loss": 0.0941,
      "step": 4655
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5958378253806123,
      "learning_rate": 4.193415208709881e-05,
      "loss": 0.0967,
      "step": 4656
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4275581328459668,
      "learning_rate": 4.1930502571059306e-05,
      "loss": 0.3019,
      "step": 4657
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8081194484779529,
      "learning_rate": 4.192685238845378e-05,
      "loss": 0.0474,
      "step": 4658
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0723702422202852,
      "learning_rate": 4.192320153942594e-05,
      "loss": 0.2861,
      "step": 4659
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.773189128831636,
      "learning_rate": 4.191955002411953e-05,
      "loss": 0.4946,
      "step": 4660
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7709454564601905,
      "learning_rate": 4.191589784267831e-05,
      "loss": 0.1121,
      "step": 4661
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7022318800984927,
      "learning_rate": 4.191224499524606e-05,
      "loss": 0.1237,
      "step": 4662
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.272141706626429,
      "learning_rate": 4.190859148196662e-05,
      "loss": 0.3069,
      "step": 4663
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.175263060149436,
      "learning_rate": 4.190493730298381e-05,
      "loss": 0.3154,
      "step": 4664
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7964605963188544,
      "learning_rate": 4.1901282458441494e-05,
      "loss": 0.1068,
      "step": 4665
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.083273005646654,
      "learning_rate": 4.189762694848358e-05,
      "loss": 0.2778,
      "step": 4666
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.412992793417958,
      "learning_rate": 4.1893970773253986e-05,
      "loss": 0.4525,
      "step": 4667
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8448294003536347,
      "learning_rate": 4.189031393289665e-05,
      "loss": 0.1131,
      "step": 4668
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.807307118782662,
      "learning_rate": 4.188665642755555e-05,
      "loss": 0.4697,
      "step": 4669
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7736187302988977,
      "learning_rate": 4.188299825737469e-05,
      "loss": 0.4155,
      "step": 4670
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9029301283664434,
      "learning_rate": 4.1879339422498087e-05,
      "loss": 0.1829,
      "step": 4671
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9716269162182406,
      "learning_rate": 4.1875679923069786e-05,
      "loss": 0.2421,
      "step": 4672
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.594417413865695,
      "learning_rate": 4.187201975923388e-05,
      "loss": 0.0667,
      "step": 4673
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8958376672830604,
      "learning_rate": 4.186835893113447e-05,
      "loss": 0.199,
      "step": 4674
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7915710824220056,
      "learning_rate": 4.1864697438915666e-05,
      "loss": 0.1109,
      "step": 4675
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6511543484642452,
      "learning_rate": 4.186103528272165e-05,
      "loss": 0.5735,
      "step": 4676
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1113704997514677,
      "learning_rate": 4.1857372462696586e-05,
      "loss": 0.1311,
      "step": 4677
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.927762125893953,
      "learning_rate": 4.1853708978984686e-05,
      "loss": 0.4366,
      "step": 4678
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.2547592330054917,
      "learning_rate": 4.185004483173019e-05,
      "loss": 0.3064,
      "step": 4679
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9568241130228066,
      "learning_rate": 4.184638002107734e-05,
      "loss": 0.1906,
      "step": 4680
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.605449511969841,
      "learning_rate": 4.1842714547170446e-05,
      "loss": 0.4219,
      "step": 4681
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9243995291805333,
      "learning_rate": 4.1839048410153806e-05,
      "loss": 0.0848,
      "step": 4682
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9006119956002424,
      "learning_rate": 4.183538161017176e-05,
      "loss": 0.135,
      "step": 4683
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1015028493645718,
      "learning_rate": 4.183171414736868e-05,
      "loss": 0.1942,
      "step": 4684
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6667899169602229,
      "learning_rate": 4.182804602188895e-05,
      "loss": 0.1183,
      "step": 4685
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4048487054205632,
      "learning_rate": 4.182437723387699e-05,
      "loss": 0.2215,
      "step": 4686
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7804013737237177,
      "learning_rate": 4.182070778347723e-05,
      "loss": 0.1346,
      "step": 4687
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8640266679834445,
      "learning_rate": 4.181703767083416e-05,
      "loss": 0.1911,
      "step": 4688
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7956011835860921,
      "learning_rate": 4.181336689609226e-05,
      "loss": 0.184,
      "step": 4689
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9853671565812627,
      "learning_rate": 4.180969545939606e-05,
      "loss": 0.2216,
      "step": 4690
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7498938221255625,
      "learning_rate": 4.18060233608901e-05,
      "loss": 0.1059,
      "step": 4691
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9501630934270936,
      "learning_rate": 4.180235060071895e-05,
      "loss": 0.2035,
      "step": 4692
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9159209890893918,
      "learning_rate": 4.1798677179027224e-05,
      "loss": 0.1805,
      "step": 4693
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6058653423402046,
      "learning_rate": 4.1795003095959526e-05,
      "loss": 0.0736,
      "step": 4694
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7577559788352769,
      "learning_rate": 4.179132835166054e-05,
      "loss": 0.1353,
      "step": 4695
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1587103173738893,
      "learning_rate": 4.17876529462749e-05,
      "loss": 0.3607,
      "step": 4696
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1075928921686746,
      "learning_rate": 4.178397687994735e-05,
      "loss": 0.187,
      "step": 4697
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.738831308139645,
      "learning_rate": 4.1780300152822596e-05,
      "loss": 0.1758,
      "step": 4698
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.269862557764646,
      "learning_rate": 4.17766227650454e-05,
      "loss": 0.2765,
      "step": 4699
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6963716322285978,
      "learning_rate": 4.177294471676053e-05,
      "loss": 0.2736,
      "step": 4700
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.547550892261491,
      "learning_rate": 4.176926600811282e-05,
      "loss": 0.6324,
      "step": 4701
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.725230170179934,
      "learning_rate": 4.176558663924709e-05,
      "loss": 0.3362,
      "step": 4702
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9921814470565022,
      "learning_rate": 4.176190661030819e-05,
      "loss": 0.2871,
      "step": 4703
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.956848851702113,
      "learning_rate": 4.175822592144102e-05,
      "loss": 1.0912,
      "step": 4704
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.938374056980877,
      "learning_rate": 4.1754544572790474e-05,
      "loss": 0.1943,
      "step": 4705
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7570751174226545,
      "learning_rate": 4.175086256450151e-05,
      "loss": 0.1657,
      "step": 4706
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5935943472253966,
      "learning_rate": 4.174717989671909e-05,
      "loss": 0.3021,
      "step": 4707
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9997181213455724,
      "learning_rate": 4.174349656958818e-05,
      "loss": 0.3529,
      "step": 4708
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.952210105919042,
      "learning_rate": 4.1739812583253825e-05,
      "loss": 0.3001,
      "step": 4709
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6821948153772469,
      "learning_rate": 4.173612793786104e-05,
      "loss": 0.0972,
      "step": 4710
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7493829885386188,
      "learning_rate": 4.173244263355491e-05,
      "loss": 0.0917,
      "step": 4711
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8358682785692201,
      "learning_rate": 4.1728756670480516e-05,
      "loss": 0.191,
      "step": 4712
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0252761445494054,
      "learning_rate": 4.172507004878299e-05,
      "loss": 0.2008,
      "step": 4713
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9716524272552789,
      "learning_rate": 4.172138276860746e-05,
      "loss": 0.1375,
      "step": 4714
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.068205544790608,
      "learning_rate": 4.171769483009912e-05,
      "loss": 0.1438,
      "step": 4715
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6826005270864931,
      "learning_rate": 4.171400623340314e-05,
      "loss": 0.089,
      "step": 4716
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8468444602615054,
      "learning_rate": 4.1710316978664764e-05,
      "loss": 0.116,
      "step": 4717
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3779998950092325,
      "learning_rate": 4.1706627066029226e-05,
      "loss": 0.2945,
      "step": 4718
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0629569571625737,
      "learning_rate": 4.170293649564181e-05,
      "loss": 0.1295,
      "step": 4719
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7671280579206715,
      "learning_rate": 4.1699245267647814e-05,
      "loss": 0.0882,
      "step": 4720
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2011686187780026,
      "learning_rate": 4.169555338219257e-05,
      "loss": 0.1544,
      "step": 4721
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5629757901351473,
      "learning_rate": 4.169186083942142e-05,
      "loss": 0.2224,
      "step": 4722
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8256683729018237,
      "learning_rate": 4.168816763947975e-05,
      "loss": 0.0672,
      "step": 4723
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.681397907695445,
      "learning_rate": 4.168447378251295e-05,
      "loss": 0.0972,
      "step": 4724
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.017398304788362,
      "learning_rate": 4.168077926866646e-05,
      "loss": 0.3175,
      "step": 4725
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8739785189046536,
      "learning_rate": 4.167708409808574e-05,
      "loss": 0.2537,
      "step": 4726
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.494770972960886,
      "learning_rate": 4.167338827091627e-05,
      "loss": 0.2688,
      "step": 4727
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7264727245294083,
      "learning_rate": 4.166969178730355e-05,
      "loss": 0.1137,
      "step": 4728
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6801095371765943,
      "learning_rate": 4.1665994647393124e-05,
      "loss": 0.1156,
      "step": 4729
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3235858995556444,
      "learning_rate": 4.166229685133054e-05,
      "loss": 0.4849,
      "step": 4730
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6720251426946521,
      "learning_rate": 4.165859839926138e-05,
      "loss": 0.1227,
      "step": 4731
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8100932066084993,
      "learning_rate": 4.165489929133128e-05,
      "loss": 0.1889,
      "step": 4732
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1380667594074354,
      "learning_rate": 4.165119952768585e-05,
      "loss": 0.3619,
      "step": 4733
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.816024623801165,
      "learning_rate": 4.1647499108470756e-05,
      "loss": 0.3969,
      "step": 4734
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9766868118495612,
      "learning_rate": 4.164379803383169e-05,
      "loss": 0.3238,
      "step": 4735
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8604346557020741,
      "learning_rate": 4.1640096303914375e-05,
      "loss": 0.2107,
      "step": 4736
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.2639724442101565,
      "learning_rate": 4.163639391886455e-05,
      "loss": 0.7829,
      "step": 4737
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4684444393945553,
      "learning_rate": 4.163269087882796e-05,
      "loss": 0.2829,
      "step": 4738
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0234503951405065,
      "learning_rate": 4.162898718395041e-05,
      "loss": 0.3051,
      "step": 4739
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9115501645887526,
      "learning_rate": 4.162528283437772e-05,
      "loss": 0.0854,
      "step": 4740
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.053283032218074,
      "learning_rate": 4.162157783025573e-05,
      "loss": 0.2256,
      "step": 4741
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.997238272152249,
      "learning_rate": 4.161787217173031e-05,
      "loss": 0.1337,
      "step": 4742
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.884456279036856,
      "learning_rate": 4.161416585894735e-05,
      "loss": 0.2393,
      "step": 4743
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7133491334254088,
      "learning_rate": 4.161045889205278e-05,
      "loss": 0.0941,
      "step": 4744
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.475796001221654,
      "learning_rate": 4.1606751271192535e-05,
      "loss": 0.4471,
      "step": 4745
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7140492138231801,
      "learning_rate": 4.160304299651259e-05,
      "loss": 0.0928,
      "step": 4746
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.444999977835025,
      "learning_rate": 4.1599334068158945e-05,
      "loss": 0.4939,
      "step": 4747
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.859417055880234,
      "learning_rate": 4.1595624486277615e-05,
      "loss": 0.9305,
      "step": 4748
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.45850286895125,
      "learning_rate": 4.1591914251014665e-05,
      "loss": 0.2598,
      "step": 4749
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.008886135386266,
      "learning_rate": 4.1588203362516153e-05,
      "loss": 0.1609,
      "step": 4750
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0115830625751197,
      "learning_rate": 4.158449182092818e-05,
      "loss": 0.1677,
      "step": 4751
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8886122739146638,
      "learning_rate": 4.158077962639689e-05,
      "loss": 0.3474,
      "step": 4752
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8497673423256619,
      "learning_rate": 4.157706677906842e-05,
      "loss": 0.1345,
      "step": 4753
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7061847685997139,
      "learning_rate": 4.1573353279088956e-05,
      "loss": 0.085,
      "step": 4754
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9266562455050087,
      "learning_rate": 4.156963912660469e-05,
      "loss": 0.1154,
      "step": 4755
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.018360234722284,
      "learning_rate": 4.156592432176185e-05,
      "loss": 0.2112,
      "step": 4756
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3717174966752914,
      "learning_rate": 4.156220886470671e-05,
      "loss": 0.2286,
      "step": 4757
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8881912285002473,
      "learning_rate": 4.155849275558553e-05,
      "loss": 0.1984,
      "step": 4758
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5464141053253868,
      "learning_rate": 4.1554775994544625e-05,
      "loss": 0.4292,
      "step": 4759
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0118295366198888,
      "learning_rate": 4.1551058581730323e-05,
      "loss": 0.198,
      "step": 4760
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.1768114645052035,
      "learning_rate": 4.154734051728898e-05,
      "loss": 0.8295,
      "step": 4761
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9734974487016306,
      "learning_rate": 4.154362180136699e-05,
      "loss": 0.1949,
      "step": 4762
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1668790280509922,
      "learning_rate": 4.1539902434110746e-05,
      "loss": 0.1973,
      "step": 4763
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8428541014254576,
      "learning_rate": 4.1536182415666695e-05,
      "loss": 0.1655,
      "step": 4764
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5927290573760632,
      "learning_rate": 4.1532461746181284e-05,
      "loss": 0.0798,
      "step": 4765
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9835467442451618,
      "learning_rate": 4.1528740425801013e-05,
      "loss": 0.2065,
      "step": 4766
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8298431088203053,
      "learning_rate": 4.152501845467238e-05,
      "loss": 0.3833,
      "step": 4767
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7359756917263284,
      "learning_rate": 4.1521295832941936e-05,
      "loss": 0.2255,
      "step": 4768
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0323985310119916,
      "learning_rate": 4.151757256075622e-05,
      "loss": 0.2114,
      "step": 4769
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6301779873652275,
      "learning_rate": 4.151384863826186e-05,
      "loss": 0.0779,
      "step": 4770
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7369095719542083,
      "learning_rate": 4.151012406560542e-05,
      "loss": 0.0647,
      "step": 4771
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6893036332298891,
      "learning_rate": 4.1506398842933566e-05,
      "loss": 0.0963,
      "step": 4772
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8361323383257098,
      "learning_rate": 4.1502672970392964e-05,
      "loss": 0.2023,
      "step": 4773
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7760105759342583,
      "learning_rate": 4.1498946448130305e-05,
      "loss": 0.0992,
      "step": 4774
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9594264418682561,
      "learning_rate": 4.1495219276292295e-05,
      "loss": 0.1502,
      "step": 4775
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9239668312893594,
      "learning_rate": 4.1491491455025675e-05,
      "loss": 0.1482,
      "step": 4776
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8415407738387787,
      "learning_rate": 4.148776298447723e-05,
      "loss": 0.1109,
      "step": 4777
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.562267327624114,
      "learning_rate": 4.148403386479374e-05,
      "loss": 0.0637,
      "step": 4778
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7470027288356887,
      "learning_rate": 4.148030409612201e-05,
      "loss": 0.1497,
      "step": 4779
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5091753516197165,
      "learning_rate": 4.1476573678608904e-05,
      "loss": 0.0783,
      "step": 4780
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9120180230825474,
      "learning_rate": 4.147284261240128e-05,
      "loss": 0.2322,
      "step": 4781
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8393961812589008,
      "learning_rate": 4.1469110897646046e-05,
      "loss": 0.0839,
      "step": 4782
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3196079880058198,
      "learning_rate": 4.14653785344901e-05,
      "loss": 0.3296,
      "step": 4783
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6806926603343401,
      "learning_rate": 4.146164552308041e-05,
      "loss": 0.11,
      "step": 4784
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.020192083456892,
      "learning_rate": 4.145791186356394e-05,
      "loss": 0.2234,
      "step": 4785
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.582776420813136,
      "learning_rate": 4.145417755608768e-05,
      "loss": 0.0768,
      "step": 4786
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6019077871132085,
      "learning_rate": 4.1450442600798654e-05,
      "loss": 0.1486,
      "step": 4787
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.774632133541405,
      "learning_rate": 4.144670699784392e-05,
      "loss": 0.3113,
      "step": 4788
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5767753615681532,
      "learning_rate": 4.1442970747370536e-05,
      "loss": 0.2685,
      "step": 4789
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8792613445437334,
      "learning_rate": 4.1439233849525614e-05,
      "loss": 0.1471,
      "step": 4790
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0024979705593045,
      "learning_rate": 4.143549630445628e-05,
      "loss": 0.13,
      "step": 4791
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7023636925361058,
      "learning_rate": 4.143175811230967e-05,
      "loss": 0.1063,
      "step": 4792
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6184930660493716,
      "learning_rate": 4.142801927323297e-05,
      "loss": 0.1091,
      "step": 4793
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7578503743751225,
      "learning_rate": 4.142427978737338e-05,
      "loss": 0.1449,
      "step": 4794
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7867334079045805,
      "learning_rate": 4.1420539654878123e-05,
      "loss": 0.1605,
      "step": 4795
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0915377583692745,
      "learning_rate": 4.1416798875894444e-05,
      "loss": 0.2009,
      "step": 4796
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.791419856473315,
      "learning_rate": 4.1413057450569636e-05,
      "loss": 0.1142,
      "step": 4797
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7697802113604951,
      "learning_rate": 4.140931537905099e-05,
      "loss": 0.155,
      "step": 4798
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1403904335776016,
      "learning_rate": 4.140557266148584e-05,
      "loss": 0.2135,
      "step": 4799
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7553101825403226,
      "learning_rate": 4.1401829298021536e-05,
      "loss": 0.4021,
      "step": 4800
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8352580079251026,
      "learning_rate": 4.139808528880545e-05,
      "loss": 0.1543,
      "step": 4801
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2982590641947565,
      "learning_rate": 4.1394340633985005e-05,
      "loss": 0.5661,
      "step": 4802
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4035731204456532,
      "learning_rate": 4.139059533370761e-05,
      "loss": 0.2734,
      "step": 4803
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6161103563361484,
      "learning_rate": 4.138684938812074e-05,
      "loss": 0.0985,
      "step": 4804
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.088144171071132,
      "learning_rate": 4.138310279737186e-05,
      "loss": 0.5055,
      "step": 4805
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.061218516165757,
      "learning_rate": 4.137935556160848e-05,
      "loss": 0.2629,
      "step": 4806
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6255820001222673,
      "learning_rate": 4.137560768097813e-05,
      "loss": 0.1315,
      "step": 4807
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0726221631366606,
      "learning_rate": 4.1371859155628376e-05,
      "loss": 0.2323,
      "step": 4808
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.016999732397986,
      "learning_rate": 4.13681099857068e-05,
      "loss": 0.159,
      "step": 4809
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7244835310577932,
      "learning_rate": 4.1364360171360996e-05,
      "loss": 0.1099,
      "step": 4810
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8710534886899512,
      "learning_rate": 4.136060971273861e-05,
      "loss": 0.1379,
      "step": 4811
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1669141955444844,
      "learning_rate": 4.135685860998728e-05,
      "loss": 0.2271,
      "step": 4812
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.935630707382864,
      "learning_rate": 4.135310686325472e-05,
      "loss": 0.1871,
      "step": 4813
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8944979342471708,
      "learning_rate": 4.1349354472688615e-05,
      "loss": 0.231,
      "step": 4814
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6345565378979084,
      "learning_rate": 4.13456014384367e-05,
      "loss": 0.0921,
      "step": 4815
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7741645280812665,
      "learning_rate": 4.1341847760646754e-05,
      "loss": 0.0924,
      "step": 4816
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1091998341180798,
      "learning_rate": 4.133809343946654e-05,
      "loss": 0.1608,
      "step": 4817
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3116594524789962,
      "learning_rate": 4.1334338475043886e-05,
      "loss": 0.4489,
      "step": 4818
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6496747858870618,
      "learning_rate": 4.1330582867526616e-05,
      "loss": 0.094,
      "step": 4819
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7805090358530897,
      "learning_rate": 4.13268266170626e-05,
      "loss": 0.1174,
      "step": 4820
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2562416688805573,
      "learning_rate": 4.1323069723799716e-05,
      "loss": 0.2641,
      "step": 4821
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7146765527150514,
      "learning_rate": 4.1319312187885876e-05,
      "loss": 0.1046,
      "step": 4822
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2229052795022164,
      "learning_rate": 4.1315554009469024e-05,
      "loss": 0.2485,
      "step": 4823
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1747217373046257,
      "learning_rate": 4.1311795188697114e-05,
      "loss": 0.1998,
      "step": 4824
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6967521319740869,
      "learning_rate": 4.130803572571814e-05,
      "loss": 0.1099,
      "step": 4825
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9094669665267247,
      "learning_rate": 4.130427562068011e-05,
      "loss": 0.1832,
      "step": 4826
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7583425354628782,
      "learning_rate": 4.130051487373107e-05,
      "loss": 0.136,
      "step": 4827
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1964691269805476,
      "learning_rate": 4.129675348501906e-05,
      "loss": 0.3712,
      "step": 4828
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7548554532291538,
      "learning_rate": 4.12929914546922e-05,
      "loss": 0.1096,
      "step": 4829
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7334777496775255,
      "learning_rate": 4.1289228782898595e-05,
      "loss": 0.1561,
      "step": 4830
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7607246761489975,
      "learning_rate": 4.128546546978637e-05,
      "loss": 0.1352,
      "step": 4831
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.272747440501354,
      "learning_rate": 4.1281701515503694e-05,
      "loss": 0.2579,
      "step": 4832
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.4080828910289975,
      "learning_rate": 4.127793692019877e-05,
      "loss": 0.4307,
      "step": 4833
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9718198837920917,
      "learning_rate": 4.127417168401979e-05,
      "loss": 0.1354,
      "step": 4834
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8696133157977235,
      "learning_rate": 4.127040580711502e-05,
      "loss": 0.1279,
      "step": 4835
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6317539865206975,
      "learning_rate": 4.1266639289632714e-05,
      "loss": 0.0812,
      "step": 4836
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7944433476171076,
      "learning_rate": 4.1262872131721156e-05,
      "loss": 0.1565,
      "step": 4837
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.10861328125288,
      "learning_rate": 4.1259104333528666e-05,
      "loss": 0.1483,
      "step": 4838
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7233070413030899,
      "learning_rate": 4.125533589520358e-05,
      "loss": 0.0777,
      "step": 4839
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3142350222434231,
      "learning_rate": 4.1251566816894285e-05,
      "loss": 0.3016,
      "step": 4840
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2461057283187182,
      "learning_rate": 4.124779709874914e-05,
      "loss": 0.1328,
      "step": 4841
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6621293104087552,
      "learning_rate": 4.1244026740916597e-05,
      "loss": 0.0994,
      "step": 4842
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9298834027341333,
      "learning_rate": 4.124025574354507e-05,
      "loss": 0.2273,
      "step": 4843
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4790782003143246,
      "learning_rate": 4.123648410678304e-05,
      "loss": 0.1468,
      "step": 4844
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5529613642461322,
      "learning_rate": 4.123271183077899e-05,
      "loss": 0.0494,
      "step": 4845
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4297680261944707,
      "learning_rate": 4.122893891568144e-05,
      "loss": 0.4375,
      "step": 4846
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9643571890831556,
      "learning_rate": 4.122516536163894e-05,
      "loss": 0.1655,
      "step": 4847
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.495420403800279,
      "learning_rate": 4.122139116880005e-05,
      "loss": 0.3743,
      "step": 4848
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9841216425207937,
      "learning_rate": 4.121761633731336e-05,
      "loss": 0.1869,
      "step": 4849
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6318730165376172,
      "learning_rate": 4.121384086732749e-05,
      "loss": 0.3716,
      "step": 4850
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6706177182328709,
      "learning_rate": 4.121006475899109e-05,
      "loss": 0.0874,
      "step": 4851
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0577662746041774,
      "learning_rate": 4.120628801245283e-05,
      "loss": 0.189,
      "step": 4852
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.083832636371373,
      "learning_rate": 4.120251062786138e-05,
      "loss": 0.1972,
      "step": 4853
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.8173703994334374,
      "learning_rate": 4.1198732605365486e-05,
      "loss": 0.4998,
      "step": 4854
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7929133418617573,
      "learning_rate": 4.119495394511388e-05,
      "loss": 0.0991,
      "step": 4855
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2637329093545018,
      "learning_rate": 4.119117464725533e-05,
      "loss": 0.3445,
      "step": 4856
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0975115370810156,
      "learning_rate": 4.118739471193862e-05,
      "loss": 0.4921,
      "step": 4857
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4314770432730065,
      "learning_rate": 4.1183614139312595e-05,
      "loss": 0.0698,
      "step": 4858
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7805662506919775,
      "learning_rate": 4.117983292952607e-05,
      "loss": 0.2832,
      "step": 4859
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8024841159926877,
      "learning_rate": 4.117605108272794e-05,
      "loss": 0.3948,
      "step": 4860
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8435759539125348,
      "learning_rate": 4.117226859906708e-05,
      "loss": 0.208,
      "step": 4861
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0018985500686364,
      "learning_rate": 4.1168485478692406e-05,
      "loss": 0.1809,
      "step": 4862
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9121901920460236,
      "learning_rate": 4.1164701721752876e-05,
      "loss": 0.3022,
      "step": 4863
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9317678230154035,
      "learning_rate": 4.1160917328397466e-05,
      "loss": 0.2055,
      "step": 4864
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9073306658141133,
      "learning_rate": 4.1157132298775145e-05,
      "loss": 0.203,
      "step": 4865
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5408674771283735,
      "learning_rate": 4.115334663303495e-05,
      "loss": 0.0806,
      "step": 4866
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1442838684422327,
      "learning_rate": 4.114956033132592e-05,
      "loss": 0.2617,
      "step": 4867
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8449278746803808,
      "learning_rate": 4.114577339379713e-05,
      "loss": 0.1021,
      "step": 4868
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9711966462333615,
      "learning_rate": 4.1141985820597665e-05,
      "loss": 0.1162,
      "step": 4869
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1924017149575186,
      "learning_rate": 4.113819761187666e-05,
      "loss": 0.2532,
      "step": 4870
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6662090067915483,
      "learning_rate": 4.1134408767783236e-05,
      "loss": 0.1032,
      "step": 4871
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7551838804733687,
      "learning_rate": 4.1130619288466585e-05,
      "loss": 0.1288,
      "step": 4872
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7360426077908863,
      "learning_rate": 4.112682917407589e-05,
      "loss": 0.0829,
      "step": 4873
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.760017158446013,
      "learning_rate": 4.1123038424760375e-05,
      "loss": 0.1442,
      "step": 4874
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2359736034107027,
      "learning_rate": 4.111924704066927e-05,
      "loss": 0.2354,
      "step": 4875
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7235824125671544,
      "learning_rate": 4.111545502195187e-05,
      "loss": 0.1803,
      "step": 4876
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6689463940653596,
      "learning_rate": 4.111166236875745e-05,
      "loss": 0.0758,
      "step": 4877
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2237161210565644,
      "learning_rate": 4.110786908123534e-05,
      "loss": 0.2817,
      "step": 4878
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9208792738129895,
      "learning_rate": 4.110407515953488e-05,
      "loss": 0.1326,
      "step": 4879
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6612421028643097,
      "learning_rate": 4.110028060380543e-05,
      "loss": 0.1263,
      "step": 4880
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3199773594646276,
      "learning_rate": 4.109648541419641e-05,
      "loss": 0.3427,
      "step": 4881
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.387929839231073,
      "learning_rate": 4.109268959085721e-05,
      "loss": 0.2064,
      "step": 4882
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8262707949371737,
      "learning_rate": 4.108889313393729e-05,
      "loss": 0.1302,
      "step": 4883
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5207390634074677,
      "learning_rate": 4.108509604358612e-05,
      "loss": 0.0842,
      "step": 4884
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7021573819052784,
      "learning_rate": 4.108129831995319e-05,
      "loss": 0.105,
      "step": 4885
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9237278576005133,
      "learning_rate": 4.107749996318802e-05,
      "loss": 0.3046,
      "step": 4886
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7411054665552242,
      "learning_rate": 4.107370097344015e-05,
      "loss": 0.076,
      "step": 4887
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.505472026437716,
      "learning_rate": 4.106990135085916e-05,
      "loss": 0.2321,
      "step": 4888
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7620887426391103,
      "learning_rate": 4.106610109559463e-05,
      "loss": 0.095,
      "step": 4889
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7736685794839372,
      "learning_rate": 4.10623002077962e-05,
      "loss": 0.1586,
      "step": 4890
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8728060252973419,
      "learning_rate": 4.105849868761348e-05,
      "loss": 0.0998,
      "step": 4891
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5885664176892771,
      "learning_rate": 4.105469653519617e-05,
      "loss": 0.0772,
      "step": 4892
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9996620302244794,
      "learning_rate": 4.105089375069395e-05,
      "loss": 0.1505,
      "step": 4893
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7734375357943315,
      "learning_rate": 4.1047090334256536e-05,
      "loss": 0.0716,
      "step": 4894
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8375733817213313,
      "learning_rate": 4.104328628603368e-05,
      "loss": 0.154,
      "step": 4895
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7321319016354406,
      "learning_rate": 4.103948160617514e-05,
      "loss": 1.1195,
      "step": 4896
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5549076492984042,
      "learning_rate": 4.1035676294830714e-05,
      "loss": 0.5553,
      "step": 4897
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9190348095836939,
      "learning_rate": 4.103187035215023e-05,
      "loss": 0.1693,
      "step": 4898
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7329519553885877,
      "learning_rate": 4.102806377828351e-05,
      "loss": 0.1191,
      "step": 4899
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7472722041005609,
      "learning_rate": 4.102425657338044e-05,
      "loss": 0.1405,
      "step": 4900
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9084352028206579,
      "learning_rate": 4.102044873759089e-05,
      "loss": 0.2173,
      "step": 4901
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6878768893891885,
      "learning_rate": 4.1016640271064806e-05,
      "loss": 0.1093,
      "step": 4902
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7556145979224814,
      "learning_rate": 4.101283117395211e-05,
      "loss": 0.152,
      "step": 4903
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7336757937040208,
      "learning_rate": 4.100902144640277e-05,
      "loss": 0.199,
      "step": 4904
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.49080063902878013,
      "learning_rate": 4.10052110885668e-05,
      "loss": 0.0632,
      "step": 4905
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6553885318091766,
      "learning_rate": 4.100140010059418e-05,
      "loss": 0.13,
      "step": 4906
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7869374344280784,
      "learning_rate": 4.099758848263498e-05,
      "loss": 0.4568,
      "step": 4907
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7740696088563565,
      "learning_rate": 4.0993776234839256e-05,
      "loss": 0.1453,
      "step": 4908
    },
    {
      "epoch": 0.3,
      "grad_norm": 16.60997946568609,
      "learning_rate": 4.09899633573571e-05,
      "loss": 1.494,
      "step": 4909
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6842808706429164,
      "learning_rate": 4.0986149850338617e-05,
      "loss": 0.3711,
      "step": 4910
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7524082440518896,
      "learning_rate": 4.0982335713933976e-05,
      "loss": 0.134,
      "step": 4911
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7653288826665973,
      "learning_rate": 4.0978520948293306e-05,
      "loss": 0.1682,
      "step": 4912
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8269990546459529,
      "learning_rate": 4.0974705553566815e-05,
      "loss": 0.1281,
      "step": 4913
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8466478915188236,
      "learning_rate": 4.097088952990473e-05,
      "loss": 0.2072,
      "step": 4914
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.4294856210058273,
      "learning_rate": 4.0967072877457266e-05,
      "loss": 0.3842,
      "step": 4915
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9451739403560506,
      "learning_rate": 4.0963255596374714e-05,
      "loss": 0.1296,
      "step": 4916
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9971888812709835,
      "learning_rate": 4.095943768680734e-05,
      "loss": 0.1077,
      "step": 4917
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1660199255038317,
      "learning_rate": 4.095561914890547e-05,
      "loss": 0.2172,
      "step": 4918
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.646323720433634,
      "learning_rate": 4.095179998281943e-05,
      "loss": 0.2365,
      "step": 4919
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5140824303845806,
      "learning_rate": 4.0947980188699595e-05,
      "loss": 0.4253,
      "step": 4920
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7215826941746097,
      "learning_rate": 4.094415976669636e-05,
      "loss": 0.6028,
      "step": 4921
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7015620732751956,
      "learning_rate": 4.094033871696012e-05,
      "loss": 0.136,
      "step": 4922
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7696857950100722,
      "learning_rate": 4.093651703964133e-05,
      "loss": 0.8855,
      "step": 4923
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8106583232882113,
      "learning_rate": 4.093269473489043e-05,
      "loss": 0.221,
      "step": 4924
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6531043124682013,
      "learning_rate": 4.0928871802857924e-05,
      "loss": 0.0796,
      "step": 4925
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.015890915261142,
      "learning_rate": 4.0925048243694316e-05,
      "loss": 0.2139,
      "step": 4926
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7761494830603122,
      "learning_rate": 4.092122405755016e-05,
      "loss": 0.1229,
      "step": 4927
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9979077992690828,
      "learning_rate": 4.091739924457599e-05,
      "loss": 0.5668,
      "step": 4928
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8777956028739569,
      "learning_rate": 4.091357380492241e-05,
      "loss": 0.1109,
      "step": 4929
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5611352725421213,
      "learning_rate": 4.090974773874002e-05,
      "loss": 0.4442,
      "step": 4930
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0302255841838586,
      "learning_rate": 4.0905921046179465e-05,
      "loss": 0.1832,
      "step": 4931
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7161520199789697,
      "learning_rate": 4.090209372739139e-05,
      "loss": 0.1281,
      "step": 4932
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8695887649974227,
      "learning_rate": 4.0898265782526495e-05,
      "loss": 0.1409,
      "step": 4933
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9405079903311672,
      "learning_rate": 4.089443721173548e-05,
      "loss": 0.1837,
      "step": 4934
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4764009763643664,
      "learning_rate": 4.0890608015169086e-05,
      "loss": 0.293,
      "step": 4935
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.709862358368562,
      "learning_rate": 4.088677819297806e-05,
      "loss": 0.0884,
      "step": 4936
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7888812357942937,
      "learning_rate": 4.088294774531319e-05,
      "loss": 0.1264,
      "step": 4937
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0087878981095493,
      "learning_rate": 4.0879116672325284e-05,
      "loss": 0.1388,
      "step": 4938
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3857652399822806,
      "learning_rate": 4.087528497416519e-05,
      "loss": 0.2319,
      "step": 4939
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5809268780945835,
      "learning_rate": 4.087145265098372e-05,
      "loss": 0.5153,
      "step": 4940
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8481516468399305,
      "learning_rate": 4.0867619702931806e-05,
      "loss": 0.1419,
      "step": 4941
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6934581368116493,
      "learning_rate": 4.086378613016033e-05,
      "loss": 0.0601,
      "step": 4942
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8055858041833327,
      "learning_rate": 4.085995193282022e-05,
      "loss": 0.094,
      "step": 4943
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2093642317105,
      "learning_rate": 4.085611711106244e-05,
      "loss": 0.3607,
      "step": 4944
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7649901313154857,
      "learning_rate": 4.085228166503796e-05,
      "loss": 0.1098,
      "step": 4945
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1796330903994436,
      "learning_rate": 4.08484455948978e-05,
      "loss": 0.1576,
      "step": 4946
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0337410602417212,
      "learning_rate": 4.084460890079297e-05,
      "loss": 0.2469,
      "step": 4947
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1408767346418396,
      "learning_rate": 4.084077158287454e-05,
      "loss": 0.3046,
      "step": 4948
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3384150379684605,
      "learning_rate": 4.0836933641293576e-05,
      "loss": 0.1114,
      "step": 4949
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0667701258868427,
      "learning_rate": 4.083309507620118e-05,
      "loss": 0.2381,
      "step": 4950
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2698325370552088,
      "learning_rate": 4.082925588774849e-05,
      "loss": 0.2431,
      "step": 4951
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7386685568888292,
      "learning_rate": 4.082541607608664e-05,
      "loss": 0.1559,
      "step": 4952
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7862129678648905,
      "learning_rate": 4.082157564136683e-05,
      "loss": 0.1468,
      "step": 4953
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0574816037070567,
      "learning_rate": 4.0817734583740244e-05,
      "loss": 0.236,
      "step": 4954
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.665587798406701,
      "learning_rate": 4.081389290335811e-05,
      "loss": 0.3868,
      "step": 4955
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9172485132688405,
      "learning_rate": 4.0810050600371675e-05,
      "loss": 0.2086,
      "step": 4956
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2439384468275665,
      "learning_rate": 4.0806207674932225e-05,
      "loss": 0.3471,
      "step": 4957
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5704461805261076,
      "learning_rate": 4.080236412719105e-05,
      "loss": 0.1285,
      "step": 4958
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.765096770322921,
      "learning_rate": 4.079851995729948e-05,
      "loss": 0.1687,
      "step": 4959
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1479423878421622,
      "learning_rate": 4.079467516540884e-05,
      "loss": 0.1369,
      "step": 4960
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7491308047460878,
      "learning_rate": 4.079082975167053e-05,
      "loss": 0.19,
      "step": 4961
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0898526437837976,
      "learning_rate": 4.078698371623593e-05,
      "loss": 0.2293,
      "step": 4962
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5067112291256293,
      "learning_rate": 4.0783137059256474e-05,
      "loss": 0.3515,
      "step": 4963
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8879216208077216,
      "learning_rate": 4.07792897808836e-05,
      "loss": 0.1367,
      "step": 4964
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1352810329118348,
      "learning_rate": 4.0775441881268774e-05,
      "loss": 0.1438,
      "step": 4965
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7408876051735934,
      "learning_rate": 4.0771593360563496e-05,
      "loss": 0.1121,
      "step": 4966
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6393398114784873,
      "learning_rate": 4.076774421891929e-05,
      "loss": 0.1136,
      "step": 4967
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8000994605497571,
      "learning_rate": 4.0763894456487687e-05,
      "loss": 0.1164,
      "step": 4968
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2241675107504586,
      "learning_rate": 4.076004407342027e-05,
      "loss": 0.1825,
      "step": 4969
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8084354058900259,
      "learning_rate": 4.0756193069868606e-05,
      "loss": 0.1784,
      "step": 4970
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7860856622394975,
      "learning_rate": 4.0752341445984346e-05,
      "loss": 0.1546,
      "step": 4971
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9740051626993957,
      "learning_rate": 4.07484892019191e-05,
      "loss": 0.1617,
      "step": 4972
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.14933827506774,
      "learning_rate": 4.0744636337824564e-05,
      "loss": 0.3986,
      "step": 4973
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.327334109920756,
      "learning_rate": 4.074078285385241e-05,
      "loss": 0.5693,
      "step": 4974
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6610839529907934,
      "learning_rate": 4.073692875015434e-05,
      "loss": 0.1307,
      "step": 4975
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1134274699735633,
      "learning_rate": 4.073307402688211e-05,
      "loss": 0.1554,
      "step": 4976
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8974807906717899,
      "learning_rate": 4.072921868418749e-05,
      "loss": 0.1145,
      "step": 4977
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8249463260618757,
      "learning_rate": 4.072536272222225e-05,
      "loss": 0.163,
      "step": 4978
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7242984796077969,
      "learning_rate": 4.0721506141138216e-05,
      "loss": 0.1207,
      "step": 4979
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8349413001625638,
      "learning_rate": 4.071764894108721e-05,
      "loss": 0.1885,
      "step": 4980
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4406190238538552,
      "learning_rate": 4.0713791122221106e-05,
      "loss": 0.5993,
      "step": 4981
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.892325884597196,
      "learning_rate": 4.070993268469178e-05,
      "loss": 0.0831,
      "step": 4982
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9103222313232268,
      "learning_rate": 4.0706073628651156e-05,
      "loss": 0.3111,
      "step": 4983
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6051463025167653,
      "learning_rate": 4.070221395425114e-05,
      "loss": 0.1136,
      "step": 4984
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7636743839534301,
      "learning_rate": 4.0698353661643715e-05,
      "loss": 0.113,
      "step": 4985
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5726561812229922,
      "learning_rate": 4.069449275098086e-05,
      "loss": 0.1213,
      "step": 4986
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0662693529331375,
      "learning_rate": 4.069063122241458e-05,
      "loss": 0.1859,
      "step": 4987
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1017404669245836,
      "learning_rate": 4.0686769076096884e-05,
      "loss": 0.1374,
      "step": 4988
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5481176247178443,
      "learning_rate": 4.068290631217987e-05,
      "loss": 0.0835,
      "step": 4989
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6050653060007312,
      "learning_rate": 4.067904293081558e-05,
      "loss": 0.4125,
      "step": 4990
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8322186794137463,
      "learning_rate": 4.0675178932156146e-05,
      "loss": 0.1667,
      "step": 4991
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.775240659116738,
      "learning_rate": 4.067131431635367e-05,
      "loss": 0.1542,
      "step": 4992
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8937877879051165,
      "learning_rate": 4.0667449083560326e-05,
      "loss": 0.1914,
      "step": 4993
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6238972887848263,
      "learning_rate": 4.066358323392828e-05,
      "loss": 0.0932,
      "step": 4994
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.931609412544066,
      "learning_rate": 4.065971676760975e-05,
      "loss": 0.1266,
      "step": 4995
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6229384392426631,
      "learning_rate": 4.065584968475693e-05,
      "loss": 0.0793,
      "step": 4996
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.314336159615378,
      "learning_rate": 4.065198198552209e-05,
      "loss": 0.2139,
      "step": 4997
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.075792760370577,
      "learning_rate": 4.0648113670057505e-05,
      "loss": 0.5272,
      "step": 4998
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9384302415274646,
      "learning_rate": 4.064424473851548e-05,
      "loss": 0.1522,
      "step": 4999
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6956457874791874,
      "learning_rate": 4.0640375191048305e-05,
      "loss": 0.1144,
      "step": 5000
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6437400929870605,
      "learning_rate": 4.0636505027808366e-05,
      "loss": 0.0937,
      "step": 5001
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6022608021003342,
      "learning_rate": 4.063263424894802e-05,
      "loss": 0.0727,
      "step": 5002
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7001560518325318,
      "learning_rate": 4.062876285461965e-05,
      "loss": 0.1271,
      "step": 5003
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.44740818889242007,
      "learning_rate": 4.062489084497569e-05,
      "loss": 0.0651,
      "step": 5004
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.438465157561425,
      "learning_rate": 4.062101822016858e-05,
      "loss": 0.3735,
      "step": 5005
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8503761653914814,
      "learning_rate": 4.061714498035079e-05,
      "loss": 0.3094,
      "step": 5006
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3565705137196327,
      "learning_rate": 4.0613271125674814e-05,
      "loss": 0.2803,
      "step": 5007
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6965340534537867,
      "learning_rate": 4.0609396656293156e-05,
      "loss": 0.0861,
      "step": 5008
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7601544030195665,
      "learning_rate": 4.0605521572358364e-05,
      "loss": 0.1205,
      "step": 5009
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6978662009491751,
      "learning_rate": 4.0601645874023006e-05,
      "loss": 0.1617,
      "step": 5010
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7249893026556271,
      "learning_rate": 4.059776956143967e-05,
      "loss": 0.1194,
      "step": 5011
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6548197205784997,
      "learning_rate": 4.0593892634760964e-05,
      "loss": 0.1147,
      "step": 5012
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.607165122620156,
      "learning_rate": 4.059001509413953e-05,
      "loss": 0.265,
      "step": 5013
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8288950136762641,
      "learning_rate": 4.058613693972804e-05,
      "loss": 0.0874,
      "step": 5014
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9018503570844759,
      "learning_rate": 4.058225817167915e-05,
      "loss": 0.1202,
      "step": 5015
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6066270479769487,
      "learning_rate": 4.05783787901456e-05,
      "loss": 0.0832,
      "step": 5016
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2408166010102009,
      "learning_rate": 4.0574498795280105e-05,
      "loss": 0.2748,
      "step": 5017
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.15242421442028,
      "learning_rate": 4.0570618187235444e-05,
      "loss": 0.2905,
      "step": 5018
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2736690464421863,
      "learning_rate": 4.056673696616437e-05,
      "loss": 0.3346,
      "step": 5019
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8593374213754874,
      "learning_rate": 4.056285513221971e-05,
      "loss": 0.1049,
      "step": 5020
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.48044848186915856,
      "learning_rate": 4.0558972685554294e-05,
      "loss": 0.0699,
      "step": 5021
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9268733524215712,
      "learning_rate": 4.055508962632097e-05,
      "loss": 0.2166,
      "step": 5022
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6849691964339659,
      "learning_rate": 4.055120595467262e-05,
      "loss": 0.082,
      "step": 5023
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5777363615987128,
      "learning_rate": 4.054732167076215e-05,
      "loss": 0.3369,
      "step": 5024
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2418104452713534,
      "learning_rate": 4.054343677474248e-05,
      "loss": 0.2837,
      "step": 5025
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9157319381091356,
      "learning_rate": 4.0539551266766576e-05,
      "loss": 0.1244,
      "step": 5026
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7532927060567898,
      "learning_rate": 4.053566514698738e-05,
      "loss": 0.1235,
      "step": 5027
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6268549393495831,
      "learning_rate": 4.053177841555793e-05,
      "loss": 0.0923,
      "step": 5028
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.813513557523333,
      "learning_rate": 4.0527891072631216e-05,
      "loss": 0.1323,
      "step": 5029
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9601845181926087,
      "learning_rate": 4.052400311836032e-05,
      "loss": 0.2185,
      "step": 5030
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.15172966221745,
      "learning_rate": 4.0520114552898287e-05,
      "loss": 0.1569,
      "step": 5031
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0897571956189633,
      "learning_rate": 4.0516225376398225e-05,
      "loss": 0.2717,
      "step": 5032
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6793367123147431,
      "learning_rate": 4.0512335589013244e-05,
      "loss": 0.1175,
      "step": 5033
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.070864370911417,
      "learning_rate": 4.0508445190896496e-05,
      "loss": 0.2399,
      "step": 5034
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1404253506195163,
      "learning_rate": 4.050455418220115e-05,
      "loss": 0.368,
      "step": 5035
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7609677743747307,
      "learning_rate": 4.050066256308039e-05,
      "loss": 0.4224,
      "step": 5036
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9723860790138179,
      "learning_rate": 4.0496770333687436e-05,
      "loss": 0.3134,
      "step": 5037
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6435920568516987,
      "learning_rate": 4.049287749417553e-05,
      "loss": 0.0835,
      "step": 5038
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.181116064887443,
      "learning_rate": 4.048898404469793e-05,
      "loss": 0.2378,
      "step": 5039
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6121403628088362,
      "learning_rate": 4.048508998540793e-05,
      "loss": 0.2663,
      "step": 5040
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8443063672315229,
      "learning_rate": 4.048119531645884e-05,
      "loss": 0.1573,
      "step": 5041
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9217509994128977,
      "learning_rate": 4.0477300038004e-05,
      "loss": 0.1334,
      "step": 5042
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0327789594845997,
      "learning_rate": 4.047340415019676e-05,
      "loss": 0.1766,
      "step": 5043
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2265103271631872,
      "learning_rate": 4.046950765319051e-05,
      "loss": 0.3899,
      "step": 5044
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3476765276700355,
      "learning_rate": 4.046561054713866e-05,
      "loss": 0.3404,
      "step": 5045
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6842204206009692,
      "learning_rate": 4.046171283219463e-05,
      "loss": 0.0959,
      "step": 5046
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2669256149892028,
      "learning_rate": 4.0457814508511896e-05,
      "loss": 0.4644,
      "step": 5047
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7728279748184921,
      "learning_rate": 4.045391557624392e-05,
      "loss": 0.1412,
      "step": 5048
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.134573201940106,
      "learning_rate": 4.045001603554421e-05,
      "loss": 0.2007,
      "step": 5049
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6379671846127963,
      "learning_rate": 4.0446115886566306e-05,
      "loss": 0.0985,
      "step": 5050
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2948123408676218,
      "learning_rate": 4.044221512946373e-05,
      "loss": 0.0889,
      "step": 5051
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5009915481171593,
      "learning_rate": 4.04383137643901e-05,
      "loss": 0.0781,
      "step": 5052
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6608270609668062,
      "learning_rate": 4.0434411791498984e-05,
      "loss": 0.0709,
      "step": 5053
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7701352456172199,
      "learning_rate": 4.043050921094401e-05,
      "loss": 0.1621,
      "step": 5054
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0041316765297574,
      "learning_rate": 4.042660602287884e-05,
      "loss": 0.3397,
      "step": 5055
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8165449958199966,
      "learning_rate": 4.0422702227457126e-05,
      "loss": 0.435,
      "step": 5056
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5576868481242239,
      "learning_rate": 4.041879782483258e-05,
      "loss": 0.0642,
      "step": 5057
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8752808793156036,
      "learning_rate": 4.04148928151589e-05,
      "loss": 0.2254,
      "step": 5058
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7854789672193129,
      "learning_rate": 4.041098719858986e-05,
      "loss": 0.0918,
      "step": 5059
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7266780468602169,
      "learning_rate": 4.040708097527919e-05,
      "loss": 0.1151,
      "step": 5060
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3367486473372712,
      "learning_rate": 4.040317414538071e-05,
      "loss": 0.3108,
      "step": 5061
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.913713404676075,
      "learning_rate": 4.039926670904823e-05,
      "loss": 0.2702,
      "step": 5062
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8973342401199482,
      "learning_rate": 4.039535866643558e-05,
      "loss": 0.0883,
      "step": 5063
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9275127831118745,
      "learning_rate": 4.0391450017696617e-05,
      "loss": 0.1441,
      "step": 5064
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7892618559235788,
      "learning_rate": 4.0387540762985244e-05,
      "loss": 0.1435,
      "step": 5065
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4894522330577475,
      "learning_rate": 4.038363090245536e-05,
      "loss": 0.1908,
      "step": 5066
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2860614676703117,
      "learning_rate": 4.037972043626091e-05,
      "loss": 0.325,
      "step": 5067
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8977654376397441,
      "learning_rate": 4.037580936455584e-05,
      "loss": 0.1634,
      "step": 5068
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.127828348634585,
      "learning_rate": 4.037189768749413e-05,
      "loss": 0.1609,
      "step": 5069
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0423301003134624,
      "learning_rate": 4.03679854052298e-05,
      "loss": 0.3233,
      "step": 5070
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8436520402013797,
      "learning_rate": 4.0364072517916865e-05,
      "loss": 0.1296,
      "step": 5071
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7233966846629503,
      "learning_rate": 4.036015902570939e-05,
      "loss": 0.1067,
      "step": 5072
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6902369763289582,
      "learning_rate": 4.035624492876144e-05,
      "loss": 0.1325,
      "step": 5073
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0368127529157551,
      "learning_rate": 4.035233022722712e-05,
      "loss": 0.1784,
      "step": 5074
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.894489732414105,
      "learning_rate": 4.0348414921260565e-05,
      "loss": 0.6653,
      "step": 5075
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8077388903736703,
      "learning_rate": 4.034449901101591e-05,
      "loss": 0.1746,
      "step": 5076
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8890661736846592,
      "learning_rate": 4.034058249664733e-05,
      "loss": 0.2342,
      "step": 5077
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5988118038550182,
      "learning_rate": 4.0336665378309025e-05,
      "loss": 0.1306,
      "step": 5078
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.903837405636659,
      "learning_rate": 4.0332747656155214e-05,
      "loss": 0.2576,
      "step": 5079
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.774483207999381,
      "learning_rate": 4.032882933034014e-05,
      "loss": 0.11,
      "step": 5080
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3000524408867113,
      "learning_rate": 4.032491040101806e-05,
      "loss": 0.3028,
      "step": 5081
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8115780704167794,
      "learning_rate": 4.032099086834328e-05,
      "loss": 0.1683,
      "step": 5082
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6248870835970288,
      "learning_rate": 4.0317070732470107e-05,
      "loss": 0.1084,
      "step": 5083
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9796610156388479,
      "learning_rate": 4.031314999355289e-05,
      "loss": 0.1768,
      "step": 5084
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7633052390056834,
      "learning_rate": 4.030922865174598e-05,
      "loss": 0.0862,
      "step": 5085
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9123327060147348,
      "learning_rate": 4.0305306707203764e-05,
      "loss": 0.1255,
      "step": 5086
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9217642179925624,
      "learning_rate": 4.030138416008066e-05,
      "loss": 0.1708,
      "step": 5087
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5373914212234931,
      "learning_rate": 4.029746101053109e-05,
      "loss": 0.0733,
      "step": 5088
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7361769026423056,
      "learning_rate": 4.0293537258709515e-05,
      "loss": 0.0556,
      "step": 5089
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8256856904219583,
      "learning_rate": 4.028961290477042e-05,
      "loss": 0.0858,
      "step": 5090
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6431977460106135,
      "learning_rate": 4.028568794886831e-05,
      "loss": 0.1312,
      "step": 5091
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9347243689972954,
      "learning_rate": 4.028176239115771e-05,
      "loss": 0.194,
      "step": 5092
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9195834463996174,
      "learning_rate": 4.027783623179316e-05,
      "loss": 0.1931,
      "step": 5093
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3796177790735713,
      "learning_rate": 4.0273909470929264e-05,
      "loss": 0.211,
      "step": 5094
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9999604687648435,
      "learning_rate": 4.02699821087206e-05,
      "loss": 0.3841,
      "step": 5095
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8873831686764765,
      "learning_rate": 4.0266054145321795e-05,
      "loss": 0.2835,
      "step": 5096
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0737821484445347,
      "learning_rate": 4.02621255808875e-05,
      "loss": 0.17,
      "step": 5097
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2519427100974094,
      "learning_rate": 4.0258196415572385e-05,
      "loss": 0.1843,
      "step": 5098
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1322707398914196,
      "learning_rate": 4.0254266649531134e-05,
      "loss": 0.2311,
      "step": 5099
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0096044022413786,
      "learning_rate": 4.0250336282918486e-05,
      "loss": 0.1874,
      "step": 5100
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2888031106885431,
      "learning_rate": 4.0246405315889156e-05,
      "loss": 0.1942,
      "step": 5101
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7911532503320107,
      "learning_rate": 4.024247374859793e-05,
      "loss": 0.0958,
      "step": 5102
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7755697350627474,
      "learning_rate": 4.023854158119958e-05,
      "loss": 0.209,
      "step": 5103
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9570624546116717,
      "learning_rate": 4.023460881384894e-05,
      "loss": 0.1454,
      "step": 5104
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9204141246451178,
      "learning_rate": 4.023067544670082e-05,
      "loss": 0.2318,
      "step": 5105
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5379931092919445,
      "learning_rate": 4.02267414799101e-05,
      "loss": 0.0424,
      "step": 5106
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.346412098730457,
      "learning_rate": 4.022280691363166e-05,
      "loss": 0.3205,
      "step": 5107
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6211926067661545,
      "learning_rate": 4.0218871748020396e-05,
      "loss": 0.0995,
      "step": 5108
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8115829339377182,
      "learning_rate": 4.021493598323124e-05,
      "loss": 0.2347,
      "step": 5109
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.518059587426517,
      "learning_rate": 4.021099961941916e-05,
      "loss": 0.3558,
      "step": 5110
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.114153432925803,
      "learning_rate": 4.02070626567391e-05,
      "loss": 0.1558,
      "step": 5111
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7102640653905441,
      "learning_rate": 4.020312509534611e-05,
      "loss": 0.0999,
      "step": 5112
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8765850884386132,
      "learning_rate": 4.019918693539518e-05,
      "loss": 0.2204,
      "step": 5113
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0667712300383814,
      "learning_rate": 4.019524817704137e-05,
      "loss": 0.3181,
      "step": 5114
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9982405229282008,
      "learning_rate": 4.019130882043974e-05,
      "loss": 0.6719,
      "step": 5115
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9210648831102434,
      "learning_rate": 4.018736886574541e-05,
      "loss": 0.1664,
      "step": 5116
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9203134283386613,
      "learning_rate": 4.018342831311347e-05,
      "loss": 0.1381,
      "step": 5117
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.912674490952148,
      "learning_rate": 4.0179487162699086e-05,
      "loss": 0.2218,
      "step": 5118
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9468130366210679,
      "learning_rate": 4.017554541465741e-05,
      "loss": 0.4575,
      "step": 5119
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3699457445007746,
      "learning_rate": 4.017160306914364e-05,
      "loss": 0.2501,
      "step": 5120
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.56290305569588,
      "learning_rate": 4.0167660126312964e-05,
      "loss": 0.3914,
      "step": 5121
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.823609827277096,
      "learning_rate": 4.016371658632066e-05,
      "loss": 0.1915,
      "step": 5122
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7723265550706394,
      "learning_rate": 4.015977244932196e-05,
      "loss": 0.111,
      "step": 5123
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7633788747192033,
      "learning_rate": 4.015582771547215e-05,
      "loss": 0.141,
      "step": 5124
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9227127413927583,
      "learning_rate": 4.015188238492655e-05,
      "loss": 0.1404,
      "step": 5125
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7614052311135644,
      "learning_rate": 4.014793645784047e-05,
      "loss": 0.1489,
      "step": 5126
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7715749561898227,
      "learning_rate": 4.0143989934369285e-05,
      "loss": 0.1273,
      "step": 5127
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3894618865678994,
      "learning_rate": 4.014004281466836e-05,
      "loss": 0.1241,
      "step": 5128
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.820572065302051,
      "learning_rate": 4.01360950988931e-05,
      "loss": 0.1342,
      "step": 5129
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.788981437530349,
      "learning_rate": 4.013214678719893e-05,
      "loss": 0.099,
      "step": 5130
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7353616337055814,
      "learning_rate": 4.0128197879741285e-05,
      "loss": 0.1196,
      "step": 5131
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8916410008822068,
      "learning_rate": 4.0124248376675656e-05,
      "loss": 0.1185,
      "step": 5132
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7773285844040143,
      "learning_rate": 4.012029827815753e-05,
      "loss": 0.1024,
      "step": 5133
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9931516275836064,
      "learning_rate": 4.0116347584342426e-05,
      "loss": 0.1814,
      "step": 5134
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7728964471322172,
      "learning_rate": 4.0112396295385876e-05,
      "loss": 0.111,
      "step": 5135
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9668773314432714,
      "learning_rate": 4.010844441144346e-05,
      "loss": 0.1732,
      "step": 5136
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8448151421924811,
      "learning_rate": 4.0104491932670754e-05,
      "loss": 0.2005,
      "step": 5137
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6979413803410347,
      "learning_rate": 4.010053885922339e-05,
      "loss": 0.1103,
      "step": 5138
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5965106357185828,
      "learning_rate": 4.0096585191256974e-05,
      "loss": 0.0802,
      "step": 5139
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.136271894974768,
      "learning_rate": 4.009263092892718e-05,
      "loss": 0.0793,
      "step": 5140
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0356222826033117,
      "learning_rate": 4.0088676072389694e-05,
      "loss": 0.1127,
      "step": 5141
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2903267363319983,
      "learning_rate": 4.008472062180022e-05,
      "loss": 0.3185,
      "step": 5142
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9439336967039403,
      "learning_rate": 4.008076457731448e-05,
      "loss": 0.1735,
      "step": 5143
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9420722808607855,
      "learning_rate": 4.007680793908823e-05,
      "loss": 0.1598,
      "step": 5144
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3707958416899675,
      "learning_rate": 4.007285070727725e-05,
      "loss": 0.5737,
      "step": 5145
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6999336473201803,
      "learning_rate": 4.006889288203733e-05,
      "loss": 0.1383,
      "step": 5146
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9490343984532145,
      "learning_rate": 4.006493446352431e-05,
      "loss": 0.2107,
      "step": 5147
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6611936277379369,
      "learning_rate": 4.006097545189401e-05,
      "loss": 0.1299,
      "step": 5148
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7305280365730907,
      "learning_rate": 4.005701584730232e-05,
      "loss": 0.1923,
      "step": 5149
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5985405685326856,
      "learning_rate": 4.005305564990512e-05,
      "loss": 0.0712,
      "step": 5150
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9690450142198104,
      "learning_rate": 4.004909485985834e-05,
      "loss": 0.455,
      "step": 5151
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0328801110257744,
      "learning_rate": 4.0045133477317896e-05,
      "loss": 0.2201,
      "step": 5152
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.434627521045557,
      "learning_rate": 4.004117150243978e-05,
      "loss": 0.3387,
      "step": 5153
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7560466635590106,
      "learning_rate": 4.003720893537995e-05,
      "loss": 0.4746,
      "step": 5154
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5317149672502386,
      "learning_rate": 4.003324577629443e-05,
      "loss": 0.1029,
      "step": 5155
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7370514572528332,
      "learning_rate": 4.002928202533925e-05,
      "loss": 0.1343,
      "step": 5156
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.915973437575076,
      "learning_rate": 4.002531768267047e-05,
      "loss": 0.3829,
      "step": 5157
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5619036772299135,
      "learning_rate": 4.0021352748444165e-05,
      "loss": 0.1162,
      "step": 5158
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.912927513860164,
      "learning_rate": 4.001738722281644e-05,
      "loss": 0.1796,
      "step": 5159
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4609632379717559,
      "learning_rate": 4.001342110594341e-05,
      "loss": 0.307,
      "step": 5160
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0275675043621006,
      "learning_rate": 4.000945439798123e-05,
      "loss": 0.18,
      "step": 5161
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1082878744267388,
      "learning_rate": 4.000548709908607e-05,
      "loss": 0.192,
      "step": 5162
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6315352013279389,
      "learning_rate": 4.000151920941414e-05,
      "loss": 0.0953,
      "step": 5163
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2613726811609416,
      "learning_rate": 3.999755072912164e-05,
      "loss": 0.3334,
      "step": 5164
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5232163226697204,
      "learning_rate": 3.999358165836482e-05,
      "loss": 0.0517,
      "step": 5165
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4716394713768408,
      "learning_rate": 3.9989611997299944e-05,
      "loss": 0.3044,
      "step": 5166
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8144213894325049,
      "learning_rate": 3.9985641746083314e-05,
      "loss": 0.1433,
      "step": 5167
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9123339446956178,
      "learning_rate": 3.998167090487122e-05,
      "loss": 0.1823,
      "step": 5168
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7354745669127039,
      "learning_rate": 3.997769947382e-05,
      "loss": 0.1924,
      "step": 5169
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1411191821548918,
      "learning_rate": 3.997372745308602e-05,
      "loss": 0.2585,
      "step": 5170
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3100409324265188,
      "learning_rate": 3.996975484282566e-05,
      "loss": 0.304,
      "step": 5171
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9444127328195944,
      "learning_rate": 3.996578164319532e-05,
      "loss": 0.7306,
      "step": 5172
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8729431879669924,
      "learning_rate": 3.996180785435144e-05,
      "loss": 0.1675,
      "step": 5173
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.31195889473287,
      "learning_rate": 3.995783347645046e-05,
      "loss": 0.7995,
      "step": 5174
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.119622534876205,
      "learning_rate": 3.9953858509648844e-05,
      "loss": 0.1185,
      "step": 5175
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.064182011225953,
      "learning_rate": 3.9949882954103115e-05,
      "loss": 0.2292,
      "step": 5176
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0841991701761646,
      "learning_rate": 3.994590680996978e-05,
      "loss": 0.2562,
      "step": 5177
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8602777481956709,
      "learning_rate": 3.9941930077405374e-05,
      "loss": 0.1008,
      "step": 5178
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5287979246489023,
      "learning_rate": 3.993795275656648e-05,
      "loss": 0.2218,
      "step": 5179
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.958620445961172,
      "learning_rate": 3.9933974847609665e-05,
      "loss": 0.1443,
      "step": 5180
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8091704732391559,
      "learning_rate": 3.992999635069157e-05,
      "loss": 0.1168,
      "step": 5181
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5154705646373547,
      "learning_rate": 3.9926017265968816e-05,
      "loss": 0.0547,
      "step": 5182
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0072987724348745,
      "learning_rate": 3.992203759359807e-05,
      "loss": 0.2553,
      "step": 5183
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8985772978066908,
      "learning_rate": 3.9918057333736005e-05,
      "loss": 0.1488,
      "step": 5184
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6362982085540515,
      "learning_rate": 3.9914076486539336e-05,
      "loss": 0.1018,
      "step": 5185
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4255161384503978,
      "learning_rate": 3.9910095052164784e-05,
      "loss": 0.2776,
      "step": 5186
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.384995384999382,
      "learning_rate": 3.9906113030769105e-05,
      "loss": 0.3863,
      "step": 5187
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8149409550534106,
      "learning_rate": 3.990213042250907e-05,
      "loss": 0.1812,
      "step": 5188
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7574068445242383,
      "learning_rate": 3.989814722754148e-05,
      "loss": 0.1515,
      "step": 5189
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0623145862455032,
      "learning_rate": 3.9894163446023156e-05,
      "loss": 0.2148,
      "step": 5190
    },
    {
      "epoch": 0.32,
      "grad_norm": 21.094200838435935,
      "learning_rate": 3.9890179078110955e-05,
      "loss": 1.4162,
      "step": 5191
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.8529635709917,
      "learning_rate": 3.9886194123961715e-05,
      "loss": 0.2032,
      "step": 5192
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3887449719539453,
      "learning_rate": 3.988220858373236e-05,
      "loss": 0.0777,
      "step": 5193
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.5504021142636084,
      "learning_rate": 3.9878222457579776e-05,
      "loss": 0.1952,
      "step": 5194
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3892658297365836,
      "learning_rate": 3.987423574566091e-05,
      "loss": 0.1842,
      "step": 5195
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.087312666295512,
      "learning_rate": 3.987024844813273e-05,
      "loss": 0.2735,
      "step": 5196
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9575190402793986,
      "learning_rate": 3.986626056515221e-05,
      "loss": 0.1521,
      "step": 5197
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8592376630773497,
      "learning_rate": 3.9862272096876354e-05,
      "loss": 0.1389,
      "step": 5198
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0334817089172337,
      "learning_rate": 3.98582830434622e-05,
      "loss": 0.1266,
      "step": 5199
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.903656144601999,
      "learning_rate": 3.985429340506679e-05,
      "loss": 0.1967,
      "step": 5200
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6722227870207746,
      "learning_rate": 3.98503031818472e-05,
      "loss": 0.0652,
      "step": 5201
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7826444933424624,
      "learning_rate": 3.984631237396053e-05,
      "loss": 0.1367,
      "step": 5202
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6747371235352144,
      "learning_rate": 3.984232098156391e-05,
      "loss": 0.2316,
      "step": 5203
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8664551318575964,
      "learning_rate": 3.983832900481447e-05,
      "loss": 0.1481,
      "step": 5204
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7996259140776335,
      "learning_rate": 3.9834336443869385e-05,
      "loss": 0.1145,
      "step": 5205
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6688351566567758,
      "learning_rate": 3.983034329888584e-05,
      "loss": 0.4895,
      "step": 5206
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8190033800192167,
      "learning_rate": 3.9826349570021057e-05,
      "loss": 0.1234,
      "step": 5207
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.3797504362116,
      "learning_rate": 3.9822355257432256e-05,
      "loss": 0.4985,
      "step": 5208
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1563596673112715,
      "learning_rate": 3.981836036127672e-05,
      "loss": 0.1663,
      "step": 5209
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0399079015881632,
      "learning_rate": 3.98143648817117e-05,
      "loss": 0.209,
      "step": 5210
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.2869818402407756,
      "learning_rate": 3.9810368818894526e-05,
      "loss": 1.3272,
      "step": 5211
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3038957961260476,
      "learning_rate": 3.980637217298252e-05,
      "loss": 0.1764,
      "step": 5212
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9919902820213977,
      "learning_rate": 3.980237494413302e-05,
      "loss": 0.4266,
      "step": 5213
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.55796469958715,
      "learning_rate": 3.979837713250342e-05,
      "loss": 0.3421,
      "step": 5214
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9292273817024315,
      "learning_rate": 3.979437873825111e-05,
      "loss": 0.1416,
      "step": 5215
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4998255717037137,
      "learning_rate": 3.9790379761533496e-05,
      "loss": 0.3012,
      "step": 5216
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0656015636268792,
      "learning_rate": 3.9786380202508035e-05,
      "loss": 0.2946,
      "step": 5217
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9604578371626404,
      "learning_rate": 3.978238006133219e-05,
      "loss": 0.1848,
      "step": 5218
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7296908550103655,
      "learning_rate": 3.9778379338163445e-05,
      "loss": 0.1389,
      "step": 5219
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2151267936104835,
      "learning_rate": 3.977437803315932e-05,
      "loss": 0.2587,
      "step": 5220
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9111417359246446,
      "learning_rate": 3.9770376146477335e-05,
      "loss": 0.1115,
      "step": 5221
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4867966095164482,
      "learning_rate": 3.976637367827506e-05,
      "loss": 0.1638,
      "step": 5222
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1318236751610542,
      "learning_rate": 3.9762370628710065e-05,
      "loss": 0.2434,
      "step": 5223
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7665647415838175,
      "learning_rate": 3.975836699793997e-05,
      "loss": 0.1046,
      "step": 5224
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0936435495696122,
      "learning_rate": 3.975436278612238e-05,
      "loss": 0.2532,
      "step": 5225
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7069828245451202,
      "learning_rate": 3.975035799341495e-05,
      "loss": 0.1144,
      "step": 5226
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.529779450262127,
      "learning_rate": 3.974635261997537e-05,
      "loss": 0.3345,
      "step": 5227
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3372379093759548,
      "learning_rate": 3.9742346665961304e-05,
      "loss": 0.3336,
      "step": 5228
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0998907002815548,
      "learning_rate": 3.9738340131530496e-05,
      "loss": 0.1769,
      "step": 5229
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.367805496312541,
      "learning_rate": 3.973433301684066e-05,
      "loss": 0.3961,
      "step": 5230
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8258171580191542,
      "learning_rate": 3.973032532204959e-05,
      "loss": 0.1039,
      "step": 5231
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1240679901416877,
      "learning_rate": 3.972631704731504e-05,
      "loss": 0.2547,
      "step": 5232
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6471381897417017,
      "learning_rate": 3.972230819279484e-05,
      "loss": 0.1111,
      "step": 5233
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3986609447352965,
      "learning_rate": 3.9718298758646824e-05,
      "loss": 0.2517,
      "step": 5234
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1453082599749604,
      "learning_rate": 3.971428874502883e-05,
      "loss": 0.1941,
      "step": 5235
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7915856316304072,
      "learning_rate": 3.9710278152098736e-05,
      "loss": 0.1132,
      "step": 5236
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.779332247667883,
      "learning_rate": 3.9706266980014453e-05,
      "loss": 0.3398,
      "step": 5237
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6399362391362698,
      "learning_rate": 3.97022552289339e-05,
      "loss": 0.0986,
      "step": 5238
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.147450594734272,
      "learning_rate": 3.969824289901503e-05,
      "loss": 0.1964,
      "step": 5239
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2723892744943008,
      "learning_rate": 3.9694229990415794e-05,
      "loss": 0.2984,
      "step": 5240
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9743913125063888,
      "learning_rate": 3.9690216503294196e-05,
      "loss": 0.2163,
      "step": 5241
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.407222654878027,
      "learning_rate": 3.968620243780824e-05,
      "loss": 0.281,
      "step": 5242
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.18978646446063,
      "learning_rate": 3.9682187794115975e-05,
      "loss": 0.2523,
      "step": 5243
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.162317395116008,
      "learning_rate": 3.967817257237545e-05,
      "loss": 0.2739,
      "step": 5244
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7689861979151553,
      "learning_rate": 3.967415677274476e-05,
      "loss": 0.2087,
      "step": 5245
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9819071233364312,
      "learning_rate": 3.967014039538199e-05,
      "loss": 0.0972,
      "step": 5246
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6745231670520807,
      "learning_rate": 3.9666123440445294e-05,
      "loss": 0.1492,
      "step": 5247
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5651969304356708,
      "learning_rate": 3.9662105908092796e-05,
      "loss": 0.1123,
      "step": 5248
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7592807014260731,
      "learning_rate": 3.965808779848268e-05,
      "loss": 0.1581,
      "step": 5249
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7122804169266617,
      "learning_rate": 3.965406911177314e-05,
      "loss": 0.1325,
      "step": 5250
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.960225823142905,
      "learning_rate": 3.96500498481224e-05,
      "loss": 0.2281,
      "step": 5251
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8491533260641861,
      "learning_rate": 3.96460300076887e-05,
      "loss": 0.1027,
      "step": 5252
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8979966395918948,
      "learning_rate": 3.964200959063031e-05,
      "loss": 0.2679,
      "step": 5253
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7127752169738751,
      "learning_rate": 3.96379885971055e-05,
      "loss": 0.2166,
      "step": 5254
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8315564464719623,
      "learning_rate": 3.963396702727259e-05,
      "loss": 0.1865,
      "step": 5255
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1985902551628604,
      "learning_rate": 3.9629944881289906e-05,
      "loss": 0.2717,
      "step": 5256
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6495320462631836,
      "learning_rate": 3.962592215931581e-05,
      "loss": 0.0767,
      "step": 5257
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8215137720836305,
      "learning_rate": 3.9621898861508675e-05,
      "loss": 0.1267,
      "step": 5258
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8078941414047078,
      "learning_rate": 3.961787498802691e-05,
      "loss": 0.1479,
      "step": 5259
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9809892164450917,
      "learning_rate": 3.961385053902892e-05,
      "loss": 0.2289,
      "step": 5260
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6216194873083879,
      "learning_rate": 3.960982551467317e-05,
      "loss": 0.0771,
      "step": 5261
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8200809932853447,
      "learning_rate": 3.9605799915118116e-05,
      "loss": 0.1725,
      "step": 5262
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9411773828873643,
      "learning_rate": 3.960177374052225e-05,
      "loss": 0.1845,
      "step": 5263
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6849093325973926,
      "learning_rate": 3.959774699104408e-05,
      "loss": 0.1145,
      "step": 5264
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6404243752952371,
      "learning_rate": 3.9593719666842163e-05,
      "loss": 0.1431,
      "step": 5265
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6921304666187286,
      "learning_rate": 3.958969176807503e-05,
      "loss": 0.0895,
      "step": 5266
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7259028957789841,
      "learning_rate": 3.958566329490129e-05,
      "loss": 0.0986,
      "step": 5267
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1249231689678674,
      "learning_rate": 3.958163424747953e-05,
      "loss": 0.2269,
      "step": 5268
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1266137703332118,
      "learning_rate": 3.9577604625968376e-05,
      "loss": 0.2654,
      "step": 5269
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.742105365814217,
      "learning_rate": 3.9573574430526477e-05,
      "loss": 0.1918,
      "step": 5270
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0439153736396654,
      "learning_rate": 3.9569543661312516e-05,
      "loss": 0.3219,
      "step": 5271
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2904660540720614,
      "learning_rate": 3.956551231848517e-05,
      "loss": 0.2493,
      "step": 5272
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8964478296152159,
      "learning_rate": 3.956148040220317e-05,
      "loss": 0.1683,
      "step": 5273
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8205635400513607,
      "learning_rate": 3.955744791262526e-05,
      "loss": 0.1054,
      "step": 5274
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0068741277452704,
      "learning_rate": 3.955341484991018e-05,
      "loss": 0.1841,
      "step": 5275
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8974004179218842,
      "learning_rate": 3.954938121421673e-05,
      "loss": 0.1978,
      "step": 5276
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2987275165666758,
      "learning_rate": 3.9545347005703714e-05,
      "loss": 0.3226,
      "step": 5277
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8404192964452885,
      "learning_rate": 3.954131222452996e-05,
      "loss": 0.2136,
      "step": 5278
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.9255703394856005,
      "learning_rate": 3.953727687085433e-05,
      "loss": 0.4071,
      "step": 5279
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.915680853539577,
      "learning_rate": 3.953324094483569e-05,
      "loss": 0.874,
      "step": 5280
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9697476248758943,
      "learning_rate": 3.952920444663293e-05,
      "loss": 0.1239,
      "step": 5281
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0849469945324797,
      "learning_rate": 3.9525167376404984e-05,
      "loss": 0.2195,
      "step": 5282
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3316405601504155,
      "learning_rate": 3.952112973431079e-05,
      "loss": 0.1497,
      "step": 5283
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5772550405614941,
      "learning_rate": 3.9517091520509305e-05,
      "loss": 0.0923,
      "step": 5284
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.781717891014503,
      "learning_rate": 3.951305273515953e-05,
      "loss": 0.2029,
      "step": 5285
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1395482131307437,
      "learning_rate": 3.9509013378420466e-05,
      "loss": 0.3258,
      "step": 5286
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.247214641689686,
      "learning_rate": 3.950497345045115e-05,
      "loss": 0.2268,
      "step": 5287
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3042577994118811,
      "learning_rate": 3.950093295141063e-05,
      "loss": 0.304,
      "step": 5288
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7854299983921706,
      "learning_rate": 3.9496891881457986e-05,
      "loss": 0.1292,
      "step": 5289
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7150978547427383,
      "learning_rate": 3.949285024075232e-05,
      "loss": 0.1475,
      "step": 5290
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.107926294869808,
      "learning_rate": 3.948880802945275e-05,
      "loss": 0.3398,
      "step": 5291
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9443423317768787,
      "learning_rate": 3.9484765247718434e-05,
      "loss": 0.1856,
      "step": 5292
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0274180802058086,
      "learning_rate": 3.9480721895708516e-05,
      "loss": 0.2999,
      "step": 5293
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0098390411540716,
      "learning_rate": 3.947667797358221e-05,
      "loss": 0.3078,
      "step": 5294
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7991773114325805,
      "learning_rate": 3.9472633481498714e-05,
      "loss": 0.1737,
      "step": 5295
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4185278771576904,
      "learning_rate": 3.9468588419617267e-05,
      "loss": 0.406,
      "step": 5296
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8285221467124404,
      "learning_rate": 3.946454278809712e-05,
      "loss": 0.8251,
      "step": 5297
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1042650931491513,
      "learning_rate": 3.946049658709756e-05,
      "loss": 0.589,
      "step": 5298
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8688893713119705,
      "learning_rate": 3.9456449816777884e-05,
      "loss": 0.1341,
      "step": 5299
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5520022741776163,
      "learning_rate": 3.945240247729742e-05,
      "loss": 0.1869,
      "step": 5300
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3897801430276304,
      "learning_rate": 3.944835456881552e-05,
      "loss": 0.1802,
      "step": 5301
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9368699841738487,
      "learning_rate": 3.944430609149153e-05,
      "loss": 0.1734,
      "step": 5302
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6388676185804087,
      "learning_rate": 3.944025704548487e-05,
      "loss": 0.0831,
      "step": 5303
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7250935178258421,
      "learning_rate": 3.9436207430954933e-05,
      "loss": 0.1299,
      "step": 5304
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6576912726282338,
      "learning_rate": 3.943215724806117e-05,
      "loss": 0.0882,
      "step": 5305
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9090429041093518,
      "learning_rate": 3.942810649696304e-05,
      "loss": 0.1148,
      "step": 5306
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0361612160532336,
      "learning_rate": 3.9424055177819996e-05,
      "loss": 0.1078,
      "step": 5307
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0298170719044886,
      "learning_rate": 3.9420003290791575e-05,
      "loss": 0.2203,
      "step": 5308
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2157506761060026,
      "learning_rate": 3.941595083603729e-05,
      "loss": 0.2112,
      "step": 5309
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7975837539900114,
      "learning_rate": 3.941189781371669e-05,
      "loss": 0.1188,
      "step": 5310
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5041233748605651,
      "learning_rate": 3.940784422398934e-05,
      "loss": 0.2518,
      "step": 5311
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6498911466293114,
      "learning_rate": 3.940379006701483e-05,
      "loss": 0.1037,
      "step": 5312
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7741399361398431,
      "learning_rate": 3.9399735342952796e-05,
      "loss": 0.0799,
      "step": 5313
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3597628969848254,
      "learning_rate": 3.939568005196286e-05,
      "loss": 0.3662,
      "step": 5314
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.076813843897054,
      "learning_rate": 3.939162419420468e-05,
      "loss": 0.212,
      "step": 5315
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9371071431268531,
      "learning_rate": 3.938756776983794e-05,
      "loss": 0.0615,
      "step": 5316
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9014437766510206,
      "learning_rate": 3.938351077902235e-05,
      "loss": 0.1678,
      "step": 5317
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.666191703169951,
      "learning_rate": 3.937945322191763e-05,
      "loss": 0.1153,
      "step": 5318
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6085176381708275,
      "learning_rate": 3.9375395098683535e-05,
      "loss": 0.0669,
      "step": 5319
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3057793165308171,
      "learning_rate": 3.937133640947982e-05,
      "loss": 0.1533,
      "step": 5320
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.456380560361455,
      "learning_rate": 3.9367277154466306e-05,
      "loss": 0.3697,
      "step": 5321
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6984356860213139,
      "learning_rate": 3.9363217333802785e-05,
      "loss": 0.1248,
      "step": 5322
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6262942273537003,
      "learning_rate": 3.93591569476491e-05,
      "loss": 0.1145,
      "step": 5323
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8795479651554707,
      "learning_rate": 3.9355095996165124e-05,
      "loss": 0.1025,
      "step": 5324
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7807854671225118,
      "learning_rate": 3.935103447951073e-05,
      "loss": 0.1716,
      "step": 5325
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.579166172972397,
      "learning_rate": 3.9346972397845823e-05,
      "loss": 0.176,
      "step": 5326
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9504420988647423,
      "learning_rate": 3.9342909751330324e-05,
      "loss": 0.2011,
      "step": 5327
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1843425846801061,
      "learning_rate": 3.9338846540124195e-05,
      "loss": 0.2601,
      "step": 5328
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9112568676652336,
      "learning_rate": 3.93347827643874e-05,
      "loss": 0.1298,
      "step": 5329
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0055877351057483,
      "learning_rate": 3.933071842427994e-05,
      "loss": 0.3427,
      "step": 5330
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.9005860290321297,
      "learning_rate": 3.9326653519961816e-05,
      "loss": 0.4903,
      "step": 5331
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8184746536220668,
      "learning_rate": 3.932258805159308e-05,
      "loss": 0.129,
      "step": 5332
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8921352279025329,
      "learning_rate": 3.9318522019333784e-05,
      "loss": 0.1954,
      "step": 5333
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.535427596765208,
      "learning_rate": 3.931445542334402e-05,
      "loss": 0.2826,
      "step": 5334
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0071369017826166,
      "learning_rate": 3.9310388263783884e-05,
      "loss": 0.1072,
      "step": 5335
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9603996497005881,
      "learning_rate": 3.930632054081351e-05,
      "loss": 0.0766,
      "step": 5336
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7851664827426381,
      "learning_rate": 3.930225225459304e-05,
      "loss": 0.3262,
      "step": 5337
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9562792727468604,
      "learning_rate": 3.929818340528265e-05,
      "loss": 0.1996,
      "step": 5338
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1071352091948432,
      "learning_rate": 3.9294113993042524e-05,
      "loss": 0.3852,
      "step": 5339
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.698551851141821,
      "learning_rate": 3.9290044018032896e-05,
      "loss": 0.1379,
      "step": 5340
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.963146193998486,
      "learning_rate": 3.928597348041398e-05,
      "loss": 0.455,
      "step": 5341
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8642687890127162,
      "learning_rate": 3.928190238034607e-05,
      "loss": 0.1999,
      "step": 5342
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.657587956086842,
      "learning_rate": 3.9277830717989416e-05,
      "loss": 0.1058,
      "step": 5343
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4860009880362762,
      "learning_rate": 3.927375849350434e-05,
      "loss": 0.0653,
      "step": 5344
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6425849171815234,
      "learning_rate": 3.926968570705115e-05,
      "loss": 0.0938,
      "step": 5345
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7540192528362772,
      "learning_rate": 3.926561235879022e-05,
      "loss": 0.1484,
      "step": 5346
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9944831030772409,
      "learning_rate": 3.92615384488819e-05,
      "loss": 0.2117,
      "step": 5347
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.095960256033781,
      "learning_rate": 3.92574639774866e-05,
      "loss": 0.2181,
      "step": 5348
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0852512968084835,
      "learning_rate": 3.925338894476471e-05,
      "loss": 0.149,
      "step": 5349
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1845097347345317,
      "learning_rate": 3.9249313350876696e-05,
      "loss": 0.2143,
      "step": 5350
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.684089372444365,
      "learning_rate": 3.9245237195982996e-05,
      "loss": 0.1524,
      "step": 5351
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6579503104917633,
      "learning_rate": 3.9241160480244097e-05,
      "loss": 0.1123,
      "step": 5352
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5461039114913463,
      "learning_rate": 3.923708320382051e-05,
      "loss": 0.0722,
      "step": 5353
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6811288315119999,
      "learning_rate": 3.9233005366872754e-05,
      "loss": 0.1277,
      "step": 5354
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7017297854799647,
      "learning_rate": 3.922892696956138e-05,
      "loss": 0.1875,
      "step": 5355
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6010677832451006,
      "learning_rate": 3.9224848012046945e-05,
      "loss": 0.129,
      "step": 5356
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9458688301096225,
      "learning_rate": 3.922076849449004e-05,
      "loss": 0.1635,
      "step": 5357
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6719117390724915,
      "learning_rate": 3.921668841705131e-05,
      "loss": 0.1048,
      "step": 5358
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.569669118427115,
      "learning_rate": 3.921260777989135e-05,
      "loss": 0.0833,
      "step": 5359
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9351458841382679,
      "learning_rate": 3.920852658317085e-05,
      "loss": 0.159,
      "step": 5360
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0956932313158914,
      "learning_rate": 3.9204444827050465e-05,
      "loss": 0.1486,
      "step": 5361
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0739819649307532,
      "learning_rate": 3.920036251169091e-05,
      "loss": 0.3207,
      "step": 5362
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1746309155476977,
      "learning_rate": 3.919627963725291e-05,
      "loss": 0.3102,
      "step": 5363
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7494505106747075,
      "learning_rate": 3.9192196203897205e-05,
      "loss": 0.0922,
      "step": 5364
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1327143535461555,
      "learning_rate": 3.9188112211784554e-05,
      "loss": 0.4194,
      "step": 5365
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.727001985144557,
      "learning_rate": 3.918402766107577e-05,
      "loss": 0.1298,
      "step": 5366
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9801326530157651,
      "learning_rate": 3.9179942551931645e-05,
      "loss": 0.2138,
      "step": 5367
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0988391412555836,
      "learning_rate": 3.9175856884513015e-05,
      "loss": 0.3149,
      "step": 5368
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7041147476433393,
      "learning_rate": 3.917177065898076e-05,
      "loss": 0.1248,
      "step": 5369
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7217407427076326,
      "learning_rate": 3.916768387549572e-05,
      "loss": 0.1269,
      "step": 5370
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.632897550808923,
      "learning_rate": 3.9163596534218816e-05,
      "loss": 0.1277,
      "step": 5371
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7805411845278186,
      "learning_rate": 3.915950863531096e-05,
      "loss": 0.1264,
      "step": 5372
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.768541344692221,
      "learning_rate": 3.915542017893311e-05,
      "loss": 0.3819,
      "step": 5373
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7547360116511262,
      "learning_rate": 3.915133116524622e-05,
      "loss": 0.1247,
      "step": 5374
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4005966372307515,
      "learning_rate": 3.9147241594411275e-05,
      "loss": 0.356,
      "step": 5375
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.388465129203911,
      "learning_rate": 3.91431514665893e-05,
      "loss": 0.5743,
      "step": 5376
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9125706840976887,
      "learning_rate": 3.913906078194131e-05,
      "loss": 0.1927,
      "step": 5377
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1124226819887395,
      "learning_rate": 3.9134969540628366e-05,
      "loss": 0.1883,
      "step": 5378
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.0842447262061885,
      "learning_rate": 3.913087774281153e-05,
      "loss": 0.7639,
      "step": 5379
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9736244239815582,
      "learning_rate": 3.9126785388651925e-05,
      "loss": 0.2281,
      "step": 5380
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7834562518888616,
      "learning_rate": 3.912269247831065e-05,
      "loss": 0.111,
      "step": 5381
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8423192892358331,
      "learning_rate": 3.9118599011948856e-05,
      "loss": 0.2056,
      "step": 5382
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7070512184255939,
      "learning_rate": 3.9114504989727686e-05,
      "loss": 0.127,
      "step": 5383
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.87012776752991,
      "learning_rate": 3.911041041180835e-05,
      "loss": 0.124,
      "step": 5384
    },
    {
      "epoch": 0.33,
      "grad_norm": 3.1074731126436714,
      "learning_rate": 3.910631527835204e-05,
      "loss": 0.5817,
      "step": 5385
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6000634012796158,
      "learning_rate": 3.910221958951999e-05,
      "loss": 0.351,
      "step": 5386
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4313715898583748,
      "learning_rate": 3.909812334547345e-05,
      "loss": 0.2489,
      "step": 5387
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7556355259649479,
      "learning_rate": 3.909402654637368e-05,
      "loss": 0.2138,
      "step": 5388
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8098114952738195,
      "learning_rate": 3.908992919238199e-05,
      "loss": 0.1715,
      "step": 5389
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6809017674997053,
      "learning_rate": 3.9085831283659695e-05,
      "loss": 0.1661,
      "step": 5390
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7175682691042387,
      "learning_rate": 3.908173282036812e-05,
      "loss": 0.0937,
      "step": 5391
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7980687223031255,
      "learning_rate": 3.907763380266864e-05,
      "loss": 0.1227,
      "step": 5392
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.361124267865157,
      "learning_rate": 3.907353423072262e-05,
      "loss": 0.5405,
      "step": 5393
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0574130090665974,
      "learning_rate": 3.906943410469148e-05,
      "loss": 0.1874,
      "step": 5394
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6042316787616131,
      "learning_rate": 3.9065333424736625e-05,
      "loss": 0.0929,
      "step": 5395
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8523661334324493,
      "learning_rate": 3.906123219101952e-05,
      "loss": 0.1562,
      "step": 5396
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9140927349895793,
      "learning_rate": 3.9057130403701626e-05,
      "loss": 0.1675,
      "step": 5397
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6331389305413738,
      "learning_rate": 3.905302806294443e-05,
      "loss": 0.0975,
      "step": 5398
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1925429803215162,
      "learning_rate": 3.904892516890944e-05,
      "loss": 0.2998,
      "step": 5399
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6412407464065975,
      "learning_rate": 3.904482172175822e-05,
      "loss": 0.125,
      "step": 5400
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9080984502620255,
      "learning_rate": 3.9040717721652285e-05,
      "loss": 0.1441,
      "step": 5401
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1865486385567423,
      "learning_rate": 3.9036613168753236e-05,
      "loss": 0.1807,
      "step": 5402
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6024396671636394,
      "learning_rate": 3.903250806322267e-05,
      "loss": 0.4951,
      "step": 5403
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.252260566269878,
      "learning_rate": 3.90284024052222e-05,
      "loss": 0.4872,
      "step": 5404
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1629917957260758,
      "learning_rate": 3.9024296194913476e-05,
      "loss": 0.45,
      "step": 5405
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9195840589782479,
      "learning_rate": 3.9020189432458166e-05,
      "loss": 0.1937,
      "step": 5406
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.48518572944565386,
      "learning_rate": 3.901608211801794e-05,
      "loss": 0.0566,
      "step": 5407
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5406853486721,
      "learning_rate": 3.9011974251754523e-05,
      "loss": 0.0825,
      "step": 5408
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8095952606164029,
      "learning_rate": 3.900786583382963e-05,
      "loss": 0.2049,
      "step": 5409
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6086701893873889,
      "learning_rate": 3.900375686440503e-05,
      "loss": 0.1054,
      "step": 5410
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8169339737760802,
      "learning_rate": 3.899964734364247e-05,
      "loss": 0.2362,
      "step": 5411
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9303229122916999,
      "learning_rate": 3.899553727170377e-05,
      "loss": 0.1447,
      "step": 5412
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5354519849724833,
      "learning_rate": 3.8991426648750746e-05,
      "loss": 0.4332,
      "step": 5413
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.587162255023121,
      "learning_rate": 3.898731547494522e-05,
      "loss": 0.0839,
      "step": 5414
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.725303597309163,
      "learning_rate": 3.898320375044906e-05,
      "loss": 0.3424,
      "step": 5415
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.162357129480643,
      "learning_rate": 3.8979091475424146e-05,
      "loss": 0.2769,
      "step": 5416
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7104510825756086,
      "learning_rate": 3.897497865003238e-05,
      "loss": 0.1544,
      "step": 5417
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5658395166920347,
      "learning_rate": 3.8970865274435694e-05,
      "loss": 0.0713,
      "step": 5418
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9970867091448753,
      "learning_rate": 3.8966751348796025e-05,
      "loss": 0.2601,
      "step": 5419
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1849324608590384,
      "learning_rate": 3.8962636873275346e-05,
      "loss": 0.156,
      "step": 5420
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2651882537709982,
      "learning_rate": 3.895852184803566e-05,
      "loss": 0.2468,
      "step": 5421
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7172937302787297,
      "learning_rate": 3.895440627323894e-05,
      "loss": 0.1739,
      "step": 5422
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.636957888982441,
      "learning_rate": 3.895029014904726e-05,
      "loss": 0.2245,
      "step": 5423
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5971918678713413,
      "learning_rate": 3.8946173475622664e-05,
      "loss": 0.1098,
      "step": 5424
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7593205153452309,
      "learning_rate": 3.894205625312721e-05,
      "loss": 0.1278,
      "step": 5425
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7654992424889522,
      "learning_rate": 3.893793848172302e-05,
      "loss": 0.1487,
      "step": 5426
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0696017222048142,
      "learning_rate": 3.8933820161572197e-05,
      "loss": 0.3665,
      "step": 5427
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.626682726003424,
      "learning_rate": 3.892970129283689e-05,
      "loss": 0.6506,
      "step": 5428
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7029769713042575,
      "learning_rate": 3.892558187567926e-05,
      "loss": 0.0953,
      "step": 5429
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0911073208399464,
      "learning_rate": 3.8921461910261494e-05,
      "loss": 0.1966,
      "step": 5430
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9745235768723333,
      "learning_rate": 3.8917341396745794e-05,
      "loss": 0.3294,
      "step": 5431
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9658905438096075,
      "learning_rate": 3.891322033529439e-05,
      "loss": 0.8052,
      "step": 5432
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7861977704251165,
      "learning_rate": 3.8909098726069525e-05,
      "loss": 0.14,
      "step": 5433
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5989537217193126,
      "learning_rate": 3.890497656923348e-05,
      "loss": 0.0959,
      "step": 5434
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.26515745454948,
      "learning_rate": 3.890085386494854e-05,
      "loss": 0.3921,
      "step": 5435
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7808636803367747,
      "learning_rate": 3.889673061337702e-05,
      "loss": 0.1083,
      "step": 5436
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9134656483080907,
      "learning_rate": 3.889260681468126e-05,
      "loss": 0.1926,
      "step": 5437
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6376592505355653,
      "learning_rate": 3.8888482469023606e-05,
      "loss": 0.0645,
      "step": 5438
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9641125232646298,
      "learning_rate": 3.8884357576566456e-05,
      "loss": 0.2097,
      "step": 5439
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6317883815156062,
      "learning_rate": 3.888023213747218e-05,
      "loss": 0.137,
      "step": 5440
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6782475231450296,
      "learning_rate": 3.887610615190324e-05,
      "loss": 0.1052,
      "step": 5441
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5793033594826529,
      "learning_rate": 3.887197962002204e-05,
      "loss": 0.2388,
      "step": 5442
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6308007751819038,
      "learning_rate": 3.8867852541991056e-05,
      "loss": 0.1013,
      "step": 5443
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6736867630931216,
      "learning_rate": 3.886372491797279e-05,
      "loss": 0.0949,
      "step": 5444
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8175720307806768,
      "learning_rate": 3.8859596748129734e-05,
      "loss": 0.1751,
      "step": 5445
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8997524379819968,
      "learning_rate": 3.885546803262442e-05,
      "loss": 0.1428,
      "step": 5446
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7046947517140352,
      "learning_rate": 3.8851338771619404e-05,
      "loss": 0.1173,
      "step": 5447
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.858891634969951,
      "learning_rate": 3.884720896527725e-05,
      "loss": 0.1393,
      "step": 5448
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4416875360812986,
      "learning_rate": 3.8843078613760555e-05,
      "loss": 0.2974,
      "step": 5449
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6455218895058338,
      "learning_rate": 3.8838947717231945e-05,
      "loss": 0.1136,
      "step": 5450
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.140558304959924,
      "learning_rate": 3.883481627585403e-05,
      "loss": 0.3532,
      "step": 5451
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6739469965624808,
      "learning_rate": 3.883068428978949e-05,
      "loss": 0.1053,
      "step": 5452
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6982827372535104,
      "learning_rate": 3.8826551759200995e-05,
      "loss": 0.1597,
      "step": 5453
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3774278800685094,
      "learning_rate": 3.882241868425125e-05,
      "loss": 0.406,
      "step": 5454
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8038801940523322,
      "learning_rate": 3.881828506510298e-05,
      "loss": 0.1622,
      "step": 5455
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7832682193171148,
      "learning_rate": 3.881415090191891e-05,
      "loss": 0.1967,
      "step": 5456
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6388322359399132,
      "learning_rate": 3.8810016194861834e-05,
      "loss": 0.0819,
      "step": 5457
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.876305997478092,
      "learning_rate": 3.880588094409451e-05,
      "loss": 0.1849,
      "step": 5458
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6609942494694866,
      "learning_rate": 3.880174514977977e-05,
      "loss": 0.1252,
      "step": 5459
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0476030450227338,
      "learning_rate": 3.879760881208042e-05,
      "loss": 0.1144,
      "step": 5460
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0951582909747726,
      "learning_rate": 3.879347193115934e-05,
      "loss": 0.383,
      "step": 5461
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9422671263998159,
      "learning_rate": 3.878933450717937e-05,
      "loss": 0.1436,
      "step": 5462
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3806518672731043,
      "learning_rate": 3.878519654030342e-05,
      "loss": 0.4429,
      "step": 5463
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0802572915354676,
      "learning_rate": 3.8781058030694406e-05,
      "loss": 0.2803,
      "step": 5464
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7713271461932626,
      "learning_rate": 3.8776918978515263e-05,
      "loss": 0.1414,
      "step": 5465
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6144473027735637,
      "learning_rate": 3.8772779383928944e-05,
      "loss": 0.114,
      "step": 5466
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8939717926816441,
      "learning_rate": 3.8768639247098423e-05,
      "loss": 0.2599,
      "step": 5467
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9507516138732746,
      "learning_rate": 3.8764498568186725e-05,
      "loss": 0.314,
      "step": 5468
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4259688245024522,
      "learning_rate": 3.876035734735683e-05,
      "loss": 0.2451,
      "step": 5469
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0395043394914119,
      "learning_rate": 3.875621558477183e-05,
      "loss": 0.2118,
      "step": 5470
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7463885432964175,
      "learning_rate": 3.8752073280594745e-05,
      "loss": 0.1329,
      "step": 5471
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5301417628176203,
      "learning_rate": 3.8747930434988685e-05,
      "loss": 0.0623,
      "step": 5472
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1117269108782355,
      "learning_rate": 3.8743787048116755e-05,
      "loss": 0.2818,
      "step": 5473
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8726070440786635,
      "learning_rate": 3.873964312014207e-05,
      "loss": 0.1054,
      "step": 5474
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8434346819125804,
      "learning_rate": 3.87354986512278e-05,
      "loss": 0.2457,
      "step": 5475
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.645241210482212,
      "learning_rate": 3.8731353641537094e-05,
      "loss": 0.1,
      "step": 5476
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6292096088866175,
      "learning_rate": 3.872720809123316e-05,
      "loss": 0.0708,
      "step": 5477
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5371969063056857,
      "learning_rate": 3.87230620004792e-05,
      "loss": 0.0743,
      "step": 5478
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9072295519729786,
      "learning_rate": 3.8718915369438455e-05,
      "loss": 0.1551,
      "step": 5479
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1955646578183146,
      "learning_rate": 3.871476819827418e-05,
      "loss": 0.4461,
      "step": 5480
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7317331386368001,
      "learning_rate": 3.8710620487149664e-05,
      "loss": 0.1321,
      "step": 5481
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6281259165128755,
      "learning_rate": 3.8706472236228175e-05,
      "loss": 0.0756,
      "step": 5482
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1912498905710986,
      "learning_rate": 3.870232344567306e-05,
      "loss": 0.2699,
      "step": 5483
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.868725548438286,
      "learning_rate": 3.8698174115647655e-05,
      "loss": 0.2356,
      "step": 5484
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5995707948163198,
      "learning_rate": 3.8694024246315316e-05,
      "loss": 0.0991,
      "step": 5485
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6068329005903863,
      "learning_rate": 3.868987383783942e-05,
      "loss": 0.5954,
      "step": 5486
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.45796923692014946,
      "learning_rate": 3.868572289038339e-05,
      "loss": 0.0553,
      "step": 5487
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.44091456317983563,
      "learning_rate": 3.868157140411064e-05,
      "loss": 0.0502,
      "step": 5488
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8242705706914946,
      "learning_rate": 3.867741937918461e-05,
      "loss": 0.1465,
      "step": 5489
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.639864358082925,
      "learning_rate": 3.867326681576878e-05,
      "loss": 0.0792,
      "step": 5490
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6527715819762726,
      "learning_rate": 3.8669113714026644e-05,
      "loss": 0.1153,
      "step": 5491
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9389736357687779,
      "learning_rate": 3.86649600741217e-05,
      "loss": 0.199,
      "step": 5492
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6106057622509058,
      "learning_rate": 3.866080589621748e-05,
      "loss": 0.0855,
      "step": 5493
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6696887194267614,
      "learning_rate": 3.8656651180477545e-05,
      "loss": 0.1447,
      "step": 5494
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6363424167197084,
      "learning_rate": 3.865249592706547e-05,
      "loss": 0.0962,
      "step": 5495
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7604822564956275,
      "learning_rate": 3.864834013614484e-05,
      "loss": 0.1198,
      "step": 5496
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.882355762187409,
      "learning_rate": 3.864418380787928e-05,
      "loss": 0.2234,
      "step": 5497
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7199010992214312,
      "learning_rate": 3.8640026942432425e-05,
      "loss": 0.1076,
      "step": 5498
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8755006440981377,
      "learning_rate": 3.8635869539967934e-05,
      "loss": 0.1306,
      "step": 5499
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7885111746826007,
      "learning_rate": 3.863171160064948e-05,
      "loss": 0.1605,
      "step": 5500
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7889460291853911,
      "learning_rate": 3.8627553124640784e-05,
      "loss": 0.1854,
      "step": 5501
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9275448239365535,
      "learning_rate": 3.8623394112105546e-05,
      "loss": 0.2648,
      "step": 5502
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1080075307109047,
      "learning_rate": 3.861923456320752e-05,
      "loss": 0.2045,
      "step": 5503
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5904352434127783,
      "learning_rate": 3.861507447811047e-05,
      "loss": 0.1007,
      "step": 5504
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7946628515766543,
      "learning_rate": 3.861091385697818e-05,
      "loss": 0.1789,
      "step": 5505
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1092422070431578,
      "learning_rate": 3.8606752699974455e-05,
      "loss": 0.2366,
      "step": 5506
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0701536115130594,
      "learning_rate": 3.860259100726312e-05,
      "loss": 0.2157,
      "step": 5507
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7266015180783117,
      "learning_rate": 3.859842877900804e-05,
      "loss": 0.0919,
      "step": 5508
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4323643559708845,
      "learning_rate": 3.859426601537307e-05,
      "loss": 0.3303,
      "step": 5509
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8576577766027309,
      "learning_rate": 3.85901027165221e-05,
      "loss": 0.1733,
      "step": 5510
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.654950971466282,
      "learning_rate": 3.858593888261904e-05,
      "loss": 0.168,
      "step": 5511
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0846931969388534,
      "learning_rate": 3.858177451382784e-05,
      "loss": 0.2992,
      "step": 5512
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9492350868771499,
      "learning_rate": 3.8577609610312446e-05,
      "loss": 0.1516,
      "step": 5513
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6254057497974785,
      "learning_rate": 3.8573444172236824e-05,
      "loss": 0.1111,
      "step": 5514
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6648941589240895,
      "learning_rate": 3.856927819976498e-05,
      "loss": 0.1313,
      "step": 5515
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6399894357930983,
      "learning_rate": 3.856511169306093e-05,
      "loss": 0.1246,
      "step": 5516
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6775459471186144,
      "learning_rate": 3.856094465228871e-05,
      "loss": 0.1493,
      "step": 5517
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7403472976931458,
      "learning_rate": 3.855677707761237e-05,
      "loss": 0.212,
      "step": 5518
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.124560440428697,
      "learning_rate": 3.855260896919601e-05,
      "loss": 0.1556,
      "step": 5519
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.823342969326594,
      "learning_rate": 3.8548440327203726e-05,
      "loss": 0.1604,
      "step": 5520
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6898834231328784,
      "learning_rate": 3.854427115179963e-05,
      "loss": 0.1357,
      "step": 5521
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4469407694157612,
      "learning_rate": 3.8540101443147875e-05,
      "loss": 0.4405,
      "step": 5522
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5776006129236909,
      "learning_rate": 3.853593120141262e-05,
      "loss": 0.0453,
      "step": 5523
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8335437089193494,
      "learning_rate": 3.8531760426758055e-05,
      "loss": 0.199,
      "step": 5524
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6886281098503481,
      "learning_rate": 3.852758911934838e-05,
      "loss": 0.1065,
      "step": 5525
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6415618419792499,
      "learning_rate": 3.852341727934783e-05,
      "loss": 0.0656,
      "step": 5526
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9915023343907348,
      "learning_rate": 3.851924490692065e-05,
      "loss": 0.2512,
      "step": 5527
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7767920201054875,
      "learning_rate": 3.851507200223111e-05,
      "loss": 0.1466,
      "step": 5528
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5495942383034831,
      "learning_rate": 3.85108985654435e-05,
      "loss": 0.1034,
      "step": 5529
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1942978791627206,
      "learning_rate": 3.850672459672213e-05,
      "loss": 0.2482,
      "step": 5530
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.748083562713191,
      "learning_rate": 3.850255009623133e-05,
      "loss": 0.116,
      "step": 5531
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2761995396903094,
      "learning_rate": 3.849837506413545e-05,
      "loss": 0.3555,
      "step": 5532
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6600510967900309,
      "learning_rate": 3.849419950059888e-05,
      "loss": 0.1582,
      "step": 5533
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6683392175036436,
      "learning_rate": 3.8490023405786e-05,
      "loss": 0.1487,
      "step": 5534
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6575144474824703,
      "learning_rate": 3.848584677986123e-05,
      "loss": 0.1887,
      "step": 5535
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8947136747451526,
      "learning_rate": 3.8481669622989016e-05,
      "loss": 0.2474,
      "step": 5536
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5415995848191546,
      "learning_rate": 3.8477491935333795e-05,
      "loss": 0.0593,
      "step": 5537
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9367020489387156,
      "learning_rate": 3.847331371706006e-05,
      "loss": 0.2065,
      "step": 5538
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.47422460957333346,
      "learning_rate": 3.846913496833232e-05,
      "loss": 0.0872,
      "step": 5539
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0317968314411228,
      "learning_rate": 3.8464955689315076e-05,
      "loss": 0.3853,
      "step": 5540
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5981246543965868,
      "learning_rate": 3.8460775880172875e-05,
      "loss": 0.0741,
      "step": 5541
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8516036180085997,
      "learning_rate": 3.8456595541070274e-05,
      "loss": 0.428,
      "step": 5542
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4695611699293487,
      "learning_rate": 3.845241467217187e-05,
      "loss": 0.4603,
      "step": 5543
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8662939046157313,
      "learning_rate": 3.844823327364227e-05,
      "loss": 0.0846,
      "step": 5544
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8666214749458211,
      "learning_rate": 3.8444051345646064e-05,
      "loss": 0.1966,
      "step": 5545
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1659061706489495,
      "learning_rate": 3.843986888834794e-05,
      "loss": 0.2538,
      "step": 5546
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.845491891467296,
      "learning_rate": 3.843568590191253e-05,
      "loss": 0.157,
      "step": 5547
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9114588221877841,
      "learning_rate": 3.8431502386504554e-05,
      "loss": 0.2549,
      "step": 5548
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5288188910087308,
      "learning_rate": 3.842731834228869e-05,
      "loss": 0.0704,
      "step": 5549
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3054221892901212,
      "learning_rate": 3.842313376942969e-05,
      "loss": 0.2312,
      "step": 5550
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4054822685609871,
      "learning_rate": 3.841894866809229e-05,
      "loss": 0.3232,
      "step": 5551
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7252199765937828,
      "learning_rate": 3.841476303844126e-05,
      "loss": 0.1438,
      "step": 5552
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9563155973499158,
      "learning_rate": 3.8410576880641394e-05,
      "loss": 0.2075,
      "step": 5553
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8518018714967687,
      "learning_rate": 3.8406390194857515e-05,
      "loss": 0.1739,
      "step": 5554
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6475360763392131,
      "learning_rate": 3.8402202981254434e-05,
      "loss": 0.0926,
      "step": 5555
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9409892272617648,
      "learning_rate": 3.839801523999703e-05,
      "loss": 0.128,
      "step": 5556
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.681423500859328,
      "learning_rate": 3.8393826971250156e-05,
      "loss": 0.1083,
      "step": 5557
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7224459939677711,
      "learning_rate": 3.838963817517872e-05,
      "loss": 0.1338,
      "step": 5558
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.396457866240523,
      "learning_rate": 3.838544885194763e-05,
      "loss": 0.3842,
      "step": 5559
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7136138470025581,
      "learning_rate": 3.838125900172182e-05,
      "loss": 0.2368,
      "step": 5560
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6647634927995404,
      "learning_rate": 3.8377068624666254e-05,
      "loss": 0.1209,
      "step": 5561
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0375142717994201,
      "learning_rate": 3.837287772094592e-05,
      "loss": 0.2087,
      "step": 5562
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7024645552795992,
      "learning_rate": 3.836868629072579e-05,
      "loss": 0.2885,
      "step": 5563
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6708865170063508,
      "learning_rate": 3.8364494334170915e-05,
      "loss": 0.1664,
      "step": 5564
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9330333321903056,
      "learning_rate": 3.836030185144631e-05,
      "loss": 0.2189,
      "step": 5565
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8184351872999267,
      "learning_rate": 3.8356108842717056e-05,
      "loss": 0.1794,
      "step": 5566
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2559238888411466,
      "learning_rate": 3.835191530814821e-05,
      "loss": 0.2111,
      "step": 5567
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2438187880506568,
      "learning_rate": 3.8347721247904896e-05,
      "loss": 0.7948,
      "step": 5568
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0305903943936459,
      "learning_rate": 3.834352666215224e-05,
      "loss": 0.3725,
      "step": 5569
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.983967877939049,
      "learning_rate": 3.833933155105537e-05,
      "loss": 0.3724,
      "step": 5570
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9499575190374344,
      "learning_rate": 3.833513591477945e-05,
      "loss": 0.1174,
      "step": 5571
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1785131973465102,
      "learning_rate": 3.833093975348968e-05,
      "loss": 0.1568,
      "step": 5572
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5837081717682094,
      "learning_rate": 3.832674306735126e-05,
      "loss": 0.0631,
      "step": 5573
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2898183368721445,
      "learning_rate": 3.8322545856529414e-05,
      "loss": 0.4189,
      "step": 5574
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0706966592858689,
      "learning_rate": 3.831834812118939e-05,
      "loss": 0.1468,
      "step": 5575
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.839405549808513,
      "learning_rate": 3.831414986149644e-05,
      "loss": 0.1437,
      "step": 5576
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8687509554977307,
      "learning_rate": 3.830995107761588e-05,
      "loss": 0.1765,
      "step": 5577
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8133710343151702,
      "learning_rate": 3.8305751769713016e-05,
      "loss": 0.1043,
      "step": 5578
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9116238585292433,
      "learning_rate": 3.830155193795316e-05,
      "loss": 0.3938,
      "step": 5579
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8556805057725122,
      "learning_rate": 3.8297351582501674e-05,
      "loss": 0.1612,
      "step": 5580
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.494545054122926,
      "learning_rate": 3.8293150703523914e-05,
      "loss": 0.2824,
      "step": 5581
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7349547636366904,
      "learning_rate": 3.82889493011853e-05,
      "loss": 0.1067,
      "step": 5582
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7460970427614083,
      "learning_rate": 3.828474737565122e-05,
      "loss": 0.174,
      "step": 5583
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.674207801103625,
      "learning_rate": 3.8280544927087116e-05,
      "loss": 0.1052,
      "step": 5584
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1565582323082932,
      "learning_rate": 3.8276341955658435e-05,
      "loss": 0.519,
      "step": 5585
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5536174364800397,
      "learning_rate": 3.827213846153067e-05,
      "loss": 0.1047,
      "step": 5586
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7038016604365918,
      "learning_rate": 3.826793444486929e-05,
      "loss": 0.1857,
      "step": 5587
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8066144238714403,
      "learning_rate": 3.8263729905839826e-05,
      "loss": 0.2192,
      "step": 5588
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9592437947337726,
      "learning_rate": 3.8259524844607805e-05,
      "loss": 0.1974,
      "step": 5589
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6490998661956071,
      "learning_rate": 3.82553192613388e-05,
      "loss": 0.1028,
      "step": 5590
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0920248221373081,
      "learning_rate": 3.825111315619837e-05,
      "loss": 0.2077,
      "step": 5591
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6431652673755901,
      "learning_rate": 3.824690652935212e-05,
      "loss": 0.0866,
      "step": 5592
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5254628318904347,
      "learning_rate": 3.824269938096566e-05,
      "loss": 0.0657,
      "step": 5593
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6966700893384524,
      "learning_rate": 3.823849171120464e-05,
      "loss": 0.1098,
      "step": 5594
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1429247932113644,
      "learning_rate": 3.823428352023471e-05,
      "loss": 0.334,
      "step": 5595
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6078343986259351,
      "learning_rate": 3.823007480822156e-05,
      "loss": 0.1002,
      "step": 5596
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6596039156654789,
      "learning_rate": 3.822586557533088e-05,
      "loss": 0.1314,
      "step": 5597
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8640078671870135,
      "learning_rate": 3.8221655821728384e-05,
      "loss": 0.1276,
      "step": 5598
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0067797727306929,
      "learning_rate": 3.821744554757983e-05,
      "loss": 0.0947,
      "step": 5599
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6077864276855375,
      "learning_rate": 3.821323475305098e-05,
      "loss": 0.0933,
      "step": 5600
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.48746943942273546,
      "learning_rate": 3.8209023438307596e-05,
      "loss": 0.0677,
      "step": 5601
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.552933081229142,
      "learning_rate": 3.82048116035155e-05,
      "loss": 0.0957,
      "step": 5602
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1382137936294112,
      "learning_rate": 3.8200599248840504e-05,
      "loss": 0.2218,
      "step": 5603
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0432956391406347,
      "learning_rate": 3.819638637444846e-05,
      "loss": 0.173,
      "step": 5604
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6812343234017619,
      "learning_rate": 3.8192172980505214e-05,
      "loss": 0.1215,
      "step": 5605
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8334054894098364,
      "learning_rate": 3.818795906717668e-05,
      "loss": 0.1669,
      "step": 5606
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6589794169464174,
      "learning_rate": 3.8183744634628735e-05,
      "loss": 0.0927,
      "step": 5607
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7776964296658777,
      "learning_rate": 3.817952968302732e-05,
      "loss": 0.467,
      "step": 5608
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5745584052149426,
      "learning_rate": 3.817531421253838e-05,
      "loss": 0.1061,
      "step": 5609
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.775260878931003,
      "learning_rate": 3.8171098223327866e-05,
      "loss": 0.0951,
      "step": 5610
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6637288567542043,
      "learning_rate": 3.816688171556179e-05,
      "loss": 0.1353,
      "step": 5611
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1977357052573199,
      "learning_rate": 3.8162664689406134e-05,
      "loss": 0.3735,
      "step": 5612
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.192963555124429,
      "learning_rate": 3.8158447145026934e-05,
      "loss": 0.1414,
      "step": 5613
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.5233344442534396,
      "learning_rate": 3.8154229082590246e-05,
      "loss": 0.8586,
      "step": 5614
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.600935282428104,
      "learning_rate": 3.815001050226213e-05,
      "loss": 0.4037,
      "step": 5615
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1574159102071968,
      "learning_rate": 3.8145791404208675e-05,
      "loss": 0.204,
      "step": 5616
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7426382669644885,
      "learning_rate": 3.814157178859599e-05,
      "loss": 0.1759,
      "step": 5617
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0232524001721468,
      "learning_rate": 3.813735165559021e-05,
      "loss": 0.2285,
      "step": 5618
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.025382294244797,
      "learning_rate": 3.813313100535747e-05,
      "loss": 0.1654,
      "step": 5619
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8268326093848573,
      "learning_rate": 3.8128909838063964e-05,
      "loss": 0.1132,
      "step": 5620
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8931200125385375,
      "learning_rate": 3.812468815387585e-05,
      "loss": 0.2795,
      "step": 5621
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.817738469600692,
      "learning_rate": 3.8120465952959384e-05,
      "loss": 0.1597,
      "step": 5622
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6569502456889416,
      "learning_rate": 3.811624323548075e-05,
      "loss": 0.0843,
      "step": 5623
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8500883083604107,
      "learning_rate": 3.8112020001606224e-05,
      "loss": 0.2475,
      "step": 5624
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9559062675432165,
      "learning_rate": 3.810779625150207e-05,
      "loss": 0.2547,
      "step": 5625
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3165225068820918,
      "learning_rate": 3.8103571985334596e-05,
      "loss": 0.255,
      "step": 5626
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.290128061908504,
      "learning_rate": 3.8099347203270084e-05,
      "loss": 0.2168,
      "step": 5627
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6505277035519943,
      "learning_rate": 3.80951219054749e-05,
      "loss": 0.1299,
      "step": 5628
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.654154749002884,
      "learning_rate": 3.809089609211538e-05,
      "loss": 0.1099,
      "step": 5629
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6766393420254798,
      "learning_rate": 3.8086669763357894e-05,
      "loss": 0.1252,
      "step": 5630
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7405668379386942,
      "learning_rate": 3.808244291936884e-05,
      "loss": 0.1293,
      "step": 5631
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.08226980762581,
      "learning_rate": 3.807821556031463e-05,
      "loss": 0.167,
      "step": 5632
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8434190529286246,
      "learning_rate": 3.807398768636171e-05,
      "loss": 0.1791,
      "step": 5633
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7131529723340548,
      "learning_rate": 3.806975929767652e-05,
      "loss": 0.1192,
      "step": 5634
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.989426993956243,
      "learning_rate": 3.806553039442554e-05,
      "loss": 0.1037,
      "step": 5635
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8425066792089932,
      "learning_rate": 3.806130097677526e-05,
      "loss": 0.1492,
      "step": 5636
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8546535450575002,
      "learning_rate": 3.80570710448922e-05,
      "loss": 0.1611,
      "step": 5637
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8567123575586599,
      "learning_rate": 3.8052840598942905e-05,
      "loss": 0.1644,
      "step": 5638
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6984970150965237,
      "learning_rate": 3.804860963909391e-05,
      "loss": 0.1296,
      "step": 5639
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6171100573582586,
      "learning_rate": 3.8044378165511814e-05,
      "loss": 0.065,
      "step": 5640
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5280359296858387,
      "learning_rate": 3.80401461783632e-05,
      "loss": 0.0754,
      "step": 5641
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6867778166045326,
      "learning_rate": 3.803591367781467e-05,
      "loss": 0.2708,
      "step": 5642
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8608576656735463,
      "learning_rate": 3.8031680664032884e-05,
      "loss": 0.1438,
      "step": 5643
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7683329511787209,
      "learning_rate": 3.80274471371845e-05,
      "loss": 0.261,
      "step": 5644
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8704251146269488,
      "learning_rate": 3.802321309743618e-05,
      "loss": 0.1173,
      "step": 5645
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6835465718168366,
      "learning_rate": 3.801897854495462e-05,
      "loss": 0.1594,
      "step": 5646
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8591139969975404,
      "learning_rate": 3.801474347990655e-05,
      "loss": 0.1387,
      "step": 5647
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8886510278129922,
      "learning_rate": 3.80105079024587e-05,
      "loss": 0.1239,
      "step": 5648
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.46075445084025124,
      "learning_rate": 3.800627181277783e-05,
      "loss": 0.0726,
      "step": 5649
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8143516120689787,
      "learning_rate": 3.8002035211030726e-05,
      "loss": 0.1169,
      "step": 5650
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0956100037581165,
      "learning_rate": 3.799779809738416e-05,
      "loss": 0.2449,
      "step": 5651
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.100022180361633,
      "learning_rate": 3.799356047200499e-05,
      "loss": 0.2652,
      "step": 5652
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.795809280410704,
      "learning_rate": 3.7989322335060016e-05,
      "loss": 0.1879,
      "step": 5653
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7490712184142293,
      "learning_rate": 3.7985083686716116e-05,
      "loss": 0.1575,
      "step": 5654
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8109621621181998,
      "learning_rate": 3.798084452714016e-05,
      "loss": 0.2107,
      "step": 5655
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7014438824276977,
      "learning_rate": 3.797660485649906e-05,
      "loss": 0.1485,
      "step": 5656
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7071152828676642,
      "learning_rate": 3.797236467495972e-05,
      "loss": 0.1458,
      "step": 5657
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1848676273503762,
      "learning_rate": 3.79681239826891e-05,
      "loss": 0.3645,
      "step": 5658
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7750777035486403,
      "learning_rate": 3.796388277985413e-05,
      "loss": 0.1378,
      "step": 5659
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7673998031367172,
      "learning_rate": 3.795964106662181e-05,
      "loss": 0.0813,
      "step": 5660
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5859675404427638,
      "learning_rate": 3.795539884315913e-05,
      "loss": 0.091,
      "step": 5661
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6390304155721633,
      "learning_rate": 3.795115610963311e-05,
      "loss": 0.1264,
      "step": 5662
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6744746660343366,
      "learning_rate": 3.79469128662108e-05,
      "loss": 0.1064,
      "step": 5663
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8850899660489722,
      "learning_rate": 3.794266911305925e-05,
      "loss": 0.2486,
      "step": 5664
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6562690667320041,
      "learning_rate": 3.793842485034553e-05,
      "loss": 0.1288,
      "step": 5665
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.821126198406116,
      "learning_rate": 3.793418007823677e-05,
      "loss": 0.0734,
      "step": 5666
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5721351424708613,
      "learning_rate": 3.792993479690006e-05,
      "loss": 0.1035,
      "step": 5667
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6679810007610412,
      "learning_rate": 3.792568900650255e-05,
      "loss": 0.0678,
      "step": 5668
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8137643819017026,
      "learning_rate": 3.792144270721141e-05,
      "loss": 0.1499,
      "step": 5669
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6040636437826534,
      "learning_rate": 3.79171958991938e-05,
      "loss": 0.0634,
      "step": 5670
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7229070950902675,
      "learning_rate": 3.791294858261693e-05,
      "loss": 0.0814,
      "step": 5671
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.44245002109884213,
      "learning_rate": 3.7908700757648025e-05,
      "loss": 0.1042,
      "step": 5672
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1409950666895545,
      "learning_rate": 3.790445242445432e-05,
      "loss": 0.2742,
      "step": 5673
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6618418396708471,
      "learning_rate": 3.790020358320307e-05,
      "loss": 0.1588,
      "step": 5674
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6349643583637445,
      "learning_rate": 3.789595423406156e-05,
      "loss": 0.101,
      "step": 5675
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.916414006934173,
      "learning_rate": 3.789170437719709e-05,
      "loss": 0.1223,
      "step": 5676
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5559066281575572,
      "learning_rate": 3.7887454012776984e-05,
      "loss": 0.0791,
      "step": 5677
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6859516363949042,
      "learning_rate": 3.788320314096858e-05,
      "loss": 0.1683,
      "step": 5678
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1328031546792758,
      "learning_rate": 3.787895176193923e-05,
      "loss": 0.388,
      "step": 5679
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.739561580683414,
      "learning_rate": 3.7874699875856315e-05,
      "loss": 0.2223,
      "step": 5680
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7443004655864429,
      "learning_rate": 3.787044748288724e-05,
      "loss": 0.1486,
      "step": 5681
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6797789853770735,
      "learning_rate": 3.786619458319943e-05,
      "loss": 0.1247,
      "step": 5682
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2449477952320025,
      "learning_rate": 3.786194117696032e-05,
      "loss": 0.2163,
      "step": 5683
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4651660859924892,
      "learning_rate": 3.785768726433735e-05,
      "loss": 0.0959,
      "step": 5684
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6944210682617141,
      "learning_rate": 3.7853432845498035e-05,
      "loss": 0.1904,
      "step": 5685
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.645661103988843,
      "learning_rate": 3.7849177920609855e-05,
      "loss": 0.1829,
      "step": 5686
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6284135706883693,
      "learning_rate": 3.784492248984033e-05,
      "loss": 0.5,
      "step": 5687
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5961815895358764,
      "learning_rate": 3.7840666553356994e-05,
      "loss": 0.1163,
      "step": 5688
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6872706644609758,
      "learning_rate": 3.783641011132743e-05,
      "loss": 0.0928,
      "step": 5689
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.693761604451862,
      "learning_rate": 3.783215316391918e-05,
      "loss": 0.1655,
      "step": 5690
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0508624844549692,
      "learning_rate": 3.782789571129988e-05,
      "loss": 0.4377,
      "step": 5691
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4827685447569399,
      "learning_rate": 3.782363775363712e-05,
      "loss": 0.0874,
      "step": 5692
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8987862823634658,
      "learning_rate": 3.7819379291098554e-05,
      "loss": 0.2682,
      "step": 5693
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.774761183842044,
      "learning_rate": 3.7815120323851844e-05,
      "loss": 0.1833,
      "step": 5694
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6786474777438395,
      "learning_rate": 3.781086085206466e-05,
      "loss": 0.1239,
      "step": 5695
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.781438177642778,
      "learning_rate": 3.78066008759047e-05,
      "loss": 0.4288,
      "step": 5696
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8270864326786093,
      "learning_rate": 3.780234039553969e-05,
      "loss": 0.1617,
      "step": 5697
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4668477538985554,
      "learning_rate": 3.779807941113736e-05,
      "loss": 0.0695,
      "step": 5698
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5572686227383339,
      "learning_rate": 3.779381792286548e-05,
      "loss": 0.5423,
      "step": 5699
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9776219827586209,
      "learning_rate": 3.778955593089181e-05,
      "loss": 0.3541,
      "step": 5700
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8217099752072924,
      "learning_rate": 3.7785293435384164e-05,
      "loss": 0.2152,
      "step": 5701
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.551998663580787,
      "learning_rate": 3.778103043651034e-05,
      "loss": 0.1036,
      "step": 5702
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.690516394434957,
      "learning_rate": 3.77767669344382e-05,
      "loss": 0.0735,
      "step": 5703
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7262960310983908,
      "learning_rate": 3.777250292933558e-05,
      "loss": 0.1175,
      "step": 5704
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5002707299713353,
      "learning_rate": 3.776823842137037e-05,
      "loss": 0.0892,
      "step": 5705
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6524563096149529,
      "learning_rate": 3.7763973410710466e-05,
      "loss": 0.1083,
      "step": 5706
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8411302633439356,
      "learning_rate": 3.775970789752378e-05,
      "loss": 0.2042,
      "step": 5707
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8190577317971428,
      "learning_rate": 3.775544188197825e-05,
      "loss": 0.1177,
      "step": 5708
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.902990760401708,
      "learning_rate": 3.775117536424183e-05,
      "loss": 0.3497,
      "step": 5709
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5235271974062057,
      "learning_rate": 3.77469083444825e-05,
      "loss": 0.0872,
      "step": 5710
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9315538443307659,
      "learning_rate": 3.7742640822868245e-05,
      "loss": 0.2073,
      "step": 5711
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5792479783510175,
      "learning_rate": 3.773837279956709e-05,
      "loss": 0.0778,
      "step": 5712
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6524633512699839,
      "learning_rate": 3.7734104274747076e-05,
      "loss": 0.1391,
      "step": 5713
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0073078264734454,
      "learning_rate": 3.7729835248576253e-05,
      "loss": 0.0989,
      "step": 5714
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7146523005276934,
      "learning_rate": 3.7725565721222684e-05,
      "loss": 0.178,
      "step": 5715
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8514516881621835,
      "learning_rate": 3.772129569285447e-05,
      "loss": 0.1798,
      "step": 5716
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7678568766301741,
      "learning_rate": 3.771702516363974e-05,
      "loss": 0.1653,
      "step": 5717
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2730645982781281,
      "learning_rate": 3.77127541337466e-05,
      "loss": 0.5553,
      "step": 5718
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9351982174919347,
      "learning_rate": 3.770848260334323e-05,
      "loss": 0.1467,
      "step": 5719
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6360618699586359,
      "learning_rate": 3.770421057259779e-05,
      "loss": 0.123,
      "step": 5720
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8995328197976249,
      "learning_rate": 3.769993804167847e-05,
      "loss": 0.301,
      "step": 5721
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9005171747754361,
      "learning_rate": 3.769566501075349e-05,
      "loss": 0.216,
      "step": 5722
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.483835274076931,
      "learning_rate": 3.769139147999109e-05,
      "loss": 0.2121,
      "step": 5723
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9902070253262222,
      "learning_rate": 3.76871174495595e-05,
      "loss": 0.1179,
      "step": 5724
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8752980006519743,
      "learning_rate": 3.768284291962701e-05,
      "loss": 0.2039,
      "step": 5725
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3564566358801806,
      "learning_rate": 3.76785678903619e-05,
      "loss": 0.2412,
      "step": 5726
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.640686578196213,
      "learning_rate": 3.767429236193249e-05,
      "loss": 0.089,
      "step": 5727
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6170758478817148,
      "learning_rate": 3.7670016334507094e-05,
      "loss": 0.1048,
      "step": 5728
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7867631296916859,
      "learning_rate": 3.76657398082541e-05,
      "loss": 0.1504,
      "step": 5729
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5755419099968399,
      "learning_rate": 3.766146278334183e-05,
      "loss": 0.1024,
      "step": 5730
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3508174922019416,
      "learning_rate": 3.7657185259938706e-05,
      "loss": 0.3325,
      "step": 5731
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6099957205984479,
      "learning_rate": 3.765290723821312e-05,
      "loss": 0.0927,
      "step": 5732
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7931021419398753,
      "learning_rate": 3.7648628718333515e-05,
      "loss": 0.1952,
      "step": 5733
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6408892790100522,
      "learning_rate": 3.764434970046833e-05,
      "loss": 0.098,
      "step": 5734
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7240382547314018,
      "learning_rate": 3.7640070184786035e-05,
      "loss": 0.1581,
      "step": 5735
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4961647799826966,
      "learning_rate": 3.763579017145512e-05,
      "loss": 0.0806,
      "step": 5736
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2613112195056464,
      "learning_rate": 3.7631509660644084e-05,
      "loss": 0.2085,
      "step": 5737
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.848959565673274,
      "learning_rate": 3.762722865252146e-05,
      "loss": 0.184,
      "step": 5738
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8629063589831799,
      "learning_rate": 3.762294714725581e-05,
      "loss": 0.1302,
      "step": 5739
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.661285678401701,
      "learning_rate": 3.761866514501567e-05,
      "loss": 0.1009,
      "step": 5740
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6717878993649771,
      "learning_rate": 3.7614382645969634e-05,
      "loss": 0.1653,
      "step": 5741
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8487546560783934,
      "learning_rate": 3.761009965028633e-05,
      "loss": 0.1442,
      "step": 5742
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1642320361906429,
      "learning_rate": 3.760581615813435e-05,
      "loss": 0.448,
      "step": 5743
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5578058347491209,
      "learning_rate": 3.7601532169682366e-05,
      "loss": 0.0728,
      "step": 5744
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.895505394157174,
      "learning_rate": 3.7597247685099026e-05,
      "loss": 0.8535,
      "step": 5745
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5005091239105984,
      "learning_rate": 3.7592962704553014e-05,
      "loss": 0.0833,
      "step": 5746
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7637634929944143,
      "learning_rate": 3.7588677228213036e-05,
      "loss": 0.2239,
      "step": 5747
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7120846730981834,
      "learning_rate": 3.758439125624781e-05,
      "loss": 0.1182,
      "step": 5748
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5933426165875401,
      "learning_rate": 3.758010478882609e-05,
      "loss": 0.0661,
      "step": 5749
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7369632656761057,
      "learning_rate": 3.757581782611662e-05,
      "loss": 0.157,
      "step": 5750
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.2341888832693826,
      "learning_rate": 3.757153036828819e-05,
      "loss": 0.2202,
      "step": 5751
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.724660530802293,
      "learning_rate": 3.75672424155096e-05,
      "loss": 0.1243,
      "step": 5752
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.829459601068939,
      "learning_rate": 3.7562953967949676e-05,
      "loss": 0.1312,
      "step": 5753
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4052234442623694,
      "learning_rate": 3.7558665025777246e-05,
      "loss": 0.3083,
      "step": 5754
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1047901006866752,
      "learning_rate": 3.755437558916117e-05,
      "loss": 0.2334,
      "step": 5755
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7320177835380527,
      "learning_rate": 3.755008565827034e-05,
      "loss": 0.1867,
      "step": 5756
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6875442853113811,
      "learning_rate": 3.7545795233273635e-05,
      "loss": 0.1455,
      "step": 5757
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.911037838684414,
      "learning_rate": 3.754150431433998e-05,
      "loss": 0.1983,
      "step": 5758
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5956216421261481,
      "learning_rate": 3.753721290163832e-05,
      "loss": 0.0792,
      "step": 5759
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.172930116258676,
      "learning_rate": 3.75329209953376e-05,
      "loss": 0.7058,
      "step": 5760
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7789493004898342,
      "learning_rate": 3.7528628595606784e-05,
      "loss": 0.1275,
      "step": 5761
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3598757178238414,
      "learning_rate": 3.75243357026149e-05,
      "loss": 0.3381,
      "step": 5762
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0464444497840757,
      "learning_rate": 3.7520042316530935e-05,
      "loss": 0.1177,
      "step": 5763
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1155197291135748,
      "learning_rate": 3.751574843752393e-05,
      "loss": 0.2791,
      "step": 5764
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5848687846553543,
      "learning_rate": 3.751145406576294e-05,
      "loss": 0.0672,
      "step": 5765
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5630579470114114,
      "learning_rate": 3.750715920141704e-05,
      "loss": 0.3033,
      "step": 5766
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3408346098579043,
      "learning_rate": 3.750286384465532e-05,
      "loss": 0.3071,
      "step": 5767
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0322367273560544,
      "learning_rate": 3.749856799564688e-05,
      "loss": 0.1357,
      "step": 5768
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.267159574024439,
      "learning_rate": 3.7494271654560877e-05,
      "loss": 0.2167,
      "step": 5769
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8344114685653186,
      "learning_rate": 3.748997482156643e-05,
      "loss": 0.1213,
      "step": 5770
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3886474499381516,
      "learning_rate": 3.748567749683273e-05,
      "loss": 0.2814,
      "step": 5771
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1393522756055399,
      "learning_rate": 3.7481379680528956e-05,
      "loss": 0.206,
      "step": 5772
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1155726018476164,
      "learning_rate": 3.747708137282432e-05,
      "loss": 0.1818,
      "step": 5773
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6636132636412586,
      "learning_rate": 3.747278257388805e-05,
      "loss": 0.5104,
      "step": 5774
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9531616265564059,
      "learning_rate": 3.746848328388939e-05,
      "loss": 0.1638,
      "step": 5775
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8757526897685612,
      "learning_rate": 3.746418350299761e-05,
      "loss": 0.1807,
      "step": 5776
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2959924653076689,
      "learning_rate": 3.7459883231381995e-05,
      "loss": 0.176,
      "step": 5777
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6133388475858765,
      "learning_rate": 3.745558246921184e-05,
      "loss": 0.1177,
      "step": 5778
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5233817036706857,
      "learning_rate": 3.745128121665647e-05,
      "loss": 0.0787,
      "step": 5779
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7733995171730609,
      "learning_rate": 3.7446979473885245e-05,
      "loss": 0.13,
      "step": 5780
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8446039875821855,
      "learning_rate": 3.744267724106752e-05,
      "loss": 0.1452,
      "step": 5781
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6196968046834268,
      "learning_rate": 3.743837451837267e-05,
      "loss": 0.1069,
      "step": 5782
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9265142341337342,
      "learning_rate": 3.74340713059701e-05,
      "loss": 0.2425,
      "step": 5783
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9585045539744618,
      "learning_rate": 3.742976760402924e-05,
      "loss": 0.2177,
      "step": 5784
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5581044823024972,
      "learning_rate": 3.7425463412719516e-05,
      "loss": 0.0696,
      "step": 5785
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6349652798603171,
      "learning_rate": 3.7421158732210386e-05,
      "loss": 0.1271,
      "step": 5786
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.196945268463804,
      "learning_rate": 3.741685356267134e-05,
      "loss": 0.3195,
      "step": 5787
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9729119729385388,
      "learning_rate": 3.741254790427187e-05,
      "loss": 0.2006,
      "step": 5788
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6299516385985373,
      "learning_rate": 3.740824175718149e-05,
      "loss": 0.1362,
      "step": 5789
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.658789475338759,
      "learning_rate": 3.7403935121569745e-05,
      "loss": 0.1751,
      "step": 5790
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.702147826118084,
      "learning_rate": 3.739962799760619e-05,
      "loss": 0.1085,
      "step": 5791
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7697840812392485,
      "learning_rate": 3.739532038546038e-05,
      "loss": 0.12,
      "step": 5792
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2703643031615568,
      "learning_rate": 3.7391012285301935e-05,
      "loss": 0.3522,
      "step": 5793
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.2693989297067043,
      "learning_rate": 3.738670369730045e-05,
      "loss": 0.5412,
      "step": 5794
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2638284030119773,
      "learning_rate": 3.738239462162556e-05,
      "loss": 0.1549,
      "step": 5795
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2372290329875226,
      "learning_rate": 3.737808505844692e-05,
      "loss": 0.159,
      "step": 5796
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7282456170366902,
      "learning_rate": 3.737377500793421e-05,
      "loss": 0.1142,
      "step": 5797
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1563099440701563,
      "learning_rate": 3.7369464470257104e-05,
      "loss": 0.315,
      "step": 5798
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6601484154930971,
      "learning_rate": 3.736515344558532e-05,
      "loss": 0.0654,
      "step": 5799
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8404077661524059,
      "learning_rate": 3.7360841934088576e-05,
      "loss": 0.1924,
      "step": 5800
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8628429294133865,
      "learning_rate": 3.735652993593663e-05,
      "loss": 0.1996,
      "step": 5801
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.909622329709821,
      "learning_rate": 3.735221745129924e-05,
      "loss": 0.1811,
      "step": 5802
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2196379315497314,
      "learning_rate": 3.734790448034621e-05,
      "loss": 0.2168,
      "step": 5803
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.614734889695496,
      "learning_rate": 3.7343591023247315e-05,
      "loss": 0.4119,
      "step": 5804
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6099077412120177,
      "learning_rate": 3.733927708017241e-05,
      "loss": 0.0896,
      "step": 5805
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8296961483349327,
      "learning_rate": 3.7334962651291314e-05,
      "loss": 0.1923,
      "step": 5806
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6903749767636441,
      "learning_rate": 3.73306477367739e-05,
      "loss": 0.1151,
      "step": 5807
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7421005128053461,
      "learning_rate": 3.732633233679005e-05,
      "loss": 0.088,
      "step": 5808
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0922036084146414,
      "learning_rate": 3.7322016451509655e-05,
      "loss": 0.2788,
      "step": 5809
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9892637145761023,
      "learning_rate": 3.731770008110265e-05,
      "loss": 0.2679,
      "step": 5810
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8375458040869823,
      "learning_rate": 3.7313383225738964e-05,
      "loss": 0.369,
      "step": 5811
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2028477409062435,
      "learning_rate": 3.730906588558855e-05,
      "loss": 0.2239,
      "step": 5812
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9108601894312758,
      "learning_rate": 3.730474806082139e-05,
      "loss": 0.1259,
      "step": 5813
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0675660811541041,
      "learning_rate": 3.730042975160749e-05,
      "loss": 0.2207,
      "step": 5814
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8361243377196546,
      "learning_rate": 3.7296110958116844e-05,
      "loss": 0.1257,
      "step": 5815
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7247814192396532,
      "learning_rate": 3.729179168051951e-05,
      "loss": 0.103,
      "step": 5816
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8468239594647646,
      "learning_rate": 3.728747191898551e-05,
      "loss": 0.1755,
      "step": 5817
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6423836370352319,
      "learning_rate": 3.7283151673684946e-05,
      "loss": 0.0756,
      "step": 5818
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8136240278326354,
      "learning_rate": 3.7278830944787897e-05,
      "loss": 0.5992,
      "step": 5819
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2194384658020718,
      "learning_rate": 3.727450973246447e-05,
      "loss": 0.1896,
      "step": 5820
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7866211301390797,
      "learning_rate": 3.7270188036884803e-05,
      "loss": 0.0723,
      "step": 5821
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8985768631974386,
      "learning_rate": 3.726586585821904e-05,
      "loss": 0.2104,
      "step": 5822
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7148455849658606,
      "learning_rate": 3.726154319663734e-05,
      "loss": 0.5778,
      "step": 5823
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8739646577639373,
      "learning_rate": 3.72572200523099e-05,
      "loss": 0.1437,
      "step": 5824
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6734993811046669,
      "learning_rate": 3.725289642540693e-05,
      "loss": 0.2333,
      "step": 5825
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.98088714224671,
      "learning_rate": 3.724857231609864e-05,
      "loss": 0.1031,
      "step": 5826
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9108319490712421,
      "learning_rate": 3.724424772455528e-05,
      "loss": 0.2201,
      "step": 5827
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7768661018265389,
      "learning_rate": 3.723992265094711e-05,
      "loss": 0.124,
      "step": 5828
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9536270244805661,
      "learning_rate": 3.7235597095444416e-05,
      "loss": 0.2363,
      "step": 5829
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0848713512660142,
      "learning_rate": 3.72312710582175e-05,
      "loss": 0.201,
      "step": 5830
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.597675077850244,
      "learning_rate": 3.722694453943667e-05,
      "loss": 0.0832,
      "step": 5831
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9320807681037259,
      "learning_rate": 3.722261753927227e-05,
      "loss": 0.232,
      "step": 5832
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.0429013726185903,
      "learning_rate": 3.7218290057894656e-05,
      "loss": 0.4637,
      "step": 5833
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5296119129898194,
      "learning_rate": 3.7213962095474216e-05,
      "loss": 0.0596,
      "step": 5834
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7556839735249968,
      "learning_rate": 3.7209633652181326e-05,
      "loss": 0.1105,
      "step": 5835
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9027459358460935,
      "learning_rate": 3.720530472818641e-05,
      "loss": 0.2627,
      "step": 5836
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7709161579829287,
      "learning_rate": 3.7200975323659905e-05,
      "loss": 0.0869,
      "step": 5837
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5097067492399245,
      "learning_rate": 3.719664543877225e-05,
      "loss": 0.5392,
      "step": 5838
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.46571723494506645,
      "learning_rate": 3.719231507369393e-05,
      "loss": 0.0703,
      "step": 5839
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5748889191367041,
      "learning_rate": 3.718798422859542e-05,
      "loss": 0.0949,
      "step": 5840
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.37362104206756397,
      "learning_rate": 3.7183652903647243e-05,
      "loss": 0.0404,
      "step": 5841
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6915740070451061,
      "learning_rate": 3.7179321099019916e-05,
      "loss": 0.132,
      "step": 5842
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7644183441993259,
      "learning_rate": 3.717498881488399e-05,
      "loss": 0.2012,
      "step": 5843
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2621390587244528,
      "learning_rate": 3.717065605141003e-05,
      "loss": 0.2986,
      "step": 5844
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5032382032682903,
      "learning_rate": 3.716632280876862e-05,
      "loss": 0.2472,
      "step": 5845
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.658884966529726,
      "learning_rate": 3.716198908713035e-05,
      "loss": 0.0794,
      "step": 5846
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6871923788751951,
      "learning_rate": 3.715765488666587e-05,
      "loss": 0.0903,
      "step": 5847
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.417212755761196,
      "learning_rate": 3.7153320207545797e-05,
      "loss": 0.3448,
      "step": 5848
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9676197274144608,
      "learning_rate": 3.714898504994079e-05,
      "loss": 0.2356,
      "step": 5849
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2035682834359394,
      "learning_rate": 3.7144649414021543e-05,
      "loss": 0.4012,
      "step": 5850
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7082472251952184,
      "learning_rate": 3.714031329995874e-05,
      "loss": 0.1645,
      "step": 5851
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7575329272212554,
      "learning_rate": 3.7135976707923106e-05,
      "loss": 0.1505,
      "step": 5852
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7549356768825909,
      "learning_rate": 3.713163963808537e-05,
      "loss": 0.1851,
      "step": 5853
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3582054113044595,
      "learning_rate": 3.712730209061629e-05,
      "loss": 0.195,
      "step": 5854
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0660696757296035,
      "learning_rate": 3.7122964065686624e-05,
      "loss": 0.1351,
      "step": 5855
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.953357863323552,
      "learning_rate": 3.711862556346719e-05,
      "loss": 0.1659,
      "step": 5856
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6431401743372209,
      "learning_rate": 3.7114286584128774e-05,
      "loss": 0.0774,
      "step": 5857
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6641332942787987,
      "learning_rate": 3.710994712784221e-05,
      "loss": 0.1452,
      "step": 5858
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8208433393307164,
      "learning_rate": 3.7105607194778355e-05,
      "loss": 0.1896,
      "step": 5859
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9826815800538653,
      "learning_rate": 3.710126678510807e-05,
      "loss": 0.1582,
      "step": 5860
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6701244607565111,
      "learning_rate": 3.7096925899002236e-05,
      "loss": 0.0908,
      "step": 5861
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7485851833674537,
      "learning_rate": 3.709258453663176e-05,
      "loss": 0.1816,
      "step": 5862
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6903851178070433,
      "learning_rate": 3.708824269816756e-05,
      "loss": 0.1143,
      "step": 5863
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5991885617845312,
      "learning_rate": 3.70839003837806e-05,
      "loss": 0.0732,
      "step": 5864
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8652424895546607,
      "learning_rate": 3.70795575936418e-05,
      "loss": 0.2387,
      "step": 5865
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0309412455321465,
      "learning_rate": 3.707521432792217e-05,
      "loss": 0.1377,
      "step": 5866
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4902855577457567,
      "learning_rate": 3.7070870586792695e-05,
      "loss": 0.0631,
      "step": 5867
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.196454430442754,
      "learning_rate": 3.70665263704244e-05,
      "loss": 0.433,
      "step": 5868
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8106093038574576,
      "learning_rate": 3.7062181678988306e-05,
      "loss": 0.2053,
      "step": 5869
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8971144502964362,
      "learning_rate": 3.705783651265548e-05,
      "loss": 0.2635,
      "step": 5870
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6761443992826156,
      "learning_rate": 3.7053490871596984e-05,
      "loss": 0.0963,
      "step": 5871
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4947857869633702,
      "learning_rate": 3.7049144755983917e-05,
      "loss": 0.0514,
      "step": 5872
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4990102438527667,
      "learning_rate": 3.7044798165987384e-05,
      "loss": 0.0863,
      "step": 5873
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5586893661146048,
      "learning_rate": 3.704045110177852e-05,
      "loss": 0.0763,
      "step": 5874
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7463608255116795,
      "learning_rate": 3.703610356352847e-05,
      "loss": 0.118,
      "step": 5875
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8448643300445375,
      "learning_rate": 3.703175555140839e-05,
      "loss": 0.1596,
      "step": 5876
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4891800916069142,
      "learning_rate": 3.702740706558947e-05,
      "loss": 0.072,
      "step": 5877
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0564312937343128,
      "learning_rate": 3.7023058106242915e-05,
      "loss": 0.1364,
      "step": 5878
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6609715561362718,
      "learning_rate": 3.701870867353994e-05,
      "loss": 0.1489,
      "step": 5879
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8999556354266367,
      "learning_rate": 3.7014358767651805e-05,
      "loss": 0.1325,
      "step": 5880
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1085245634702057,
      "learning_rate": 3.701000838874974e-05,
      "loss": 0.1797,
      "step": 5881
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7872579851485498,
      "learning_rate": 3.700565753700505e-05,
      "loss": 0.1878,
      "step": 5882
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7090565295743924,
      "learning_rate": 3.700130621258901e-05,
      "loss": 0.1576,
      "step": 5883
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5751983799063596,
      "learning_rate": 3.699695441567295e-05,
      "loss": 0.7018,
      "step": 5884
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8331922225655298,
      "learning_rate": 3.699260214642819e-05,
      "loss": 0.1483,
      "step": 5885
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4209864581064647,
      "learning_rate": 3.69882494050261e-05,
      "loss": 0.0766,
      "step": 5886
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7185515644984339,
      "learning_rate": 3.6983896191638025e-05,
      "loss": 0.1202,
      "step": 5887
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8308642051356014,
      "learning_rate": 3.697954250643538e-05,
      "loss": 0.1961,
      "step": 5888
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.665700024398781,
      "learning_rate": 3.697518834958955e-05,
      "loss": 0.1148,
      "step": 5889
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6511094218579977,
      "learning_rate": 3.697083372127198e-05,
      "loss": 0.1212,
      "step": 5890
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5399471696167492,
      "learning_rate": 3.696647862165411e-05,
      "loss": 0.0798,
      "step": 5891
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6832760716544293,
      "learning_rate": 3.696212305090739e-05,
      "loss": 0.0883,
      "step": 5892
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7720641310843215,
      "learning_rate": 3.6957767009203315e-05,
      "loss": 0.1458,
      "step": 5893
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8487110778933401,
      "learning_rate": 3.695341049671339e-05,
      "loss": 0.2024,
      "step": 5894
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7465306766256014,
      "learning_rate": 3.694905351360912e-05,
      "loss": 0.118,
      "step": 5895
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.234216729606689,
      "learning_rate": 3.694469606006206e-05,
      "loss": 0.2685,
      "step": 5896
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6998033220729931,
      "learning_rate": 3.6940338136243745e-05,
      "loss": 0.1276,
      "step": 5897
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8801571315200494,
      "learning_rate": 3.6935979742325766e-05,
      "loss": 0.4427,
      "step": 5898
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8208787412869949,
      "learning_rate": 3.693162087847971e-05,
      "loss": 0.1345,
      "step": 5899
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8615744575237955,
      "learning_rate": 3.692726154487719e-05,
      "loss": 0.1134,
      "step": 5900
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9370416108051448,
      "learning_rate": 3.6922901741689826e-05,
      "loss": 0.1133,
      "step": 5901
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8194040963694127,
      "learning_rate": 3.691854146908929e-05,
      "loss": 0.2264,
      "step": 5902
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8124318151252623,
      "learning_rate": 3.691418072724721e-05,
      "loss": 0.1231,
      "step": 5903
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6557913560598317,
      "learning_rate": 3.690981951633532e-05,
      "loss": 0.1237,
      "step": 5904
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.691540001380015,
      "learning_rate": 3.69054578365253e-05,
      "loss": 0.1851,
      "step": 5905
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7622725440932383,
      "learning_rate": 3.690109568798886e-05,
      "loss": 0.149,
      "step": 5906
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9254850811280535,
      "learning_rate": 3.689673307089776e-05,
      "loss": 0.188,
      "step": 5907
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6510822050090461,
      "learning_rate": 3.689236998542376e-05,
      "loss": 0.3177,
      "step": 5908
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0441261124341013,
      "learning_rate": 3.688800643173862e-05,
      "loss": 0.145,
      "step": 5909
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6793924474949923,
      "learning_rate": 3.6883642410014154e-05,
      "loss": 0.085,
      "step": 5910
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6376921534219611,
      "learning_rate": 3.6879277920422153e-05,
      "loss": 0.0809,
      "step": 5911
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9633269222124324,
      "learning_rate": 3.687491296313449e-05,
      "loss": 0.1746,
      "step": 5912
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7720340995653072,
      "learning_rate": 3.6870547538322974e-05,
      "loss": 0.0979,
      "step": 5913
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2829939086730555,
      "learning_rate": 3.68661816461595e-05,
      "loss": 0.2266,
      "step": 5914
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8404301624588854,
      "learning_rate": 3.686181528681596e-05,
      "loss": 0.0957,
      "step": 5915
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.154076421425643,
      "learning_rate": 3.685744846046424e-05,
      "loss": 1.0052,
      "step": 5916
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5217550749246644,
      "learning_rate": 3.6853081167276284e-05,
      "loss": 0.0672,
      "step": 5917
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6499487572911438,
      "learning_rate": 3.684871340742402e-05,
      "loss": 0.1094,
      "step": 5918
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8043135563697817,
      "learning_rate": 3.684434518107942e-05,
      "loss": 0.1953,
      "step": 5919
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8252090143785422,
      "learning_rate": 3.6839976488414465e-05,
      "loss": 0.263,
      "step": 5920
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7768546922065759,
      "learning_rate": 3.683560732960115e-05,
      "loss": 0.1257,
      "step": 5921
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.724797574301473,
      "learning_rate": 3.683123770481149e-05,
      "loss": 0.1141,
      "step": 5922
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7623211637748263,
      "learning_rate": 3.682686761421752e-05,
      "loss": 0.1496,
      "step": 5923
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.554571902069361,
      "learning_rate": 3.68224970579913e-05,
      "loss": 0.1297,
      "step": 5924
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5707982589547348,
      "learning_rate": 3.681812603630489e-05,
      "loss": 0.082,
      "step": 5925
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8300765427065299,
      "learning_rate": 3.68137545493304e-05,
      "loss": 0.096,
      "step": 5926
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.84596547094812,
      "learning_rate": 3.680938259723991e-05,
      "loss": 0.1422,
      "step": 5927
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4139544879611663,
      "learning_rate": 3.6805010180205594e-05,
      "loss": 0.0638,
      "step": 5928
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8386655385969816,
      "learning_rate": 3.680063729839954e-05,
      "loss": 0.1365,
      "step": 5929
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9086228454643289,
      "learning_rate": 3.679626395199395e-05,
      "loss": 0.2951,
      "step": 5930
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7296269891547448,
      "learning_rate": 3.6791890141160983e-05,
      "loss": 0.1421,
      "step": 5931
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7382258953020969,
      "learning_rate": 3.6787515866072866e-05,
      "loss": 0.1818,
      "step": 5932
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7293990915721583,
      "learning_rate": 3.678314112690179e-05,
      "loss": 0.148,
      "step": 5933
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3252392665629906,
      "learning_rate": 3.6778765923820006e-05,
      "loss": 0.3349,
      "step": 5934
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7311465332156017,
      "learning_rate": 3.677439025699976e-05,
      "loss": 0.1297,
      "step": 5935
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4197636032615903,
      "learning_rate": 3.677001412661335e-05,
      "loss": 0.4163,
      "step": 5936
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8560995870988715,
      "learning_rate": 3.676563753283303e-05,
      "loss": 0.1816,
      "step": 5937
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3412852859892694,
      "learning_rate": 3.676126047583113e-05,
      "loss": 0.6147,
      "step": 5938
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.484483582051584,
      "learning_rate": 3.675688295577999e-05,
      "loss": 0.0393,
      "step": 5939
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8338485899927983,
      "learning_rate": 3.675250497285193e-05,
      "loss": 0.1276,
      "step": 5940
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2429384550136457,
      "learning_rate": 3.6748126527219326e-05,
      "loss": 0.3356,
      "step": 5941
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4071370553413811,
      "learning_rate": 3.674374761905457e-05,
      "loss": 0.2349,
      "step": 5942
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6672538034527218,
      "learning_rate": 3.673936824853004e-05,
      "loss": 0.1926,
      "step": 5943
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8463502099370429,
      "learning_rate": 3.6734988415818185e-05,
      "loss": 0.2042,
      "step": 5944
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7472565608253992,
      "learning_rate": 3.6730608121091416e-05,
      "loss": 0.2194,
      "step": 5945
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0599007258053923,
      "learning_rate": 3.672622736452219e-05,
      "loss": 0.1961,
      "step": 5946
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.65821055196083,
      "learning_rate": 3.672184614628301e-05,
      "loss": 0.1207,
      "step": 5947
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9223394543941839,
      "learning_rate": 3.671746446654633e-05,
      "loss": 0.1995,
      "step": 5948
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.063809835653731,
      "learning_rate": 3.6713082325484685e-05,
      "loss": 0.1739,
      "step": 5949
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9650385874459997,
      "learning_rate": 3.6708699723270585e-05,
      "loss": 0.2606,
      "step": 5950
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0370831070802837,
      "learning_rate": 3.670431666007659e-05,
      "loss": 0.3012,
      "step": 5951
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3089419690379214,
      "learning_rate": 3.669993313607526e-05,
      "loss": 0.2563,
      "step": 5952
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6469924459661315,
      "learning_rate": 3.6695549151439186e-05,
      "loss": 0.1064,
      "step": 5953
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6119526399005883,
      "learning_rate": 3.6691164706340944e-05,
      "loss": 0.095,
      "step": 5954
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9373304434142856,
      "learning_rate": 3.668677980095319e-05,
      "loss": 0.1826,
      "step": 5955
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9722583019003013,
      "learning_rate": 3.668239443544852e-05,
      "loss": 0.1505,
      "step": 5956
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8504997878773,
      "learning_rate": 3.667800860999961e-05,
      "loss": 0.1203,
      "step": 5957
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8810286623876059,
      "learning_rate": 3.6673622324779145e-05,
      "loss": 0.1963,
      "step": 5958
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.498173444042941,
      "learning_rate": 3.666923557995979e-05,
      "loss": 0.3987,
      "step": 5959
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6570997382377043,
      "learning_rate": 3.666484837571427e-05,
      "loss": 0.1021,
      "step": 5960
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6787496950192565,
      "learning_rate": 3.666046071221532e-05,
      "loss": 0.1329,
      "step": 5961
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5695518995867334,
      "learning_rate": 3.665607258963566e-05,
      "loss": 0.116,
      "step": 5962
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6556353685201076,
      "learning_rate": 3.665168400814808e-05,
      "loss": 0.204,
      "step": 5963
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6090848248370634,
      "learning_rate": 3.6647294967925346e-05,
      "loss": 0.0976,
      "step": 5964
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0051533180566161,
      "learning_rate": 3.664290546914026e-05,
      "loss": 0.1426,
      "step": 5965
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6922949984585499,
      "learning_rate": 3.6638515511965646e-05,
      "loss": 0.0869,
      "step": 5966
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7417139017665607,
      "learning_rate": 3.663412509657433e-05,
      "loss": 0.2062,
      "step": 5967
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5964159051497947,
      "learning_rate": 3.662973422313918e-05,
      "loss": 0.0827,
      "step": 5968
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8196375864310196,
      "learning_rate": 3.6625342891833046e-05,
      "loss": 0.1807,
      "step": 5969
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3160689565872177,
      "learning_rate": 3.662095110282884e-05,
      "loss": 0.2405,
      "step": 5970
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1161285480463068,
      "learning_rate": 3.661655885629946e-05,
      "loss": 0.1861,
      "step": 5971
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.99282518022206,
      "learning_rate": 3.661216615241783e-05,
      "loss": 0.2292,
      "step": 5972
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.68188630421536,
      "learning_rate": 3.6607772991356894e-05,
      "loss": 0.9297,
      "step": 5973
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6727625231168606,
      "learning_rate": 3.6603379373289614e-05,
      "loss": 0.1061,
      "step": 5974
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7745843019115364,
      "learning_rate": 3.6598985298388984e-05,
      "loss": 0.1557,
      "step": 5975
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4768225873009697,
      "learning_rate": 3.6594590766827976e-05,
      "loss": 0.2736,
      "step": 5976
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8222209103044367,
      "learning_rate": 3.6590195778779636e-05,
      "loss": 0.1643,
      "step": 5977
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.859454495888544,
      "learning_rate": 3.658580033441697e-05,
      "loss": 0.1834,
      "step": 5978
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9096864636092025,
      "learning_rate": 3.6581404433913043e-05,
      "loss": 0.2183,
      "step": 5979
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6768419152147789,
      "learning_rate": 3.657700807744092e-05,
      "loss": 0.1816,
      "step": 5980
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8107351673473239,
      "learning_rate": 3.65726112651737e-05,
      "loss": 0.1658,
      "step": 5981
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7910168385466124,
      "learning_rate": 3.656821399728448e-05,
      "loss": 0.1789,
      "step": 5982
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9796541934403749,
      "learning_rate": 3.656381627394638e-05,
      "loss": 0.1411,
      "step": 5983
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8328446459695723,
      "learning_rate": 3.6559418095332545e-05,
      "loss": 0.1872,
      "step": 5984
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7366657231069738,
      "learning_rate": 3.655501946161614e-05,
      "loss": 0.159,
      "step": 5985
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4516361882789151,
      "learning_rate": 3.6550620372970335e-05,
      "loss": 0.0585,
      "step": 5986
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6460022329535126,
      "learning_rate": 3.654622082956833e-05,
      "loss": 0.1506,
      "step": 5987
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.075447994526598,
      "learning_rate": 3.654182083158333e-05,
      "loss": 0.1724,
      "step": 5988
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6407747446708287,
      "learning_rate": 3.653742037918858e-05,
      "loss": 0.165,
      "step": 5989
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6912568416616798,
      "learning_rate": 3.653301947255732e-05,
      "loss": 0.1526,
      "step": 5990
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9776353255617766,
      "learning_rate": 3.6528618111862814e-05,
      "loss": 0.1802,
      "step": 5991
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.549880541815743,
      "learning_rate": 3.6524216297278345e-05,
      "loss": 0.0718,
      "step": 5992
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6172330166740323,
      "learning_rate": 3.6519814028977235e-05,
      "loss": 0.125,
      "step": 5993
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1451654100499375,
      "learning_rate": 3.651541130713279e-05,
      "loss": 0.1608,
      "step": 5994
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.695125903390768,
      "learning_rate": 3.651100813191834e-05,
      "loss": 0.0738,
      "step": 5995
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.443568717746922,
      "learning_rate": 3.650660450350725e-05,
      "loss": 0.2394,
      "step": 5996
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1223899094410161,
      "learning_rate": 3.65022004220729e-05,
      "loss": 0.2933,
      "step": 5997
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5306033782146764,
      "learning_rate": 3.649779588778867e-05,
      "loss": 0.0607,
      "step": 5998
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6653311148644966,
      "learning_rate": 3.649339090082799e-05,
      "loss": 0.1738,
      "step": 5999
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6112312773888282,
      "learning_rate": 3.648898546136426e-05,
      "loss": 0.2742,
      "step": 6000
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7455463791541536,
      "learning_rate": 3.6484579569570936e-05,
      "loss": 0.1008,
      "step": 6001
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.548446975172694,
      "learning_rate": 3.6480173225621486e-05,
      "loss": 0.1535,
      "step": 6002
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.044650382503656,
      "learning_rate": 3.647576642968939e-05,
      "loss": 0.3774,
      "step": 6003
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5901552278986598,
      "learning_rate": 3.647135918194815e-05,
      "loss": 0.1264,
      "step": 6004
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9194432626232764,
      "learning_rate": 3.646695148257128e-05,
      "loss": 0.197,
      "step": 6005
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7232669019600829,
      "learning_rate": 3.6462543331732305e-05,
      "loss": 0.1438,
      "step": 6006
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7447277136060436,
      "learning_rate": 3.645813472960479e-05,
      "loss": 0.1432,
      "step": 6007
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8548391220472523,
      "learning_rate": 3.645372567636229e-05,
      "loss": 0.1271,
      "step": 6008
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6113451378310826,
      "learning_rate": 3.644931617217841e-05,
      "loss": 0.1298,
      "step": 6009
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6751163470434276,
      "learning_rate": 3.644490621722674e-05,
      "loss": 0.1041,
      "step": 6010
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5518067179256013,
      "learning_rate": 3.644049581168091e-05,
      "loss": 0.0483,
      "step": 6011
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6952634382312818,
      "learning_rate": 3.6436084955714564e-05,
      "loss": 0.1335,
      "step": 6012
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.50965530804865,
      "learning_rate": 3.643167364950136e-05,
      "loss": 0.0687,
      "step": 6013
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5336516460999597,
      "learning_rate": 3.642726189321496e-05,
      "loss": 0.1079,
      "step": 6014
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4436219920748126,
      "learning_rate": 3.642284968702909e-05,
      "loss": 0.0409,
      "step": 6015
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0802037009410688,
      "learning_rate": 3.641843703111743e-05,
      "loss": 0.2092,
      "step": 6016
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6394444367553627,
      "learning_rate": 3.6414023925653716e-05,
      "loss": 0.0776,
      "step": 6017
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7226672134028245,
      "learning_rate": 3.6409610370811704e-05,
      "loss": 0.1027,
      "step": 6018
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5595493607338727,
      "learning_rate": 3.6405196366765155e-05,
      "loss": 0.1048,
      "step": 6019
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.190233089203992,
      "learning_rate": 3.640078191368785e-05,
      "loss": 0.2779,
      "step": 6020
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0393880583030841,
      "learning_rate": 3.63963670117536e-05,
      "loss": 0.197,
      "step": 6021
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9720016627772772,
      "learning_rate": 3.639195166113619e-05,
      "loss": 0.3064,
      "step": 6022
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6564793355047227,
      "learning_rate": 3.6387535862009505e-05,
      "loss": 0.1335,
      "step": 6023
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5701729962309044,
      "learning_rate": 3.638311961454736e-05,
      "loss": 0.0663,
      "step": 6024
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7482764874747615,
      "learning_rate": 3.637870291892363e-05,
      "loss": 0.1859,
      "step": 6025
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7664520546669382,
      "learning_rate": 3.637428577531222e-05,
      "loss": 0.1339,
      "step": 6026
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7499360114475792,
      "learning_rate": 3.6369868183887036e-05,
      "loss": 0.1128,
      "step": 6027
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6563933884140478,
      "learning_rate": 3.636545014482198e-05,
      "loss": 0.1044,
      "step": 6028
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8390850016953658,
      "learning_rate": 3.636103165829101e-05,
      "loss": 0.1636,
      "step": 6029
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7239916307348119,
      "learning_rate": 3.635661272446808e-05,
      "loss": 0.1199,
      "step": 6030
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6243510678523858,
      "learning_rate": 3.6352193343527176e-05,
      "loss": 0.3261,
      "step": 6031
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8823366702106684,
      "learning_rate": 3.6347773515642275e-05,
      "loss": 0.1266,
      "step": 6032
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5556158292638482,
      "learning_rate": 3.63433532409874e-05,
      "loss": 0.0799,
      "step": 6033
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9401725753889316,
      "learning_rate": 3.633893251973658e-05,
      "loss": 0.2619,
      "step": 6034
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6684225486553134,
      "learning_rate": 3.6334511352063863e-05,
      "loss": 0.1589,
      "step": 6035
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6519317893447701,
      "learning_rate": 3.633008973814331e-05,
      "loss": 0.1456,
      "step": 6036
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8810560567284913,
      "learning_rate": 3.6325667678149e-05,
      "loss": 0.1552,
      "step": 6037
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5625530250781346,
      "learning_rate": 3.632124517225504e-05,
      "loss": 0.0726,
      "step": 6038
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7672581578219209,
      "learning_rate": 3.6316822220635546e-05,
      "loss": 0.1228,
      "step": 6039
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7528582320030923,
      "learning_rate": 3.6312398823464644e-05,
      "loss": 0.1684,
      "step": 6040
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0434189264445095,
      "learning_rate": 3.63079749809165e-05,
      "loss": 0.2724,
      "step": 6041
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8840185980955634,
      "learning_rate": 3.6303550693165264e-05,
      "loss": 0.2758,
      "step": 6042
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0689071833185464,
      "learning_rate": 3.6299125960385154e-05,
      "loss": 0.2421,
      "step": 6043
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8259263221521488,
      "learning_rate": 3.6294700782750346e-05,
      "loss": 0.1804,
      "step": 6044
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6455049029752831,
      "learning_rate": 3.629027516043508e-05,
      "loss": 0.1134,
      "step": 6045
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6451975525709984,
      "learning_rate": 3.6285849093613585e-05,
      "loss": 0.1914,
      "step": 6046
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.593846977854369,
      "learning_rate": 3.628142258246012e-05,
      "loss": 0.1154,
      "step": 6047
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6404489762137975,
      "learning_rate": 3.6276995627148966e-05,
      "loss": 0.1726,
      "step": 6048
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4182173361816794,
      "learning_rate": 3.627256822785441e-05,
      "loss": 0.2482,
      "step": 6049
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8386083312677097,
      "learning_rate": 3.626814038475077e-05,
      "loss": 0.1793,
      "step": 6050
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7808221255210543,
      "learning_rate": 3.626371209801237e-05,
      "loss": 0.1314,
      "step": 6051
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7123623760412658,
      "learning_rate": 3.625928336781354e-05,
      "loss": 0.1694,
      "step": 6052
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6576114157211495,
      "learning_rate": 3.625485419432866e-05,
      "loss": 0.0996,
      "step": 6053
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6636954214966758,
      "learning_rate": 3.62504245777321e-05,
      "loss": 0.115,
      "step": 6054
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.767896387118706,
      "learning_rate": 3.624599451819828e-05,
      "loss": 0.1315,
      "step": 6055
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5583453440531565,
      "learning_rate": 3.624156401590158e-05,
      "loss": 0.5068,
      "step": 6056
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.71546560505059,
      "learning_rate": 3.623713307101645e-05,
      "loss": 0.1789,
      "step": 6057
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4913342745806469,
      "learning_rate": 3.623270168371734e-05,
      "loss": 0.1388,
      "step": 6058
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4967692694236042,
      "learning_rate": 3.622826985417871e-05,
      "loss": 0.0752,
      "step": 6059
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2333934969722387,
      "learning_rate": 3.622383758257505e-05,
      "loss": 0.2197,
      "step": 6060
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6880774965628473,
      "learning_rate": 3.621940486908086e-05,
      "loss": 0.1273,
      "step": 6061
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2970279595230465,
      "learning_rate": 3.6214971713870666e-05,
      "loss": 0.3221,
      "step": 6062
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.704717133842486,
      "learning_rate": 3.6210538117119e-05,
      "loss": 0.2028,
      "step": 6063
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1299793015552682,
      "learning_rate": 3.62061040790004e-05,
      "loss": 0.2488,
      "step": 6064
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6960298612124417,
      "learning_rate": 3.6201669599689465e-05,
      "loss": 0.1499,
      "step": 6065
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7101936840263846,
      "learning_rate": 3.6197234679360755e-05,
      "loss": 0.1055,
      "step": 6066
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5162520784587488,
      "learning_rate": 3.619279931818891e-05,
      "loss": 0.0585,
      "step": 6067
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8186723574853062,
      "learning_rate": 3.618836351634852e-05,
      "loss": 0.1542,
      "step": 6068
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.684702419859217,
      "learning_rate": 3.618392727401425e-05,
      "loss": 0.1008,
      "step": 6069
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1094387733634383,
      "learning_rate": 3.617949059136074e-05,
      "loss": 0.3111,
      "step": 6070
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.16045105864691,
      "learning_rate": 3.617505346856268e-05,
      "loss": 0.1356,
      "step": 6071
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6492837247182698,
      "learning_rate": 3.617061590579475e-05,
      "loss": 0.1136,
      "step": 6072
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8381999941231206,
      "learning_rate": 3.6166177903231676e-05,
      "loss": 0.1692,
      "step": 6073
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6158895570315369,
      "learning_rate": 3.616173946104816e-05,
      "loss": 0.1384,
      "step": 6074
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5978614646878344,
      "learning_rate": 3.615730057941897e-05,
      "loss": 0.4027,
      "step": 6075
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5976780265999625,
      "learning_rate": 3.6152861258518864e-05,
      "loss": 0.0858,
      "step": 6076
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5330084336619635,
      "learning_rate": 3.614842149852261e-05,
      "loss": 0.1155,
      "step": 6077
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7552186686807562,
      "learning_rate": 3.614398129960501e-05,
      "loss": 0.138,
      "step": 6078
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8065861955945582,
      "learning_rate": 3.6139540661940884e-05,
      "loss": 0.124,
      "step": 6079
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5416829097533147,
      "learning_rate": 3.6135099585705065e-05,
      "loss": 0.2823,
      "step": 6080
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.09403453707402,
      "learning_rate": 3.613065807107239e-05,
      "loss": 0.2835,
      "step": 6081
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7865247840902679,
      "learning_rate": 3.612621611821773e-05,
      "loss": 0.1434,
      "step": 6082
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5388374383335207,
      "learning_rate": 3.612177372731596e-05,
      "loss": 0.0684,
      "step": 6083
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1587215034040241,
      "learning_rate": 3.6117330898542e-05,
      "loss": 0.2233,
      "step": 6084
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8661938528358775,
      "learning_rate": 3.6112887632070746e-05,
      "loss": 0.1342,
      "step": 6085
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.704307920389712,
      "learning_rate": 3.610844392807715e-05,
      "loss": 0.1201,
      "step": 6086
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6869881884647118,
      "learning_rate": 3.6103999786736144e-05,
      "loss": 0.1436,
      "step": 6087
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0527739248039059,
      "learning_rate": 3.609955520822272e-05,
      "loss": 0.4327,
      "step": 6088
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.091149171509354,
      "learning_rate": 3.609511019271184e-05,
      "loss": 0.3717,
      "step": 6089
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5603020697581231,
      "learning_rate": 3.6090664740378534e-05,
      "loss": 0.1081,
      "step": 6090
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1540359738486337,
      "learning_rate": 3.60862188513978e-05,
      "loss": 0.334,
      "step": 6091
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5617119027491078,
      "learning_rate": 3.6081772525944686e-05,
      "loss": 0.245,
      "step": 6092
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5487461492705794,
      "learning_rate": 3.6077325764194246e-05,
      "loss": 0.0839,
      "step": 6093
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9200507570020591,
      "learning_rate": 3.607287856632156e-05,
      "loss": 0.2515,
      "step": 6094
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7984463224478904,
      "learning_rate": 3.606843093250169e-05,
      "loss": 0.129,
      "step": 6095
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8091767967280837,
      "learning_rate": 3.6063982862909777e-05,
      "loss": 0.2076,
      "step": 6096
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7420776521301934,
      "learning_rate": 3.605953435772093e-05,
      "loss": 0.088,
      "step": 6097
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.778835847822675,
      "learning_rate": 3.605508541711028e-05,
      "loss": 0.1273,
      "step": 6098
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9977594231427304,
      "learning_rate": 3.605063604125299e-05,
      "loss": 0.193,
      "step": 6099
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.558830903985157,
      "learning_rate": 3.604618623032426e-05,
      "loss": 0.0642,
      "step": 6100
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7086627909594022,
      "learning_rate": 3.6041735984499246e-05,
      "loss": 0.1262,
      "step": 6101
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5589476134287917,
      "learning_rate": 3.603728530395318e-05,
      "loss": 0.0705,
      "step": 6102
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7799567680780364,
      "learning_rate": 3.603283418886128e-05,
      "loss": 0.1901,
      "step": 6103
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5761490913394597,
      "learning_rate": 3.602838263939879e-05,
      "loss": 0.1105,
      "step": 6104
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7692194731969125,
      "learning_rate": 3.6023930655740966e-05,
      "loss": 0.1971,
      "step": 6105
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.011360490724273,
      "learning_rate": 3.6019478238063105e-05,
      "loss": 0.2032,
      "step": 6106
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7646310774094993,
      "learning_rate": 3.601502538654047e-05,
      "loss": 0.1037,
      "step": 6107
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3084279675089299,
      "learning_rate": 3.6010572101348414e-05,
      "loss": 0.284,
      "step": 6108
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7730618770597468,
      "learning_rate": 3.600611838266222e-05,
      "loss": 0.239,
      "step": 6109
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9727705682495001,
      "learning_rate": 3.6001664230657276e-05,
      "loss": 0.182,
      "step": 6110
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5928017636867122,
      "learning_rate": 3.5997209645508914e-05,
      "loss": 0.0973,
      "step": 6111
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.079681372333988,
      "learning_rate": 3.599275462739253e-05,
      "loss": 0.1577,
      "step": 6112
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8027660797792682,
      "learning_rate": 3.598829917648352e-05,
      "loss": 0.0686,
      "step": 6113
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5713994005336233,
      "learning_rate": 3.59838432929573e-05,
      "loss": 0.114,
      "step": 6114
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.527266007461284,
      "learning_rate": 3.5979386976989294e-05,
      "loss": 0.353,
      "step": 6115
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8034067221898702,
      "learning_rate": 3.597493022875496e-05,
      "loss": 0.1755,
      "step": 6116
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5565947728317187,
      "learning_rate": 3.597047304842974e-05,
      "loss": 0.0703,
      "step": 6117
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6790103371491277,
      "learning_rate": 3.5966015436189146e-05,
      "loss": 0.1021,
      "step": 6118
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7014796186448828,
      "learning_rate": 3.5961557392208666e-05,
      "loss": 0.0985,
      "step": 6119
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8141887767735739,
      "learning_rate": 3.5957098916663815e-05,
      "loss": 0.205,
      "step": 6120
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.49267844142969935,
      "learning_rate": 3.5952640009730126e-05,
      "loss": 0.0695,
      "step": 6121
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7388840784614077,
      "learning_rate": 3.594818067158315e-05,
      "loss": 0.1118,
      "step": 6122
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0440041690775694,
      "learning_rate": 3.594372090239845e-05,
      "loss": 0.223,
      "step": 6123
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8429720723548976,
      "learning_rate": 3.593926070235163e-05,
      "loss": 0.1396,
      "step": 6124
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0031261971878003,
      "learning_rate": 3.593480007161826e-05,
      "loss": 0.2304,
      "step": 6125
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7534814643963574,
      "learning_rate": 3.593033901037399e-05,
      "loss": 0.3057,
      "step": 6126
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6134477345043892,
      "learning_rate": 3.5925877518794424e-05,
      "loss": 0.0667,
      "step": 6127
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.516536947589831,
      "learning_rate": 3.5921415597055234e-05,
      "loss": 0.0686,
      "step": 6128
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4734991674913768,
      "learning_rate": 3.591695324533209e-05,
      "loss": 0.0717,
      "step": 6129
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.799258218526546,
      "learning_rate": 3.591249046380067e-05,
      "loss": 0.1821,
      "step": 6130
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7602745759256954,
      "learning_rate": 3.590802725263668e-05,
      "loss": 0.128,
      "step": 6131
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7687218770592658,
      "learning_rate": 3.590356361201584e-05,
      "loss": 0.171,
      "step": 6132
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.724799644804349,
      "learning_rate": 3.589909954211388e-05,
      "loss": 0.0739,
      "step": 6133
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6072456236750453,
      "learning_rate": 3.589463504310657e-05,
      "loss": 0.1379,
      "step": 6134
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5969059128248209,
      "learning_rate": 3.589017011516967e-05,
      "loss": 0.5325,
      "step": 6135
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5621121440789416,
      "learning_rate": 3.588570475847896e-05,
      "loss": 0.0966,
      "step": 6136
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0545421124500707,
      "learning_rate": 3.588123897321026e-05,
      "loss": 0.2296,
      "step": 6137
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6995052494878892,
      "learning_rate": 3.5876772759539375e-05,
      "loss": 0.1267,
      "step": 6138
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5744497266658988,
      "learning_rate": 3.587230611764216e-05,
      "loss": 0.0747,
      "step": 6139
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6505893172373763,
      "learning_rate": 3.5867839047694456e-05,
      "loss": 0.0717,
      "step": 6140
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6214024479072536,
      "learning_rate": 3.586337154987214e-05,
      "loss": 0.1096,
      "step": 6141
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7930018430955883,
      "learning_rate": 3.58589036243511e-05,
      "loss": 0.1813,
      "step": 6142
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8004511777227892,
      "learning_rate": 3.585443527130724e-05,
      "loss": 0.1715,
      "step": 6143
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0810673290948003,
      "learning_rate": 3.5849966490916496e-05,
      "loss": 0.1711,
      "step": 6144
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9083412239856175,
      "learning_rate": 3.584549728335478e-05,
      "loss": 0.1598,
      "step": 6145
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1164591172502853,
      "learning_rate": 3.584102764879808e-05,
      "loss": 0.29,
      "step": 6146
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8428389335521143,
      "learning_rate": 3.5836557587422334e-05,
      "loss": 0.0862,
      "step": 6147
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6329027456518457,
      "learning_rate": 3.583208709940356e-05,
      "loss": 0.451,
      "step": 6148
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7526658620674279,
      "learning_rate": 3.582761618491775e-05,
      "loss": 0.1009,
      "step": 6149
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6554223276890292,
      "learning_rate": 3.582314484414093e-05,
      "loss": 0.136,
      "step": 6150
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9770582665553205,
      "learning_rate": 3.581867307724914e-05,
      "loss": 0.1823,
      "step": 6151
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5126567583896384,
      "learning_rate": 3.581420088441845e-05,
      "loss": 0.0642,
      "step": 6152
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6878782204919122,
      "learning_rate": 3.5809728265824906e-05,
      "loss": 0.1059,
      "step": 6153
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4654477696337198,
      "learning_rate": 3.580525522164462e-05,
      "loss": 0.0744,
      "step": 6154
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5357481636807488,
      "learning_rate": 3.580078175205369e-05,
      "loss": 0.0506,
      "step": 6155
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7811199061747065,
      "learning_rate": 3.579630785722824e-05,
      "loss": 0.1484,
      "step": 6156
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6087202029401257,
      "learning_rate": 3.5791833537344416e-05,
      "loss": 0.0967,
      "step": 6157
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8459248491891371,
      "learning_rate": 3.578735879257837e-05,
      "loss": 0.2861,
      "step": 6158
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8897871027031571,
      "learning_rate": 3.5782883623106275e-05,
      "loss": 0.13,
      "step": 6159
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.47569671483919607,
      "learning_rate": 3.5778408029104335e-05,
      "loss": 0.062,
      "step": 6160
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6019892418516138,
      "learning_rate": 3.5773932010748734e-05,
      "loss": 0.0705,
      "step": 6161
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7105698147992917,
      "learning_rate": 3.5769455568215716e-05,
      "loss": 0.1502,
      "step": 6162
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6239062887004697,
      "learning_rate": 3.5764978701681504e-05,
      "loss": 0.0872,
      "step": 6163
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0213174640630478,
      "learning_rate": 3.576050141132238e-05,
      "loss": 0.2035,
      "step": 6164
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.45896335643802433,
      "learning_rate": 3.5756023697314586e-05,
      "loss": 0.0894,
      "step": 6165
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5638952485318814,
      "learning_rate": 3.575154555983444e-05,
      "loss": 0.0788,
      "step": 6166
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8098797404532442,
      "learning_rate": 3.574706699905824e-05,
      "loss": 0.112,
      "step": 6167
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7756230846546167,
      "learning_rate": 3.5742588015162316e-05,
      "loss": 0.2046,
      "step": 6168
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.998327619161038,
      "learning_rate": 3.5738108608322994e-05,
      "loss": 0.211,
      "step": 6169
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5837722638895254,
      "learning_rate": 3.5733628778716646e-05,
      "loss": 0.0918,
      "step": 6170
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9489561306920193,
      "learning_rate": 3.572914852651964e-05,
      "loss": 0.1133,
      "step": 6171
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6228055113569952,
      "learning_rate": 3.5724667851908364e-05,
      "loss": 0.127,
      "step": 6172
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5621879469281061,
      "learning_rate": 3.572018675505923e-05,
      "loss": 0.0881,
      "step": 6173
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6172262511377076,
      "learning_rate": 3.571570523614866e-05,
      "loss": 0.1516,
      "step": 6174
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6030278635593047,
      "learning_rate": 3.571122329535309e-05,
      "loss": 0.1161,
      "step": 6175
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7652245278713651,
      "learning_rate": 3.570674093284899e-05,
      "loss": 0.1929,
      "step": 6176
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8481295423301838,
      "learning_rate": 3.5702258148812824e-05,
      "loss": 0.2111,
      "step": 6177
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1966080917543807,
      "learning_rate": 3.5697774943421086e-05,
      "loss": 0.4507,
      "step": 6178
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0148022380280106,
      "learning_rate": 3.569329131685027e-05,
      "loss": 0.1466,
      "step": 6179
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8249842262578717,
      "learning_rate": 3.568880726927693e-05,
      "loss": 0.1186,
      "step": 6180
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5540123381881029,
      "learning_rate": 3.568432280087758e-05,
      "loss": 0.1005,
      "step": 6181
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9812427315488739,
      "learning_rate": 3.5679837911828784e-05,
      "loss": 0.281,
      "step": 6182
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6110596788547321,
      "learning_rate": 3.5675352602307113e-05,
      "loss": 0.0858,
      "step": 6183
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6373609162544532,
      "learning_rate": 3.567086687248916e-05,
      "loss": 0.0923,
      "step": 6184
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7425205514070368,
      "learning_rate": 3.566638072255153e-05,
      "loss": 0.1078,
      "step": 6185
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.502314288724323,
      "learning_rate": 3.5661894152670847e-05,
      "loss": 0.4472,
      "step": 6186
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5165045402902557,
      "learning_rate": 3.565740716302375e-05,
      "loss": 0.0938,
      "step": 6187
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5799402874955306,
      "learning_rate": 3.56529197537869e-05,
      "loss": 0.0537,
      "step": 6188
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5158946877807241,
      "learning_rate": 3.564843192513696e-05,
      "loss": 0.0713,
      "step": 6189
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5955840014921312,
      "learning_rate": 3.5643943677250624e-05,
      "loss": 0.0853,
      "step": 6190
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9390196179191561,
      "learning_rate": 3.5639455010304596e-05,
      "loss": 0.3482,
      "step": 6191
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5117121312979231,
      "learning_rate": 3.56349659244756e-05,
      "loss": 0.0821,
      "step": 6192
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3401778107012703,
      "learning_rate": 3.563047641994037e-05,
      "loss": 0.2672,
      "step": 6193
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0261617926821587,
      "learning_rate": 3.5625986496875675e-05,
      "loss": 0.1301,
      "step": 6194
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0931069877811908,
      "learning_rate": 3.562149615545827e-05,
      "loss": 0.3655,
      "step": 6195
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.758767470356073,
      "learning_rate": 3.5617005395864956e-05,
      "loss": 0.1175,
      "step": 6196
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7376232872948091,
      "learning_rate": 3.561251421827252e-05,
      "loss": 0.2439,
      "step": 6197
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8095629065708896,
      "learning_rate": 3.560802262285779e-05,
      "loss": 0.109,
      "step": 6198
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5768096361083451,
      "learning_rate": 3.560353060979762e-05,
      "loss": 0.0736,
      "step": 6199
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9361485700573587,
      "learning_rate": 3.559903817926885e-05,
      "loss": 0.2282,
      "step": 6200
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6327912326055318,
      "learning_rate": 3.5594545331448345e-05,
      "loss": 0.117,
      "step": 6201
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7753547355623224,
      "learning_rate": 3.5590052066512997e-05,
      "loss": 0.1418,
      "step": 6202
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5763932864706687,
      "learning_rate": 3.558555838463972e-05,
      "loss": 0.0814,
      "step": 6203
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0319391094848724,
      "learning_rate": 3.55810642860054e-05,
      "loss": 0.1547,
      "step": 6204
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6320712253467334,
      "learning_rate": 3.5576569770787015e-05,
      "loss": 0.0921,
      "step": 6205
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6482181587249436,
      "learning_rate": 3.55720748391615e-05,
      "loss": 0.1292,
      "step": 6206
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4757492253043771,
      "learning_rate": 3.556757949130581e-05,
      "loss": 0.0736,
      "step": 6207
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.3727267578779085,
      "learning_rate": 3.556308372739695e-05,
      "loss": 0.4029,
      "step": 6208
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8687482472984229,
      "learning_rate": 3.555858754761191e-05,
      "loss": 0.1588,
      "step": 6209
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.783444211256555,
      "learning_rate": 3.555409095212771e-05,
      "loss": 0.1684,
      "step": 6210
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2764490953662713,
      "learning_rate": 3.5549593941121384e-05,
      "loss": 0.4658,
      "step": 6211
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6661245883011834,
      "learning_rate": 3.5545096514769985e-05,
      "loss": 0.2067,
      "step": 6212
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6136198185678903,
      "learning_rate": 3.554059867325058e-05,
      "loss": 0.0953,
      "step": 6213
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9286763471115638,
      "learning_rate": 3.553610041674025e-05,
      "loss": 0.1681,
      "step": 6214
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.695873084960394,
      "learning_rate": 3.5531601745416105e-05,
      "loss": 0.237,
      "step": 6215
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.609910040581679,
      "learning_rate": 3.552710265945524e-05,
      "loss": 0.1009,
      "step": 6216
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1750591432268602,
      "learning_rate": 3.5522603159034806e-05,
      "loss": 0.259,
      "step": 6217
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5846647940398771,
      "learning_rate": 3.5518103244331935e-05,
      "loss": 0.0941,
      "step": 6218
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7559115679494491,
      "learning_rate": 3.551360291552381e-05,
      "loss": 0.1213,
      "step": 6219
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7721428025878869,
      "learning_rate": 3.55091021727876e-05,
      "loss": 0.128,
      "step": 6220
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4860257961922195,
      "learning_rate": 3.550460101630051e-05,
      "loss": 0.0855,
      "step": 6221
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7251594257398327,
      "learning_rate": 3.5500099446239745e-05,
      "loss": 0.1063,
      "step": 6222
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0219862065328942,
      "learning_rate": 3.549559746278254e-05,
      "loss": 0.1927,
      "step": 6223
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5937698162704483,
      "learning_rate": 3.549109506610614e-05,
      "loss": 0.0798,
      "step": 6224
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8116139762116875,
      "learning_rate": 3.5486592256387805e-05,
      "loss": 0.1451,
      "step": 6225
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.825739997861172,
      "learning_rate": 3.548208903380483e-05,
      "loss": 0.1546,
      "step": 6226
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8172306108039717,
      "learning_rate": 3.547758539853448e-05,
      "loss": 0.2123,
      "step": 6227
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7412660854912959,
      "learning_rate": 3.54730813507541e-05,
      "loss": 0.1827,
      "step": 6228
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8315016062405945,
      "learning_rate": 3.5468576890641e-05,
      "loss": 0.2098,
      "step": 6229
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5912485297020933,
      "learning_rate": 3.546407201837251e-05,
      "loss": 0.1081,
      "step": 6230
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6308538219111494,
      "learning_rate": 3.5459566734126026e-05,
      "loss": 0.0878,
      "step": 6231
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6858014214473167,
      "learning_rate": 3.5455061038078893e-05,
      "loss": 0.1311,
      "step": 6232
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6528357296771949,
      "learning_rate": 3.545055493040851e-05,
      "loss": 0.1445,
      "step": 6233
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6523019292300328,
      "learning_rate": 3.54460484112923e-05,
      "loss": 0.1377,
      "step": 6234
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7788256293509341,
      "learning_rate": 3.544154148090767e-05,
      "loss": 0.1141,
      "step": 6235
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5933060155402679,
      "learning_rate": 3.543703413943207e-05,
      "loss": 0.0722,
      "step": 6236
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4884079156228261,
      "learning_rate": 3.543252638704296e-05,
      "loss": 0.0889,
      "step": 6237
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9905709631478625,
      "learning_rate": 3.5428018223917804e-05,
      "loss": 0.1728,
      "step": 6238
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3256606636548358,
      "learning_rate": 3.54235096502341e-05,
      "loss": 0.4537,
      "step": 6239
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8288852074274846,
      "learning_rate": 3.5419000666169344e-05,
      "loss": 0.1033,
      "step": 6240
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9420839465335178,
      "learning_rate": 3.541449127190107e-05,
      "loss": 0.1673,
      "step": 6241
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8492043188841769,
      "learning_rate": 3.540998146760681e-05,
      "loss": 0.2026,
      "step": 6242
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6403107207812533,
      "learning_rate": 3.5405471253464126e-05,
      "loss": 0.0948,
      "step": 6243
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8462683315272486,
      "learning_rate": 3.5400960629650574e-05,
      "loss": 0.1171,
      "step": 6244
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7338750703244127,
      "learning_rate": 3.539644959634375e-05,
      "loss": 0.1238,
      "step": 6245
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7647581474185008,
      "learning_rate": 3.539193815372125e-05,
      "loss": 0.1584,
      "step": 6246
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8837587030076812,
      "learning_rate": 3.538742630196071e-05,
      "loss": 0.1873,
      "step": 6247
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.818492966766535,
      "learning_rate": 3.538291404123974e-05,
      "loss": 0.7466,
      "step": 6248
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9613701713990813,
      "learning_rate": 3.537840137173601e-05,
      "loss": 0.241,
      "step": 6249
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9074696808297097,
      "learning_rate": 3.5373888293627174e-05,
      "loss": 0.1149,
      "step": 6250
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5581914668223994,
      "learning_rate": 3.536937480709093e-05,
      "loss": 0.0627,
      "step": 6251
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6491372038518688,
      "learning_rate": 3.5364860912304955e-05,
      "loss": 0.0615,
      "step": 6252
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0130162270840561,
      "learning_rate": 3.536034660944699e-05,
      "loss": 0.2384,
      "step": 6253
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6719824498891304,
      "learning_rate": 3.535583189869474e-05,
      "loss": 0.1216,
      "step": 6254
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6444026472097829,
      "learning_rate": 3.535131678022598e-05,
      "loss": 0.141,
      "step": 6255
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9531194327460465,
      "learning_rate": 3.5346801254218455e-05,
      "loss": 0.2574,
      "step": 6256
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5940865252311271,
      "learning_rate": 3.534228532084995e-05,
      "loss": 0.0924,
      "step": 6257
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8207105225405104,
      "learning_rate": 3.533776898029826e-05,
      "loss": 0.0949,
      "step": 6258
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1605144289619573,
      "learning_rate": 3.533325223274119e-05,
      "loss": 0.4833,
      "step": 6259
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5696711678578935,
      "learning_rate": 3.532873507835658e-05,
      "loss": 0.1366,
      "step": 6260
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0204440265561021,
      "learning_rate": 3.532421751732227e-05,
      "loss": 0.2152,
      "step": 6261
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5117158657519449,
      "learning_rate": 3.531969954981612e-05,
      "loss": 0.0631,
      "step": 6262
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7057114890201225,
      "learning_rate": 3.531518117601599e-05,
      "loss": 0.1709,
      "step": 6263
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5985763923596124,
      "learning_rate": 3.531066239609979e-05,
      "loss": 0.2622,
      "step": 6264
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.724284732067528,
      "learning_rate": 3.530614321024543e-05,
      "loss": 0.0959,
      "step": 6265
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6052344944880121,
      "learning_rate": 3.530162361863082e-05,
      "loss": 0.0985,
      "step": 6266
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6683068309695649,
      "learning_rate": 3.52971036214339e-05,
      "loss": 0.0796,
      "step": 6267
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6374708070743502,
      "learning_rate": 3.5292583218832644e-05,
      "loss": 0.0933,
      "step": 6268
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6462253278779753,
      "learning_rate": 3.5288062411005e-05,
      "loss": 0.2703,
      "step": 6269
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.576080207796838,
      "learning_rate": 3.528354119812897e-05,
      "loss": 0.1016,
      "step": 6270
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5751648978780626,
      "learning_rate": 3.527901958038255e-05,
      "loss": 0.0833,
      "step": 6271
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6477294889297304,
      "learning_rate": 3.527449755794376e-05,
      "loss": 0.1054,
      "step": 6272
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8143713066797094,
      "learning_rate": 3.526997513099064e-05,
      "loss": 0.1594,
      "step": 6273
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7404506931381427,
      "learning_rate": 3.526545229970124e-05,
      "loss": 0.1717,
      "step": 6274
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5258679207318104,
      "learning_rate": 3.526092906425363e-05,
      "loss": 0.4495,
      "step": 6275
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7576147604878809,
      "learning_rate": 3.5256405424825875e-05,
      "loss": 0.2464,
      "step": 6276
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.49349950637210316,
      "learning_rate": 3.525188138159611e-05,
      "loss": 0.0491,
      "step": 6277
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5639595396941566,
      "learning_rate": 3.5247356934742406e-05,
      "loss": 0.2363,
      "step": 6278
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6869027613341747,
      "learning_rate": 3.5242832084442925e-05,
      "loss": 0.1811,
      "step": 6279
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0540014799236301,
      "learning_rate": 3.523830683087581e-05,
      "loss": 0.2137,
      "step": 6280
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6372097743157339,
      "learning_rate": 3.523378117421921e-05,
      "loss": 0.0883,
      "step": 6281
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8645333036160603,
      "learning_rate": 3.5229255114651305e-05,
      "loss": 0.1354,
      "step": 6282
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.021813688022122,
      "learning_rate": 3.522472865235031e-05,
      "loss": 0.1758,
      "step": 6283
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4827208497835883,
      "learning_rate": 3.5220201787494407e-05,
      "loss": 0.0547,
      "step": 6284
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7397848239473598,
      "learning_rate": 3.5215674520261844e-05,
      "loss": 0.0916,
      "step": 6285
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7924444702326141,
      "learning_rate": 3.5211146850830845e-05,
      "loss": 0.1929,
      "step": 6286
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7405152796457446,
      "learning_rate": 3.5206618779379686e-05,
      "loss": 0.151,
      "step": 6287
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6573901048959829,
      "learning_rate": 3.520209030608662e-05,
      "loss": 0.1083,
      "step": 6288
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6721097898317027,
      "learning_rate": 3.519756143112996e-05,
      "loss": 0.0902,
      "step": 6289
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1777984097806713,
      "learning_rate": 3.519303215468799e-05,
      "loss": 0.6985,
      "step": 6290
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9048223382670103,
      "learning_rate": 3.518850247693905e-05,
      "loss": 0.2126,
      "step": 6291
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0593465925320835,
      "learning_rate": 3.518397239806145e-05,
      "loss": 0.2934,
      "step": 6292
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9093152019421331,
      "learning_rate": 3.5179441918233564e-05,
      "loss": 0.2844,
      "step": 6293
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.492959177417465,
      "learning_rate": 3.517491103763375e-05,
      "loss": 0.3242,
      "step": 6294
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0093696162363013,
      "learning_rate": 3.51703797564404e-05,
      "loss": 0.1964,
      "step": 6295
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6540649246897206,
      "learning_rate": 3.516584807483191e-05,
      "loss": 0.0989,
      "step": 6296
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0434644342837427,
      "learning_rate": 3.51613159929867e-05,
      "loss": 0.2622,
      "step": 6297
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0938323011951243,
      "learning_rate": 3.515678351108319e-05,
      "loss": 0.3163,
      "step": 6298
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1868750664802987,
      "learning_rate": 3.5152250629299844e-05,
      "loss": 0.3039,
      "step": 6299
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6767578031880753,
      "learning_rate": 3.5147717347815115e-05,
      "loss": 0.1733,
      "step": 6300
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9755212908957686,
      "learning_rate": 3.514318366680747e-05,
      "loss": 0.214,
      "step": 6301
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9688174705891146,
      "learning_rate": 3.513864958645542e-05,
      "loss": 0.3442,
      "step": 6302
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7659661366150481,
      "learning_rate": 3.513411510693747e-05,
      "loss": 0.1551,
      "step": 6303
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7307411803570261,
      "learning_rate": 3.512958022843214e-05,
      "loss": 0.1448,
      "step": 6304
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8560287817446919,
      "learning_rate": 3.512504495111799e-05,
      "loss": 0.3848,
      "step": 6305
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8705857914160249,
      "learning_rate": 3.512050927517355e-05,
      "loss": 0.3567,
      "step": 6306
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8632436126736731,
      "learning_rate": 3.511597320077742e-05,
      "loss": 0.2458,
      "step": 6307
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6909886977702299,
      "learning_rate": 3.511143672810816e-05,
      "loss": 0.0924,
      "step": 6308
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8682225946629567,
      "learning_rate": 3.51068998573444e-05,
      "loss": 0.2347,
      "step": 6309
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5765711067264397,
      "learning_rate": 3.5102362588664744e-05,
      "loss": 0.1273,
      "step": 6310
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8794295353671937,
      "learning_rate": 3.509782492224784e-05,
      "loss": 0.1478,
      "step": 6311
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.692729910425165,
      "learning_rate": 3.509328685827232e-05,
      "loss": 0.0928,
      "step": 6312
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4885861727927154,
      "learning_rate": 3.5088748396916874e-05,
      "loss": 0.382,
      "step": 6313
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.829930000852684,
      "learning_rate": 3.508420953836017e-05,
      "loss": 0.0885,
      "step": 6314
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8707317050274161,
      "learning_rate": 3.5079670282780906e-05,
      "loss": 0.1658,
      "step": 6315
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7258055138038314,
      "learning_rate": 3.50751306303578e-05,
      "loss": 0.1133,
      "step": 6316
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8773769940864066,
      "learning_rate": 3.5070590581269586e-05,
      "loss": 0.213,
      "step": 6317
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.971208198196552,
      "learning_rate": 3.506605013569499e-05,
      "loss": 0.1514,
      "step": 6318
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6162552462705845,
      "learning_rate": 3.5061509293812806e-05,
      "loss": 0.1141,
      "step": 6319
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6212177287578382,
      "learning_rate": 3.505696805580178e-05,
      "loss": 0.0671,
      "step": 6320
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6442075806208512,
      "learning_rate": 3.505242642184071e-05,
      "loss": 0.1412,
      "step": 6321
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7718470392365574,
      "learning_rate": 3.504788439210841e-05,
      "loss": 0.2019,
      "step": 6322
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9846646597265656,
      "learning_rate": 3.50433419667837e-05,
      "loss": 0.1191,
      "step": 6323
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.141439468493823,
      "learning_rate": 3.5038799146045425e-05,
      "loss": 0.2444,
      "step": 6324
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5890259718478775,
      "learning_rate": 3.5034255930072427e-05,
      "loss": 0.091,
      "step": 6325
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7977524517930679,
      "learning_rate": 3.5029712319043586e-05,
      "loss": 0.172,
      "step": 6326
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.634596104636893,
      "learning_rate": 3.502516831313778e-05,
      "loss": 0.0979,
      "step": 6327
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0117582590950702,
      "learning_rate": 3.5020623912533915e-05,
      "loss": 0.159,
      "step": 6328
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8191540595017417,
      "learning_rate": 3.5016079117410904e-05,
      "loss": 0.1977,
      "step": 6329
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6712846863383481,
      "learning_rate": 3.501153392794768e-05,
      "loss": 0.1097,
      "step": 6330
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8961652511217405,
      "learning_rate": 3.500698834432319e-05,
      "loss": 0.1634,
      "step": 6331
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1892683005688118,
      "learning_rate": 3.500244236671639e-05,
      "loss": 0.3577,
      "step": 6332
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2872016438870113,
      "learning_rate": 3.4997895995306273e-05,
      "loss": 0.3511,
      "step": 6333
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4585818647027358,
      "learning_rate": 3.499334923027183e-05,
      "loss": 0.2828,
      "step": 6334
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9519086299412953,
      "learning_rate": 3.498880207179206e-05,
      "loss": 0.2326,
      "step": 6335
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6902795819208621,
      "learning_rate": 3.498425452004599e-05,
      "loss": 0.1517,
      "step": 6336
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7934379882417815,
      "learning_rate": 3.497970657521267e-05,
      "loss": 0.0884,
      "step": 6337
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9494678109767299,
      "learning_rate": 3.497515823747115e-05,
      "loss": 0.3425,
      "step": 6338
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7156201319079321,
      "learning_rate": 3.49706095070005e-05,
      "loss": 0.1188,
      "step": 6339
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8536746331696882,
      "learning_rate": 3.49660603839798e-05,
      "loss": 0.1602,
      "step": 6340
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7625325025626507,
      "learning_rate": 3.4961510868588175e-05,
      "loss": 0.1465,
      "step": 6341
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.131786845859634,
      "learning_rate": 3.495696096100472e-05,
      "loss": 0.5847,
      "step": 6342
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8287912042983961,
      "learning_rate": 3.495241066140858e-05,
      "loss": 0.1192,
      "step": 6343
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8027114415075336,
      "learning_rate": 3.494785996997889e-05,
      "loss": 0.1007,
      "step": 6344
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.694626092235532,
      "learning_rate": 3.494330888689483e-05,
      "loss": 0.358,
      "step": 6345
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6246341371284296,
      "learning_rate": 3.493875741233558e-05,
      "loss": 0.1493,
      "step": 6346
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7220216765795199,
      "learning_rate": 3.4934205546480316e-05,
      "loss": 0.1131,
      "step": 6347
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8932850365130285,
      "learning_rate": 3.492965328950826e-05,
      "loss": 0.2062,
      "step": 6348
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.737903575061426,
      "learning_rate": 3.492510064159864e-05,
      "loss": 0.1238,
      "step": 6349
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3749293960120514,
      "learning_rate": 3.4920547602930704e-05,
      "loss": 0.4276,
      "step": 6350
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6612379876074477,
      "learning_rate": 3.4915994173683685e-05,
      "loss": 0.0832,
      "step": 6351
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7511173813354972,
      "learning_rate": 3.4911440354036884e-05,
      "loss": 0.1455,
      "step": 6352
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8604741756599324,
      "learning_rate": 3.490688614416956e-05,
      "loss": 0.1332,
      "step": 6353
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8773674298365556,
      "learning_rate": 3.490233154426104e-05,
      "loss": 0.1983,
      "step": 6354
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6802043054605802,
      "learning_rate": 3.489777655449062e-05,
      "loss": 0.1102,
      "step": 6355
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0894339091070147,
      "learning_rate": 3.489322117503765e-05,
      "loss": 0.2269,
      "step": 6356
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9123067357689837,
      "learning_rate": 3.488866540608147e-05,
      "loss": 0.2089,
      "step": 6357
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6760450885819224,
      "learning_rate": 3.4884109247801445e-05,
      "loss": 0.0944,
      "step": 6358
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1346900608919694,
      "learning_rate": 3.487955270037695e-05,
      "loss": 0.4128,
      "step": 6359
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7791250140646717,
      "learning_rate": 3.48749957639874e-05,
      "loss": 0.1574,
      "step": 6360
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8191007865608689,
      "learning_rate": 3.487043843881217e-05,
      "loss": 0.1808,
      "step": 6361
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3488082391034653,
      "learning_rate": 3.486588072503072e-05,
      "loss": 0.1974,
      "step": 6362
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6962672735241253,
      "learning_rate": 3.486132262282247e-05,
      "loss": 0.1704,
      "step": 6363
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.791214019638825,
      "learning_rate": 3.485676413236688e-05,
      "loss": 0.4666,
      "step": 6364
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0096622287354617,
      "learning_rate": 3.485220525384342e-05,
      "loss": 0.3857,
      "step": 6365
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.485104146257041,
      "learning_rate": 3.4847645987431584e-05,
      "loss": 0.7857,
      "step": 6366
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7215910613577327,
      "learning_rate": 3.484308633331087e-05,
      "loss": 0.1637,
      "step": 6367
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.710689386432153,
      "learning_rate": 3.483852629166079e-05,
      "loss": 0.0731,
      "step": 6368
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.39401963661824335,
      "learning_rate": 3.483396586266087e-05,
      "loss": 0.0677,
      "step": 6369
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7218273039177729,
      "learning_rate": 3.482940504649067e-05,
      "loss": 0.0819,
      "step": 6370
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9784861013758215,
      "learning_rate": 3.4824843843329756e-05,
      "loss": 0.1195,
      "step": 6371
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6815346771863762,
      "learning_rate": 3.482028225335769e-05,
      "loss": 0.0951,
      "step": 6372
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.620154624651133,
      "learning_rate": 3.4815720276754084e-05,
      "loss": 0.0905,
      "step": 6373
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6302276797619528,
      "learning_rate": 3.481115791369853e-05,
      "loss": 0.103,
      "step": 6374
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0397207893473697,
      "learning_rate": 3.480659516437066e-05,
      "loss": 0.1559,
      "step": 6375
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7611263262240205,
      "learning_rate": 3.4802032028950106e-05,
      "loss": 0.1125,
      "step": 6376
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7170717270842593,
      "learning_rate": 3.4797468507616524e-05,
      "loss": 0.0974,
      "step": 6377
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4755584554719126,
      "learning_rate": 3.47929046005496e-05,
      "loss": 0.0949,
      "step": 6378
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6030701489768006,
      "learning_rate": 3.4788340307928985e-05,
      "loss": 0.088,
      "step": 6379
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.48904252161366685,
      "learning_rate": 3.4783775629934414e-05,
      "loss": 0.0581,
      "step": 6380
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5549567334175747,
      "learning_rate": 3.4779210566745565e-05,
      "loss": 0.0997,
      "step": 6381
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.151333585380553,
      "learning_rate": 3.477464511854221e-05,
      "loss": 0.7482,
      "step": 6382
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9166367390514948,
      "learning_rate": 3.477007928550405e-05,
      "loss": 0.1856,
      "step": 6383
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8823160285690363,
      "learning_rate": 3.4765513067810876e-05,
      "loss": 0.1666,
      "step": 6384
    },
    {
      "epoch": 0.39,
      "grad_norm": 9.591374356533208,
      "learning_rate": 3.476094646564245e-05,
      "loss": 0.3784,
      "step": 6385
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5992496660063613,
      "learning_rate": 3.475637947917858e-05,
      "loss": 0.1369,
      "step": 6386
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.230220626694032,
      "learning_rate": 3.4751812108599044e-05,
      "loss": 0.1446,
      "step": 6387
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.419071731285655,
      "learning_rate": 3.4747244354083684e-05,
      "loss": 0.3588,
      "step": 6388
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1860465594538225,
      "learning_rate": 3.474267621581232e-05,
      "loss": 0.1575,
      "step": 6389
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7454650131297151,
      "learning_rate": 3.4738107693964815e-05,
      "loss": 0.1186,
      "step": 6390
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6398842514834663,
      "learning_rate": 3.473353878872103e-05,
      "loss": 0.1077,
      "step": 6391
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3808075651318308,
      "learning_rate": 3.472896950026086e-05,
      "loss": 0.3889,
      "step": 6392
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7278720437283338,
      "learning_rate": 3.472439982876417e-05,
      "loss": 0.1106,
      "step": 6393
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8140183847095208,
      "learning_rate": 3.47198297744109e-05,
      "loss": 0.1763,
      "step": 6394
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8091066512500342,
      "learning_rate": 3.471525933738096e-05,
      "loss": 0.1096,
      "step": 6395
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6597600629568023,
      "learning_rate": 3.47106885178543e-05,
      "loss": 0.1185,
      "step": 6396
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6628499674662373,
      "learning_rate": 3.470611731601088e-05,
      "loss": 0.1176,
      "step": 6397
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1970692475193776,
      "learning_rate": 3.470154573203065e-05,
      "loss": 0.1202,
      "step": 6398
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1283624676714905,
      "learning_rate": 3.4696973766093626e-05,
      "loss": 0.1767,
      "step": 6399
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6225557725929406,
      "learning_rate": 3.4692401418379784e-05,
      "loss": 0.1194,
      "step": 6400
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6992745535362721,
      "learning_rate": 3.468782868906916e-05,
      "loss": 0.0977,
      "step": 6401
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6311222695637594,
      "learning_rate": 3.468325557834178e-05,
      "loss": 0.1006,
      "step": 6402
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.705106036603063,
      "learning_rate": 3.467868208637768e-05,
      "loss": 0.096,
      "step": 6403
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7003977469655096,
      "learning_rate": 3.4674108213356936e-05,
      "loss": 0.1187,
      "step": 6404
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5778207890336512,
      "learning_rate": 3.466953395945961e-05,
      "loss": 0.1056,
      "step": 6405
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7971579000994276,
      "learning_rate": 3.466495932486581e-05,
      "loss": 0.144,
      "step": 6406
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0452822772347599,
      "learning_rate": 3.4660384309755626e-05,
      "loss": 0.167,
      "step": 6407
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0095638045828044,
      "learning_rate": 3.46558089143092e-05,
      "loss": 0.1883,
      "step": 6408
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9279504210731547,
      "learning_rate": 3.4651233138706654e-05,
      "loss": 0.2608,
      "step": 6409
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3937314437803754,
      "learning_rate": 3.464665698312814e-05,
      "loss": 0.2442,
      "step": 6410
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3173111713741732,
      "learning_rate": 3.4642080447753824e-05,
      "loss": 0.3929,
      "step": 6411
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7275606552551742,
      "learning_rate": 3.463750353276389e-05,
      "loss": 0.1213,
      "step": 6412
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1264966844781275,
      "learning_rate": 3.463292623833854e-05,
      "loss": 0.1613,
      "step": 6413
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.020589285385968,
      "learning_rate": 3.462834856465798e-05,
      "loss": 0.2116,
      "step": 6414
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0103726251141532,
      "learning_rate": 3.4623770511902434e-05,
      "loss": 0.1891,
      "step": 6415
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9009758593029189,
      "learning_rate": 3.461919208025215e-05,
      "loss": 0.2514,
      "step": 6416
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7103479710249707,
      "learning_rate": 3.461461326988737e-05,
      "loss": 0.0981,
      "step": 6417
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8562517370444738,
      "learning_rate": 3.461003408098839e-05,
      "loss": 0.1516,
      "step": 6418
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.6553003288896204,
      "learning_rate": 3.460545451373547e-05,
      "loss": 0.349,
      "step": 6419
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7468831032673992,
      "learning_rate": 3.460087456830893e-05,
      "loss": 0.1154,
      "step": 6420
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5052850697841936,
      "learning_rate": 3.4596294244889066e-05,
      "loss": 0.0662,
      "step": 6421
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7188100035527581,
      "learning_rate": 3.459171354365623e-05,
      "loss": 0.0955,
      "step": 6422
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6560604843372572,
      "learning_rate": 3.458713246479075e-05,
      "loss": 0.1081,
      "step": 6423
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9211465568305667,
      "learning_rate": 3.4582551008473e-05,
      "loss": 0.1836,
      "step": 6424
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7467755603392433,
      "learning_rate": 3.457796917488335e-05,
      "loss": 0.1425,
      "step": 6425
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.228655196466473,
      "learning_rate": 3.4573386964202187e-05,
      "loss": 0.3201,
      "step": 6426
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0909744374942696,
      "learning_rate": 3.4568804376609914e-05,
      "loss": 0.1872,
      "step": 6427
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.7396541150483213,
      "learning_rate": 3.456422141228696e-05,
      "loss": 0.3419,
      "step": 6428
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7039490850067063,
      "learning_rate": 3.455963807141376e-05,
      "loss": 0.177,
      "step": 6429
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4884599710290074,
      "learning_rate": 3.4555054354170745e-05,
      "loss": 0.4111,
      "step": 6430
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.619076593934459,
      "learning_rate": 3.4550470260738403e-05,
      "loss": 0.1242,
      "step": 6431
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9878707666475783,
      "learning_rate": 3.454588579129719e-05,
      "loss": 0.2843,
      "step": 6432
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.3826103735798076,
      "learning_rate": 3.454130094602763e-05,
      "loss": 0.3956,
      "step": 6433
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2113992645259022,
      "learning_rate": 3.453671572511021e-05,
      "loss": 0.3243,
      "step": 6434
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7140689563692417,
      "learning_rate": 3.453213012872545e-05,
      "loss": 0.104,
      "step": 6435
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8859814930246253,
      "learning_rate": 3.452754415705389e-05,
      "loss": 0.1965,
      "step": 6436
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7574202426739258,
      "learning_rate": 3.45229578102761e-05,
      "loss": 0.1588,
      "step": 6437
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.713844348449138,
      "learning_rate": 3.4518371088572635e-05,
      "loss": 0.1641,
      "step": 6438
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0389498700488335,
      "learning_rate": 3.451378399212407e-05,
      "loss": 0.2119,
      "step": 6439
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1870852617284784,
      "learning_rate": 3.4509196521111014e-05,
      "loss": 0.3037,
      "step": 6440
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.843656506942778,
      "learning_rate": 3.450460867571408e-05,
      "loss": 0.1644,
      "step": 6441
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0405533240764915,
      "learning_rate": 3.4500020456113886e-05,
      "loss": 0.2283,
      "step": 6442
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8386108089619875,
      "learning_rate": 3.449543186249108e-05,
      "loss": 0.129,
      "step": 6443
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8804148092845476,
      "learning_rate": 3.449084289502632e-05,
      "loss": 0.1633,
      "step": 6444
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8849638323331306,
      "learning_rate": 3.448625355390027e-05,
      "loss": 0.1076,
      "step": 6445
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2089792751108372,
      "learning_rate": 3.448166383929362e-05,
      "loss": 0.2801,
      "step": 6446
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5686555430339223,
      "learning_rate": 3.447707375138706e-05,
      "loss": 0.1446,
      "step": 6447
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4926602168158674,
      "learning_rate": 3.447248329036133e-05,
      "loss": 0.4528,
      "step": 6448
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7613146276867976,
      "learning_rate": 3.4467892456397134e-05,
      "loss": 0.1436,
      "step": 6449
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9378140836688895,
      "learning_rate": 3.4463301249675235e-05,
      "loss": 0.1453,
      "step": 6450
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1140122687729777,
      "learning_rate": 3.445870967037638e-05,
      "loss": 0.3477,
      "step": 6451
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0141114036172914,
      "learning_rate": 3.4454117718681346e-05,
      "loss": 0.1733,
      "step": 6452
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6337826561477844,
      "learning_rate": 3.444952539477092e-05,
      "loss": 0.1379,
      "step": 6453
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9560622366547853,
      "learning_rate": 3.4444932698825906e-05,
      "loss": 0.2111,
      "step": 6454
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.021416902056131,
      "learning_rate": 3.444033963102712e-05,
      "loss": 0.2137,
      "step": 6455
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0944529354995878,
      "learning_rate": 3.443574619155541e-05,
      "loss": 0.1884,
      "step": 6456
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9082495741220463,
      "learning_rate": 3.4431152380591594e-05,
      "loss": 0.2201,
      "step": 6457
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3938185675101977,
      "learning_rate": 3.442655819831656e-05,
      "loss": 0.3831,
      "step": 6458
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5582823879694163,
      "learning_rate": 3.442196364491117e-05,
      "loss": 0.094,
      "step": 6459
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1440853955129227,
      "learning_rate": 3.441736872055632e-05,
      "loss": 0.2927,
      "step": 6460
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8331160114921483,
      "learning_rate": 3.441277342543291e-05,
      "loss": 0.1484,
      "step": 6461
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0373542120797976,
      "learning_rate": 3.440817775972188e-05,
      "loss": 0.2185,
      "step": 6462
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7018929540574993,
      "learning_rate": 3.440358172360413e-05,
      "loss": 0.1166,
      "step": 6463
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1686456047157436,
      "learning_rate": 3.439898531726064e-05,
      "loss": 0.1336,
      "step": 6464
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6812530529361197,
      "learning_rate": 3.439438854087236e-05,
      "loss": 0.1562,
      "step": 6465
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.1667531397526356,
      "learning_rate": 3.4389791394620266e-05,
      "loss": 0.6029,
      "step": 6466
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7056026911486966,
      "learning_rate": 3.438519387868536e-05,
      "loss": 0.1993,
      "step": 6467
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.497242929655026,
      "learning_rate": 3.438059599324864e-05,
      "loss": 0.9525,
      "step": 6468
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5486023886507502,
      "learning_rate": 3.437599773849114e-05,
      "loss": 0.1413,
      "step": 6469
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8229981273233415,
      "learning_rate": 3.4371399114593886e-05,
      "loss": 0.1679,
      "step": 6470
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7481315046300421,
      "learning_rate": 3.436680012173793e-05,
      "loss": 0.1233,
      "step": 6471
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2138004093024417,
      "learning_rate": 3.436220076010435e-05,
      "loss": 0.3748,
      "step": 6472
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9235178446739168,
      "learning_rate": 3.4357601029874195e-05,
      "loss": 0.0552,
      "step": 6473
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.42034841063863987,
      "learning_rate": 3.43530009312286e-05,
      "loss": 0.0654,
      "step": 6474
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7443805305883331,
      "learning_rate": 3.434840046434865e-05,
      "loss": 0.0843,
      "step": 6475
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8644292194112556,
      "learning_rate": 3.434379962941548e-05,
      "loss": 0.138,
      "step": 6476
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6941264913756905,
      "learning_rate": 3.433919842661021e-05,
      "loss": 0.0892,
      "step": 6477
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6566869276790627,
      "learning_rate": 3.433459685611401e-05,
      "loss": 0.1683,
      "step": 6478
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1014744248889967,
      "learning_rate": 3.4329994918108046e-05,
      "loss": 0.2466,
      "step": 6479
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6410721466065487,
      "learning_rate": 3.432539261277349e-05,
      "loss": 0.1254,
      "step": 6480
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7740527062061091,
      "learning_rate": 3.432078994029155e-05,
      "loss": 0.1042,
      "step": 6481
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6635775185191322,
      "learning_rate": 3.4316186900843416e-05,
      "loss": 0.0904,
      "step": 6482
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9281914818785001,
      "learning_rate": 3.4311583494610346e-05,
      "loss": 0.1989,
      "step": 6483
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.020907555292438,
      "learning_rate": 3.4306979721773543e-05,
      "loss": 0.1887,
      "step": 6484
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7525277382626133,
      "learning_rate": 3.430237558251429e-05,
      "loss": 0.1441,
      "step": 6485
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8175209187895226,
      "learning_rate": 3.429777107701384e-05,
      "loss": 0.1274,
      "step": 6486
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.620105390245389,
      "learning_rate": 3.429316620545348e-05,
      "loss": 0.1748,
      "step": 6487
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.729192813518475,
      "learning_rate": 3.428856096801451e-05,
      "loss": 0.1666,
      "step": 6488
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.918854527967581,
      "learning_rate": 3.428395536487823e-05,
      "loss": 0.4126,
      "step": 6489
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7042582636259902,
      "learning_rate": 3.4279349396225976e-05,
      "loss": 0.4526,
      "step": 6490
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5952126739504825,
      "learning_rate": 3.427474306223909e-05,
      "loss": 0.0917,
      "step": 6491
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7130132781710717,
      "learning_rate": 3.4270136363098915e-05,
      "loss": 0.107,
      "step": 6492
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9496695835803516,
      "learning_rate": 3.426552929898684e-05,
      "loss": 0.2234,
      "step": 6493
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6712060907556793,
      "learning_rate": 3.4260921870084224e-05,
      "loss": 0.1351,
      "step": 6494
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7091739514545128,
      "learning_rate": 3.425631407657248e-05,
      "loss": 0.1203,
      "step": 6495
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7643130006908614,
      "learning_rate": 3.4251705918633025e-05,
      "loss": 0.1582,
      "step": 6496
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6552937638697892,
      "learning_rate": 3.424709739644727e-05,
      "loss": 0.1217,
      "step": 6497
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7644110032020983,
      "learning_rate": 3.4242488510196665e-05,
      "loss": 0.1449,
      "step": 6498
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7020138511150856,
      "learning_rate": 3.4237879260062665e-05,
      "loss": 0.1265,
      "step": 6499
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7314714571447879,
      "learning_rate": 3.4233269646226736e-05,
      "loss": 0.1203,
      "step": 6500
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9589629162128261,
      "learning_rate": 3.422865966887037e-05,
      "loss": 0.1757,
      "step": 6501
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9619279008432274,
      "learning_rate": 3.4224049328175045e-05,
      "loss": 0.3198,
      "step": 6502
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.032660522158123,
      "learning_rate": 3.42194386243223e-05,
      "loss": 0.2839,
      "step": 6503
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7436177073837171,
      "learning_rate": 3.421482755749364e-05,
      "loss": 0.1109,
      "step": 6504
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7705966865389831,
      "learning_rate": 3.421021612787062e-05,
      "loss": 0.1204,
      "step": 6505
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7375548408829986,
      "learning_rate": 3.420560433563479e-05,
      "loss": 0.1272,
      "step": 6506
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.228846875664852,
      "learning_rate": 3.420099218096772e-05,
      "loss": 0.2549,
      "step": 6507
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7777690741617578,
      "learning_rate": 3.419637966405099e-05,
      "loss": 0.2226,
      "step": 6508
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5141173368051422,
      "learning_rate": 3.4191766785066206e-05,
      "loss": 0.085,
      "step": 6509
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7118457829922809,
      "learning_rate": 3.4187153544194974e-05,
      "loss": 0.1182,
      "step": 6510
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4548541028545659,
      "learning_rate": 3.418253994161892e-05,
      "loss": 0.07,
      "step": 6511
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9778378215477664,
      "learning_rate": 3.417792597751969e-05,
      "loss": 0.1979,
      "step": 6512
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6019502613911744,
      "learning_rate": 3.417331165207894e-05,
      "loss": 0.0717,
      "step": 6513
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5579343506863,
      "learning_rate": 3.4168696965478334e-05,
      "loss": 0.0811,
      "step": 6514
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2321650765542964,
      "learning_rate": 3.416408191789956e-05,
      "loss": 0.2273,
      "step": 6515
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7124337476641168,
      "learning_rate": 3.4159466509524305e-05,
      "loss": 0.0903,
      "step": 6516
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.003989744392147,
      "learning_rate": 3.41548507405343e-05,
      "loss": 0.2428,
      "step": 6517
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.989657348000941,
      "learning_rate": 3.4150234611111256e-05,
      "loss": 0.2242,
      "step": 6518
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0901230528398056,
      "learning_rate": 3.4145618121436914e-05,
      "loss": 0.3471,
      "step": 6519
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1504513663229448,
      "learning_rate": 3.4141001271693035e-05,
      "loss": 0.3328,
      "step": 6520
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6519693728600374,
      "learning_rate": 3.413638406206139e-05,
      "loss": 0.1123,
      "step": 6521
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9460867520503059,
      "learning_rate": 3.413176649272375e-05,
      "loss": 0.3157,
      "step": 6522
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6779465449217923,
      "learning_rate": 3.4127148563861924e-05,
      "loss": 0.0852,
      "step": 6523
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6874746953310834,
      "learning_rate": 3.4122530275657704e-05,
      "loss": 0.1137,
      "step": 6524
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6210971394517542,
      "learning_rate": 3.411791162829294e-05,
      "loss": 0.1132,
      "step": 6525
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.742991744484894,
      "learning_rate": 3.411329262194946e-05,
      "loss": 0.1524,
      "step": 6526
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7734446017153447,
      "learning_rate": 3.410867325680912e-05,
      "loss": 0.2518,
      "step": 6527
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6940333674606756,
      "learning_rate": 3.410405353305378e-05,
      "loss": 0.1189,
      "step": 6528
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8140418586695984,
      "learning_rate": 3.4099433450865336e-05,
      "loss": 0.2531,
      "step": 6529
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8826827897695698,
      "learning_rate": 3.4094813010425674e-05,
      "loss": 0.3048,
      "step": 6530
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8159505804794659,
      "learning_rate": 3.40901922119167e-05,
      "loss": 0.1547,
      "step": 6531
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.581874215279988,
      "learning_rate": 3.408557105552035e-05,
      "loss": 0.1192,
      "step": 6532
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5969597828239575,
      "learning_rate": 3.4080949541418556e-05,
      "loss": 0.0651,
      "step": 6533
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8208127486003975,
      "learning_rate": 3.407632766979326e-05,
      "loss": 0.2431,
      "step": 6534
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8666389388532394,
      "learning_rate": 3.4071705440826454e-05,
      "loss": 0.2228,
      "step": 6535
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1710144577391837,
      "learning_rate": 3.4067082854700095e-05,
      "loss": 0.2399,
      "step": 6536
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.850924077838126,
      "learning_rate": 3.4062459911596186e-05,
      "loss": 0.1227,
      "step": 6537
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5636523587899372,
      "learning_rate": 3.405783661169674e-05,
      "loss": 0.1076,
      "step": 6538
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2260779911746642,
      "learning_rate": 3.405321295518377e-05,
      "loss": 0.3226,
      "step": 6539
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6877605380412565,
      "learning_rate": 3.4048588942239313e-05,
      "loss": 0.1771,
      "step": 6540
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2343226221335335,
      "learning_rate": 3.404396457304543e-05,
      "loss": 0.3961,
      "step": 6541
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2661104376387338,
      "learning_rate": 3.403933984778418e-05,
      "loss": 0.3496,
      "step": 6542
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6008090208191049,
      "learning_rate": 3.403471476663765e-05,
      "loss": 0.0723,
      "step": 6543
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2917723012344333,
      "learning_rate": 3.403008932978791e-05,
      "loss": 0.2277,
      "step": 6544
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8152703752669852,
      "learning_rate": 3.402546353741709e-05,
      "loss": 0.1452,
      "step": 6545
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2001654603405956,
      "learning_rate": 3.40208373897073e-05,
      "loss": 0.262,
      "step": 6546
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6340135559646314,
      "learning_rate": 3.4016210886840686e-05,
      "loss": 0.0883,
      "step": 6547
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7159657561403205,
      "learning_rate": 3.401158402899937e-05,
      "loss": 0.1544,
      "step": 6548
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0330441552140606,
      "learning_rate": 3.4006956816365545e-05,
      "loss": 0.1533,
      "step": 6549
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6582817751398292,
      "learning_rate": 3.4002329249121366e-05,
      "loss": 0.0922,
      "step": 6550
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2087289754005393,
      "learning_rate": 3.3997701327449044e-05,
      "loss": 0.2321,
      "step": 6551
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7115403695292789,
      "learning_rate": 3.3993073051530765e-05,
      "loss": 0.1142,
      "step": 6552
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8797654423050941,
      "learning_rate": 3.398844442154876e-05,
      "loss": 0.2547,
      "step": 6553
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7622933436530444,
      "learning_rate": 3.3983815437685254e-05,
      "loss": 0.1645,
      "step": 6554
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0472239238184309,
      "learning_rate": 3.3979186100122496e-05,
      "loss": 0.163,
      "step": 6555
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5130377338018342,
      "learning_rate": 3.397455640904274e-05,
      "loss": 0.0823,
      "step": 6556
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9405913047012435,
      "learning_rate": 3.3969926364628276e-05,
      "loss": 0.2371,
      "step": 6557
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1112371976428657,
      "learning_rate": 3.396529596706137e-05,
      "loss": 0.2719,
      "step": 6558
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6662700339753049,
      "learning_rate": 3.396066521652435e-05,
      "loss": 0.0899,
      "step": 6559
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.711837755178543,
      "learning_rate": 3.395603411319951e-05,
      "loss": 0.1475,
      "step": 6560
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8252668453821543,
      "learning_rate": 3.3951402657269195e-05,
      "loss": 0.5258,
      "step": 6561
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8464066730390806,
      "learning_rate": 3.394677084891574e-05,
      "loss": 0.1472,
      "step": 6562
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7760514145323746,
      "learning_rate": 3.39421386883215e-05,
      "loss": 0.1193,
      "step": 6563
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.561790733716711,
      "learning_rate": 3.393750617566886e-05,
      "loss": 0.4283,
      "step": 6564
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6565357100879279,
      "learning_rate": 3.393287331114019e-05,
      "loss": 0.1061,
      "step": 6565
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2069806450971023,
      "learning_rate": 3.3928240094917896e-05,
      "loss": 0.1672,
      "step": 6566
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6545406945928978,
      "learning_rate": 3.3923606527184394e-05,
      "loss": 0.0676,
      "step": 6567
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9141704622183996,
      "learning_rate": 3.39189726081221e-05,
      "loss": 0.1532,
      "step": 6568
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6957381612213516,
      "learning_rate": 3.391433833791347e-05,
      "loss": 0.1601,
      "step": 6569
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7258549469942075,
      "learning_rate": 3.390970371674095e-05,
      "loss": 0.1087,
      "step": 6570
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.627052864889394,
      "learning_rate": 3.390506874478701e-05,
      "loss": 0.0959,
      "step": 6571
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5439842275132649,
      "learning_rate": 3.390043342223413e-05,
      "loss": 0.0785,
      "step": 6572
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.48256451980339876,
      "learning_rate": 3.3895797749264806e-05,
      "loss": 0.0587,
      "step": 6573
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6938388178233726,
      "learning_rate": 3.389116172606155e-05,
      "loss": 0.1063,
      "step": 6574
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6404408889601013,
      "learning_rate": 3.3886525352806885e-05,
      "loss": 0.101,
      "step": 6575
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9827582762236124,
      "learning_rate": 3.388188862968334e-05,
      "loss": 0.3794,
      "step": 6576
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6899093025221601,
      "learning_rate": 3.387725155687349e-05,
      "loss": 0.2019,
      "step": 6577
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6394811035496843,
      "learning_rate": 3.3872614134559874e-05,
      "loss": 0.4604,
      "step": 6578
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8333853422520112,
      "learning_rate": 3.3867976362925086e-05,
      "loss": 0.2402,
      "step": 6579
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7688347099660258,
      "learning_rate": 3.386333824215171e-05,
      "loss": 0.1178,
      "step": 6580
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8436702681819174,
      "learning_rate": 3.385869977242236e-05,
      "loss": 0.1575,
      "step": 6581
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6642414918934084,
      "learning_rate": 3.3854060953919644e-05,
      "loss": 0.1938,
      "step": 6582
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5733335902879038,
      "learning_rate": 3.384942178682621e-05,
      "loss": 0.1159,
      "step": 6583
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7135510476709434,
      "learning_rate": 3.3844782271324686e-05,
      "loss": 0.1791,
      "step": 6584
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8815805335530624,
      "learning_rate": 3.384014240759775e-05,
      "loss": 0.1975,
      "step": 6585
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.712832507508419,
      "learning_rate": 3.383550219582808e-05,
      "loss": 0.1034,
      "step": 6586
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8181571147734379,
      "learning_rate": 3.383086163619835e-05,
      "loss": 0.1471,
      "step": 6587
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6324979357215447,
      "learning_rate": 3.3826220728891265e-05,
      "loss": 0.1104,
      "step": 6588
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8584568582953362,
      "learning_rate": 3.3821579474089556e-05,
      "loss": 0.175,
      "step": 6589
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7045835039179027,
      "learning_rate": 3.3816937871975926e-05,
      "loss": 0.134,
      "step": 6590
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6961780567741034,
      "learning_rate": 3.3812295922733143e-05,
      "loss": 0.0805,
      "step": 6591
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.17397922323873,
      "learning_rate": 3.380765362654396e-05,
      "loss": 0.2606,
      "step": 6592
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6879124322277601,
      "learning_rate": 3.3803010983591134e-05,
      "loss": 0.1402,
      "step": 6593
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9518387402854632,
      "learning_rate": 3.379836799405745e-05,
      "loss": 0.303,
      "step": 6594
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7869501431042359,
      "learning_rate": 3.379372465812572e-05,
      "loss": 0.1586,
      "step": 6595
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6173130654131282,
      "learning_rate": 3.378908097597875e-05,
      "loss": 0.0977,
      "step": 6596
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0046827239775504,
      "learning_rate": 3.378443694779936e-05,
      "loss": 0.1742,
      "step": 6597
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.893378939048308,
      "learning_rate": 3.377979257377039e-05,
      "loss": 0.1427,
      "step": 6598
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6290049402667041,
      "learning_rate": 3.3775147854074694e-05,
      "loss": 0.1445,
      "step": 6599
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5110833254670157,
      "learning_rate": 3.3770502788895144e-05,
      "loss": 0.4374,
      "step": 6600
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6894350346839763,
      "learning_rate": 3.376585737841461e-05,
      "loss": 0.1038,
      "step": 6601
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1289548889648053,
      "learning_rate": 3.376121162281598e-05,
      "loss": 0.2879,
      "step": 6602
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.598560762159763,
      "learning_rate": 3.375656552228219e-05,
      "loss": 0.1055,
      "step": 6603
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7612971256026219,
      "learning_rate": 3.375191907699612e-05,
      "loss": 0.1121,
      "step": 6604
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7820403296270251,
      "learning_rate": 3.374727228714073e-05,
      "loss": 0.1397,
      "step": 6605
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7738217332600172,
      "learning_rate": 3.374262515289896e-05,
      "loss": 0.1976,
      "step": 6606
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2684950873608,
      "learning_rate": 3.373797767445378e-05,
      "loss": 0.2426,
      "step": 6607
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0758937921742813,
      "learning_rate": 3.373332985198815e-05,
      "loss": 0.2354,
      "step": 6608
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5608914166807022,
      "learning_rate": 3.3728681685685066e-05,
      "loss": 0.1246,
      "step": 6609
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4742322842450664,
      "learning_rate": 3.372403317572752e-05,
      "loss": 0.0801,
      "step": 6610
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8400708501972708,
      "learning_rate": 3.371938432229854e-05,
      "loss": 0.1174,
      "step": 6611
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1633833967550835,
      "learning_rate": 3.371473512558116e-05,
      "loss": 0.368,
      "step": 6612
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4189649267868167,
      "learning_rate": 3.3710085585758404e-05,
      "loss": 0.3807,
      "step": 6613
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.018685913327767,
      "learning_rate": 3.3705435703013345e-05,
      "loss": 0.4163,
      "step": 6614
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6987364956989575,
      "learning_rate": 3.370078547752903e-05,
      "loss": 0.1284,
      "step": 6615
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5838942853417862,
      "learning_rate": 3.3696134909488566e-05,
      "loss": 0.1233,
      "step": 6616
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8345869267733751,
      "learning_rate": 3.369148399907503e-05,
      "loss": 0.2181,
      "step": 6617
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6296807266747385,
      "learning_rate": 3.368683274647155e-05,
      "loss": 0.0706,
      "step": 6618
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8699268365200931,
      "learning_rate": 3.3682181151861226e-05,
      "loss": 0.2318,
      "step": 6619
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7335917636705054,
      "learning_rate": 3.3677529215427214e-05,
      "loss": 0.174,
      "step": 6620
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.119815296037404,
      "learning_rate": 3.367287693735266e-05,
      "loss": 0.254,
      "step": 6621
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5818538175274867,
      "learning_rate": 3.366822431782073e-05,
      "loss": 0.1272,
      "step": 6622
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.565596097868088,
      "learning_rate": 3.366357135701459e-05,
      "loss": 0.0623,
      "step": 6623
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.676740411669929,
      "learning_rate": 3.3658918055117437e-05,
      "loss": 0.0613,
      "step": 6624
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5057997603236709,
      "learning_rate": 3.365426441231247e-05,
      "loss": 0.1072,
      "step": 6625
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.747665633267416,
      "learning_rate": 3.3649610428782916e-05,
      "loss": 0.1646,
      "step": 6626
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0289539770906135,
      "learning_rate": 3.3644956104712e-05,
      "loss": 0.2839,
      "step": 6627
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7140720765738863,
      "learning_rate": 3.3640301440282976e-05,
      "loss": 0.122,
      "step": 6628
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5519567485518323,
      "learning_rate": 3.363564643567908e-05,
      "loss": 0.1076,
      "step": 6629
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6210059321559858,
      "learning_rate": 3.36309910910836e-05,
      "loss": 0.0937,
      "step": 6630
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4653259811090138,
      "learning_rate": 3.362633540667981e-05,
      "loss": 0.0848,
      "step": 6631
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4325226755151539,
      "learning_rate": 3.3621679382651016e-05,
      "loss": 0.0608,
      "step": 6632
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6190422623656708,
      "learning_rate": 3.361702301918053e-05,
      "loss": 0.1175,
      "step": 6633
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6261096917004607,
      "learning_rate": 3.361236631645167e-05,
      "loss": 0.0833,
      "step": 6634
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2057523819073988,
      "learning_rate": 3.360770927464776e-05,
      "loss": 0.1322,
      "step": 6635
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7091910139840644,
      "learning_rate": 3.3603051893952185e-05,
      "loss": 0.1472,
      "step": 6636
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6075147564800846,
      "learning_rate": 3.3598394174548286e-05,
      "loss": 0.0856,
      "step": 6637
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6847608525767352,
      "learning_rate": 3.359373611661945e-05,
      "loss": 0.1161,
      "step": 6638
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7376875804046604,
      "learning_rate": 3.3589077720349064e-05,
      "loss": 0.1996,
      "step": 6639
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5617088215201678,
      "learning_rate": 3.358441898592053e-05,
      "loss": 0.1301,
      "step": 6640
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.562256372592564,
      "learning_rate": 3.357975991351726e-05,
      "loss": 0.0934,
      "step": 6641
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.613917714036675,
      "learning_rate": 3.35751005033227e-05,
      "loss": 0.0919,
      "step": 6642
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5995647086400793,
      "learning_rate": 3.3570440755520296e-05,
      "loss": 0.1126,
      "step": 6643
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9991679052218524,
      "learning_rate": 3.356578067029349e-05,
      "loss": 0.0947,
      "step": 6644
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5369246875759928,
      "learning_rate": 3.3561120247825754e-05,
      "loss": 0.0949,
      "step": 6645
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7322202358978923,
      "learning_rate": 3.355645948830059e-05,
      "loss": 0.1147,
      "step": 6646
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.623477484171453,
      "learning_rate": 3.355179839190147e-05,
      "loss": 0.0751,
      "step": 6647
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3993862240864392,
      "learning_rate": 3.354713695881193e-05,
      "loss": 0.4833,
      "step": 6648
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6544051572856502,
      "learning_rate": 3.3542475189215476e-05,
      "loss": 0.0947,
      "step": 6649
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5822517880629849,
      "learning_rate": 3.353781308329566e-05,
      "loss": 0.1016,
      "step": 6650
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8894563479836666,
      "learning_rate": 3.3533150641236016e-05,
      "loss": 0.1091,
      "step": 6651
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0049432552207318,
      "learning_rate": 3.352848786322012e-05,
      "loss": 0.2019,
      "step": 6652
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.657023438029038,
      "learning_rate": 3.352382474943154e-05,
      "loss": 0.0982,
      "step": 6653
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.034170997634604,
      "learning_rate": 3.3519161300053875e-05,
      "loss": 0.203,
      "step": 6654
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0985186530827709,
      "learning_rate": 3.351449751527072e-05,
      "loss": 0.2716,
      "step": 6655
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4992969786163211,
      "learning_rate": 3.350983339526571e-05,
      "loss": 0.0714,
      "step": 6656
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.556215685620983,
      "learning_rate": 3.350516894022244e-05,
      "loss": 0.0969,
      "step": 6657
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5512508500734659,
      "learning_rate": 3.3500504150324584e-05,
      "loss": 0.0665,
      "step": 6658
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8765486876088029,
      "learning_rate": 3.3495839025755784e-05,
      "loss": 0.301,
      "step": 6659
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7826575542349271,
      "learning_rate": 3.349117356669972e-05,
      "loss": 0.1792,
      "step": 6660
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5816845833700469,
      "learning_rate": 3.348650777334006e-05,
      "loss": 0.0835,
      "step": 6661
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6107141387757699,
      "learning_rate": 3.348184164586051e-05,
      "loss": 0.1504,
      "step": 6662
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8105536334155318,
      "learning_rate": 3.3477175184444775e-05,
      "loss": 0.207,
      "step": 6663
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0406640321596767,
      "learning_rate": 3.347250838927658e-05,
      "loss": 0.3969,
      "step": 6664
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6806241915701208,
      "learning_rate": 3.3467841260539654e-05,
      "loss": 0.1557,
      "step": 6665
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7959183601097465,
      "learning_rate": 3.346317379841775e-05,
      "loss": 0.1259,
      "step": 6666
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7018190310730825,
      "learning_rate": 3.3458506003094626e-05,
      "loss": 0.1118,
      "step": 6667
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5751984902333381,
      "learning_rate": 3.345383787475406e-05,
      "loss": 0.0883,
      "step": 6668
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8026933847211565,
      "learning_rate": 3.344916941357984e-05,
      "loss": 0.1909,
      "step": 6669
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6178272015054199,
      "learning_rate": 3.344450061975577e-05,
      "loss": 0.1153,
      "step": 6670
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5901649017859942,
      "learning_rate": 3.343983149346564e-05,
      "loss": 0.0795,
      "step": 6671
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.892552197347639,
      "learning_rate": 3.3435162034893316e-05,
      "loss": 0.1938,
      "step": 6672
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7408214661708556,
      "learning_rate": 3.3430492244222614e-05,
      "loss": 0.1792,
      "step": 6673
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0978019155594712,
      "learning_rate": 3.342582212163738e-05,
      "loss": 0.3244,
      "step": 6674
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0609958036850105,
      "learning_rate": 3.3421151667321494e-05,
      "loss": 0.1697,
      "step": 6675
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.49556561193184717,
      "learning_rate": 3.341648088145884e-05,
      "loss": 0.0634,
      "step": 6676
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3402565432653095,
      "learning_rate": 3.341180976423329e-05,
      "loss": 0.3381,
      "step": 6677
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6301701633489495,
      "learning_rate": 3.340713831582877e-05,
      "loss": 0.1247,
      "step": 6678
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.924110160521155,
      "learning_rate": 3.340246653642918e-05,
      "loss": 0.2831,
      "step": 6679
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5228790339181497,
      "learning_rate": 3.339779442621848e-05,
      "loss": 0.0608,
      "step": 6680
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8057997778078044,
      "learning_rate": 3.339312198538057e-05,
      "loss": 0.1488,
      "step": 6681
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3331505633881324,
      "learning_rate": 3.338844921409945e-05,
      "loss": 0.379,
      "step": 6682
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.50049114135708,
      "learning_rate": 3.338377611255906e-05,
      "loss": 0.2894,
      "step": 6683
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5495258417386015,
      "learning_rate": 3.3379102680943406e-05,
      "loss": 0.4688,
      "step": 6684
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7649206320948603,
      "learning_rate": 3.337442891943648e-05,
      "loss": 0.2597,
      "step": 6685
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.696007542729428,
      "learning_rate": 3.336975482822227e-05,
      "loss": 0.1993,
      "step": 6686
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.12546846476186,
      "learning_rate": 3.336508040748482e-05,
      "loss": 0.132,
      "step": 6687
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9432520238830208,
      "learning_rate": 3.336040565740816e-05,
      "loss": 0.2712,
      "step": 6688
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.713496371756705,
      "learning_rate": 3.335573057817633e-05,
      "loss": 0.086,
      "step": 6689
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5052312479877988,
      "learning_rate": 3.3351055169973414e-05,
      "loss": 0.1172,
      "step": 6690
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5813076359377738,
      "learning_rate": 3.334637943298346e-05,
      "loss": 0.0975,
      "step": 6691
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.706455259003298,
      "learning_rate": 3.3341703367390574e-05,
      "loss": 0.1592,
      "step": 6692
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.039082353708047,
      "learning_rate": 3.3337026973378835e-05,
      "loss": 0.2853,
      "step": 6693
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6527413575789087,
      "learning_rate": 3.333235025113238e-05,
      "loss": 0.101,
      "step": 6694
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6513942617085116,
      "learning_rate": 3.332767320083533e-05,
      "loss": 0.1153,
      "step": 6695
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6336646068550011,
      "learning_rate": 3.332299582267181e-05,
      "loss": 0.1698,
      "step": 6696
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6960227552685573,
      "learning_rate": 3.331831811682598e-05,
      "loss": 0.129,
      "step": 6697
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7439979640570915,
      "learning_rate": 3.3313640083482004e-05,
      "loss": 0.1504,
      "step": 6698
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4798703666059904,
      "learning_rate": 3.330896172282406e-05,
      "loss": 0.0982,
      "step": 6699
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9006021061225347,
      "learning_rate": 3.330428303503634e-05,
      "loss": 0.174,
      "step": 6700
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5600016034834624,
      "learning_rate": 3.329960402030304e-05,
      "loss": 0.1416,
      "step": 6701
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7250572794705118,
      "learning_rate": 3.329492467880838e-05,
      "loss": 0.1548,
      "step": 6702
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1898862369200935,
      "learning_rate": 3.3290245010736595e-05,
      "loss": 0.2566,
      "step": 6703
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6749928964555726,
      "learning_rate": 3.3285565016271926e-05,
      "loss": 0.1337,
      "step": 6704
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8656448075603023,
      "learning_rate": 3.328088469559862e-05,
      "loss": 0.1468,
      "step": 6705
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7923385912014176,
      "learning_rate": 3.327620404890095e-05,
      "loss": 0.1417,
      "step": 6706
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6779207440806874,
      "learning_rate": 3.327152307636319e-05,
      "loss": 0.205,
      "step": 6707
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.543619413790541,
      "learning_rate": 3.3266841778169647e-05,
      "loss": 0.0798,
      "step": 6708
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6652976470134057,
      "learning_rate": 3.3262160154504616e-05,
      "loss": 0.1876,
      "step": 6709
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1397078562182101,
      "learning_rate": 3.325747820555242e-05,
      "loss": 0.4316,
      "step": 6710
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4727990654300024,
      "learning_rate": 3.325279593149738e-05,
      "loss": 0.0582,
      "step": 6711
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6075795289704746,
      "learning_rate": 3.3248113332523864e-05,
      "loss": 0.1728,
      "step": 6712
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8023340874790706,
      "learning_rate": 3.32434304088162e-05,
      "loss": 0.1536,
      "step": 6713
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7484768667154301,
      "learning_rate": 3.323874716055878e-05,
      "loss": 0.0831,
      "step": 6714
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8298654547099924,
      "learning_rate": 3.3234063587935983e-05,
      "loss": 0.097,
      "step": 6715
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.003118034035035,
      "learning_rate": 3.32293796911322e-05,
      "loss": 0.1425,
      "step": 6716
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.615441817697074,
      "learning_rate": 3.3224695470331836e-05,
      "loss": 0.0713,
      "step": 6717
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5513025410050596,
      "learning_rate": 3.322001092571932e-05,
      "loss": 0.0923,
      "step": 6718
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8068732544992478,
      "learning_rate": 3.321532605747908e-05,
      "loss": 0.1824,
      "step": 6719
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6103875028228968,
      "learning_rate": 3.3210640865795574e-05,
      "loss": 0.1298,
      "step": 6720
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1805394768220425,
      "learning_rate": 3.320595535085324e-05,
      "loss": 0.2725,
      "step": 6721
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.051606645840759,
      "learning_rate": 3.320126951283658e-05,
      "loss": 0.345,
      "step": 6722
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7547535837601668,
      "learning_rate": 3.319658335193005e-05,
      "loss": 0.2502,
      "step": 6723
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7378807904734421,
      "learning_rate": 3.319189686831816e-05,
      "loss": 0.1131,
      "step": 6724
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7496403159949857,
      "learning_rate": 3.318721006218541e-05,
      "loss": 0.153,
      "step": 6725
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7833872837930137,
      "learning_rate": 3.318252293371635e-05,
      "loss": 0.1625,
      "step": 6726
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5779117909538991,
      "learning_rate": 3.3177835483095485e-05,
      "loss": 0.1181,
      "step": 6727
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7476170313049951,
      "learning_rate": 3.317314771050738e-05,
      "loss": 0.0961,
      "step": 6728
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5761570927444732,
      "learning_rate": 3.31684596161366e-05,
      "loss": 0.3306,
      "step": 6729
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3764100318959906,
      "learning_rate": 3.31637712001677e-05,
      "loss": 0.3132,
      "step": 6730
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3884006623628868,
      "learning_rate": 3.3159082462785283e-05,
      "loss": 0.5128,
      "step": 6731
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5390480242033221,
      "learning_rate": 3.3154393404173944e-05,
      "loss": 0.0842,
      "step": 6732
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5258096208152758,
      "learning_rate": 3.3149704024518287e-05,
      "loss": 0.0739,
      "step": 6733
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5795716707198091,
      "learning_rate": 3.3145014324002944e-05,
      "loss": 0.1196,
      "step": 6734
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6123190978376704,
      "learning_rate": 3.314032430281255e-05,
      "loss": 0.1633,
      "step": 6735
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9744222515893863,
      "learning_rate": 3.313563396113175e-05,
      "loss": 0.1886,
      "step": 6736
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6415575758404275,
      "learning_rate": 3.313094329914521e-05,
      "loss": 0.1294,
      "step": 6737
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9944537854925544,
      "learning_rate": 3.312625231703761e-05,
      "loss": 0.1969,
      "step": 6738
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7999567234972806,
      "learning_rate": 3.312156101499363e-05,
      "loss": 0.1735,
      "step": 6739
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5900007032560128,
      "learning_rate": 3.311686939319798e-05,
      "loss": 0.091,
      "step": 6740
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.608305033198539,
      "learning_rate": 3.3112177451835355e-05,
      "loss": 0.0692,
      "step": 6741
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9338241948065524,
      "learning_rate": 3.31074851910905e-05,
      "loss": 0.134,
      "step": 6742
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9158079946521585,
      "learning_rate": 3.310279261114813e-05,
      "loss": 0.1459,
      "step": 6743
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6776737951190467,
      "learning_rate": 3.309809971219301e-05,
      "loss": 0.1191,
      "step": 6744
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1165026498507695,
      "learning_rate": 3.309340649440991e-05,
      "loss": 0.2237,
      "step": 6745
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5770475915633738,
      "learning_rate": 3.3088712957983584e-05,
      "loss": 0.0779,
      "step": 6746
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7611774894793373,
      "learning_rate": 3.308401910309885e-05,
      "loss": 0.1956,
      "step": 6747
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3535271446527473,
      "learning_rate": 3.307932492994047e-05,
      "loss": 0.3742,
      "step": 6748
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3752990530094384,
      "learning_rate": 3.307463043869329e-05,
      "loss": 0.2309,
      "step": 6749
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.116850772982653,
      "learning_rate": 3.3069935629542116e-05,
      "loss": 0.3283,
      "step": 6750
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5890435411393586,
      "learning_rate": 3.30652405026718e-05,
      "loss": 0.0907,
      "step": 6751
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8945719639222621,
      "learning_rate": 3.3060545058267176e-05,
      "loss": 0.159,
      "step": 6752
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.409810621945485,
      "learning_rate": 3.305584929651312e-05,
      "loss": 0.0916,
      "step": 6753
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8297589978005213,
      "learning_rate": 3.30511532175945e-05,
      "loss": 0.1105,
      "step": 6754
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7661493770342664,
      "learning_rate": 3.304645682169622e-05,
      "loss": 0.1363,
      "step": 6755
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8120470216472541,
      "learning_rate": 3.304176010900316e-05,
      "loss": 0.177,
      "step": 6756
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8189629878418894,
      "learning_rate": 3.3037063079700245e-05,
      "loss": 0.196,
      "step": 6757
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.409072628812459,
      "learning_rate": 3.3032365733972396e-05,
      "loss": 0.4885,
      "step": 6758
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7793227890983581,
      "learning_rate": 3.3027668072004544e-05,
      "loss": 0.1948,
      "step": 6759
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7801726909430176,
      "learning_rate": 3.3022970093981656e-05,
      "loss": 0.1277,
      "step": 6760
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0460158665402846,
      "learning_rate": 3.3018271800088676e-05,
      "loss": 0.2819,
      "step": 6761
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5797918144609164,
      "learning_rate": 3.3013573190510596e-05,
      "loss": 0.1353,
      "step": 6762
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.518290810735041,
      "learning_rate": 3.30088742654324e-05,
      "loss": 0.4957,
      "step": 6763
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6392920761159985,
      "learning_rate": 3.3004175025039076e-05,
      "loss": 0.074,
      "step": 6764
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.073777920866926,
      "learning_rate": 3.299947546951565e-05,
      "loss": 0.1052,
      "step": 6765
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9301911781152902,
      "learning_rate": 3.2994775599047135e-05,
      "loss": 0.1595,
      "step": 6766
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7696644724491815,
      "learning_rate": 3.299007541381858e-05,
      "loss": 0.5792,
      "step": 6767
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5212157272552276,
      "learning_rate": 3.298537491401503e-05,
      "loss": 0.1082,
      "step": 6768
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7575382368652681,
      "learning_rate": 3.298067409982155e-05,
      "loss": 0.1455,
      "step": 6769
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7743312536908009,
      "learning_rate": 3.29759729714232e-05,
      "loss": 0.1157,
      "step": 6770
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8506175740217486,
      "learning_rate": 3.297127152900509e-05,
      "loss": 0.1796,
      "step": 6771
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6334905181276655,
      "learning_rate": 3.29665697727523e-05,
      "loss": 0.1191,
      "step": 6772
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0581075809339477,
      "learning_rate": 3.296186770284995e-05,
      "loss": 0.2765,
      "step": 6773
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8559111686777137,
      "learning_rate": 3.295716531948316e-05,
      "loss": 0.2566,
      "step": 6774
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0334963666640813,
      "learning_rate": 3.295246262283708e-05,
      "loss": 0.2036,
      "step": 6775
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.4643357422621217,
      "learning_rate": 3.294775961309684e-05,
      "loss": 0.4924,
      "step": 6776
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7348762935242267,
      "learning_rate": 3.2943056290447604e-05,
      "loss": 0.1729,
      "step": 6777
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8644952165903689,
      "learning_rate": 3.293835265507455e-05,
      "loss": 0.2142,
      "step": 6778
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7170012838714017,
      "learning_rate": 3.293364870716287e-05,
      "loss": 0.1665,
      "step": 6779
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7899894948887061,
      "learning_rate": 3.292894444689775e-05,
      "loss": 0.1555,
      "step": 6780
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.733291082635338,
      "learning_rate": 3.29242398744644e-05,
      "loss": 0.1445,
      "step": 6781
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.457462331743346,
      "learning_rate": 3.291953499004805e-05,
      "loss": 0.047,
      "step": 6782
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6375956718368567,
      "learning_rate": 3.2914829793833936e-05,
      "loss": 0.1213,
      "step": 6783
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7737849143546901,
      "learning_rate": 3.2910124286007296e-05,
      "loss": 0.111,
      "step": 6784
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6151090995550477,
      "learning_rate": 3.2905418466753393e-05,
      "loss": 0.0919,
      "step": 6785
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.449117234392591,
      "learning_rate": 3.29007123362575e-05,
      "loss": 0.705,
      "step": 6786
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.774235385212502,
      "learning_rate": 3.289600589470491e-05,
      "loss": 0.1836,
      "step": 6787
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.428602809776487,
      "learning_rate": 3.289129914228089e-05,
      "loss": 0.4259,
      "step": 6788
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8771395104419575,
      "learning_rate": 3.288659207917078e-05,
      "loss": 0.1633,
      "step": 6789
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6786517091989024,
      "learning_rate": 3.288188470555989e-05,
      "loss": 0.0919,
      "step": 6790
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7824059540497971,
      "learning_rate": 3.287717702163355e-05,
      "loss": 0.1432,
      "step": 6791
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.243475265044767,
      "learning_rate": 3.2872469027577105e-05,
      "loss": 0.2078,
      "step": 6792
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7921147496062272,
      "learning_rate": 3.2867760723575904e-05,
      "loss": 0.2354,
      "step": 6793
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8701780218814491,
      "learning_rate": 3.286305210981534e-05,
      "loss": 0.115,
      "step": 6794
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.31253215844356,
      "learning_rate": 3.285834318648077e-05,
      "loss": 0.4672,
      "step": 6795
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5639506677192142,
      "learning_rate": 3.285363395375759e-05,
      "loss": 0.1209,
      "step": 6796
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8635274393075827,
      "learning_rate": 3.2848924411831216e-05,
      "loss": 0.6694,
      "step": 6797
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0789676744691712,
      "learning_rate": 3.284421456088707e-05,
      "loss": 0.4048,
      "step": 6798
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6834789103449195,
      "learning_rate": 3.283950440111057e-05,
      "loss": 0.1112,
      "step": 6799
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7918167080398401,
      "learning_rate": 3.2834793932687155e-05,
      "loss": 0.1676,
      "step": 6800
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.096192530991711,
      "learning_rate": 3.283008315580231e-05,
      "loss": 0.3203,
      "step": 6801
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5891847234854722,
      "learning_rate": 3.282537207064146e-05,
      "loss": 0.096,
      "step": 6802
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8852521458938134,
      "learning_rate": 3.282066067739011e-05,
      "loss": 0.3134,
      "step": 6803
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8170061512115893,
      "learning_rate": 3.281594897623374e-05,
      "loss": 0.1412,
      "step": 6804
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5550872473191905,
      "learning_rate": 3.281123696735787e-05,
      "loss": 0.1043,
      "step": 6805
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6563490786618202,
      "learning_rate": 3.280652465094798e-05,
      "loss": 0.0635,
      "step": 6806
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2212064762236803,
      "learning_rate": 3.2801812027189644e-05,
      "loss": 0.2635,
      "step": 6807
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9940868988977636,
      "learning_rate": 3.2797099096268366e-05,
      "loss": 0.347,
      "step": 6808
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7130236877447668,
      "learning_rate": 3.2792385858369704e-05,
      "loss": 0.1301,
      "step": 6809
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7242641384179621,
      "learning_rate": 3.278767231367923e-05,
      "loss": 0.1567,
      "step": 6810
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9326713254205462,
      "learning_rate": 3.278295846238252e-05,
      "loss": 0.1957,
      "step": 6811
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9163728080391659,
      "learning_rate": 3.277824430466515e-05,
      "loss": 0.2044,
      "step": 6812
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8530653216718992,
      "learning_rate": 3.277352984071273e-05,
      "loss": 0.1027,
      "step": 6813
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.075828376881628,
      "learning_rate": 3.276881507071087e-05,
      "loss": 0.6197,
      "step": 6814
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5389750115321916,
      "learning_rate": 3.27640999948452e-05,
      "loss": 0.1217,
      "step": 6815
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0495249738230814,
      "learning_rate": 3.275938461330134e-05,
      "loss": 0.2719,
      "step": 6816
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5404324714350132,
      "learning_rate": 3.275466892626495e-05,
      "loss": 0.0785,
      "step": 6817
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6232644928641402,
      "learning_rate": 3.274995293392169e-05,
      "loss": 0.3159,
      "step": 6818
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0356761873898286,
      "learning_rate": 3.274523663645722e-05,
      "loss": 0.3045,
      "step": 6819
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7740767278478763,
      "learning_rate": 3.274052003405724e-05,
      "loss": 0.1434,
      "step": 6820
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7063567935446609,
      "learning_rate": 3.273580312690744e-05,
      "loss": 0.0944,
      "step": 6821
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7829865063728603,
      "learning_rate": 3.273108591519352e-05,
      "loss": 0.1971,
      "step": 6822
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6137939399157047,
      "learning_rate": 3.2726368399101215e-05,
      "loss": 0.1428,
      "step": 6823
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1295841451441244,
      "learning_rate": 3.2721650578816236e-05,
      "loss": 0.2214,
      "step": 6824
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8216121303150938,
      "learning_rate": 3.271693245452435e-05,
      "loss": 0.1336,
      "step": 6825
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5822603434318047,
      "learning_rate": 3.271221402641131e-05,
      "loss": 0.0874,
      "step": 6826
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8953292436684921,
      "learning_rate": 3.270749529466286e-05,
      "loss": 0.2171,
      "step": 6827
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5079971565462926,
      "learning_rate": 3.270277625946481e-05,
      "loss": 0.0648,
      "step": 6828
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.2956428930516375,
      "learning_rate": 3.2698056921002936e-05,
      "loss": 0.8253,
      "step": 6829
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5404643168936639,
      "learning_rate": 3.269333727946303e-05,
      "loss": 0.1282,
      "step": 6830
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5451459979519313,
      "learning_rate": 3.268861733503094e-05,
      "loss": 0.0998,
      "step": 6831
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7089254134846688,
      "learning_rate": 3.268389708789246e-05,
      "loss": 0.1383,
      "step": 6832
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.682752736318811,
      "learning_rate": 3.267917653823346e-05,
      "loss": 0.1104,
      "step": 6833
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7845665575933455,
      "learning_rate": 3.2674455686239765e-05,
      "loss": 0.1365,
      "step": 6834
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6947451052174962,
      "learning_rate": 3.266973453209725e-05,
      "loss": 0.1254,
      "step": 6835
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7204688531732443,
      "learning_rate": 3.266501307599179e-05,
      "loss": 0.1691,
      "step": 6836
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6967409768830618,
      "learning_rate": 3.266029131810928e-05,
      "loss": 0.1462,
      "step": 6837
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1374239728351814,
      "learning_rate": 3.26555692586356e-05,
      "loss": 0.1932,
      "step": 6838
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5942357423708847,
      "learning_rate": 3.2650846897756684e-05,
      "loss": 0.1103,
      "step": 6839
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.618078962356927,
      "learning_rate": 3.2646124235658435e-05,
      "loss": 0.0825,
      "step": 6840
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5487744665945443,
      "learning_rate": 3.26414012725268e-05,
      "loss": 0.0987,
      "step": 6841
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2519736430199369,
      "learning_rate": 3.2636678008547706e-05,
      "loss": 0.4171,
      "step": 6842
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4797215150512991,
      "learning_rate": 3.263195444390714e-05,
      "loss": 0.0814,
      "step": 6843
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7324341602267322,
      "learning_rate": 3.262723057879105e-05,
      "loss": 0.1633,
      "step": 6844
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6965782802336574,
      "learning_rate": 3.2622506413385436e-05,
      "loss": 0.0436,
      "step": 6845
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.514911095023144,
      "learning_rate": 3.261778194787627e-05,
      "loss": 0.5682,
      "step": 6846
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8463198700912538,
      "learning_rate": 3.261305718244957e-05,
      "loss": 0.1071,
      "step": 6847
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7551818963248869,
      "learning_rate": 3.260833211729136e-05,
      "loss": 0.1214,
      "step": 6848
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.935584621933423,
      "learning_rate": 3.260360675258766e-05,
      "loss": 0.1313,
      "step": 6849
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4921782886652803,
      "learning_rate": 3.2598881088524516e-05,
      "loss": 0.0864,
      "step": 6850
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8772849818925202,
      "learning_rate": 3.2594155125287974e-05,
      "loss": 0.1787,
      "step": 6851
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4602515757919785,
      "learning_rate": 3.2589428863064096e-05,
      "loss": 0.0475,
      "step": 6852
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8492514049675779,
      "learning_rate": 3.258470230203897e-05,
      "loss": 0.1077,
      "step": 6853
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6825295701562608,
      "learning_rate": 3.257997544239868e-05,
      "loss": 0.0914,
      "step": 6854
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4131460747776704,
      "learning_rate": 3.257524828432933e-05,
      "loss": 0.0895,
      "step": 6855
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5605163274520146,
      "learning_rate": 3.257052082801701e-05,
      "loss": 0.0727,
      "step": 6856
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5494119067432313,
      "learning_rate": 3.256579307364787e-05,
      "loss": 0.1337,
      "step": 6857
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.278643639721745,
      "learning_rate": 3.256106502140803e-05,
      "loss": 0.193,
      "step": 6858
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5357484656315556,
      "learning_rate": 3.255633667148365e-05,
      "loss": 0.0606,
      "step": 6859
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9493304043237946,
      "learning_rate": 3.2551608024060876e-05,
      "loss": 0.1945,
      "step": 6860
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.631375654567976,
      "learning_rate": 3.2546879079325877e-05,
      "loss": 0.0866,
      "step": 6861
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6274249854169527,
      "learning_rate": 3.2542149837464845e-05,
      "loss": 0.1201,
      "step": 6862
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6668917501052958,
      "learning_rate": 3.2537420298663966e-05,
      "loss": 0.116,
      "step": 6863
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7611126695427725,
      "learning_rate": 3.253269046310945e-05,
      "loss": 0.1383,
      "step": 6864
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8373197137440219,
      "learning_rate": 3.252796033098752e-05,
      "loss": 0.1503,
      "step": 6865
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5497078723316341,
      "learning_rate": 3.252322990248437e-05,
      "loss": 0.1191,
      "step": 6866
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6476319348034163,
      "learning_rate": 3.2518499177786294e-05,
      "loss": 0.093,
      "step": 6867
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5417191285287752,
      "learning_rate": 3.251376815707951e-05,
      "loss": 0.1239,
      "step": 6868
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7599963359642844,
      "learning_rate": 3.250903684055028e-05,
      "loss": 0.0772,
      "step": 6869
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8013720935891294,
      "learning_rate": 3.25043052283849e-05,
      "loss": 0.2349,
      "step": 6870
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.663608260827743,
      "learning_rate": 3.249957332076964e-05,
      "loss": 0.3964,
      "step": 6871
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.155032011656046,
      "learning_rate": 3.24948411178908e-05,
      "loss": 0.3478,
      "step": 6872
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6932204758874225,
      "learning_rate": 3.24901086199347e-05,
      "loss": 0.1562,
      "step": 6873
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8131082621664198,
      "learning_rate": 3.248537582708765e-05,
      "loss": 0.2981,
      "step": 6874
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6499101242397304,
      "learning_rate": 3.248064273953599e-05,
      "loss": 0.1055,
      "step": 6875
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9054132503810887,
      "learning_rate": 3.247590935746607e-05,
      "loss": 0.1989,
      "step": 6876
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5616116845784642,
      "learning_rate": 3.247117568106423e-05,
      "loss": 0.1,
      "step": 6877
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5649339623755679,
      "learning_rate": 3.246644171051686e-05,
      "loss": 0.1682,
      "step": 6878
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6731597768520485,
      "learning_rate": 3.246170744601032e-05,
      "loss": 0.1342,
      "step": 6879
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0720846212927948,
      "learning_rate": 3.245697288773102e-05,
      "loss": 0.1917,
      "step": 6880
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.007566991797188,
      "learning_rate": 3.245223803586534e-05,
      "loss": 0.3254,
      "step": 6881
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9244240004774987,
      "learning_rate": 3.244750289059972e-05,
      "loss": 0.1637,
      "step": 6882
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.2750021473454773,
      "learning_rate": 3.244276745212056e-05,
      "loss": 0.5833,
      "step": 6883
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6643611818053898,
      "learning_rate": 3.2438031720614325e-05,
      "loss": 0.154,
      "step": 6884
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7581152723527975,
      "learning_rate": 3.2433295696267445e-05,
      "loss": 0.1722,
      "step": 6885
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7925927935222087,
      "learning_rate": 3.242855937926639e-05,
      "loss": 0.1206,
      "step": 6886
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.065651124057069,
      "learning_rate": 3.242382276979762e-05,
      "loss": 0.3278,
      "step": 6887
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6217301957105976,
      "learning_rate": 3.241908586804764e-05,
      "loss": 0.0903,
      "step": 6888
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9167843045689386,
      "learning_rate": 3.241434867420292e-05,
      "loss": 0.2603,
      "step": 6889
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6111761841486941,
      "learning_rate": 3.240961118844999e-05,
      "loss": 0.1084,
      "step": 6890
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7763470617022283,
      "learning_rate": 3.2404873410975356e-05,
      "loss": 0.1849,
      "step": 6891
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.665141975142801,
      "learning_rate": 3.2400135341965535e-05,
      "loss": 0.0832,
      "step": 6892
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0452253535865599,
      "learning_rate": 3.2395396981607096e-05,
      "loss": 0.2163,
      "step": 6893
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7070458455121894,
      "learning_rate": 3.239065833008658e-05,
      "loss": 0.1523,
      "step": 6894
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9597153317343668,
      "learning_rate": 3.2385919387590536e-05,
      "loss": 0.2631,
      "step": 6895
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7512313658139186,
      "learning_rate": 3.2381180154305564e-05,
      "loss": 0.1578,
      "step": 6896
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8737837084114738,
      "learning_rate": 3.237644063041822e-05,
      "loss": 0.1465,
      "step": 6897
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9264429930407415,
      "learning_rate": 3.237170081611515e-05,
      "loss": 0.228,
      "step": 6898
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.66055603569121,
      "learning_rate": 3.2366960711582913e-05,
      "loss": 0.088,
      "step": 6899
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2134158302311624,
      "learning_rate": 3.236222031700816e-05,
      "loss": 0.2656,
      "step": 6900
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5432672899837221,
      "learning_rate": 3.235747963257752e-05,
      "loss": 0.0861,
      "step": 6901
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6724978536229549,
      "learning_rate": 3.2352738658477636e-05,
      "loss": 0.1602,
      "step": 6902
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5398459964894287,
      "learning_rate": 3.234799739489516e-05,
      "loss": 0.0863,
      "step": 6903
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9590758547394782,
      "learning_rate": 3.234325584201675e-05,
      "loss": 0.2569,
      "step": 6904
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6955689204148561,
      "learning_rate": 3.2338514000029104e-05,
      "loss": 0.1552,
      "step": 6905
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6289968910333523,
      "learning_rate": 3.2333771869118904e-05,
      "loss": 0.3998,
      "step": 6906
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.730798871155309,
      "learning_rate": 3.232902944947285e-05,
      "loss": 0.0826,
      "step": 6907
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6504589168758192,
      "learning_rate": 3.232428674127765e-05,
      "loss": 0.1299,
      "step": 6908
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6564625109410454,
      "learning_rate": 3.231954374472003e-05,
      "loss": 0.1278,
      "step": 6909
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6657759833246917,
      "learning_rate": 3.2314800459986735e-05,
      "loss": 0.118,
      "step": 6910
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8485580771585823,
      "learning_rate": 3.231005688726449e-05,
      "loss": 0.1628,
      "step": 6911
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7933641924823798,
      "learning_rate": 3.230531302674008e-05,
      "loss": 0.1585,
      "step": 6912
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7114145643974188,
      "learning_rate": 3.230056887860026e-05,
      "loss": 0.1489,
      "step": 6913
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8583324901368111,
      "learning_rate": 3.2295824443031803e-05,
      "loss": 0.1999,
      "step": 6914
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6052566362593487,
      "learning_rate": 3.229107972022151e-05,
      "loss": 0.0814,
      "step": 6915
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7941419157266799,
      "learning_rate": 3.2286334710356195e-05,
      "loss": 0.3185,
      "step": 6916
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6172317048399116,
      "learning_rate": 3.2281589413622644e-05,
      "loss": 0.1151,
      "step": 6917
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8686494353747832,
      "learning_rate": 3.227684383020771e-05,
      "loss": 0.142,
      "step": 6918
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9542169596767216,
      "learning_rate": 3.227209796029821e-05,
      "loss": 0.2604,
      "step": 6919
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5801869717520624,
      "learning_rate": 3.2267351804081006e-05,
      "loss": 0.0551,
      "step": 6920
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6741256642626237,
      "learning_rate": 3.2262605361742956e-05,
      "loss": 0.1271,
      "step": 6921
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3822364477899054,
      "learning_rate": 3.225785863347093e-05,
      "loss": 0.3198,
      "step": 6922
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8233905769245474,
      "learning_rate": 3.2253111619451804e-05,
      "loss": 0.1347,
      "step": 6923
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5122255352060805,
      "learning_rate": 3.2248364319872475e-05,
      "loss": 0.1331,
      "step": 6924
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7057047262231796,
      "learning_rate": 3.224361673491985e-05,
      "loss": 0.2042,
      "step": 6925
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7450105373264516,
      "learning_rate": 3.223886886478084e-05,
      "loss": 0.1356,
      "step": 6926
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5100208684498623,
      "learning_rate": 3.223412070964238e-05,
      "loss": 0.1931,
      "step": 6927
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3965896991140352,
      "learning_rate": 3.2229372269691404e-05,
      "loss": 0.3147,
      "step": 6928
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7300382929896256,
      "learning_rate": 3.222462354511485e-05,
      "loss": 0.1607,
      "step": 6929
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.45386519779512075,
      "learning_rate": 3.2219874536099704e-05,
      "loss": 0.0342,
      "step": 6930
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7288720954419401,
      "learning_rate": 3.221512524283291e-05,
      "loss": 0.1,
      "step": 6931
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6324220920692692,
      "learning_rate": 3.221037566550147e-05,
      "loss": 0.0668,
      "step": 6932
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7175812304396196,
      "learning_rate": 3.220562580429237e-05,
      "loss": 0.0878,
      "step": 6933
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5764912272000813,
      "learning_rate": 3.220087565939262e-05,
      "loss": 0.209,
      "step": 6934
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7015578732631533,
      "learning_rate": 3.2196125230989236e-05,
      "loss": 0.1134,
      "step": 6935
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.43338309885894305,
      "learning_rate": 3.2191374519269246e-05,
      "loss": 0.0605,
      "step": 6936
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8122131959567219,
      "learning_rate": 3.218662352441968e-05,
      "loss": 0.1348,
      "step": 6937
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7834165149880523,
      "learning_rate": 3.2181872246627596e-05,
      "loss": 0.148,
      "step": 6938
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2472392815391828,
      "learning_rate": 3.217712068608006e-05,
      "loss": 0.3753,
      "step": 6939
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.65881973917986,
      "learning_rate": 3.217236884296414e-05,
      "loss": 0.1054,
      "step": 6940
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0374611364016897,
      "learning_rate": 3.2167616717466916e-05,
      "loss": 0.2131,
      "step": 6941
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9295116194833841,
      "learning_rate": 3.2162864309775485e-05,
      "loss": 0.2541,
      "step": 6942
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5515329179595122,
      "learning_rate": 3.215811162007695e-05,
      "loss": 0.099,
      "step": 6943
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7220661302604127,
      "learning_rate": 3.215335864855844e-05,
      "loss": 0.1143,
      "step": 6944
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5109196711798107,
      "learning_rate": 3.2148605395407055e-05,
      "loss": 0.0735,
      "step": 6945
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7394812993015996,
      "learning_rate": 3.2143851860809974e-05,
      "loss": 0.0949,
      "step": 6946
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7853461055527869,
      "learning_rate": 3.213909804495431e-05,
      "loss": 0.1344,
      "step": 6947
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5809067218322905,
      "learning_rate": 3.213434394802724e-05,
      "loss": 0.0726,
      "step": 6948
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.321344823010996,
      "learning_rate": 3.212958957021594e-05,
      "loss": 0.2578,
      "step": 6949
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5989090898886118,
      "learning_rate": 3.212483491170759e-05,
      "loss": 0.1094,
      "step": 6950
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0676184537372153,
      "learning_rate": 3.2120079972689384e-05,
      "loss": 0.233,
      "step": 6951
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0626375670086425,
      "learning_rate": 3.211532475334853e-05,
      "loss": 0.141,
      "step": 6952
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2656855518124548,
      "learning_rate": 3.2110569253872226e-05,
      "loss": 0.2274,
      "step": 6953
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6484920812021124,
      "learning_rate": 3.210581347444773e-05,
      "loss": 0.1583,
      "step": 6954
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6175157397957239,
      "learning_rate": 3.210105741526226e-05,
      "loss": 0.126,
      "step": 6955
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6025049899984279,
      "learning_rate": 3.209630107650306e-05,
      "loss": 0.0911,
      "step": 6956
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6098568755101827,
      "learning_rate": 3.209154445835742e-05,
      "loss": 0.0685,
      "step": 6957
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.523754550759638,
      "learning_rate": 3.2086787561012584e-05,
      "loss": 0.083,
      "step": 6958
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5109929448323658,
      "learning_rate": 3.208203038465584e-05,
      "loss": 0.3637,
      "step": 6959
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.561581548173568,
      "learning_rate": 3.207727292947449e-05,
      "loss": 0.0768,
      "step": 6960
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7506035176214425,
      "learning_rate": 3.2072515195655825e-05,
      "loss": 0.1619,
      "step": 6961
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9132016426614551,
      "learning_rate": 3.2067757183387174e-05,
      "loss": 0.3172,
      "step": 6962
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9349904968782413,
      "learning_rate": 3.206299889285585e-05,
      "loss": 0.4418,
      "step": 6963
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5049082601171521,
      "learning_rate": 3.205824032424921e-05,
      "loss": 0.0532,
      "step": 6964
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.622821040407007,
      "learning_rate": 3.2053481477754584e-05,
      "loss": 0.0916,
      "step": 6965
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.557266023121403,
      "learning_rate": 3.2048722353559335e-05,
      "loss": 0.1046,
      "step": 6966
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9325595309564573,
      "learning_rate": 3.204396295185084e-05,
      "loss": 0.291,
      "step": 6967
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6647613103660871,
      "learning_rate": 3.203920327281647e-05,
      "loss": 0.2235,
      "step": 6968
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7440835224007686,
      "learning_rate": 3.203444331664363e-05,
      "loss": 0.1785,
      "step": 6969
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3349446912746112,
      "learning_rate": 3.20296830835197e-05,
      "loss": 0.3063,
      "step": 6970
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3873333858235817,
      "learning_rate": 3.2024922573632124e-05,
      "loss": 0.327,
      "step": 6971
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7201426468458586,
      "learning_rate": 3.2020161787168306e-05,
      "loss": 0.1352,
      "step": 6972
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6292417398128873,
      "learning_rate": 3.201540072431568e-05,
      "loss": 0.1577,
      "step": 6973
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5295195758834099,
      "learning_rate": 3.2010639385261716e-05,
      "loss": 0.1154,
      "step": 6974
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7590226420929993,
      "learning_rate": 3.200587777019384e-05,
      "loss": 0.1196,
      "step": 6975
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6352232989792767,
      "learning_rate": 3.200111587929955e-05,
      "loss": 0.1488,
      "step": 6976
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6884469960201225,
      "learning_rate": 3.1996353712766294e-05,
      "loss": 0.1526,
      "step": 6977
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7162669341301828,
      "learning_rate": 3.199159127078159e-05,
      "loss": 0.109,
      "step": 6978
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5202570260644028,
      "learning_rate": 3.198682855353292e-05,
      "loss": 0.0576,
      "step": 6979
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.115423430550948,
      "learning_rate": 3.198206556120781e-05,
      "loss": 0.3444,
      "step": 6980
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2683541233308226,
      "learning_rate": 3.197730229399377e-05,
      "loss": 0.3093,
      "step": 6981
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7086774216969892,
      "learning_rate": 3.1972538752078335e-05,
      "loss": 0.1124,
      "step": 6982
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9423497345038021,
      "learning_rate": 3.196777493564905e-05,
      "loss": 0.1786,
      "step": 6983
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.883745832905533,
      "learning_rate": 3.196301084489349e-05,
      "loss": 0.1396,
      "step": 6984
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8061426919251072,
      "learning_rate": 3.195824647999918e-05,
      "loss": 0.1462,
      "step": 6985
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8604241742789289,
      "learning_rate": 3.195348184115374e-05,
      "loss": 0.211,
      "step": 6986
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7324792858143271,
      "learning_rate": 3.194871692854472e-05,
      "loss": 0.2045,
      "step": 6987
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8183629772733386,
      "learning_rate": 3.194395174235974e-05,
      "loss": 0.1308,
      "step": 6988
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4652579928405376,
      "learning_rate": 3.1939186282786404e-05,
      "loss": 0.0546,
      "step": 6989
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.48787852874005044,
      "learning_rate": 3.193442055001234e-05,
      "loss": 0.0676,
      "step": 6990
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9193593323406261,
      "learning_rate": 3.192965454422516e-05,
      "loss": 0.4284,
      "step": 6991
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6208553081860233,
      "learning_rate": 3.1924888265612514e-05,
      "loss": 0.1205,
      "step": 6992
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.004162075988448,
      "learning_rate": 3.192012171436206e-05,
      "loss": 0.1316,
      "step": 6993
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.833832829729887,
      "learning_rate": 3.191535489066144e-05,
      "loss": 0.1958,
      "step": 6994
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5565207383873287,
      "learning_rate": 3.191058779469836e-05,
      "loss": 0.0545,
      "step": 6995
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7319327358205663,
      "learning_rate": 3.190582042666048e-05,
      "loss": 0.1263,
      "step": 6996
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6130711144276453,
      "learning_rate": 3.19010527867355e-05,
      "loss": 0.118,
      "step": 6997
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9168498369385158,
      "learning_rate": 3.189628487511112e-05,
      "loss": 0.1689,
      "step": 6998
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7285143905388478,
      "learning_rate": 3.1891516691975064e-05,
      "loss": 0.1455,
      "step": 6999
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8205591168572028,
      "learning_rate": 3.1886748237515064e-05,
      "loss": 0.232,
      "step": 7000
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1838973455924355,
      "learning_rate": 3.188197951191884e-05,
      "loss": 0.4746,
      "step": 7001
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9232193322615673,
      "learning_rate": 3.187721051537416e-05,
      "loss": 0.1916,
      "step": 7002
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.598965531452698,
      "learning_rate": 3.1872441248068766e-05,
      "loss": 0.0772,
      "step": 7003
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9668600616654834,
      "learning_rate": 3.186767171019044e-05,
      "loss": 0.1989,
      "step": 7004
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1123078862158275,
      "learning_rate": 3.1862901901926956e-05,
      "loss": 0.2605,
      "step": 7005
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.096492815371059,
      "learning_rate": 3.18581318234661e-05,
      "loss": 0.1969,
      "step": 7006
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6507979823992471,
      "learning_rate": 3.185336147499568e-05,
      "loss": 0.1239,
      "step": 7007
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9076727233492198,
      "learning_rate": 3.184859085670352e-05,
      "loss": 0.2187,
      "step": 7008
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.48427785323214806,
      "learning_rate": 3.184381996877741e-05,
      "loss": 0.0742,
      "step": 7009
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8967831844197807,
      "learning_rate": 3.183904881140521e-05,
      "loss": 0.1358,
      "step": 7010
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.726244456115221,
      "learning_rate": 3.183427738477477e-05,
      "loss": 0.1758,
      "step": 7011
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5361141167420336,
      "learning_rate": 3.182950568907392e-05,
      "loss": 0.3453,
      "step": 7012
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8363032341465182,
      "learning_rate": 3.182473372449054e-05,
      "loss": 0.2075,
      "step": 7013
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5548223672379279,
      "learning_rate": 3.1819961491212496e-05,
      "loss": 0.091,
      "step": 7014
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4784676052310456,
      "learning_rate": 3.181518898942769e-05,
      "loss": 0.0847,
      "step": 7015
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8353382244983291,
      "learning_rate": 3.1810416219324e-05,
      "loss": 0.1217,
      "step": 7016
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6376910500671523,
      "learning_rate": 3.180564318108935e-05,
      "loss": 0.1439,
      "step": 7017
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7236909145540985,
      "learning_rate": 3.1800869874911645e-05,
      "loss": 0.1569,
      "step": 7018
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5768808461461231,
      "learning_rate": 3.179609630097882e-05,
      "loss": 0.0975,
      "step": 7019
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7005050768025305,
      "learning_rate": 3.179132245947882e-05,
      "loss": 0.2019,
      "step": 7020
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7261615852810153,
      "learning_rate": 3.1786548350599584e-05,
      "loss": 0.1406,
      "step": 7021
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2871343994303839,
      "learning_rate": 3.178177397452907e-05,
      "loss": 0.4548,
      "step": 7022
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3689422467352839,
      "learning_rate": 3.177699933145526e-05,
      "loss": 0.8526,
      "step": 7023
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7760638133651484,
      "learning_rate": 3.177222442156613e-05,
      "loss": 0.1412,
      "step": 7024
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8517822504975786,
      "learning_rate": 3.176744924504967e-05,
      "loss": 0.1511,
      "step": 7025
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.9700336263325005,
      "learning_rate": 3.176267380209388e-05,
      "loss": 0.2504,
      "step": 7026
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7943084009724365,
      "learning_rate": 3.1757898092886776e-05,
      "loss": 0.2731,
      "step": 7027
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9832937924800887,
      "learning_rate": 3.175312211761638e-05,
      "loss": 0.1895,
      "step": 7028
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6815280198883673,
      "learning_rate": 3.174834587647073e-05,
      "loss": 0.1226,
      "step": 7029
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7244749187997913,
      "learning_rate": 3.174356936963787e-05,
      "loss": 0.2385,
      "step": 7030
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.48602488344586603,
      "learning_rate": 3.173879259730584e-05,
      "loss": 0.0445,
      "step": 7031
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.326067876450611,
      "learning_rate": 3.173401555966272e-05,
      "loss": 0.2548,
      "step": 7032
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7143095889239506,
      "learning_rate": 3.172923825689659e-05,
      "loss": 0.1587,
      "step": 7033
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9239645986709258,
      "learning_rate": 3.1724460689195515e-05,
      "loss": 0.1423,
      "step": 7034
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2020368272272646,
      "learning_rate": 3.1719682856747614e-05,
      "loss": 0.4946,
      "step": 7035
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0120293895022074,
      "learning_rate": 3.171490475974097e-05,
      "loss": 0.2587,
      "step": 7036
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7914945887598919,
      "learning_rate": 3.1710126398363726e-05,
      "loss": 0.1959,
      "step": 7037
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5449383241774312,
      "learning_rate": 3.170534777280399e-05,
      "loss": 0.0787,
      "step": 7038
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5683842913706293,
      "learning_rate": 3.1700568883249906e-05,
      "loss": 0.1203,
      "step": 7039
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6287126254605543,
      "learning_rate": 3.1695789729889624e-05,
      "loss": 0.1272,
      "step": 7040
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6245389207751486,
      "learning_rate": 3.1691010312911304e-05,
      "loss": 0.0841,
      "step": 7041
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9939201057645233,
      "learning_rate": 3.1686230632503114e-05,
      "loss": 0.2097,
      "step": 7042
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6399093578788425,
      "learning_rate": 3.168145068885323e-05,
      "loss": 0.1345,
      "step": 7043
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7640363563300089,
      "learning_rate": 3.167667048214985e-05,
      "loss": 0.1106,
      "step": 7044
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6113513797563145,
      "learning_rate": 3.167189001258116e-05,
      "loss": 0.0917,
      "step": 7045
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0271821624648028,
      "learning_rate": 3.166710928033538e-05,
      "loss": 0.2744,
      "step": 7046
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7408821000761635,
      "learning_rate": 3.1662328285600726e-05,
      "loss": 0.2543,
      "step": 7047
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3131152925572387,
      "learning_rate": 3.1657547028565436e-05,
      "loss": 0.3732,
      "step": 7048
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.44366550565910556,
      "learning_rate": 3.165276550941775e-05,
      "loss": 0.0801,
      "step": 7049
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8027145467288122,
      "learning_rate": 3.164798372834592e-05,
      "loss": 0.4412,
      "step": 7050
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8288765113652561,
      "learning_rate": 3.1643201685538206e-05,
      "loss": 0.1984,
      "step": 7051
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7178781895300338,
      "learning_rate": 3.163841938118287e-05,
      "loss": 0.1771,
      "step": 7052
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6443222572034696,
      "learning_rate": 3.163363681546822e-05,
      "loss": 0.1122,
      "step": 7053
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6000980538433655,
      "learning_rate": 3.162885398858252e-05,
      "loss": 0.1154,
      "step": 7054
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0641845031275188,
      "learning_rate": 3.16240709007141e-05,
      "loss": 0.2212,
      "step": 7055
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5972854855088682,
      "learning_rate": 3.161928755205126e-05,
      "loss": 0.1195,
      "step": 7056
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3196137093800835,
      "learning_rate": 3.161450394278232e-05,
      "loss": 0.3913,
      "step": 7057
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9356617501061584,
      "learning_rate": 3.160972007309562e-05,
      "loss": 0.1115,
      "step": 7058
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.41172409165256857,
      "learning_rate": 3.160493594317951e-05,
      "loss": 0.0629,
      "step": 7059
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7576776058794195,
      "learning_rate": 3.1600151553222326e-05,
      "loss": 0.1786,
      "step": 7060
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6769800662850965,
      "learning_rate": 3.159536690341246e-05,
      "loss": 0.1566,
      "step": 7061
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9498750894157199,
      "learning_rate": 3.159058199393826e-05,
      "loss": 0.2031,
      "step": 7062
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7023226342083783,
      "learning_rate": 3.1585796824988136e-05,
      "loss": 0.1552,
      "step": 7063
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6578273121256081,
      "learning_rate": 3.158101139675046e-05,
      "loss": 0.1749,
      "step": 7064
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.124334886172319,
      "learning_rate": 3.1576225709413654e-05,
      "loss": 1.1627,
      "step": 7065
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7414518847008146,
      "learning_rate": 3.157143976316612e-05,
      "loss": 0.206,
      "step": 7066
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4611780839830264,
      "learning_rate": 3.15666535581963e-05,
      "loss": 0.0577,
      "step": 7067
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5576540639309602,
      "learning_rate": 3.156186709469263e-05,
      "loss": 0.0857,
      "step": 7068
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8214251360682553,
      "learning_rate": 3.155708037284354e-05,
      "loss": 0.2509,
      "step": 7069
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7242050180748607,
      "learning_rate": 3.1552293392837496e-05,
      "loss": 0.1039,
      "step": 7070
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5672355133359581,
      "learning_rate": 3.154750615486297e-05,
      "loss": 0.1041,
      "step": 7071
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5612242774937755,
      "learning_rate": 3.154271865910844e-05,
      "loss": 0.1086,
      "step": 7072
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2105347997901466,
      "learning_rate": 3.1537930905762374e-05,
      "loss": 0.5244,
      "step": 7073
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0190039456646638,
      "learning_rate": 3.153314289501329e-05,
      "loss": 0.2797,
      "step": 7074
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.0305026850971437,
      "learning_rate": 3.152835462704969e-05,
      "loss": 0.4257,
      "step": 7075
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6311552363700184,
      "learning_rate": 3.152356610206008e-05,
      "loss": 0.1746,
      "step": 7076
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4966799644495625,
      "learning_rate": 3.151877732023301e-05,
      "loss": 0.1004,
      "step": 7077
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8005512119240046,
      "learning_rate": 3.1513988281756996e-05,
      "loss": 0.219,
      "step": 7078
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7787344748053685,
      "learning_rate": 3.15091989868206e-05,
      "loss": 0.2047,
      "step": 7079
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5316912842748538,
      "learning_rate": 3.150440943561237e-05,
      "loss": 0.0636,
      "step": 7080
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0617752887921015,
      "learning_rate": 3.1499619628320886e-05,
      "loss": 0.4566,
      "step": 7081
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.419673753377176,
      "learning_rate": 3.149482956513471e-05,
      "loss": 0.3448,
      "step": 7082
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7401871995567336,
      "learning_rate": 3.149003924624245e-05,
      "loss": 0.114,
      "step": 7083
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.285177106137643,
      "learning_rate": 3.1485248671832686e-05,
      "loss": 0.2647,
      "step": 7084
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5042228643691717,
      "learning_rate": 3.1480457842094034e-05,
      "loss": 0.0376,
      "step": 7085
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4769831891248662,
      "learning_rate": 3.147566675721512e-05,
      "loss": 0.0666,
      "step": 7086
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.640860873918318,
      "learning_rate": 3.1470875417384554e-05,
      "loss": 0.0905,
      "step": 7087
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6235581239392973,
      "learning_rate": 3.1466083822790987e-05,
      "loss": 0.1174,
      "step": 7088
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9172878510679877,
      "learning_rate": 3.146129197362308e-05,
      "loss": 0.1934,
      "step": 7089
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4629816613770177,
      "learning_rate": 3.145649987006946e-05,
      "loss": 0.0871,
      "step": 7090
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5911573638336575,
      "learning_rate": 3.145170751231882e-05,
      "loss": 0.1131,
      "step": 7091
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8901129283103963,
      "learning_rate": 3.144691490055983e-05,
      "loss": 0.2316,
      "step": 7092
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2137140541860787,
      "learning_rate": 3.144212203498119e-05,
      "loss": 0.2614,
      "step": 7093
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7447923557920415,
      "learning_rate": 3.1437328915771576e-05,
      "loss": 0.1694,
      "step": 7094
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4368345342434903,
      "learning_rate": 3.143253554311971e-05,
      "loss": 0.0698,
      "step": 7095
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1009586438304029,
      "learning_rate": 3.142774191721431e-05,
      "loss": 0.1859,
      "step": 7096
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4750753033758617,
      "learning_rate": 3.142294803824412e-05,
      "loss": 0.0821,
      "step": 7097
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6608374326063229,
      "learning_rate": 3.141815390639784e-05,
      "loss": 0.1171,
      "step": 7098
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4113138323781633,
      "learning_rate": 3.141335952186425e-05,
      "loss": 0.4472,
      "step": 7099
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7294177963848341,
      "learning_rate": 3.1408564884832095e-05,
      "loss": 0.1308,
      "step": 7100
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5398494220393522,
      "learning_rate": 3.1403769995490156e-05,
      "loss": 0.0962,
      "step": 7101
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.689282583037345,
      "learning_rate": 3.1398974854027194e-05,
      "loss": 0.1565,
      "step": 7102
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5655050736671351,
      "learning_rate": 3.1394179460632015e-05,
      "loss": 0.0733,
      "step": 7103
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8602677350582175,
      "learning_rate": 3.13893838154934e-05,
      "loss": 0.0946,
      "step": 7104
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0757188159398947,
      "learning_rate": 3.1384587918800174e-05,
      "loss": 0.2304,
      "step": 7105
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6273724134003763,
      "learning_rate": 3.137979177074114e-05,
      "loss": 0.1252,
      "step": 7106
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7045726174092901,
      "learning_rate": 3.137499537150513e-05,
      "loss": 0.1306,
      "step": 7107
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6064122773568882,
      "learning_rate": 3.137019872128099e-05,
      "loss": 0.0897,
      "step": 7108
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.264986656546956,
      "learning_rate": 3.136540182025756e-05,
      "loss": 0.5008,
      "step": 7109
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0884289514240104,
      "learning_rate": 3.1360604668623695e-05,
      "loss": 0.4615,
      "step": 7110
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.024059249974634,
      "learning_rate": 3.135580726656826e-05,
      "loss": 0.3117,
      "step": 7111
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5287488918249565,
      "learning_rate": 3.135100961428015e-05,
      "loss": 0.0771,
      "step": 7112
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8096916806334249,
      "learning_rate": 3.134621171194823e-05,
      "loss": 0.1425,
      "step": 7113
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6145959704885435,
      "learning_rate": 3.134141355976141e-05,
      "loss": 0.0873,
      "step": 7114
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8023724926212255,
      "learning_rate": 3.13366151579086e-05,
      "loss": 0.1032,
      "step": 7115
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5770949452992585,
      "learning_rate": 3.13318165065787e-05,
      "loss": 0.1179,
      "step": 7116
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5899744851322255,
      "learning_rate": 3.132701760596065e-05,
      "loss": 0.1182,
      "step": 7117
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8743358452090416,
      "learning_rate": 3.132221845624338e-05,
      "loss": 0.1762,
      "step": 7118
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9781750594351958,
      "learning_rate": 3.131741905761585e-05,
      "loss": 0.3582,
      "step": 7119
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6366969770197662,
      "learning_rate": 3.131261941026699e-05,
      "loss": 0.17,
      "step": 7120
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.547820761868924,
      "learning_rate": 3.1307819514385786e-05,
      "loss": 0.0909,
      "step": 7121
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6122939903297173,
      "learning_rate": 3.1303019370161204e-05,
      "loss": 0.0796,
      "step": 7122
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8238399497754096,
      "learning_rate": 3.1298218977782234e-05,
      "loss": 0.2399,
      "step": 7123
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5858731511623881,
      "learning_rate": 3.129341833743787e-05,
      "loss": 0.1132,
      "step": 7124
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0710210918024339,
      "learning_rate": 3.1288617449317115e-05,
      "loss": 0.1991,
      "step": 7125
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6686206253653343,
      "learning_rate": 3.128381631360899e-05,
      "loss": 0.0979,
      "step": 7126
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5296778073692426,
      "learning_rate": 3.127901493050251e-05,
      "loss": 0.1169,
      "step": 7127
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6093718385455134,
      "learning_rate": 3.12742133001867e-05,
      "loss": 0.1165,
      "step": 7128
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2415825069151243,
      "learning_rate": 3.126941142285064e-05,
      "loss": 0.2371,
      "step": 7129
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7039844721786407,
      "learning_rate": 3.1264609298683336e-05,
      "loss": 0.1444,
      "step": 7130
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5094815339853134,
      "learning_rate": 3.1259806927873885e-05,
      "loss": 0.1081,
      "step": 7131
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.776363571568647,
      "learning_rate": 3.125500431061135e-05,
      "loss": 0.1357,
      "step": 7132
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9087797469245381,
      "learning_rate": 3.125020144708481e-05,
      "loss": 0.1842,
      "step": 7133
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.871370112607259,
      "learning_rate": 3.1245398337483356e-05,
      "loss": 0.1752,
      "step": 7134
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9576362415013303,
      "learning_rate": 3.12405949819961e-05,
      "loss": 0.2246,
      "step": 7135
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7890203222348069,
      "learning_rate": 3.123579138081214e-05,
      "loss": 0.1399,
      "step": 7136
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5387123082388983,
      "learning_rate": 3.123098753412061e-05,
      "loss": 0.0755,
      "step": 7137
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.40923194703622,
      "learning_rate": 3.122618344211063e-05,
      "loss": 0.0536,
      "step": 7138
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6335495693532102,
      "learning_rate": 3.122137910497135e-05,
      "loss": 0.0828,
      "step": 7139
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9675473135620088,
      "learning_rate": 3.1216574522891915e-05,
      "loss": 0.1751,
      "step": 7140
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9506926700105858,
      "learning_rate": 3.121176969606149e-05,
      "loss": 0.2132,
      "step": 7141
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6368762588869198,
      "learning_rate": 3.120696462466923e-05,
      "loss": 0.0905,
      "step": 7142
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9319571575664892,
      "learning_rate": 3.120215930890433e-05,
      "loss": 0.1815,
      "step": 7143
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6206134806617665,
      "learning_rate": 3.119735374895598e-05,
      "loss": 0.1342,
      "step": 7144
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7214338828961286,
      "learning_rate": 3.119254794501335e-05,
      "loss": 0.1721,
      "step": 7145
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7203408773098583,
      "learning_rate": 3.118774189726569e-05,
      "loss": 0.2389,
      "step": 7146
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.007249358252609,
      "learning_rate": 3.1182935605902195e-05,
      "loss": 0.3078,
      "step": 7147
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0747288323081263,
      "learning_rate": 3.117812907111209e-05,
      "loss": 0.2165,
      "step": 7148
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9407662484293291,
      "learning_rate": 3.117332229308461e-05,
      "loss": 0.2199,
      "step": 7149
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6042935077787162,
      "learning_rate": 3.116851527200902e-05,
      "loss": 0.1157,
      "step": 7150
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5700994100674024,
      "learning_rate": 3.116370800807455e-05,
      "loss": 0.1311,
      "step": 7151
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7158148162308403,
      "learning_rate": 3.115890050147049e-05,
      "loss": 0.1241,
      "step": 7152
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.645595754770379,
      "learning_rate": 3.11540927523861e-05,
      "loss": 0.1153,
      "step": 7153
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6465588752771543,
      "learning_rate": 3.114928476101066e-05,
      "loss": 0.1213,
      "step": 7154
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.486856390403663,
      "learning_rate": 3.1144476527533476e-05,
      "loss": 0.0991,
      "step": 7155
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4976010382121064,
      "learning_rate": 3.113966805214385e-05,
      "loss": 0.0753,
      "step": 7156
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5611159207121326,
      "learning_rate": 3.1134859335031096e-05,
      "loss": 0.1206,
      "step": 7157
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7406546762328666,
      "learning_rate": 3.113005037638453e-05,
      "loss": 0.125,
      "step": 7158
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7953848061544688,
      "learning_rate": 3.112524117639348e-05,
      "loss": 0.1197,
      "step": 7159
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8965845288073628,
      "learning_rate": 3.1120431735247305e-05,
      "loss": 0.09,
      "step": 7160
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1245172415156182,
      "learning_rate": 3.111562205313534e-05,
      "loss": 0.4562,
      "step": 7161
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6093687505972996,
      "learning_rate": 3.1110812130246955e-05,
      "loss": 0.0958,
      "step": 7162
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.592512911079189,
      "learning_rate": 3.1106001966771516e-05,
      "loss": 0.1956,
      "step": 7163
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5018278963149992,
      "learning_rate": 3.110119156289841e-05,
      "loss": 0.0623,
      "step": 7164
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.859880093270631,
      "learning_rate": 3.1096380918817005e-05,
      "loss": 0.1228,
      "step": 7165
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7613354551787689,
      "learning_rate": 3.1091570034716726e-05,
      "loss": 0.164,
      "step": 7166
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5153466636628304,
      "learning_rate": 3.108675891078697e-05,
      "loss": 0.0957,
      "step": 7167
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6622021969346715,
      "learning_rate": 3.108194754721715e-05,
      "loss": 0.1725,
      "step": 7168
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6951310844768575,
      "learning_rate": 3.107713594419669e-05,
      "loss": 0.181,
      "step": 7169
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1443959466640306,
      "learning_rate": 3.107232410191504e-05,
      "loss": 0.304,
      "step": 7170
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7436666716073808,
      "learning_rate": 3.1067512020561636e-05,
      "loss": 0.744,
      "step": 7171
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.772251192049805,
      "learning_rate": 3.106269970032594e-05,
      "loss": 0.2476,
      "step": 7172
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9149579086768326,
      "learning_rate": 3.10578871413974e-05,
      "loss": 0.2796,
      "step": 7173
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.43100644422807755,
      "learning_rate": 3.105307434396551e-05,
      "loss": 0.061,
      "step": 7174
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7005113711280286,
      "learning_rate": 3.1048261308219744e-05,
      "loss": 0.1688,
      "step": 7175
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6114707509827816,
      "learning_rate": 3.10434480343496e-05,
      "loss": 0.0793,
      "step": 7176
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5596718171450042,
      "learning_rate": 3.103863452254457e-05,
      "loss": 0.0784,
      "step": 7177
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6255708566725351,
      "learning_rate": 3.103382077299417e-05,
      "loss": 0.1417,
      "step": 7178
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2587754701960998,
      "learning_rate": 3.102900678588792e-05,
      "loss": 0.5808,
      "step": 7179
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6940268889683742,
      "learning_rate": 3.102419256141536e-05,
      "loss": 0.228,
      "step": 7180
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6946260992679548,
      "learning_rate": 3.101937809976602e-05,
      "loss": 0.1481,
      "step": 7181
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.39506296802373114,
      "learning_rate": 3.1014563401129445e-05,
      "loss": 0.0757,
      "step": 7182
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1242391538114225,
      "learning_rate": 3.10097484656952e-05,
      "loss": 0.1572,
      "step": 7183
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6082944404043727,
      "learning_rate": 3.1004933293652854e-05,
      "loss": 0.1003,
      "step": 7184
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5950202942715237,
      "learning_rate": 3.100011788519198e-05,
      "loss": 0.1093,
      "step": 7185
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8895327127269516,
      "learning_rate": 3.099530224050217e-05,
      "loss": 0.1502,
      "step": 7186
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5451608804806045,
      "learning_rate": 3.0990486359773005e-05,
      "loss": 0.0897,
      "step": 7187
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6994195451261276,
      "learning_rate": 3.09856702431941e-05,
      "loss": 0.5306,
      "step": 7188
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5561357221643259,
      "learning_rate": 3.0980853890955066e-05,
      "loss": 0.0698,
      "step": 7189
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4249741010579513,
      "learning_rate": 3.0976037303245545e-05,
      "loss": 0.4261,
      "step": 7190
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6476732707393759,
      "learning_rate": 3.097122048025513e-05,
      "loss": 0.1028,
      "step": 7191
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7101706824848925,
      "learning_rate": 3.0966403422173497e-05,
      "loss": 0.1041,
      "step": 7192
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.48997109862426946,
      "learning_rate": 3.096158612919028e-05,
      "loss": 0.0609,
      "step": 7193
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5721424278154089,
      "learning_rate": 3.095676860149515e-05,
      "loss": 0.1337,
      "step": 7194
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8896489167986569,
      "learning_rate": 3.0951950839277765e-05,
      "loss": 0.1734,
      "step": 7195
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8277865466651066,
      "learning_rate": 3.094713284272782e-05,
      "loss": 0.1566,
      "step": 7196
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8106942443079278,
      "learning_rate": 3.094231461203498e-05,
      "loss": 0.1327,
      "step": 7197
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6779475914303812,
      "learning_rate": 3.093749614738896e-05,
      "loss": 0.3894,
      "step": 7198
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6175648359789019,
      "learning_rate": 3.093267744897946e-05,
      "loss": 0.0841,
      "step": 7199
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4964823244025756,
      "learning_rate": 3.0927858516996206e-05,
      "loss": 0.4051,
      "step": 7200
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8294044394876428,
      "learning_rate": 3.0923039351628905e-05,
      "loss": 0.1554,
      "step": 7201
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6357548566150839,
      "learning_rate": 3.09182199530673e-05,
      "loss": 0.2108,
      "step": 7202
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6655448910074221,
      "learning_rate": 3.0913400321501134e-05,
      "loss": 0.0993,
      "step": 7203
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6394156166982197,
      "learning_rate": 3.0908580457120164e-05,
      "loss": 0.1056,
      "step": 7204
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0003469715910238,
      "learning_rate": 3.0903760360114144e-05,
      "loss": 0.2132,
      "step": 7205
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0992749980585659,
      "learning_rate": 3.089894003067285e-05,
      "loss": 0.2042,
      "step": 7206
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5085784101708404,
      "learning_rate": 3.0894119468986054e-05,
      "loss": 0.2601,
      "step": 7207
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4126702380011937,
      "learning_rate": 3.0889298675243556e-05,
      "loss": 0.2836,
      "step": 7208
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4495303410413485,
      "learning_rate": 3.088447764963514e-05,
      "loss": 0.0587,
      "step": 7209
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7649883131741841,
      "learning_rate": 3.087965639235063e-05,
      "loss": 0.0952,
      "step": 7210
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6314249265045428,
      "learning_rate": 3.087483490357983e-05,
      "loss": 0.0926,
      "step": 7211
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6739724108006283,
      "learning_rate": 3.087001318351258e-05,
      "loss": 0.1233,
      "step": 7212
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.057107227928025,
      "learning_rate": 3.0865191232338684e-05,
      "loss": 0.2672,
      "step": 7213
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7016328938921091,
      "learning_rate": 3.086036905024802e-05,
      "loss": 0.1378,
      "step": 7214
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.731186236218433,
      "learning_rate": 3.085554663743042e-05,
      "loss": 0.105,
      "step": 7215
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5238070085581513,
      "learning_rate": 3.085072399407576e-05,
      "loss": 0.0515,
      "step": 7216
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3589529763088877,
      "learning_rate": 3.08459011203739e-05,
      "loss": 0.5553,
      "step": 7217
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.47490921162919464,
      "learning_rate": 3.084107801651472e-05,
      "loss": 0.0676,
      "step": 7218
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7642436632126077,
      "learning_rate": 3.0836254682688114e-05,
      "loss": 0.1995,
      "step": 7219
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9445680970288715,
      "learning_rate": 3.083143111908398e-05,
      "loss": 0.3154,
      "step": 7220
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.139448797529303,
      "learning_rate": 3.0826607325892226e-05,
      "loss": 0.5025,
      "step": 7221
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0196877015379728,
      "learning_rate": 3.082178330330276e-05,
      "loss": 0.151,
      "step": 7222
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5684657466203914,
      "learning_rate": 3.081695905150552e-05,
      "loss": 0.0876,
      "step": 7223
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6410541682110892,
      "learning_rate": 3.081213457069043e-05,
      "loss": 0.1039,
      "step": 7224
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9517521484401817,
      "learning_rate": 3.0807309861047436e-05,
      "loss": 0.1587,
      "step": 7225
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6347650218231162,
      "learning_rate": 3.0802484922766486e-05,
      "loss": 0.1104,
      "step": 7226
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9895696283744707,
      "learning_rate": 3.079765975603755e-05,
      "loss": 0.3671,
      "step": 7227
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6529181042447697,
      "learning_rate": 3.07928343610506e-05,
      "loss": 0.0977,
      "step": 7228
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9110112372143924,
      "learning_rate": 3.07880087379956e-05,
      "loss": 0.1033,
      "step": 7229
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5938438909005397,
      "learning_rate": 3.0783182887062557e-05,
      "loss": 0.0803,
      "step": 7230
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1071595937546528,
      "learning_rate": 3.077835680844145e-05,
      "loss": 0.3979,
      "step": 7231
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.782006330474864,
      "learning_rate": 3.07735305023223e-05,
      "loss": 0.0948,
      "step": 7232
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8328604388859538,
      "learning_rate": 3.0768703968895114e-05,
      "loss": 0.1437,
      "step": 7233
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8284720258853367,
      "learning_rate": 3.076387720834992e-05,
      "loss": 0.215,
      "step": 7234
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6216660460091036,
      "learning_rate": 3.075905022087675e-05,
      "loss": 0.152,
      "step": 7235
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0690981577935308,
      "learning_rate": 3.075422300666564e-05,
      "loss": 0.2114,
      "step": 7236
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3408981533417024,
      "learning_rate": 3.074939556590665e-05,
      "loss": 0.3977,
      "step": 7237
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9724557315675764,
      "learning_rate": 3.0744567898789845e-05,
      "loss": 0.5838,
      "step": 7238
    },
    {
      "epoch": 0.44,
      "grad_norm": 15.074490695916854,
      "learning_rate": 3.073974000550527e-05,
      "loss": 1.0228,
      "step": 7239
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.352083623407205,
      "learning_rate": 3.073491188624303e-05,
      "loss": 0.2304,
      "step": 7240
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8741181861688612,
      "learning_rate": 3.073008354119319e-05,
      "loss": 0.1576,
      "step": 7241
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1978821892757354,
      "learning_rate": 3.072525497054586e-05,
      "loss": 0.2802,
      "step": 7242
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6208317911647484,
      "learning_rate": 3.072042617449113e-05,
      "loss": 0.1473,
      "step": 7243
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7277307163743179,
      "learning_rate": 3.071559715321913e-05,
      "loss": 0.161,
      "step": 7244
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7150972508591118,
      "learning_rate": 3.0710767906919966e-05,
      "loss": 0.1316,
      "step": 7245
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4260391320409107,
      "learning_rate": 3.0705938435783785e-05,
      "loss": 0.4284,
      "step": 7246
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.718162810282657,
      "learning_rate": 3.070110874000072e-05,
      "loss": 0.1673,
      "step": 7247
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7056971707450875,
      "learning_rate": 3.069627881976092e-05,
      "loss": 0.1459,
      "step": 7248
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7906708871897599,
      "learning_rate": 3.0691448675254523e-05,
      "loss": 0.0865,
      "step": 7249
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7844060298378911,
      "learning_rate": 3.0686618306671736e-05,
      "loss": 0.1084,
      "step": 7250
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3490408067704835,
      "learning_rate": 3.068178771420269e-05,
      "loss": 0.4108,
      "step": 7251
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9885032974954031,
      "learning_rate": 3.06769568980376e-05,
      "loss": 0.1363,
      "step": 7252
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0956555847925618,
      "learning_rate": 3.0672125858366644e-05,
      "loss": 0.121,
      "step": 7253
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6101369151083851,
      "learning_rate": 3.0667294595380035e-05,
      "loss": 0.1134,
      "step": 7254
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.80963558317715,
      "learning_rate": 3.0662463109267974e-05,
      "loss": 0.1078,
      "step": 7255
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9081205107316229,
      "learning_rate": 3.0657631400220676e-05,
      "loss": 0.156,
      "step": 7256
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6675904497856344,
      "learning_rate": 3.0652799468428386e-05,
      "loss": 0.1336,
      "step": 7257
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6062509464091327,
      "learning_rate": 3.0647967314081326e-05,
      "loss": 0.0876,
      "step": 7258
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7650616372535147,
      "learning_rate": 3.064313493736974e-05,
      "loss": 0.1548,
      "step": 7259
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7976107573085782,
      "learning_rate": 3.063830233848389e-05,
      "loss": 0.1585,
      "step": 7260
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.260186547929294,
      "learning_rate": 3.0633469517614034e-05,
      "loss": 0.0944,
      "step": 7261
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5129166394859731,
      "learning_rate": 3.062863647495045e-05,
      "loss": 0.0896,
      "step": 7262
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2384429975748168,
      "learning_rate": 3.062380321068341e-05,
      "loss": 0.2008,
      "step": 7263
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4671814208858631,
      "learning_rate": 3.061896972500321e-05,
      "loss": 0.5103,
      "step": 7264
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6967053986696647,
      "learning_rate": 3.061413601810014e-05,
      "loss": 0.1545,
      "step": 7265
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6120832004289983,
      "learning_rate": 3.060930209016451e-05,
      "loss": 0.0866,
      "step": 7266
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7989707077880702,
      "learning_rate": 3.060446794138665e-05,
      "loss": 0.1874,
      "step": 7267
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.362527378297739,
      "learning_rate": 3.0599633571956864e-05,
      "loss": 0.4634,
      "step": 7268
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0575888315228406,
      "learning_rate": 3.0594798982065485e-05,
      "loss": 0.4512,
      "step": 7269
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6456205299391129,
      "learning_rate": 3.058996417190287e-05,
      "loss": 0.0797,
      "step": 7270
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4763528816741687,
      "learning_rate": 3.058512914165935e-05,
      "loss": 0.3576,
      "step": 7271
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8272930044084233,
      "learning_rate": 3.0580293891525294e-05,
      "loss": 0.1708,
      "step": 7272
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4006263284543634,
      "learning_rate": 3.0575458421691074e-05,
      "loss": 0.2056,
      "step": 7273
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7585019998496831,
      "learning_rate": 3.057062273234706e-05,
      "loss": 0.1435,
      "step": 7274
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6947259910696162,
      "learning_rate": 3.056578682368363e-05,
      "loss": 0.2131,
      "step": 7275
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5960367328314345,
      "learning_rate": 3.056095069589119e-05,
      "loss": 0.0612,
      "step": 7276
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.0166216530012844,
      "learning_rate": 3.055611434916013e-05,
      "loss": 0.7053,
      "step": 7277
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7731447232609109,
      "learning_rate": 3.055127778368087e-05,
      "loss": 0.1652,
      "step": 7278
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8999932626963497,
      "learning_rate": 3.054644099964383e-05,
      "loss": 0.1243,
      "step": 7279
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8150289641765761,
      "learning_rate": 3.054160399723942e-05,
      "loss": 0.1922,
      "step": 7280
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6631413735636,
      "learning_rate": 3.0536766776658105e-05,
      "loss": 0.1199,
      "step": 7281
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7096245419889577,
      "learning_rate": 3.0531929338090305e-05,
      "loss": 0.117,
      "step": 7282
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6575856946939223,
      "learning_rate": 3.052709168172648e-05,
      "loss": 0.1029,
      "step": 7283
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0050951953040748,
      "learning_rate": 3.0522253807757096e-05,
      "loss": 0.1619,
      "step": 7284
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7265636199449869,
      "learning_rate": 3.0517415716372628e-05,
      "loss": 0.1087,
      "step": 7285
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9067025228023261,
      "learning_rate": 3.0512577407763548e-05,
      "loss": 0.4363,
      "step": 7286
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5101096084333331,
      "learning_rate": 3.0507738882120342e-05,
      "loss": 0.054,
      "step": 7287
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7207334975446117,
      "learning_rate": 3.0502900139633506e-05,
      "loss": 0.2332,
      "step": 7288
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7327066646033862,
      "learning_rate": 3.0498061180493555e-05,
      "loss": 0.156,
      "step": 7289
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5522664286954413,
      "learning_rate": 3.0493222004890987e-05,
      "loss": 0.0722,
      "step": 7290
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8958998740250556,
      "learning_rate": 3.0488382613016336e-05,
      "loss": 0.2473,
      "step": 7291
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.349265578116554,
      "learning_rate": 3.048354300506013e-05,
      "loss": 0.2615,
      "step": 7292
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6950365037832403,
      "learning_rate": 3.0478703181212902e-05,
      "loss": 0.0992,
      "step": 7293
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.828508491488891,
      "learning_rate": 3.0473863141665205e-05,
      "loss": 0.187,
      "step": 7294
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9021678338686713,
      "learning_rate": 3.046902288660759e-05,
      "loss": 0.3671,
      "step": 7295
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7853426194498094,
      "learning_rate": 3.0464182416230624e-05,
      "loss": 0.3067,
      "step": 7296
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.5889267488147576,
      "learning_rate": 3.045934173072489e-05,
      "loss": 0.4466,
      "step": 7297
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9506123027552776,
      "learning_rate": 3.0454500830280948e-05,
      "loss": 0.1745,
      "step": 7298
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5683233043391107,
      "learning_rate": 3.0449659715089407e-05,
      "loss": 0.0777,
      "step": 7299
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5016474939033824,
      "learning_rate": 3.044481838534085e-05,
      "loss": 0.0726,
      "step": 7300
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6572249079948482,
      "learning_rate": 3.0439976841225897e-05,
      "loss": 0.1501,
      "step": 7301
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6559391772456721,
      "learning_rate": 3.0435135082935156e-05,
      "loss": 0.1124,
      "step": 7302
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.941048333739013,
      "learning_rate": 3.0430293110659254e-05,
      "loss": 0.1919,
      "step": 7303
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7559590913025608,
      "learning_rate": 3.042545092458881e-05,
      "loss": 0.1781,
      "step": 7304
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.624510214929492,
      "learning_rate": 3.042060852491449e-05,
      "loss": 1.0718,
      "step": 7305
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.559775260507635,
      "learning_rate": 3.0415765911826914e-05,
      "loss": 0.0927,
      "step": 7306
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7877539400389943,
      "learning_rate": 3.041092308551676e-05,
      "loss": 0.1169,
      "step": 7307
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8570978480471426,
      "learning_rate": 3.0406080046174685e-05,
      "loss": 0.0979,
      "step": 7308
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9886033794431,
      "learning_rate": 3.0401236793991366e-05,
      "loss": 0.2563,
      "step": 7309
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9510505884055749,
      "learning_rate": 3.039639332915748e-05,
      "loss": 0.1506,
      "step": 7310
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6827680465238872,
      "learning_rate": 3.0391549651863728e-05,
      "loss": 0.1037,
      "step": 7311
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8407841617147443,
      "learning_rate": 3.0386705762300788e-05,
      "loss": 0.1485,
      "step": 7312
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7491296854225983,
      "learning_rate": 3.0381861660659395e-05,
      "loss": 0.0964,
      "step": 7313
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6169716114619198,
      "learning_rate": 3.0377017347130243e-05,
      "loss": 0.1014,
      "step": 7314
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.814991779601196,
      "learning_rate": 3.0372172821904072e-05,
      "loss": 0.1556,
      "step": 7315
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8630886162402445,
      "learning_rate": 3.03673280851716e-05,
      "loss": 0.115,
      "step": 7316
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1888700725342953,
      "learning_rate": 3.036248313712358e-05,
      "loss": 0.1974,
      "step": 7317
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0300838865001152,
      "learning_rate": 3.035763797795076e-05,
      "loss": 0.0815,
      "step": 7318
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7474991107922511,
      "learning_rate": 3.0352792607843884e-05,
      "loss": 0.1873,
      "step": 7319
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5355667209792796,
      "learning_rate": 3.0347947026993727e-05,
      "loss": 0.3826,
      "step": 7320
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.618557328332486,
      "learning_rate": 3.0343101235591064e-05,
      "loss": 0.1014,
      "step": 7321
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6526305995015348,
      "learning_rate": 3.0338255233826685e-05,
      "loss": 0.0853,
      "step": 7322
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.963658536212752,
      "learning_rate": 3.033340902189136e-05,
      "loss": 0.1459,
      "step": 7323
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.714036603861026,
      "learning_rate": 3.0328562599975906e-05,
      "loss": 0.092,
      "step": 7324
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0402624196019805,
      "learning_rate": 3.032371596827112e-05,
      "loss": 0.1638,
      "step": 7325
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.70204390455334,
      "learning_rate": 3.031886912696782e-05,
      "loss": 0.1407,
      "step": 7326
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8034399895912384,
      "learning_rate": 3.0314022076256836e-05,
      "loss": 0.2539,
      "step": 7327
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6766455566669751,
      "learning_rate": 3.030917481632899e-05,
      "loss": 0.0783,
      "step": 7328
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.286006519464085,
      "learning_rate": 3.0304327347375128e-05,
      "loss": 0.2719,
      "step": 7329
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0534650795329874,
      "learning_rate": 3.0299479669586093e-05,
      "loss": 0.1366,
      "step": 7330
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6781991874409335,
      "learning_rate": 3.0294631783152756e-05,
      "loss": 0.1596,
      "step": 7331
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7123279253805425,
      "learning_rate": 3.0289783688265962e-05,
      "loss": 0.1308,
      "step": 7332
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7022370613026655,
      "learning_rate": 3.02849353851166e-05,
      "loss": 0.1548,
      "step": 7333
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.403609411391562,
      "learning_rate": 3.028008687389553e-05,
      "loss": 0.6654,
      "step": 7334
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5256374862391542,
      "learning_rate": 3.0275238154793674e-05,
      "loss": 0.1059,
      "step": 7335
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.45255382919003057,
      "learning_rate": 3.0270389228001893e-05,
      "loss": 0.0574,
      "step": 7336
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6448789700431434,
      "learning_rate": 3.0265540093711127e-05,
      "loss": 0.0951,
      "step": 7337
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4309317516344258,
      "learning_rate": 3.0260690752112262e-05,
      "loss": 0.4004,
      "step": 7338
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.936022378053887,
      "learning_rate": 3.025584120339624e-05,
      "loss": 0.1174,
      "step": 7339
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9471652507167624,
      "learning_rate": 3.025099144775398e-05,
      "loss": 0.2854,
      "step": 7340
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9520842798101397,
      "learning_rate": 3.0246141485376422e-05,
      "loss": 0.2589,
      "step": 7341
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5299916455899524,
      "learning_rate": 3.0241291316454518e-05,
      "loss": 0.0961,
      "step": 7342
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8229949953345709,
      "learning_rate": 3.0236440941179216e-05,
      "loss": 0.1653,
      "step": 7343
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6960574534400148,
      "learning_rate": 3.023159035974148e-05,
      "loss": 0.1207,
      "step": 7344
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6935180071035816,
      "learning_rate": 3.0226739572332285e-05,
      "loss": 0.0989,
      "step": 7345
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1107228585278923,
      "learning_rate": 3.0221888579142603e-05,
      "loss": 0.2219,
      "step": 7346
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8331340085341821,
      "learning_rate": 3.0217037380363435e-05,
      "loss": 0.2067,
      "step": 7347
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5231950269241292,
      "learning_rate": 3.0212185976185754e-05,
      "loss": 0.5033,
      "step": 7348
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.659719817584222,
      "learning_rate": 3.020733436680058e-05,
      "loss": 0.1359,
      "step": 7349
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7272520042499532,
      "learning_rate": 3.0202482552398925e-05,
      "loss": 0.1259,
      "step": 7350
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6062260036709944,
      "learning_rate": 3.01976305331718e-05,
      "loss": 0.107,
      "step": 7351
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6063761071670122,
      "learning_rate": 3.019277830931024e-05,
      "loss": 0.1192,
      "step": 7352
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8883792151090361,
      "learning_rate": 3.0187925881005268e-05,
      "loss": 0.1078,
      "step": 7353
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6218722497877888,
      "learning_rate": 3.0183073248447942e-05,
      "loss": 0.1645,
      "step": 7354
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.483836905804374,
      "learning_rate": 3.0178220411829306e-05,
      "loss": 0.0999,
      "step": 7355
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5921394351770981,
      "learning_rate": 3.017336737134042e-05,
      "loss": 0.1051,
      "step": 7356
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8304179819421619,
      "learning_rate": 3.0168514127172354e-05,
      "loss": 0.1791,
      "step": 7357
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1248244426503868,
      "learning_rate": 3.016366067951618e-05,
      "loss": 0.4261,
      "step": 7358
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5494266651915521,
      "learning_rate": 3.015880702856299e-05,
      "loss": 0.0834,
      "step": 7359
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5957312421518993,
      "learning_rate": 3.015395317450387e-05,
      "loss": 0.1023,
      "step": 7360
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8372679542440503,
      "learning_rate": 3.0149099117529917e-05,
      "loss": 0.1368,
      "step": 7361
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8215455155403533,
      "learning_rate": 3.0144244857832237e-05,
      "loss": 0.1434,
      "step": 7362
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9092862884704459,
      "learning_rate": 3.0139390395601958e-05,
      "loss": 0.1804,
      "step": 7363
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4199534619442187,
      "learning_rate": 3.013453573103019e-05,
      "loss": 0.0722,
      "step": 7364
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6632488762602188,
      "learning_rate": 3.0129680864308075e-05,
      "loss": 0.1448,
      "step": 7365
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5934670127820618,
      "learning_rate": 3.0124825795626743e-05,
      "loss": 0.1524,
      "step": 7366
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8936509386116446,
      "learning_rate": 3.0119970525177355e-05,
      "loss": 0.1815,
      "step": 7367
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6072780889503393,
      "learning_rate": 3.0115115053151043e-05,
      "loss": 0.118,
      "step": 7368
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5154342358762419,
      "learning_rate": 3.0110259379739002e-05,
      "loss": 0.0927,
      "step": 7369
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7414162316217129,
      "learning_rate": 3.010540350513238e-05,
      "loss": 0.1444,
      "step": 7370
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.600179319350206,
      "learning_rate": 3.0100547429522364e-05,
      "loss": 0.1421,
      "step": 7371
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9411003524462657,
      "learning_rate": 3.0095691153100137e-05,
      "loss": 0.2581,
      "step": 7372
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.46475419149368286,
      "learning_rate": 3.00908346760569e-05,
      "loss": 0.076,
      "step": 7373
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7288970168237654,
      "learning_rate": 3.008597799858385e-05,
      "loss": 0.1264,
      "step": 7374
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0241607309780012,
      "learning_rate": 3.0081121120872208e-05,
      "loss": 0.3173,
      "step": 7375
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5945101482914218,
      "learning_rate": 3.0076264043113178e-05,
      "loss": 0.0798,
      "step": 7376
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5824479490141159,
      "learning_rate": 3.0071406765498e-05,
      "loss": 0.1046,
      "step": 7377
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8642556271272012,
      "learning_rate": 3.00665492882179e-05,
      "loss": 0.1699,
      "step": 7378
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7655765237726927,
      "learning_rate": 3.0061691611464126e-05,
      "loss": 0.147,
      "step": 7379
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5960845174753683,
      "learning_rate": 3.0056833735427924e-05,
      "loss": 0.0922,
      "step": 7380
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4943401200360342,
      "learning_rate": 3.0051975660300558e-05,
      "loss": 0.0517,
      "step": 7381
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9397580132687379,
      "learning_rate": 3.004711738627328e-05,
      "loss": 0.2043,
      "step": 7382
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4991927956466304,
      "learning_rate": 3.0042258913537387e-05,
      "loss": 0.1046,
      "step": 7383
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.66725970923421,
      "learning_rate": 3.0037400242284137e-05,
      "loss": 0.0916,
      "step": 7384
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5400330117488547,
      "learning_rate": 3.0032541372704837e-05,
      "loss": 0.1554,
      "step": 7385
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5502908179731195,
      "learning_rate": 3.0027682304990772e-05,
      "loss": 0.0754,
      "step": 7386
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4379571497255034,
      "learning_rate": 3.002282303933325e-05,
      "loss": 0.0737,
      "step": 7387
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5414310855131473,
      "learning_rate": 3.0017963575923585e-05,
      "loss": 0.1421,
      "step": 7388
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6425260376795057,
      "learning_rate": 3.0013103914953094e-05,
      "loss": 0.0911,
      "step": 7389
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4281192360090279,
      "learning_rate": 3.0008244056613123e-05,
      "loss": 0.089,
      "step": 7390
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6713159506850783,
      "learning_rate": 3.000338400109498e-05,
      "loss": 0.1176,
      "step": 7391
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5463555800832403,
      "learning_rate": 2.999852374859003e-05,
      "loss": 0.0896,
      "step": 7392
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7021657506191858,
      "learning_rate": 2.999366329928962e-05,
      "loss": 0.1597,
      "step": 7393
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4835224289098072,
      "learning_rate": 2.9988802653385097e-05,
      "loss": 0.0478,
      "step": 7394
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6643190948968772,
      "learning_rate": 2.9983941811067852e-05,
      "loss": 0.0886,
      "step": 7395
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.048369887541838,
      "learning_rate": 2.9979080772529235e-05,
      "loss": 0.151,
      "step": 7396
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5691497723733093,
      "learning_rate": 2.997421953796064e-05,
      "loss": 0.1302,
      "step": 7397
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1576777079387344,
      "learning_rate": 2.996935810755346e-05,
      "loss": 0.2792,
      "step": 7398
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5749648158901726,
      "learning_rate": 2.9964496481499087e-05,
      "loss": 0.0573,
      "step": 7399
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5930568539669735,
      "learning_rate": 2.9959634659988933e-05,
      "loss": 0.1093,
      "step": 7400
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6007283937204033,
      "learning_rate": 2.9954772643214408e-05,
      "loss": 0.1029,
      "step": 7401
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.707767320585573,
      "learning_rate": 2.9949910431366927e-05,
      "loss": 0.1388,
      "step": 7402
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6104470306027452,
      "learning_rate": 2.9945048024637935e-05,
      "loss": 0.1377,
      "step": 7403
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2360515530324003,
      "learning_rate": 2.9940185423218854e-05,
      "loss": 0.2564,
      "step": 7404
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.683273176344083,
      "learning_rate": 2.9935322627301133e-05,
      "loss": 0.1873,
      "step": 7405
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6616153981749254,
      "learning_rate": 2.993045963707622e-05,
      "loss": 0.1926,
      "step": 7406
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.734126259358664,
      "learning_rate": 2.9925596452735583e-05,
      "loss": 0.1515,
      "step": 7407
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6501222442330132,
      "learning_rate": 2.9920733074470676e-05,
      "loss": 0.1299,
      "step": 7408
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6139806706009059,
      "learning_rate": 2.9915869502472983e-05,
      "loss": 0.1864,
      "step": 7409
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6034635585494589,
      "learning_rate": 2.9911005736933988e-05,
      "loss": 0.0906,
      "step": 7410
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5168302569953388,
      "learning_rate": 2.990614177804517e-05,
      "loss": 0.0653,
      "step": 7411
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7969708843252068,
      "learning_rate": 2.9901277625998048e-05,
      "loss": 0.1424,
      "step": 7412
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4973410395603693,
      "learning_rate": 2.9896413280984105e-05,
      "loss": 0.075,
      "step": 7413
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8619167219720878,
      "learning_rate": 2.9891548743194864e-05,
      "loss": 0.2405,
      "step": 7414
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.45111376924255947,
      "learning_rate": 2.9886684012821835e-05,
      "loss": 0.0441,
      "step": 7415
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5122139500561799,
      "learning_rate": 2.9881819090056568e-05,
      "loss": 0.1055,
      "step": 7416
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4874110285829841,
      "learning_rate": 2.9876953975090583e-05,
      "loss": 0.0659,
      "step": 7417
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5745190304430071,
      "learning_rate": 2.987208866811541e-05,
      "loss": 0.1419,
      "step": 7418
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7280015173151763,
      "learning_rate": 2.9867223169322632e-05,
      "loss": 0.25,
      "step": 7419
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7450176942757514,
      "learning_rate": 2.9862357478903785e-05,
      "loss": 0.2027,
      "step": 7420
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0237847151048247,
      "learning_rate": 2.985749159705043e-05,
      "loss": 0.3266,
      "step": 7421
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7274129118545821,
      "learning_rate": 2.985262552395416e-05,
      "loss": 0.1567,
      "step": 7422
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8546452344023775,
      "learning_rate": 2.984775925980654e-05,
      "loss": 0.0885,
      "step": 7423
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7815250419738902,
      "learning_rate": 2.9842892804799166e-05,
      "loss": 0.1555,
      "step": 7424
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4767360702379309,
      "learning_rate": 2.983802615912363e-05,
      "loss": 0.0597,
      "step": 7425
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.166718230505339,
      "learning_rate": 2.9833159322971537e-05,
      "loss": 0.4776,
      "step": 7426
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7125120961150326,
      "learning_rate": 2.9828292296534493e-05,
      "loss": 0.1958,
      "step": 7427
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6144205398813672,
      "learning_rate": 2.982342508000413e-05,
      "loss": 0.0967,
      "step": 7428
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6358075348705078,
      "learning_rate": 2.9818557673572057e-05,
      "loss": 0.1037,
      "step": 7429
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8562737366301942,
      "learning_rate": 2.9813690077429923e-05,
      "loss": 0.1609,
      "step": 7430
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7167256077135298,
      "learning_rate": 2.9808822291769357e-05,
      "loss": 0.1318,
      "step": 7431
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0609412299032477,
      "learning_rate": 2.9803954316782013e-05,
      "loss": 0.2848,
      "step": 7432
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7166199053424717,
      "learning_rate": 2.9799086152659538e-05,
      "loss": 0.1511,
      "step": 7433
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7078808654545701,
      "learning_rate": 2.9794217799593603e-05,
      "loss": 0.1435,
      "step": 7434
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6122747818348071,
      "learning_rate": 2.9789349257775884e-05,
      "loss": 0.1073,
      "step": 7435
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8416189028747824,
      "learning_rate": 2.9784480527398052e-05,
      "loss": 0.1863,
      "step": 7436
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7619099187782093,
      "learning_rate": 2.9779611608651786e-05,
      "loss": 0.1879,
      "step": 7437
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8746145277486354,
      "learning_rate": 2.9774742501728792e-05,
      "loss": 0.1452,
      "step": 7438
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5735326338896988,
      "learning_rate": 2.9769873206820754e-05,
      "loss": 0.0856,
      "step": 7439
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2257959447301743,
      "learning_rate": 2.97650037241194e-05,
      "loss": 0.2202,
      "step": 7440
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.667833307208343,
      "learning_rate": 2.976013405381643e-05,
      "loss": 0.1331,
      "step": 7441
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3150650751128776,
      "learning_rate": 2.975526419610357e-05,
      "loss": 0.3419,
      "step": 7442
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6286113767072102,
      "learning_rate": 2.975039415117255e-05,
      "loss": 0.1531,
      "step": 7443
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7018087831643491,
      "learning_rate": 2.974552391921511e-05,
      "loss": 0.1526,
      "step": 7444
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.550746788310049,
      "learning_rate": 2.9740653500422992e-05,
      "loss": 0.1058,
      "step": 7445
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1875467650457616,
      "learning_rate": 2.9735782894987953e-05,
      "loss": 0.3023,
      "step": 7446
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4352306118530978,
      "learning_rate": 2.9730912103101738e-05,
      "loss": 0.0458,
      "step": 7447
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6917417403464187,
      "learning_rate": 2.972604112495613e-05,
      "loss": 0.1031,
      "step": 7448
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9103512306551529,
      "learning_rate": 2.972116996074289e-05,
      "loss": 0.3299,
      "step": 7449
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8773712151888706,
      "learning_rate": 2.9716298610653814e-05,
      "loss": 0.1647,
      "step": 7450
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.44332883049660465,
      "learning_rate": 2.9711427074880676e-05,
      "loss": 0.1134,
      "step": 7451
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8535153290496169,
      "learning_rate": 2.9706555353615278e-05,
      "loss": 0.2867,
      "step": 7452
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7776175846277418,
      "learning_rate": 2.970168344704942e-05,
      "loss": 0.1387,
      "step": 7453
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6363670098013472,
      "learning_rate": 2.969681135537492e-05,
      "loss": 0.1108,
      "step": 7454
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9061288014061927,
      "learning_rate": 2.9691939078783588e-05,
      "loss": 0.1772,
      "step": 7455
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8375191847896611,
      "learning_rate": 2.968706661746726e-05,
      "loss": 0.1458,
      "step": 7456
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5678262906491983,
      "learning_rate": 2.9682193971617744e-05,
      "loss": 0.1032,
      "step": 7457
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6220661054544824,
      "learning_rate": 2.9677321141426905e-05,
      "loss": 0.1551,
      "step": 7458
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8832890375368847,
      "learning_rate": 2.9672448127086582e-05,
      "loss": 0.2372,
      "step": 7459
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6438792244303907,
      "learning_rate": 2.9667574928788628e-05,
      "loss": 0.1924,
      "step": 7460
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5191513613703147,
      "learning_rate": 2.9662701546724896e-05,
      "loss": 0.0958,
      "step": 7461
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.47542177744395236,
      "learning_rate": 2.965782798108727e-05,
      "loss": 0.0659,
      "step": 7462
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8308864560950988,
      "learning_rate": 2.9652954232067613e-05,
      "loss": 0.1263,
      "step": 7463
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6017654704194706,
      "learning_rate": 2.9648080299857815e-05,
      "loss": 0.0597,
      "step": 7464
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6452260983773799,
      "learning_rate": 2.9643206184649768e-05,
      "loss": 0.1099,
      "step": 7465
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5028669433465633,
      "learning_rate": 2.9638331886635368e-05,
      "loss": 0.0782,
      "step": 7466
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.890788623522704,
      "learning_rate": 2.963345740600651e-05,
      "loss": 0.1979,
      "step": 7467
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5585499349966957,
      "learning_rate": 2.9628582742955112e-05,
      "loss": 0.0689,
      "step": 7468
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.0867743644863435,
      "learning_rate": 2.9623707897673098e-05,
      "loss": 0.4866,
      "step": 7469
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8444004126810808,
      "learning_rate": 2.9618832870352397e-05,
      "loss": 0.2249,
      "step": 7470
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7156652858940601,
      "learning_rate": 2.9613957661184927e-05,
      "loss": 0.1028,
      "step": 7471
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5581688849590227,
      "learning_rate": 2.9609082270362642e-05,
      "loss": 0.0888,
      "step": 7472
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5691644052715865,
      "learning_rate": 2.9604206698077474e-05,
      "loss": 0.0999,
      "step": 7473
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6016278134824117,
      "learning_rate": 2.9599330944521403e-05,
      "loss": 0.0812,
      "step": 7474
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6636632871601396,
      "learning_rate": 2.9594455009886368e-05,
      "loss": 0.2641,
      "step": 7475
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0555413501565936,
      "learning_rate": 2.958957889436435e-05,
      "loss": 0.237,
      "step": 7476
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7520589601609488,
      "learning_rate": 2.958470259814732e-05,
      "loss": 0.1764,
      "step": 7477
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6076498538214637,
      "learning_rate": 2.9579826121427267e-05,
      "loss": 0.1249,
      "step": 7478
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7282687032469389,
      "learning_rate": 2.957494946439617e-05,
      "loss": 0.0771,
      "step": 7479
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6242314187846587,
      "learning_rate": 2.957007262724604e-05,
      "loss": 0.1075,
      "step": 7480
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7362855030072308,
      "learning_rate": 2.9565195610168873e-05,
      "loss": 0.1703,
      "step": 7481
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7199993954403017,
      "learning_rate": 2.9560318413356684e-05,
      "loss": 0.2101,
      "step": 7482
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6876615934089687,
      "learning_rate": 2.9555441037001486e-05,
      "loss": 0.1143,
      "step": 7483
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7265342317952551,
      "learning_rate": 2.955056348129532e-05,
      "loss": 0.2523,
      "step": 7484
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9705195705333501,
      "learning_rate": 2.95456857464302e-05,
      "loss": 0.1608,
      "step": 7485
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5116411386727882,
      "learning_rate": 2.9540807832598177e-05,
      "loss": 0.0794,
      "step": 7486
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4989621058727552,
      "learning_rate": 2.95359297399913e-05,
      "loss": 0.0832,
      "step": 7487
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7823871535007529,
      "learning_rate": 2.953105146880161e-05,
      "loss": 0.1015,
      "step": 7488
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5879209840682943,
      "learning_rate": 2.952617301922118e-05,
      "loss": 0.1095,
      "step": 7489
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6430027835773946,
      "learning_rate": 2.952129439144208e-05,
      "loss": 0.1264,
      "step": 7490
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5988880220275644,
      "learning_rate": 2.9516415585656366e-05,
      "loss": 0.1305,
      "step": 7491
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9868245762070976,
      "learning_rate": 2.9511536602056144e-05,
      "loss": 0.187,
      "step": 7492
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7101830299925008,
      "learning_rate": 2.9506657440833485e-05,
      "loss": 0.1784,
      "step": 7493
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.777508796285592,
      "learning_rate": 2.9501778102180504e-05,
      "loss": 0.1922,
      "step": 7494
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8958969272389445,
      "learning_rate": 2.9496898586289284e-05,
      "loss": 0.9822,
      "step": 7495
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6756459356134158,
      "learning_rate": 2.949201889335195e-05,
      "loss": 0.1233,
      "step": 7496
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0018132044698393,
      "learning_rate": 2.9487139023560605e-05,
      "loss": 0.1427,
      "step": 7497
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7322296493748011,
      "learning_rate": 2.9482258977107385e-05,
      "loss": 0.1605,
      "step": 7498
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5464477358106761,
      "learning_rate": 2.9477378754184414e-05,
      "loss": 0.0899,
      "step": 7499
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7573550150012847,
      "learning_rate": 2.947249835498384e-05,
      "loss": 0.14,
      "step": 7500
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.397644498494794,
      "learning_rate": 2.9467617779697794e-05,
      "loss": 0.2986,
      "step": 7501
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.639534444088886,
      "learning_rate": 2.946273702851844e-05,
      "loss": 0.1191,
      "step": 7502
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.45815057192943404,
      "learning_rate": 2.945785610163792e-05,
      "loss": 0.0779,
      "step": 7503
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2120127777515626,
      "learning_rate": 2.9452974999248423e-05,
      "loss": 0.1594,
      "step": 7504
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7864284463282286,
      "learning_rate": 2.9448093721542097e-05,
      "loss": 0.165,
      "step": 7505
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6433320913718618,
      "learning_rate": 2.9443212268711146e-05,
      "loss": 0.0995,
      "step": 7506
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5526931086115836,
      "learning_rate": 2.943833064094773e-05,
      "loss": 0.1414,
      "step": 7507
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8853513368240888,
      "learning_rate": 2.9433448838444067e-05,
      "loss": 0.225,
      "step": 7508
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6858500634876159,
      "learning_rate": 2.9428566861392336e-05,
      "loss": 0.1276,
      "step": 7509
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5875745140288811,
      "learning_rate": 2.9423684709984765e-05,
      "loss": 0.1343,
      "step": 7510
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7663354571381998,
      "learning_rate": 2.9418802384413552e-05,
      "loss": 0.1633,
      "step": 7511
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5864894593338386,
      "learning_rate": 2.941391988487092e-05,
      "loss": 0.0803,
      "step": 7512
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5415032921377597,
      "learning_rate": 2.94090372115491e-05,
      "loss": 0.0854,
      "step": 7513
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7457284767134365,
      "learning_rate": 2.9404154364640328e-05,
      "loss": 0.1559,
      "step": 7514
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.036122550450473,
      "learning_rate": 2.939927134433684e-05,
      "loss": 0.1637,
      "step": 7515
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.801879793823593,
      "learning_rate": 2.9394388150830887e-05,
      "loss": 0.16,
      "step": 7516
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.841337271554699,
      "learning_rate": 2.938950478431472e-05,
      "loss": 0.2011,
      "step": 7517
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.49134189681558066,
      "learning_rate": 2.9384621244980608e-05,
      "loss": 0.0636,
      "step": 7518
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5629333429193708,
      "learning_rate": 2.9379737533020812e-05,
      "loss": 0.1153,
      "step": 7519
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.217126387202034,
      "learning_rate": 2.937485364862761e-05,
      "loss": 0.691,
      "step": 7520
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.632420595927106,
      "learning_rate": 2.936996959199329e-05,
      "loss": 0.1553,
      "step": 7521
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6715560039600054,
      "learning_rate": 2.9365085363310134e-05,
      "loss": 0.083,
      "step": 7522
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5097698388358596,
      "learning_rate": 2.9360200962770435e-05,
      "loss": 0.0649,
      "step": 7523
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5097612110076224,
      "learning_rate": 2.93553163905665e-05,
      "loss": 0.3329,
      "step": 7524
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7741758957411063,
      "learning_rate": 2.935043164689064e-05,
      "loss": 0.1145,
      "step": 7525
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5683895525964403,
      "learning_rate": 2.934554673193517e-05,
      "loss": 0.1009,
      "step": 7526
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6154202160256824,
      "learning_rate": 2.9340661645892402e-05,
      "loss": 0.0587,
      "step": 7527
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1550737590811642,
      "learning_rate": 2.9335776388954682e-05,
      "loss": 0.3409,
      "step": 7528
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6202565544090031,
      "learning_rate": 2.9330890961314334e-05,
      "loss": 0.1154,
      "step": 7529
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7279545680540028,
      "learning_rate": 2.932600536316371e-05,
      "loss": 0.0956,
      "step": 7530
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9582161790791398,
      "learning_rate": 2.9321119594695146e-05,
      "loss": 0.2975,
      "step": 7531
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4975752982053029,
      "learning_rate": 2.9316233656101017e-05,
      "loss": 0.099,
      "step": 7532
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1114657129828456,
      "learning_rate": 2.9311347547573665e-05,
      "loss": 0.3165,
      "step": 7533
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5303031843766733,
      "learning_rate": 2.930646126930548e-05,
      "loss": 0.0799,
      "step": 7534
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6475007584495909,
      "learning_rate": 2.930157482148882e-05,
      "loss": 0.162,
      "step": 7535
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.628602584447114,
      "learning_rate": 2.9296688204316087e-05,
      "loss": 0.154,
      "step": 7536
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6360231349970188,
      "learning_rate": 2.929180141797965e-05,
      "loss": 0.1383,
      "step": 7537
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4892391257672652,
      "learning_rate": 2.928691446267192e-05,
      "loss": 0.0923,
      "step": 7538
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3296936313598982,
      "learning_rate": 2.9282027338585296e-05,
      "loss": 0.0707,
      "step": 7539
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.806040683444565,
      "learning_rate": 2.927714004591218e-05,
      "loss": 0.2294,
      "step": 7540
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9817205294364678,
      "learning_rate": 2.9272252584845e-05,
      "loss": 0.1745,
      "step": 7541
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6404819601968936,
      "learning_rate": 2.9267364955576173e-05,
      "loss": 0.1208,
      "step": 7542
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0941131261372976,
      "learning_rate": 2.9262477158298135e-05,
      "loss": 0.2207,
      "step": 7543
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5996476175266281,
      "learning_rate": 2.9257589193203305e-05,
      "loss": 0.1131,
      "step": 7544
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7482783578531397,
      "learning_rate": 2.925270106048415e-05,
      "loss": 0.0917,
      "step": 7545
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4900357017250199,
      "learning_rate": 2.92478127603331e-05,
      "loss": 0.0855,
      "step": 7546
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.480208926107963,
      "learning_rate": 2.924292429294262e-05,
      "loss": 0.0504,
      "step": 7547
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7392092883301324,
      "learning_rate": 2.923803565850517e-05,
      "loss": 0.1109,
      "step": 7548
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4852989722774302,
      "learning_rate": 2.923314685721321e-05,
      "loss": 0.0681,
      "step": 7549
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9628077511420637,
      "learning_rate": 2.9228257889259235e-05,
      "loss": 0.2262,
      "step": 7550
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4673953989636423,
      "learning_rate": 2.9223368754835716e-05,
      "loss": 0.0745,
      "step": 7551
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9463168650325802,
      "learning_rate": 2.9218479454135133e-05,
      "loss": 0.267,
      "step": 7552
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5984054747127526,
      "learning_rate": 2.9213589987350004e-05,
      "loss": 0.1055,
      "step": 7553
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6540335922994628,
      "learning_rate": 2.9208700354672807e-05,
      "loss": 0.1305,
      "step": 7554
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6304587003080121,
      "learning_rate": 2.920381055629606e-05,
      "loss": 0.1278,
      "step": 7555
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.48810905252042786,
      "learning_rate": 2.9198920592412283e-05,
      "loss": 0.0601,
      "step": 7556
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.932082808241126,
      "learning_rate": 2.9194030463213995e-05,
      "loss": 0.1778,
      "step": 7557
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5085337039637003,
      "learning_rate": 2.918914016889372e-05,
      "loss": 0.0973,
      "step": 7558
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0397693864297846,
      "learning_rate": 2.9184249709643995e-05,
      "loss": 0.3491,
      "step": 7559
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9135319873465982,
      "learning_rate": 2.9179359085657354e-05,
      "loss": 0.2049,
      "step": 7560
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8860007675065337,
      "learning_rate": 2.917446829712636e-05,
      "loss": 0.258,
      "step": 7561
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5353746012246232,
      "learning_rate": 2.9169577344243548e-05,
      "loss": 0.1043,
      "step": 7562
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4896640887726316,
      "learning_rate": 2.9164686227201498e-05,
      "loss": 0.1371,
      "step": 7563
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9740895862945316,
      "learning_rate": 2.915979494619276e-05,
      "loss": 0.2138,
      "step": 7564
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.6064611124156274,
      "learning_rate": 2.9154903501409914e-05,
      "loss": 0.5157,
      "step": 7565
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7518646919606625,
      "learning_rate": 2.9150011893045536e-05,
      "loss": 0.1335,
      "step": 7566
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5798772136406477,
      "learning_rate": 2.9145120121292226e-05,
      "loss": 0.0849,
      "step": 7567
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5000354753536101,
      "learning_rate": 2.9140228186342555e-05,
      "loss": 0.0681,
      "step": 7568
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5851768845413686,
      "learning_rate": 2.9135336088389142e-05,
      "loss": 0.1135,
      "step": 7569
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7275409935355809,
      "learning_rate": 2.9130443827624575e-05,
      "loss": 0.1048,
      "step": 7570
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.760056742521553,
      "learning_rate": 2.912555140424148e-05,
      "loss": 0.0933,
      "step": 7571
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0120905373693199,
      "learning_rate": 2.9120658818432465e-05,
      "loss": 0.2296,
      "step": 7572
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5475354876360639,
      "learning_rate": 2.9115766070390173e-05,
      "loss": 0.3548,
      "step": 7573
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.58497172049597,
      "learning_rate": 2.9110873160307205e-05,
      "loss": 0.1375,
      "step": 7574
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8879597121464655,
      "learning_rate": 2.9105980088376224e-05,
      "loss": 0.1824,
      "step": 7575
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5637107334901897,
      "learning_rate": 2.9101086854789855e-05,
      "loss": 0.0782,
      "step": 7576
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5382376862265588,
      "learning_rate": 2.9096193459740768e-05,
      "loss": 0.115,
      "step": 7577
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.647141215856581,
      "learning_rate": 2.90912999034216e-05,
      "loss": 0.1045,
      "step": 7578
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8619393561723304,
      "learning_rate": 2.9086406186025027e-05,
      "loss": 0.1966,
      "step": 7579
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9764785303925932,
      "learning_rate": 2.9081512307743714e-05,
      "loss": 0.3015,
      "step": 7580
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.321876172612123,
      "learning_rate": 2.9076618268770335e-05,
      "loss": 0.2262,
      "step": 7581
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0785430058612762,
      "learning_rate": 2.907172406929758e-05,
      "loss": 0.3369,
      "step": 7582
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6273242301692443,
      "learning_rate": 2.9066829709518118e-05,
      "loss": 0.1438,
      "step": 7583
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8075893434943786,
      "learning_rate": 2.9061935189624666e-05,
      "loss": 0.1589,
      "step": 7584
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1783587420592048,
      "learning_rate": 2.9057040509809913e-05,
      "loss": 0.2099,
      "step": 7585
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2213668626866752,
      "learning_rate": 2.9052145670266562e-05,
      "loss": 0.2163,
      "step": 7586
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8723524127766239,
      "learning_rate": 2.9047250671187337e-05,
      "loss": 0.2576,
      "step": 7587
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6310545501955866,
      "learning_rate": 2.9042355512764946e-05,
      "loss": 0.1685,
      "step": 7588
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.83545732231185,
      "learning_rate": 2.9037460195192127e-05,
      "loss": 0.2093,
      "step": 7589
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6925672766033278,
      "learning_rate": 2.9032564718661603e-05,
      "loss": 0.1548,
      "step": 7590
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3658767580044529,
      "learning_rate": 2.9027669083366123e-05,
      "loss": 0.3641,
      "step": 7591
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8218320373554346,
      "learning_rate": 2.9022773289498418e-05,
      "loss": 0.153,
      "step": 7592
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1417204440168318,
      "learning_rate": 2.901787733725125e-05,
      "loss": 0.3368,
      "step": 7593
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5296251378344676,
      "learning_rate": 2.9012981226817364e-05,
      "loss": 0.0648,
      "step": 7594
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5742290713488838,
      "learning_rate": 2.9008084958389537e-05,
      "loss": 0.1061,
      "step": 7595
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6964109718815353,
      "learning_rate": 2.900318853216053e-05,
      "loss": 0.1308,
      "step": 7596
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7252365902081923,
      "learning_rate": 2.8998291948323124e-05,
      "loss": 0.1286,
      "step": 7597
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5846006878498694,
      "learning_rate": 2.8993395207070095e-05,
      "loss": 0.1091,
      "step": 7598
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4666722412787967,
      "learning_rate": 2.898849830859424e-05,
      "loss": 0.0872,
      "step": 7599
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.508281164130938,
      "learning_rate": 2.8983601253088343e-05,
      "loss": 0.1052,
      "step": 7600
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9691762523561444,
      "learning_rate": 2.897870404074522e-05,
      "loss": 0.1974,
      "step": 7601
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5443948198496406,
      "learning_rate": 2.897380667175766e-05,
      "loss": 0.0885,
      "step": 7602
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5840733374019631,
      "learning_rate": 2.8968909146318478e-05,
      "loss": 0.0954,
      "step": 7603
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7187901958137912,
      "learning_rate": 2.8964011464620505e-05,
      "loss": 0.1563,
      "step": 7604
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7085292641496014,
      "learning_rate": 2.895911362685656e-05,
      "loss": 0.1038,
      "step": 7605
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.49899504501245034,
      "learning_rate": 2.8954215633219472e-05,
      "loss": 0.0732,
      "step": 7606
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7809339663949137,
      "learning_rate": 2.8949317483902083e-05,
      "loss": 0.5932,
      "step": 7607
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6746217867494517,
      "learning_rate": 2.894441917909723e-05,
      "loss": 0.1086,
      "step": 7608
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6152385564858781,
      "learning_rate": 2.893952071899777e-05,
      "loss": 0.1136,
      "step": 7609
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8400666414750755,
      "learning_rate": 2.8934622103796555e-05,
      "loss": 0.1213,
      "step": 7610
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.429517755732576,
      "learning_rate": 2.8929723333686452e-05,
      "loss": 0.0673,
      "step": 7611
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6870849352373286,
      "learning_rate": 2.8924824408860317e-05,
      "loss": 0.1613,
      "step": 7612
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9450848722515237,
      "learning_rate": 2.8919925329511045e-05,
      "loss": 0.3079,
      "step": 7613
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4064649731288404,
      "learning_rate": 2.8915026095831486e-05,
      "loss": 0.0581,
      "step": 7614
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6833926317588613,
      "learning_rate": 2.8910126708014557e-05,
      "loss": 0.131,
      "step": 7615
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6401373410052676,
      "learning_rate": 2.8905227166253136e-05,
      "loss": 0.1141,
      "step": 7616
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7124397745342,
      "learning_rate": 2.8900327470740118e-05,
      "loss": 0.09,
      "step": 7617
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8244243396945599,
      "learning_rate": 2.889542762166842e-05,
      "loss": 0.1462,
      "step": 7618
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5780518864029123,
      "learning_rate": 2.889052761923094e-05,
      "loss": 0.1501,
      "step": 7619
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4874585472042697,
      "learning_rate": 2.8885627463620598e-05,
      "loss": 0.0736,
      "step": 7620
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7025673669232616,
      "learning_rate": 2.888072715503032e-05,
      "loss": 0.1299,
      "step": 7621
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6113574249135205,
      "learning_rate": 2.8875826693653024e-05,
      "loss": 0.0825,
      "step": 7622
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6188456971269973,
      "learning_rate": 2.8870926079681666e-05,
      "loss": 0.0973,
      "step": 7623
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9284758907678079,
      "learning_rate": 2.8866025313309165e-05,
      "loss": 0.1368,
      "step": 7624
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.792635857909605,
      "learning_rate": 2.8861124394728483e-05,
      "loss": 0.1161,
      "step": 7625
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5375122180717514,
      "learning_rate": 2.885622332413256e-05,
      "loss": 0.0771,
      "step": 7626
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5429620364859465,
      "learning_rate": 2.885132210171437e-05,
      "loss": 0.1134,
      "step": 7627
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9028089131960015,
      "learning_rate": 2.8846420727666855e-05,
      "loss": 0.338,
      "step": 7628
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2801595845993994,
      "learning_rate": 2.8841519202183014e-05,
      "loss": 0.2137,
      "step": 7629
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.589419516295134,
      "learning_rate": 2.88366175254558e-05,
      "loss": 0.079,
      "step": 7630
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8669623781881086,
      "learning_rate": 2.8831715697678208e-05,
      "loss": 0.1926,
      "step": 7631
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8642828882912231,
      "learning_rate": 2.8826813719043222e-05,
      "loss": 0.1695,
      "step": 7632
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7855214843138693,
      "learning_rate": 2.882191158974384e-05,
      "loss": 0.1873,
      "step": 7633
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.037041729488922,
      "learning_rate": 2.8817009309973052e-05,
      "loss": 0.2532,
      "step": 7634
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8600207144892328,
      "learning_rate": 2.8812106879923878e-05,
      "loss": 0.2521,
      "step": 7635
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.43928345055509777,
      "learning_rate": 2.880720429978932e-05,
      "loss": 0.0689,
      "step": 7636
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6657711264999902,
      "learning_rate": 2.8802301569762407e-05,
      "loss": 0.0612,
      "step": 7637
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7411480266743212,
      "learning_rate": 2.8797398690036147e-05,
      "loss": 0.0768,
      "step": 7638
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6040375361667949,
      "learning_rate": 2.879249566080359e-05,
      "loss": 0.1206,
      "step": 7639
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9399775580470979,
      "learning_rate": 2.878759248225775e-05,
      "loss": 0.3756,
      "step": 7640
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2271261488337843,
      "learning_rate": 2.878268915459169e-05,
      "loss": 0.202,
      "step": 7641
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3290201638218597,
      "learning_rate": 2.8777785677998435e-05,
      "loss": 0.391,
      "step": 7642
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8157728077747562,
      "learning_rate": 2.8772882052671064e-05,
      "loss": 0.2727,
      "step": 7643
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8157322080278188,
      "learning_rate": 2.8767978278802615e-05,
      "loss": 0.1866,
      "step": 7644
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.846823682519772,
      "learning_rate": 2.876307435658616e-05,
      "loss": 0.147,
      "step": 7645
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.723558989109233,
      "learning_rate": 2.875817028621477e-05,
      "loss": 0.1218,
      "step": 7646
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5649004617678101,
      "learning_rate": 2.875326606788153e-05,
      "loss": 0.0729,
      "step": 7647
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1186660375708646,
      "learning_rate": 2.874836170177951e-05,
      "loss": 0.3019,
      "step": 7648
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6111852876434014,
      "learning_rate": 2.8743457188101802e-05,
      "loss": 0.1453,
      "step": 7649
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7994986768487595,
      "learning_rate": 2.8738552527041506e-05,
      "loss": 0.2762,
      "step": 7650
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6641740247260147,
      "learning_rate": 2.8733647718791713e-05,
      "loss": 0.1311,
      "step": 7651
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4097173407036271,
      "learning_rate": 2.8728742763545534e-05,
      "loss": 0.0655,
      "step": 7652
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6869757587078519,
      "learning_rate": 2.8723837661496085e-05,
      "loss": 0.185,
      "step": 7653
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4891146314541137,
      "learning_rate": 2.8718932412836474e-05,
      "loss": 0.056,
      "step": 7654
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5116278028802096,
      "learning_rate": 2.871402701775983e-05,
      "loss": 0.0412,
      "step": 7655
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0279330058774028,
      "learning_rate": 2.870912147645928e-05,
      "loss": 0.142,
      "step": 7656
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6506880801657215,
      "learning_rate": 2.870421578912796e-05,
      "loss": 0.1056,
      "step": 7657
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.612409851386112,
      "learning_rate": 2.8699309955959002e-05,
      "loss": 0.1382,
      "step": 7658
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0791279514659449,
      "learning_rate": 2.8694403977145572e-05,
      "loss": 0.269,
      "step": 7659
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6273050619236485,
      "learning_rate": 2.8689497852880804e-05,
      "loss": 0.1146,
      "step": 7660
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.776455025828485,
      "learning_rate": 2.868459158335786e-05,
      "loss": 0.1583,
      "step": 7661
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.019594013190885,
      "learning_rate": 2.8679685168769903e-05,
      "loss": 0.1886,
      "step": 7662
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.106998300492409,
      "learning_rate": 2.867477860931011e-05,
      "loss": 0.3843,
      "step": 7663
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5168330431141198,
      "learning_rate": 2.8669871905171648e-05,
      "loss": 0.0899,
      "step": 7664
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5536592556335885,
      "learning_rate": 2.8664965056547698e-05,
      "loss": 0.1732,
      "step": 7665
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3844588431287045,
      "learning_rate": 2.8660058063631445e-05,
      "loss": 0.0483,
      "step": 7666
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4863586421819782,
      "learning_rate": 2.8655150926616087e-05,
      "loss": 0.0712,
      "step": 7667
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4450797510504696,
      "learning_rate": 2.8650243645694815e-05,
      "loss": 0.5353,
      "step": 7668
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6212155885756371,
      "learning_rate": 2.8645336221060837e-05,
      "loss": 0.111,
      "step": 7669
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7931892924680467,
      "learning_rate": 2.8640428652907352e-05,
      "loss": 0.1138,
      "step": 7670
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6852965288089075,
      "learning_rate": 2.8635520941427592e-05,
      "loss": 0.1052,
      "step": 7671
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4590683995320093,
      "learning_rate": 2.8630613086814757e-05,
      "loss": 0.0749,
      "step": 7672
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2275788150843627,
      "learning_rate": 2.862570508926209e-05,
      "loss": 0.2193,
      "step": 7673
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9793810603539419,
      "learning_rate": 2.8620796948962813e-05,
      "loss": 0.2895,
      "step": 7674
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6490897475915388,
      "learning_rate": 2.8615888666110164e-05,
      "loss": 0.1174,
      "step": 7675
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2539909019050197,
      "learning_rate": 2.861098024089739e-05,
      "loss": 0.2291,
      "step": 7676
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3897705316401496,
      "learning_rate": 2.860607167351773e-05,
      "loss": 0.1569,
      "step": 7677
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0810623128412666,
      "learning_rate": 2.860116296416445e-05,
      "loss": 0.3371,
      "step": 7678
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6296395687916942,
      "learning_rate": 2.8596254113030807e-05,
      "loss": 0.1992,
      "step": 7679
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6313087716659587,
      "learning_rate": 2.8591345120310054e-05,
      "loss": 0.0999,
      "step": 7680
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.48602383351983736,
      "learning_rate": 2.858643598619547e-05,
      "loss": 0.0679,
      "step": 7681
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.49729445252511817,
      "learning_rate": 2.8581526710880334e-05,
      "loss": 0.1024,
      "step": 7682
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1705211234653272,
      "learning_rate": 2.8576617294557923e-05,
      "loss": 0.272,
      "step": 7683
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6660740387556068,
      "learning_rate": 2.8571707737421533e-05,
      "loss": 0.0784,
      "step": 7684
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5990787184920042,
      "learning_rate": 2.856679803966444e-05,
      "loss": 0.0899,
      "step": 7685
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7759787028759082,
      "learning_rate": 2.856188820147996e-05,
      "loss": 0.1669,
      "step": 7686
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8334789024870294,
      "learning_rate": 2.8556978223061378e-05,
      "loss": 0.183,
      "step": 7687
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5178559301816426,
      "learning_rate": 2.8552068104602026e-05,
      "loss": 0.1044,
      "step": 7688
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6969980087396161,
      "learning_rate": 2.8547157846295198e-05,
      "loss": 0.1006,
      "step": 7689
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9048260106083852,
      "learning_rate": 2.8542247448334225e-05,
      "loss": 0.2452,
      "step": 7690
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1237712119340197,
      "learning_rate": 2.8537336910912432e-05,
      "loss": 0.1931,
      "step": 7691
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7638241583437098,
      "learning_rate": 2.8532426234223152e-05,
      "loss": 0.136,
      "step": 7692
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.058509002562369,
      "learning_rate": 2.8527515418459716e-05,
      "loss": 0.2444,
      "step": 7693
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.919386343699045,
      "learning_rate": 2.8522604463815477e-05,
      "loss": 0.1971,
      "step": 7694
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6973142273499892,
      "learning_rate": 2.8517693370483765e-05,
      "loss": 0.1272,
      "step": 7695
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3528086064488793,
      "learning_rate": 2.8512782138657955e-05,
      "loss": 0.0534,
      "step": 7696
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8730834338349193,
      "learning_rate": 2.8507870768531387e-05,
      "loss": 0.161,
      "step": 7697
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5903235576937924,
      "learning_rate": 2.8502959260297435e-05,
      "loss": 0.0888,
      "step": 7698
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6900463541262416,
      "learning_rate": 2.8498047614149464e-05,
      "loss": 0.0919,
      "step": 7699
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3268375899711133,
      "learning_rate": 2.849313583028086e-05,
      "loss": 0.2799,
      "step": 7700
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6932118745234709,
      "learning_rate": 2.8488223908884986e-05,
      "loss": 0.1362,
      "step": 7701
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6209204625371672,
      "learning_rate": 2.8483311850155242e-05,
      "loss": 0.141,
      "step": 7702
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8492840172663265,
      "learning_rate": 2.847839965428501e-05,
      "loss": 0.2229,
      "step": 7703
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6603953338729348,
      "learning_rate": 2.8473487321467695e-05,
      "loss": 0.1779,
      "step": 7704
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7417361073314649,
      "learning_rate": 2.846857485189669e-05,
      "loss": 0.0867,
      "step": 7705
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7183390953448952,
      "learning_rate": 2.846366224576541e-05,
      "loss": 0.2679,
      "step": 7706
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5949163806942378,
      "learning_rate": 2.845874950326726e-05,
      "loss": 0.0908,
      "step": 7707
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5135714094599791,
      "learning_rate": 2.8453836624595675e-05,
      "loss": 0.098,
      "step": 7708
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9344564515112666,
      "learning_rate": 2.8448923609944056e-05,
      "loss": 0.2365,
      "step": 7709
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6258447026819399,
      "learning_rate": 2.8444010459505848e-05,
      "loss": 0.151,
      "step": 7710
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6245520423087859,
      "learning_rate": 2.8439097173474477e-05,
      "loss": 0.0654,
      "step": 7711
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9023027923273507,
      "learning_rate": 2.8434183752043393e-05,
      "loss": 0.2071,
      "step": 7712
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5164620393963132,
      "learning_rate": 2.8429270195406026e-05,
      "loss": 0.1238,
      "step": 7713
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5580070513492434,
      "learning_rate": 2.8424356503755835e-05,
      "loss": 0.0842,
      "step": 7714
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9116159046216912,
      "learning_rate": 2.841944267728628e-05,
      "loss": 0.1874,
      "step": 7715
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6299420438676919,
      "learning_rate": 2.8414528716190803e-05,
      "loss": 0.1235,
      "step": 7716
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7192288291083578,
      "learning_rate": 2.8409614620662896e-05,
      "loss": 0.1266,
      "step": 7717
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.015929512221488,
      "learning_rate": 2.8404700390896018e-05,
      "loss": 0.3766,
      "step": 7718
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1127319597744465,
      "learning_rate": 2.8399786027083635e-05,
      "loss": 0.29,
      "step": 7719
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7088128245309236,
      "learning_rate": 2.8394871529419247e-05,
      "loss": 0.1149,
      "step": 7720
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4915552507878482,
      "learning_rate": 2.8389956898096332e-05,
      "loss": 0.094,
      "step": 7721
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4949837193885824,
      "learning_rate": 2.8385042133308388e-05,
      "loss": 0.0977,
      "step": 7722
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7773173653361006,
      "learning_rate": 2.83801272352489e-05,
      "loss": 0.1802,
      "step": 7723
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5883730315554142,
      "learning_rate": 2.837521220411139e-05,
      "loss": 0.0997,
      "step": 7724
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7250680224195332,
      "learning_rate": 2.837029704008935e-05,
      "loss": 0.1584,
      "step": 7725
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.96373277877245,
      "learning_rate": 2.8365381743376307e-05,
      "loss": 0.2656,
      "step": 7726
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9999295673681041,
      "learning_rate": 2.836046631416576e-05,
      "loss": 0.253,
      "step": 7727
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6362782218253435,
      "learning_rate": 2.835555075265126e-05,
      "loss": 0.1391,
      "step": 7728
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5374122538775464,
      "learning_rate": 2.8350635059026316e-05,
      "loss": 0.0612,
      "step": 7729
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7682373915982664,
      "learning_rate": 2.834571923348447e-05,
      "loss": 0.1633,
      "step": 7730
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5386706518867647,
      "learning_rate": 2.8340803276219256e-05,
      "loss": 0.0975,
      "step": 7731
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5722528580970773,
      "learning_rate": 2.8335887187424226e-05,
      "loss": 0.1181,
      "step": 7732
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6210005580697181,
      "learning_rate": 2.8330970967292925e-05,
      "loss": 0.1445,
      "step": 7733
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4542273210573228,
      "learning_rate": 2.8326054616018915e-05,
      "loss": 0.1036,
      "step": 7734
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5089669496552389,
      "learning_rate": 2.8321138133795745e-05,
      "loss": 0.114,
      "step": 7735
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3653653963375678,
      "learning_rate": 2.8316221520816987e-05,
      "loss": 0.6137,
      "step": 7736
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6866870529271487,
      "learning_rate": 2.8311304777276215e-05,
      "loss": 0.1443,
      "step": 7737
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5075340085301973,
      "learning_rate": 2.8306387903367e-05,
      "loss": 0.0897,
      "step": 7738
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6102178559838968,
      "learning_rate": 2.8301470899282918e-05,
      "loss": 0.0775,
      "step": 7739
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7053935788574318,
      "learning_rate": 2.829655376521757e-05,
      "loss": 0.1016,
      "step": 7740
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.49114334407106464,
      "learning_rate": 2.8291636501364528e-05,
      "loss": 0.1116,
      "step": 7741
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7921098773898496,
      "learning_rate": 2.82867191079174e-05,
      "loss": 0.1459,
      "step": 7742
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4645199622912022,
      "learning_rate": 2.828180158506979e-05,
      "loss": 0.0525,
      "step": 7743
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5108506276460106,
      "learning_rate": 2.8276883933015296e-05,
      "loss": 0.0917,
      "step": 7744
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5978794298760957,
      "learning_rate": 2.8271966151947536e-05,
      "loss": 0.126,
      "step": 7745
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6841434176251947,
      "learning_rate": 2.8267048242060122e-05,
      "loss": 0.198,
      "step": 7746
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7422895114529451,
      "learning_rate": 2.8262130203546683e-05,
      "loss": 0.1857,
      "step": 7747
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7091155982812795,
      "learning_rate": 2.8257212036600826e-05,
      "loss": 0.1391,
      "step": 7748
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3455304298425275,
      "learning_rate": 2.8252293741416214e-05,
      "loss": 0.2888,
      "step": 7749
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6721929449268697,
      "learning_rate": 2.8247375318186454e-05,
      "loss": 0.1499,
      "step": 7750
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9637979381443996,
      "learning_rate": 2.8242456767105206e-05,
      "loss": 0.3361,
      "step": 7751
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6343712763385239,
      "learning_rate": 2.8237538088366112e-05,
      "loss": 0.1701,
      "step": 7752
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8014910194962783,
      "learning_rate": 2.8232619282162816e-05,
      "loss": 0.0964,
      "step": 7753
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7077555835949477,
      "learning_rate": 2.822770034868899e-05,
      "loss": 0.1697,
      "step": 7754
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5770956481585887,
      "learning_rate": 2.8222781288138282e-05,
      "loss": 0.1376,
      "step": 7755
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7863752835771265,
      "learning_rate": 2.8217862100704373e-05,
      "loss": 0.2125,
      "step": 7756
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6305388346062651,
      "learning_rate": 2.8212942786580914e-05,
      "loss": 0.0776,
      "step": 7757
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9539950234325875,
      "learning_rate": 2.820802334596161e-05,
      "loss": 0.2485,
      "step": 7758
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7338592055406437,
      "learning_rate": 2.8203103779040114e-05,
      "loss": 0.1034,
      "step": 7759
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.46234325390227066,
      "learning_rate": 2.8198184086010138e-05,
      "loss": 0.09,
      "step": 7760
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5650414641291174,
      "learning_rate": 2.819326426706535e-05,
      "loss": 0.1955,
      "step": 7761
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6555260409853512,
      "learning_rate": 2.8188344322399472e-05,
      "loss": 0.1215,
      "step": 7762
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7804949235965476,
      "learning_rate": 2.8183424252206185e-05,
      "loss": 0.1559,
      "step": 7763
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9550073350638523,
      "learning_rate": 2.8178504056679212e-05,
      "loss": 0.1854,
      "step": 7764
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7056011063851622,
      "learning_rate": 2.8173583736012245e-05,
      "loss": 0.1034,
      "step": 7765
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6250038502460197,
      "learning_rate": 2.816866329039902e-05,
      "loss": 0.088,
      "step": 7766
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5316975557996702,
      "learning_rate": 2.8163742720033243e-05,
      "loss": 0.1054,
      "step": 7767
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1875130562983152,
      "learning_rate": 2.8158822025108654e-05,
      "loss": 0.288,
      "step": 7768
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7950005821266388,
      "learning_rate": 2.815390120581898e-05,
      "loss": 0.1027,
      "step": 7769
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6514045262172374,
      "learning_rate": 2.814898026235795e-05,
      "loss": 0.0934,
      "step": 7770
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0840471249816828,
      "learning_rate": 2.8144059194919308e-05,
      "loss": 0.4745,
      "step": 7771
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6076220390829669,
      "learning_rate": 2.8139138003696815e-05,
      "loss": 0.1223,
      "step": 7772
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.500963045421803,
      "learning_rate": 2.8134216688884198e-05,
      "loss": 0.077,
      "step": 7773
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.797236374659382,
      "learning_rate": 2.8129295250675225e-05,
      "loss": 0.2335,
      "step": 7774
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5245711782975184,
      "learning_rate": 2.8124373689263656e-05,
      "loss": 0.0518,
      "step": 7775
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5932385752879974,
      "learning_rate": 2.8119452004843256e-05,
      "loss": 0.1129,
      "step": 7776
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0992825098074241,
      "learning_rate": 2.811453019760779e-05,
      "loss": 0.3197,
      "step": 7777
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.42750630426949976,
      "learning_rate": 2.8109608267751046e-05,
      "loss": 0.0633,
      "step": 7778
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1827045151897386,
      "learning_rate": 2.810468621546679e-05,
      "loss": 0.4682,
      "step": 7779
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0768749969357747,
      "learning_rate": 2.809976404094882e-05,
      "loss": 0.2522,
      "step": 7780
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5197233480471702,
      "learning_rate": 2.809484174439092e-05,
      "loss": 0.1344,
      "step": 7781
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8991233318042453,
      "learning_rate": 2.8089919325986873e-05,
      "loss": 0.3671,
      "step": 7782
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6052704500975125,
      "learning_rate": 2.8084996785930496e-05,
      "loss": 0.0902,
      "step": 7783
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9568751819997192,
      "learning_rate": 2.808007412441558e-05,
      "loss": 0.2402,
      "step": 7784
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7597583678814606,
      "learning_rate": 2.8075151341635934e-05,
      "loss": 0.1485,
      "step": 7785
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5642849136468637,
      "learning_rate": 2.8070228437785385e-05,
      "loss": 0.1004,
      "step": 7786
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7215949916279703,
      "learning_rate": 2.806530541305773e-05,
      "loss": 0.1544,
      "step": 7787
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5658653975063107,
      "learning_rate": 2.8060382267646816e-05,
      "loss": 0.1342,
      "step": 7788
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4546567948073974,
      "learning_rate": 2.8055459001746454e-05,
      "loss": 0.0823,
      "step": 7789
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2374945406041171,
      "learning_rate": 2.8050535615550482e-05,
      "loss": 0.2954,
      "step": 7790
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7059291704506054,
      "learning_rate": 2.8045612109252733e-05,
      "loss": 0.1363,
      "step": 7791
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5913084364133191,
      "learning_rate": 2.8040688483047055e-05,
      "loss": 0.1271,
      "step": 7792
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.71771975956505,
      "learning_rate": 2.803576473712729e-05,
      "loss": 0.1376,
      "step": 7793
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.46015310256313513,
      "learning_rate": 2.8030840871687297e-05,
      "loss": 0.0829,
      "step": 7794
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9929844887438456,
      "learning_rate": 2.802591688692092e-05,
      "loss": 0.3828,
      "step": 7795
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7466150591129567,
      "learning_rate": 2.8020992783022037e-05,
      "loss": 0.0945,
      "step": 7796
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.41828175834595316,
      "learning_rate": 2.8016068560184494e-05,
      "loss": 0.0602,
      "step": 7797
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.599258189427001,
      "learning_rate": 2.801114421860217e-05,
      "loss": 0.1041,
      "step": 7798
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7048712910965431,
      "learning_rate": 2.8006219758468942e-05,
      "loss": 0.1434,
      "step": 7799
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7408200521264168,
      "learning_rate": 2.8001295179978694e-05,
      "loss": 0.1544,
      "step": 7800
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7051707308799389,
      "learning_rate": 2.7996370483325297e-05,
      "loss": 0.2309,
      "step": 7801
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3348289247150464,
      "learning_rate": 2.799144566870265e-05,
      "loss": 0.5651,
      "step": 7802
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4671270681140126,
      "learning_rate": 2.7986520736304635e-05,
      "loss": 0.1078,
      "step": 7803
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4266667539213034,
      "learning_rate": 2.798159568632517e-05,
      "loss": 0.0739,
      "step": 7804
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6316141439371961,
      "learning_rate": 2.7976670518958138e-05,
      "loss": 0.0894,
      "step": 7805
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9812031807370909,
      "learning_rate": 2.7971745234397457e-05,
      "loss": 0.3252,
      "step": 7806
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0576429013719169,
      "learning_rate": 2.796681983283703e-05,
      "loss": 0.3495,
      "step": 7807
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5370350374826864,
      "learning_rate": 2.796189431447078e-05,
      "loss": 0.0965,
      "step": 7808
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6462814508901853,
      "learning_rate": 2.795696867949264e-05,
      "loss": 0.1305,
      "step": 7809
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6202364994714336,
      "learning_rate": 2.7952042928096515e-05,
      "loss": 0.1127,
      "step": 7810
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4170960503560765,
      "learning_rate": 2.7947117060476352e-05,
      "loss": 0.0793,
      "step": 7811
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7454178416857656,
      "learning_rate": 2.7942191076826075e-05,
      "loss": 0.194,
      "step": 7812
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0051498012814482,
      "learning_rate": 2.793726497733962e-05,
      "loss": 0.3798,
      "step": 7813
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7213374986043093,
      "learning_rate": 2.7932338762210947e-05,
      "loss": 0.2003,
      "step": 7814
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2369375310123611,
      "learning_rate": 2.7927412431634e-05,
      "loss": 0.4151,
      "step": 7815
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6348325031609909,
      "learning_rate": 2.792248598580271e-05,
      "loss": 0.1293,
      "step": 7816
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0810269906405758,
      "learning_rate": 2.7917559424911066e-05,
      "loss": 0.4065,
      "step": 7817
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4616158037782111,
      "learning_rate": 2.7912632749153007e-05,
      "loss": 0.0885,
      "step": 7818
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5155709185193588,
      "learning_rate": 2.790770595872252e-05,
      "loss": 0.0955,
      "step": 7819
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6245952013034327,
      "learning_rate": 2.7902779053813554e-05,
      "loss": 0.1201,
      "step": 7820
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0575520705876054,
      "learning_rate": 2.7897852034620105e-05,
      "loss": 0.169,
      "step": 7821
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6178817201604758,
      "learning_rate": 2.7892924901336136e-05,
      "loss": 0.0649,
      "step": 7822
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6931944017364088,
      "learning_rate": 2.788799765415565e-05,
      "loss": 0.1052,
      "step": 7823
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5026040082846313,
      "learning_rate": 2.7883070293272624e-05,
      "loss": 0.0683,
      "step": 7824
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5007126461646135,
      "learning_rate": 2.7878142818881053e-05,
      "loss": 0.0844,
      "step": 7825
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1431623546094984,
      "learning_rate": 2.7873215231174933e-05,
      "loss": 0.3877,
      "step": 7826
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5055228705875293,
      "learning_rate": 2.7868287530348274e-05,
      "loss": 0.1658,
      "step": 7827
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9641636870259291,
      "learning_rate": 2.7863359716595077e-05,
      "loss": 0.1457,
      "step": 7828
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6583039664026412,
      "learning_rate": 2.7858431790109364e-05,
      "loss": 0.1432,
      "step": 7829
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6882798471688015,
      "learning_rate": 2.785350375108513e-05,
      "loss": 0.113,
      "step": 7830
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.924794771615839,
      "learning_rate": 2.7848575599716414e-05,
      "loss": 0.239,
      "step": 7831
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5971773628968028,
      "learning_rate": 2.7843647336197237e-05,
      "loss": 0.182,
      "step": 7832
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5801544847771475,
      "learning_rate": 2.7838718960721626e-05,
      "loss": 0.0805,
      "step": 7833
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8565722863081638,
      "learning_rate": 2.783379047348361e-05,
      "loss": 0.2029,
      "step": 7834
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6969083411957583,
      "learning_rate": 2.7828861874677244e-05,
      "loss": 0.1175,
      "step": 7835
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.44531900584713013,
      "learning_rate": 2.7823933164496546e-05,
      "loss": 0.0769,
      "step": 7836
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5988518841859948,
      "learning_rate": 2.7819004343135578e-05,
      "loss": 0.1416,
      "step": 7837
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.35922364066207885,
      "learning_rate": 2.7814075410788394e-05,
      "loss": 0.0508,
      "step": 7838
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2653565857847342,
      "learning_rate": 2.780914636764904e-05,
      "loss": 0.4913,
      "step": 7839
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.941147106084415,
      "learning_rate": 2.780421721391158e-05,
      "loss": 0.3104,
      "step": 7840
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6198137771232926,
      "learning_rate": 2.7799287949770086e-05,
      "loss": 0.1302,
      "step": 7841
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7770779838324383,
      "learning_rate": 2.779435857541861e-05,
      "loss": 0.1954,
      "step": 7842
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5689202271077057,
      "learning_rate": 2.778942909105124e-05,
      "loss": 0.1234,
      "step": 7843
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4818690966868108,
      "learning_rate": 2.7784499496862044e-05,
      "loss": 0.0765,
      "step": 7844
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7639091806590116,
      "learning_rate": 2.7779569793045113e-05,
      "loss": 0.1281,
      "step": 7845
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5983847786639943,
      "learning_rate": 2.777463997979452e-05,
      "loss": 0.1126,
      "step": 7846
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5557503382097753,
      "learning_rate": 2.7769710057304364e-05,
      "loss": 0.0803,
      "step": 7847
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7954987297132066,
      "learning_rate": 2.7764780025768745e-05,
      "loss": 0.3122,
      "step": 7848
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2609670944896945,
      "learning_rate": 2.775984988538175e-05,
      "loss": 0.381,
      "step": 7849
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8038091859636513,
      "learning_rate": 2.7754919636337486e-05,
      "loss": 0.2087,
      "step": 7850
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6337949261239263,
      "learning_rate": 2.7749989278830063e-05,
      "loss": 0.0888,
      "step": 7851
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.57621420034377,
      "learning_rate": 2.774505881305359e-05,
      "loss": 0.4375,
      "step": 7852
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7394249003446126,
      "learning_rate": 2.7740128239202184e-05,
      "loss": 0.0889,
      "step": 7853
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.216329962445807,
      "learning_rate": 2.773519755746996e-05,
      "loss": 0.2302,
      "step": 7854
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5959515966037842,
      "learning_rate": 2.773026676805106e-05,
      "loss": 0.0748,
      "step": 7855
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3795338636887762,
      "learning_rate": 2.772533587113959e-05,
      "loss": 0.0546,
      "step": 7856
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4613536400186554,
      "learning_rate": 2.77204048669297e-05,
      "loss": 0.0535,
      "step": 7857
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8154346139522298,
      "learning_rate": 2.771547375561552e-05,
      "loss": 0.1676,
      "step": 7858
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6274213362462077,
      "learning_rate": 2.7710542537391193e-05,
      "loss": 0.1123,
      "step": 7859
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6524282858957058,
      "learning_rate": 2.7705611212450854e-05,
      "loss": 0.1369,
      "step": 7860
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0357334125682218,
      "learning_rate": 2.7700679780988677e-05,
      "loss": 0.172,
      "step": 7861
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4493022506640734,
      "learning_rate": 2.7695748243198786e-05,
      "loss": 0.055,
      "step": 7862
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4149667264490057,
      "learning_rate": 2.769081659927537e-05,
      "loss": 0.0717,
      "step": 7863
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7533359407708243,
      "learning_rate": 2.7685884849412563e-05,
      "loss": 0.111,
      "step": 7864
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5134756538263665,
      "learning_rate": 2.7680952993804553e-05,
      "loss": 0.1096,
      "step": 7865
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4413548769610203,
      "learning_rate": 2.7676021032645498e-05,
      "loss": 0.0702,
      "step": 7866
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6311387502264117,
      "learning_rate": 2.7671088966129582e-05,
      "loss": 0.0803,
      "step": 7867
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.452302512619557,
      "learning_rate": 2.7666156794450977e-05,
      "loss": 0.0602,
      "step": 7868
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8512591993339915,
      "learning_rate": 2.766122451780387e-05,
      "loss": 0.2154,
      "step": 7869
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2308549934778084,
      "learning_rate": 2.7656292136382438e-05,
      "loss": 0.3892,
      "step": 7870
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6159414906890479,
      "learning_rate": 2.765135965038089e-05,
      "loss": 0.13,
      "step": 7871
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.45555496534998036,
      "learning_rate": 2.7646427059993406e-05,
      "loss": 0.0571,
      "step": 7872
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.761447103343016,
      "learning_rate": 2.76414943654142e-05,
      "loss": 0.5182,
      "step": 7873
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5463218655325214,
      "learning_rate": 2.763656156683746e-05,
      "loss": 0.1647,
      "step": 7874
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6636609674211158,
      "learning_rate": 2.763162866445741e-05,
      "loss": 0.1768,
      "step": 7875
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6587666243497352,
      "learning_rate": 2.7626695658468245e-05,
      "loss": 0.1715,
      "step": 7876
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.668291048909082,
      "learning_rate": 2.7621762549064195e-05,
      "loss": 1.0855,
      "step": 7877
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5660474914967595,
      "learning_rate": 2.761682933643947e-05,
      "loss": 0.6077,
      "step": 7878
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5341754488754485,
      "learning_rate": 2.7611896020788304e-05,
      "loss": 0.1195,
      "step": 7879
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.036996518352076,
      "learning_rate": 2.760696260230492e-05,
      "loss": 0.3056,
      "step": 7880
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.821619975810957,
      "learning_rate": 2.7602029081183538e-05,
      "loss": 0.1741,
      "step": 7881
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7087858968087895,
      "learning_rate": 2.759709545761842e-05,
      "loss": 0.2225,
      "step": 7882
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5778129839949723,
      "learning_rate": 2.759216173180379e-05,
      "loss": 0.1372,
      "step": 7883
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6572313481825489,
      "learning_rate": 2.7587227903933886e-05,
      "loss": 0.15,
      "step": 7884
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5116898365777344,
      "learning_rate": 2.7582293974202977e-05,
      "loss": 0.0611,
      "step": 7885
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8566978196805087,
      "learning_rate": 2.7577359942805298e-05,
      "loss": 0.1829,
      "step": 7886
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7884313608409556,
      "learning_rate": 2.757242580993511e-05,
      "loss": 0.1675,
      "step": 7887
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6526996663425539,
      "learning_rate": 2.7567491575786674e-05,
      "loss": 0.14,
      "step": 7888
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5749053822636427,
      "learning_rate": 2.756255724055426e-05,
      "loss": 0.0829,
      "step": 7889
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0270401152936477,
      "learning_rate": 2.7557622804432122e-05,
      "loss": 0.1827,
      "step": 7890
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.211757825091442,
      "learning_rate": 2.7552688267614552e-05,
      "loss": 0.7145,
      "step": 7891
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.48566285449309343,
      "learning_rate": 2.7547753630295802e-05,
      "loss": 0.081,
      "step": 7892
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8014642013579392,
      "learning_rate": 2.7542818892670174e-05,
      "loss": 0.2843,
      "step": 7893
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6339411768499044,
      "learning_rate": 2.7537884054931946e-05,
      "loss": 0.1779,
      "step": 7894
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3913889183117325,
      "learning_rate": 2.7532949117275404e-05,
      "loss": 0.0478,
      "step": 7895
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.508483458582816,
      "learning_rate": 2.752801407989483e-05,
      "loss": 0.2579,
      "step": 7896
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0182742337680948,
      "learning_rate": 2.7523078942984542e-05,
      "loss": 0.1338,
      "step": 7897
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2170409626974152,
      "learning_rate": 2.751814370673882e-05,
      "loss": 0.2354,
      "step": 7898
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.392821193053337,
      "learning_rate": 2.7513208371351978e-05,
      "loss": 0.467,
      "step": 7899
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6425513550964765,
      "learning_rate": 2.750827293701832e-05,
      "loss": 0.1254,
      "step": 7900
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7723528008435402,
      "learning_rate": 2.750333740393216e-05,
      "loss": 0.2385,
      "step": 7901
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5074599295722471,
      "learning_rate": 2.7498401772287813e-05,
      "loss": 0.0925,
      "step": 7902
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8335280073700081,
      "learning_rate": 2.7493466042279596e-05,
      "loss": 0.2541,
      "step": 7903
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5258585888121077,
      "learning_rate": 2.7488530214101836e-05,
      "loss": 0.076,
      "step": 7904
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4661682324387033,
      "learning_rate": 2.748359428794886e-05,
      "loss": 0.1109,
      "step": 7905
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7388180999483486,
      "learning_rate": 2.747865826401499e-05,
      "loss": 0.1308,
      "step": 7906
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.66911828581831,
      "learning_rate": 2.7473722142494575e-05,
      "loss": 0.2959,
      "step": 7907
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6041895116709372,
      "learning_rate": 2.746878592358194e-05,
      "loss": 0.1155,
      "step": 7908
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8350864357343031,
      "learning_rate": 2.7463849607471437e-05,
      "loss": 0.2813,
      "step": 7909
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6893183121573281,
      "learning_rate": 2.745891319435741e-05,
      "loss": 0.1027,
      "step": 7910
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7742010448568557,
      "learning_rate": 2.745397668443421e-05,
      "loss": 0.129,
      "step": 7911
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5749598013973887,
      "learning_rate": 2.7449040077896186e-05,
      "loss": 0.1092,
      "step": 7912
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6091727648110017,
      "learning_rate": 2.7444103374937703e-05,
      "loss": 0.1497,
      "step": 7913
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6821044182488536,
      "learning_rate": 2.7439166575753116e-05,
      "loss": 0.1305,
      "step": 7914
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7498718122313219,
      "learning_rate": 2.743422968053679e-05,
      "loss": 0.1422,
      "step": 7915
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6785782804128153,
      "learning_rate": 2.7429292689483098e-05,
      "loss": 0.1021,
      "step": 7916
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1067039255815156,
      "learning_rate": 2.7424355602786418e-05,
      "loss": 0.3018,
      "step": 7917
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5900580710861656,
      "learning_rate": 2.741941842064111e-05,
      "loss": 0.1139,
      "step": 7918
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.559493238801669,
      "learning_rate": 2.741448114324157e-05,
      "loss": 0.0811,
      "step": 7919
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4663204471382085,
      "learning_rate": 2.7409543770782174e-05,
      "loss": 0.0581,
      "step": 7920
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.741237107869106,
      "learning_rate": 2.740460630345732e-05,
      "loss": 0.0951,
      "step": 7921
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9818932082796864,
      "learning_rate": 2.7399668741461387e-05,
      "loss": 0.3376,
      "step": 7922
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.47846370485875445,
      "learning_rate": 2.7394731084988777e-05,
      "loss": 0.0753,
      "step": 7923
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.667349440835789,
      "learning_rate": 2.738979333423389e-05,
      "loss": 0.0861,
      "step": 7924
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7130319929880511,
      "learning_rate": 2.7384855489391127e-05,
      "loss": 0.0973,
      "step": 7925
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4567267291212011,
      "learning_rate": 2.7379917550654887e-05,
      "loss": 0.0465,
      "step": 7926
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5370059327557055,
      "learning_rate": 2.7374979518219596e-05,
      "loss": 0.0709,
      "step": 7927
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.712271622487352,
      "learning_rate": 2.7370041392279656e-05,
      "loss": 0.1352,
      "step": 7928
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5040210459969368,
      "learning_rate": 2.736510317302949e-05,
      "loss": 0.0804,
      "step": 7929
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6798200026252217,
      "learning_rate": 2.736016486066351e-05,
      "loss": 0.1259,
      "step": 7930
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.49458580524604123,
      "learning_rate": 2.735522645537616e-05,
      "loss": 0.1405,
      "step": 7931
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4783536628025734,
      "learning_rate": 2.735028795736184e-05,
      "loss": 0.0541,
      "step": 7932
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8734612545750781,
      "learning_rate": 2.7345349366815015e-05,
      "loss": 0.1625,
      "step": 7933
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3799110427036116,
      "learning_rate": 2.7340410683930096e-05,
      "loss": 0.042,
      "step": 7934
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.48511217643855736,
      "learning_rate": 2.7335471908901535e-05,
      "loss": 0.1099,
      "step": 7935
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5426905344039609,
      "learning_rate": 2.7330533041923762e-05,
      "loss": 0.0754,
      "step": 7936
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2067202499292162,
      "learning_rate": 2.732559408319124e-05,
      "loss": 0.3048,
      "step": 7937
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5756903153296741,
      "learning_rate": 2.7320655032898412e-05,
      "loss": 0.0956,
      "step": 7938
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6133895552535067,
      "learning_rate": 2.7315715891239733e-05,
      "loss": 0.1371,
      "step": 7939
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5994666277958152,
      "learning_rate": 2.731077665840966e-05,
      "loss": 0.1006,
      "step": 7940
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5548442609275887,
      "learning_rate": 2.7305837334602646e-05,
      "loss": 0.1227,
      "step": 7941
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7582360434200716,
      "learning_rate": 2.7300897920013174e-05,
      "loss": 0.1354,
      "step": 7942
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8072683063495847,
      "learning_rate": 2.7295958414835703e-05,
      "loss": 0.135,
      "step": 7943
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.42274543812807197,
      "learning_rate": 2.7291018819264696e-05,
      "loss": 0.0715,
      "step": 7944
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0880999485499974,
      "learning_rate": 2.728607913349464e-05,
      "loss": 0.2092,
      "step": 7945
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5624432807684983,
      "learning_rate": 2.7281139357720014e-05,
      "loss": 0.1423,
      "step": 7946
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5715161793510779,
      "learning_rate": 2.727619949213529e-05,
      "loss": 0.11,
      "step": 7947
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8175915319579833,
      "learning_rate": 2.7271259536934967e-05,
      "loss": 0.1075,
      "step": 7948
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0831627184403094,
      "learning_rate": 2.726631949231352e-05,
      "loss": 0.1313,
      "step": 7949
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6589234744708513,
      "learning_rate": 2.7261379358465457e-05,
      "loss": 0.1003,
      "step": 7950
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7468954230425329,
      "learning_rate": 2.725643913558526e-05,
      "loss": 0.1517,
      "step": 7951
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2258553662637721,
      "learning_rate": 2.725149882386745e-05,
      "loss": 0.2389,
      "step": 7952
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6106616754309933,
      "learning_rate": 2.7246558423506502e-05,
      "loss": 0.0802,
      "step": 7953
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5125182745827814,
      "learning_rate": 2.724161793469695e-05,
      "loss": 0.0852,
      "step": 7954
    },
    {
      "epoch": 0.49,
      "grad_norm": 14.584372943845631,
      "learning_rate": 2.723667735763329e-05,
      "loss": 1.4016,
      "step": 7955
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6533377097400108,
      "learning_rate": 2.7231736692510042e-05,
      "loss": 0.5763,
      "step": 7956
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6231083218108517,
      "learning_rate": 2.7226795939521716e-05,
      "loss": 0.1287,
      "step": 7957
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0343777550799855,
      "learning_rate": 2.722185509886284e-05,
      "loss": 0.1301,
      "step": 7958
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0478376773100486,
      "learning_rate": 2.7216914170727936e-05,
      "loss": 0.1276,
      "step": 7959
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9588524330142932,
      "learning_rate": 2.7211973155311537e-05,
      "loss": 0.1614,
      "step": 7960
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.061781350648634,
      "learning_rate": 2.7207032052808164e-05,
      "loss": 0.1848,
      "step": 7961
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9701904718926392,
      "learning_rate": 2.720209086341236e-05,
      "loss": 0.1861,
      "step": 7962
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8133971796895325,
      "learning_rate": 2.719714958731866e-05,
      "loss": 0.397,
      "step": 7963
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.722424119367259,
      "learning_rate": 2.719220822472161e-05,
      "loss": 0.1525,
      "step": 7964
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0065678144882786,
      "learning_rate": 2.7187266775815745e-05,
      "loss": 0.1512,
      "step": 7965
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6975228842049165,
      "learning_rate": 2.7182325240795625e-05,
      "loss": 0.1132,
      "step": 7966
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9371286402534655,
      "learning_rate": 2.717738361985579e-05,
      "loss": 0.7249,
      "step": 7967
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0418124053673221,
      "learning_rate": 2.7172441913190804e-05,
      "loss": 0.0923,
      "step": 7968
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9611999119416187,
      "learning_rate": 2.716750012099522e-05,
      "loss": 0.2168,
      "step": 7969
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7138478396377432,
      "learning_rate": 2.716255824346361e-05,
      "loss": 0.1091,
      "step": 7970
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.608912263330368,
      "learning_rate": 2.7157616280790527e-05,
      "loss": 0.1086,
      "step": 7971
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9507233255557884,
      "learning_rate": 2.7152674233170543e-05,
      "loss": 0.3434,
      "step": 7972
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.754678525187519,
      "learning_rate": 2.7147732100798234e-05,
      "loss": 0.1116,
      "step": 7973
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8525045545890569,
      "learning_rate": 2.714278988386818e-05,
      "loss": 0.126,
      "step": 7974
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.567062994045507,
      "learning_rate": 2.7137847582574944e-05,
      "loss": 0.0765,
      "step": 7975
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9451048062774895,
      "learning_rate": 2.7132905197113118e-05,
      "loss": 0.2174,
      "step": 7976
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6749244920665117,
      "learning_rate": 2.7127962727677292e-05,
      "loss": 0.1223,
      "step": 7977
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5982631200939655,
      "learning_rate": 2.7123020174462037e-05,
      "loss": 0.1032,
      "step": 7978
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0836207694415234,
      "learning_rate": 2.7118077537661967e-05,
      "loss": 0.2166,
      "step": 7979
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9579281300470791,
      "learning_rate": 2.7113134817471663e-05,
      "loss": 0.248,
      "step": 7980
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8255681161632279,
      "learning_rate": 2.710819201408572e-05,
      "loss": 0.1489,
      "step": 7981
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6338103822464553,
      "learning_rate": 2.7103249127698754e-05,
      "loss": 0.0924,
      "step": 7982
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6380145731253479,
      "learning_rate": 2.709830615850536e-05,
      "loss": 0.1351,
      "step": 7983
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0198177793066365,
      "learning_rate": 2.7093363106700153e-05,
      "loss": 0.248,
      "step": 7984
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5171023001971765,
      "learning_rate": 2.708841997247773e-05,
      "loss": 0.1134,
      "step": 7985
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.556628636259755,
      "learning_rate": 2.708347675603273e-05,
      "loss": 0.145,
      "step": 7986
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.509782275561507,
      "learning_rate": 2.707853345755974e-05,
      "loss": 0.0539,
      "step": 7987
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5837655824154012,
      "learning_rate": 2.7073590077253413e-05,
      "loss": 0.1059,
      "step": 7988
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7074144278362264,
      "learning_rate": 2.706864661530835e-05,
      "loss": 0.4048,
      "step": 7989
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7247727138029312,
      "learning_rate": 2.7063703071919194e-05,
      "loss": 0.1371,
      "step": 7990
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6086367462875562,
      "learning_rate": 2.7058759447280563e-05,
      "loss": 0.1023,
      "step": 7991
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1506702687879151,
      "learning_rate": 2.7053815741587103e-05,
      "loss": 0.2865,
      "step": 7992
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9487820613548206,
      "learning_rate": 2.7048871955033434e-05,
      "loss": 0.2009,
      "step": 7993
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5167111041990434,
      "learning_rate": 2.704392808781422e-05,
      "loss": 0.0797,
      "step": 7994
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7643812473042746,
      "learning_rate": 2.7038984140124084e-05,
      "loss": 0.1686,
      "step": 7995
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7353037448418503,
      "learning_rate": 2.7034040112157688e-05,
      "loss": 0.2165,
      "step": 7996
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.38363671832873963,
      "learning_rate": 2.7029096004109666e-05,
      "loss": 0.0424,
      "step": 7997
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8089360924608197,
      "learning_rate": 2.702415181617469e-05,
      "loss": 0.1373,
      "step": 7998
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7183511923386888,
      "learning_rate": 2.70192075485474e-05,
      "loss": 0.1693,
      "step": 7999
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5658604031640749,
      "learning_rate": 2.701426320142247e-05,
      "loss": 0.127,
      "step": 8000
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6908231910548688,
      "learning_rate": 2.700931877499454e-05,
      "loss": 0.1734,
      "step": 8001
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.47357765565408766,
      "learning_rate": 2.7004374269458305e-05,
      "loss": 0.0954,
      "step": 8002
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9319516783872704,
      "learning_rate": 2.699942968500841e-05,
      "loss": 0.2269,
      "step": 8003
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.43092298509372956,
      "learning_rate": 2.6994485021839538e-05,
      "loss": 0.1175,
      "step": 8004
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8963628203583048,
      "learning_rate": 2.6989540280146358e-05,
      "loss": 0.1721,
      "step": 8005
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.924590066246343,
      "learning_rate": 2.6984595460123552e-05,
      "loss": 0.1269,
      "step": 8006
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.599722722903489,
      "learning_rate": 2.6979650561965803e-05,
      "loss": 0.0832,
      "step": 8007
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6429915634217005,
      "learning_rate": 2.6974705585867792e-05,
      "loss": 0.1108,
      "step": 8008
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6614550570300672,
      "learning_rate": 2.69697605320242e-05,
      "loss": 0.1254,
      "step": 8009
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9818115417528281,
      "learning_rate": 2.696481540062974e-05,
      "loss": 0.2791,
      "step": 8010
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5887621267044937,
      "learning_rate": 2.695987019187908e-05,
      "loss": 0.1032,
      "step": 8011
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.282600272179725,
      "learning_rate": 2.6954924905966922e-05,
      "loss": 0.3916,
      "step": 8012
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7375803148643193,
      "learning_rate": 2.6949979543087978e-05,
      "loss": 0.1149,
      "step": 8013
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9730123831326039,
      "learning_rate": 2.694503410343694e-05,
      "loss": 0.2065,
      "step": 8014
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0556621655827354,
      "learning_rate": 2.694008858720851e-05,
      "loss": 0.5561,
      "step": 8015
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7355974515794724,
      "learning_rate": 2.6935142994597407e-05,
      "loss": 0.0945,
      "step": 8016
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.621833085293893,
      "learning_rate": 2.6930197325798335e-05,
      "loss": 0.2115,
      "step": 8017
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6304644456536141,
      "learning_rate": 2.692525158100601e-05,
      "loss": 0.1445,
      "step": 8018
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0143721159983228,
      "learning_rate": 2.692030576041515e-05,
      "loss": 0.2242,
      "step": 8019
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9711201682934842,
      "learning_rate": 2.691535986422048e-05,
      "loss": 0.2317,
      "step": 8020
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.711523155939599,
      "learning_rate": 2.6910413892616714e-05,
      "loss": 0.1379,
      "step": 8021
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5675083502227743,
      "learning_rate": 2.6905467845798588e-05,
      "loss": 0.0756,
      "step": 8022
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5300016122943144,
      "learning_rate": 2.6900521723960824e-05,
      "loss": 0.0713,
      "step": 8023
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7857631152704957,
      "learning_rate": 2.689557552729816e-05,
      "loss": 0.1348,
      "step": 8024
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9310512949758812,
      "learning_rate": 2.689062925600533e-05,
      "loss": 0.2588,
      "step": 8025
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1848000892361854,
      "learning_rate": 2.6885682910277072e-05,
      "loss": 0.2145,
      "step": 8026
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5765939964285526,
      "learning_rate": 2.6880736490308122e-05,
      "loss": 0.1111,
      "step": 8027
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6610500288574978,
      "learning_rate": 2.6875789996293232e-05,
      "loss": 0.1525,
      "step": 8028
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3055753066221842,
      "learning_rate": 2.6870843428427138e-05,
      "loss": 0.3205,
      "step": 8029
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0106663687230681,
      "learning_rate": 2.6865896786904608e-05,
      "loss": 0.1766,
      "step": 8030
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1782941737647306,
      "learning_rate": 2.686095007192038e-05,
      "loss": 0.382,
      "step": 8031
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8476639886880817,
      "learning_rate": 2.6856003283669216e-05,
      "loss": 0.9008,
      "step": 8032
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9190405684720812,
      "learning_rate": 2.6851056422345865e-05,
      "loss": 0.2258,
      "step": 8033
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7552604513962207,
      "learning_rate": 2.684610948814511e-05,
      "loss": 0.1596,
      "step": 8034
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7646954825801587,
      "learning_rate": 2.684116248126169e-05,
      "loss": 0.1208,
      "step": 8035
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5116690891266782,
      "learning_rate": 2.6836215401890384e-05,
      "loss": 0.081,
      "step": 8036
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3649839414682368,
      "learning_rate": 2.6831268250225965e-05,
      "loss": 0.0683,
      "step": 8037
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9438334939006343,
      "learning_rate": 2.68263210264632e-05,
      "loss": 0.1686,
      "step": 8038
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7394459818505748,
      "learning_rate": 2.6821373730796872e-05,
      "loss": 0.153,
      "step": 8039
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1374319896503013,
      "learning_rate": 2.6816426363421753e-05,
      "loss": 0.3855,
      "step": 8040
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.626826725440876,
      "learning_rate": 2.6811478924532622e-05,
      "loss": 0.1155,
      "step": 8041
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6324839071374164,
      "learning_rate": 2.6806531414324275e-05,
      "loss": 0.1691,
      "step": 8042
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6133091221128898,
      "learning_rate": 2.6801583832991483e-05,
      "loss": 0.1297,
      "step": 8043
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5802120336493111,
      "learning_rate": 2.679663618072905e-05,
      "loss": 0.0625,
      "step": 8044
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6062809006275812,
      "learning_rate": 2.6791688457731767e-05,
      "loss": 0.0894,
      "step": 8045
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9433991512474004,
      "learning_rate": 2.678674066419442e-05,
      "loss": 0.1875,
      "step": 8046
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1948666259846772,
      "learning_rate": 2.6781792800311812e-05,
      "loss": 0.2737,
      "step": 8047
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.750702209788344,
      "learning_rate": 2.677684486627875e-05,
      "loss": 0.1801,
      "step": 8048
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7235599808357425,
      "learning_rate": 2.6771896862290026e-05,
      "loss": 0.1251,
      "step": 8049
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5426042015072702,
      "learning_rate": 2.676694878854045e-05,
      "loss": 0.0863,
      "step": 8050
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0001295309288325,
      "learning_rate": 2.676200064522484e-05,
      "loss": 0.1771,
      "step": 8051
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7022880465836351,
      "learning_rate": 2.6757052432538e-05,
      "loss": 0.1566,
      "step": 8052
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7546397267666353,
      "learning_rate": 2.6752104150674746e-05,
      "loss": 0.0897,
      "step": 8053
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7017619004123847,
      "learning_rate": 2.67471557998299e-05,
      "loss": 0.0861,
      "step": 8054
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6250595615238372,
      "learning_rate": 2.674220738019827e-05,
      "loss": 0.1184,
      "step": 8055
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6394454397992817,
      "learning_rate": 2.6737258891974693e-05,
      "loss": 0.1112,
      "step": 8056
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1421013873290553,
      "learning_rate": 2.6732310335353983e-05,
      "loss": 0.2487,
      "step": 8057
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7350912371214214,
      "learning_rate": 2.672736171053098e-05,
      "loss": 0.2295,
      "step": 8058
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9868342856423363,
      "learning_rate": 2.6722413017700504e-05,
      "loss": 0.2198,
      "step": 8059
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.49086367286169913,
      "learning_rate": 2.6717464257057394e-05,
      "loss": 0.0784,
      "step": 8060
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.49658609521348673,
      "learning_rate": 2.6712515428796487e-05,
      "loss": 0.095,
      "step": 8061
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5586540468555857,
      "learning_rate": 2.6707566533112622e-05,
      "loss": 0.0805,
      "step": 8062
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0172569141520529,
      "learning_rate": 2.670261757020063e-05,
      "loss": 0.1699,
      "step": 8063
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7424698511416018,
      "learning_rate": 2.669766854025537e-05,
      "loss": 0.1942,
      "step": 8064
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8894943009940571,
      "learning_rate": 2.669271944347168e-05,
      "loss": 0.1692,
      "step": 8065
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4625596307437312,
      "learning_rate": 2.668777028004442e-05,
      "loss": 0.4274,
      "step": 8066
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.874743815284976,
      "learning_rate": 2.668282105016843e-05,
      "loss": 0.4941,
      "step": 8067
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6569763007603984,
      "learning_rate": 2.667787175403857e-05,
      "loss": 0.1219,
      "step": 8068
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0917962085872353,
      "learning_rate": 2.6672922391849693e-05,
      "loss": 0.345,
      "step": 8069
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1807579319000894,
      "learning_rate": 2.6667972963796666e-05,
      "loss": 0.2033,
      "step": 8070
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8244073495381009,
      "learning_rate": 2.6663023470074343e-05,
      "loss": 0.1315,
      "step": 8071
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6850632796543984,
      "learning_rate": 2.6658073910877603e-05,
      "loss": 0.1685,
      "step": 8072
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8444294645810528,
      "learning_rate": 2.6653124286401297e-05,
      "loss": 0.3775,
      "step": 8073
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5559084830766231,
      "learning_rate": 2.6648174596840308e-05,
      "loss": 0.1257,
      "step": 8074
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.54569834246862,
      "learning_rate": 2.6643224842389497e-05,
      "loss": 0.0682,
      "step": 8075
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5453416353167667,
      "learning_rate": 2.6638275023243757e-05,
      "loss": 0.1045,
      "step": 8076
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7671296819854716,
      "learning_rate": 2.6633325139597948e-05,
      "loss": 0.1541,
      "step": 8077
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9062225341429675,
      "learning_rate": 2.662837519164696e-05,
      "loss": 0.2774,
      "step": 8078
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9831500733846008,
      "learning_rate": 2.6623425179585677e-05,
      "loss": 0.2169,
      "step": 8079
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7096192841160686,
      "learning_rate": 2.6618475103608976e-05,
      "loss": 0.1317,
      "step": 8080
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5313897672772911,
      "learning_rate": 2.661352496391175e-05,
      "loss": 0.0995,
      "step": 8081
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1678599127113656,
      "learning_rate": 2.6608574760688886e-05,
      "loss": 0.2751,
      "step": 8082
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1102897171094994,
      "learning_rate": 2.6603624494135294e-05,
      "loss": 0.2949,
      "step": 8083
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9517751716675399,
      "learning_rate": 2.659867416444584e-05,
      "loss": 0.1826,
      "step": 8084
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5913425245689694,
      "learning_rate": 2.6593723771815455e-05,
      "loss": 0.107,
      "step": 8085
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9719283186396994,
      "learning_rate": 2.658877331643901e-05,
      "loss": 0.1714,
      "step": 8086
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.881951518981339,
      "learning_rate": 2.6583822798511425e-05,
      "loss": 0.3582,
      "step": 8087
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7714402244394818,
      "learning_rate": 2.65788722182276e-05,
      "loss": 0.1508,
      "step": 8088
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8752570002653509,
      "learning_rate": 2.657392157578245e-05,
      "loss": 0.287,
      "step": 8089
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.07557368230925,
      "learning_rate": 2.6568970871370874e-05,
      "loss": 0.3051,
      "step": 8090
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.00916561013666,
      "learning_rate": 2.6564020105187794e-05,
      "loss": 0.3012,
      "step": 8091
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7898677111571576,
      "learning_rate": 2.6559069277428118e-05,
      "loss": 0.152,
      "step": 8092
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.730559010416552,
      "learning_rate": 2.655411838828677e-05,
      "loss": 0.1963,
      "step": 8093
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7863415579707598,
      "learning_rate": 2.6549167437958666e-05,
      "loss": 0.1854,
      "step": 8094
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5165081548333316,
      "learning_rate": 2.6544216426638728e-05,
      "loss": 0.1087,
      "step": 8095
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5261219707320189,
      "learning_rate": 2.6539265354521885e-05,
      "loss": 0.111,
      "step": 8096
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6348522312562731,
      "learning_rate": 2.6534314221803064e-05,
      "loss": 0.2006,
      "step": 8097
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4551321357482528,
      "learning_rate": 2.6529363028677184e-05,
      "loss": 0.0451,
      "step": 8098
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7254368802894477,
      "learning_rate": 2.6524411775339197e-05,
      "loss": 0.1604,
      "step": 8099
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6256376769266337,
      "learning_rate": 2.6519460461984014e-05,
      "loss": 0.1013,
      "step": 8100
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5702360278914964,
      "learning_rate": 2.6514509088806593e-05,
      "loss": 0.1103,
      "step": 8101
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6821872958060198,
      "learning_rate": 2.650955765600186e-05,
      "loss": 0.1244,
      "step": 8102
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6725400564690084,
      "learning_rate": 2.650460616376476e-05,
      "loss": 0.1078,
      "step": 8103
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5332186938897496,
      "learning_rate": 2.6499654612290238e-05,
      "loss": 0.0917,
      "step": 8104
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5567517953803722,
      "learning_rate": 2.649470300177324e-05,
      "loss": 0.3553,
      "step": 8105
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0832630438609794,
      "learning_rate": 2.6489751332408713e-05,
      "loss": 0.1949,
      "step": 8106
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5463129012016553,
      "learning_rate": 2.648479960439161e-05,
      "loss": 0.0636,
      "step": 8107
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5898535204333232,
      "learning_rate": 2.647984781791688e-05,
      "loss": 0.1219,
      "step": 8108
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8336779957503803,
      "learning_rate": 2.647489597317948e-05,
      "loss": 0.0586,
      "step": 8109
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6454546455723584,
      "learning_rate": 2.6469944070374376e-05,
      "loss": 0.43,
      "step": 8110
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9911901195652125,
      "learning_rate": 2.6464992109696506e-05,
      "loss": 0.1697,
      "step": 8111
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6171219619205471,
      "learning_rate": 2.646004009134086e-05,
      "loss": 0.1059,
      "step": 8112
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6500034888088926,
      "learning_rate": 2.6455088015502388e-05,
      "loss": 0.1691,
      "step": 8113
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2140495012622146,
      "learning_rate": 2.645013588237605e-05,
      "loss": 0.4808,
      "step": 8114
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4949528922062055,
      "learning_rate": 2.6445183692156828e-05,
      "loss": 0.1311,
      "step": 8115
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5845445547579053,
      "learning_rate": 2.644023144503969e-05,
      "loss": 0.12,
      "step": 8116
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.755638466802172,
      "learning_rate": 2.6435279141219603e-05,
      "loss": 0.0851,
      "step": 8117
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.002347102043293,
      "learning_rate": 2.643032678089155e-05,
      "loss": 0.1756,
      "step": 8118
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6320031378669051,
      "learning_rate": 2.6425374364250512e-05,
      "loss": 0.1033,
      "step": 8119
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9388386334064613,
      "learning_rate": 2.6420421891491453e-05,
      "loss": 0.2485,
      "step": 8120
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8460173136503044,
      "learning_rate": 2.6415469362809368e-05,
      "loss": 0.1793,
      "step": 8121
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0336367412877252,
      "learning_rate": 2.6410516778399248e-05,
      "loss": 0.316,
      "step": 8122
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2105555830667787,
      "learning_rate": 2.6405564138456068e-05,
      "loss": 0.3153,
      "step": 8123
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8553588897439331,
      "learning_rate": 2.6400611443174817e-05,
      "loss": 0.1939,
      "step": 8124
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6076425932549151,
      "learning_rate": 2.639565869275049e-05,
      "loss": 0.096,
      "step": 8125
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.846563307246946,
      "learning_rate": 2.639070588737808e-05,
      "loss": 0.1513,
      "step": 8126
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6339076583544392,
      "learning_rate": 2.6385753027252593e-05,
      "loss": 0.0941,
      "step": 8127
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6870194523800996,
      "learning_rate": 2.6380800112568998e-05,
      "loss": 0.1107,
      "step": 8128
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5990656631623307,
      "learning_rate": 2.6375847143522324e-05,
      "loss": 0.077,
      "step": 8129
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5095344844442083,
      "learning_rate": 2.637089412030756e-05,
      "loss": 0.1266,
      "step": 8130
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7147042753006687,
      "learning_rate": 2.636594104311972e-05,
      "loss": 0.1814,
      "step": 8131
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7736767403918411,
      "learning_rate": 2.636098791215379e-05,
      "loss": 0.2418,
      "step": 8132
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7199710483197063,
      "learning_rate": 2.6356034727604794e-05,
      "loss": 0.1222,
      "step": 8133
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6292534290276577,
      "learning_rate": 2.6351081489667738e-05,
      "loss": 0.1332,
      "step": 8134
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4068118952262199,
      "learning_rate": 2.6346128198537647e-05,
      "loss": 0.0579,
      "step": 8135
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8785099829915235,
      "learning_rate": 2.6341174854409506e-05,
      "loss": 0.2223,
      "step": 8136
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6535626223815444,
      "learning_rate": 2.633622145747836e-05,
      "loss": 0.1749,
      "step": 8137
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6146296884825528,
      "learning_rate": 2.6331268007939215e-05,
      "loss": 0.1272,
      "step": 8138
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5545647914446711,
      "learning_rate": 2.63263145059871e-05,
      "loss": 0.0824,
      "step": 8139
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4193858377211541,
      "learning_rate": 2.6321360951817027e-05,
      "loss": 0.4368,
      "step": 8140
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6647295992685123,
      "learning_rate": 2.631640734562403e-05,
      "loss": 0.1155,
      "step": 8141
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5729845503526938,
      "learning_rate": 2.6311453687603127e-05,
      "loss": 0.1893,
      "step": 8142
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6496335051562983,
      "learning_rate": 2.6306499977949357e-05,
      "loss": 0.0963,
      "step": 8143
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5252234440627059,
      "learning_rate": 2.630154621685775e-05,
      "loss": 0.1148,
      "step": 8144
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5354834638973163,
      "learning_rate": 2.6296592404523325e-05,
      "loss": 0.0975,
      "step": 8145
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5240742354650356,
      "learning_rate": 2.6291638541141135e-05,
      "loss": 0.0938,
      "step": 8146
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6076447381423168,
      "learning_rate": 2.628668462690621e-05,
      "loss": 0.1261,
      "step": 8147
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8069793015221888,
      "learning_rate": 2.628173066201358e-05,
      "loss": 0.0923,
      "step": 8148
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.46813054168780116,
      "learning_rate": 2.6276776646658308e-05,
      "loss": 0.0815,
      "step": 8149
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8947414526012117,
      "learning_rate": 2.6271822581035415e-05,
      "loss": 0.3441,
      "step": 8150
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.7430567794765834,
      "learning_rate": 2.6266868465339965e-05,
      "loss": 0.4899,
      "step": 8151
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.009735849092106,
      "learning_rate": 2.6261914299766988e-05,
      "loss": 0.169,
      "step": 8152
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.622550740247502,
      "learning_rate": 2.6256960084511543e-05,
      "loss": 0.0938,
      "step": 8153
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5674433630275786,
      "learning_rate": 2.6252005819768677e-05,
      "loss": 0.1964,
      "step": 8154
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5594549885428739,
      "learning_rate": 2.6247051505733443e-05,
      "loss": 0.0872,
      "step": 8155
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5268531872717536,
      "learning_rate": 2.6242097142600903e-05,
      "loss": 0.0926,
      "step": 8156
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9059867489012035,
      "learning_rate": 2.6237142730566105e-05,
      "loss": 0.3915,
      "step": 8157
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.49707690973563,
      "learning_rate": 2.623218826982411e-05,
      "loss": 0.1102,
      "step": 8158
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.494029446442353,
      "learning_rate": 2.6227233760569984e-05,
      "loss": 0.0613,
      "step": 8159
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.46933012281399844,
      "learning_rate": 2.6222279202998778e-05,
      "loss": 0.0576,
      "step": 8160
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.015166250391686,
      "learning_rate": 2.6217324597305577e-05,
      "loss": 0.4197,
      "step": 8161
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0458702371496398,
      "learning_rate": 2.6212369943685422e-05,
      "loss": 0.2368,
      "step": 8162
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8748624086749274,
      "learning_rate": 2.62074152423334e-05,
      "loss": 0.209,
      "step": 8163
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9031427690409977,
      "learning_rate": 2.6202460493444565e-05,
      "loss": 0.2088,
      "step": 8164
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6365602711262952,
      "learning_rate": 2.619750569721401e-05,
      "loss": 0.1159,
      "step": 8165
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6326242488372429,
      "learning_rate": 2.619255085383679e-05,
      "loss": 0.0918,
      "step": 8166
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3271705589769585,
      "learning_rate": 2.6187595963507993e-05,
      "loss": 0.2116,
      "step": 8167
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6103113006218673,
      "learning_rate": 2.618264102642269e-05,
      "loss": 0.0889,
      "step": 8168
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8498431221262105,
      "learning_rate": 2.617768604277596e-05,
      "loss": 0.255,
      "step": 8169
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5114506370737543,
      "learning_rate": 2.6172731012762885e-05,
      "loss": 0.057,
      "step": 8170
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.575969001651207,
      "learning_rate": 2.616777593657856e-05,
      "loss": 0.0909,
      "step": 8171
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5719687277245282,
      "learning_rate": 2.616282081441805e-05,
      "loss": 0.099,
      "step": 8172
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9278057628932039,
      "learning_rate": 2.6157865646476453e-05,
      "loss": 0.1136,
      "step": 8173
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5762994167896435,
      "learning_rate": 2.6152910432948852e-05,
      "loss": 0.1119,
      "step": 8174
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.40313940620763633,
      "learning_rate": 2.6147955174030354e-05,
      "loss": 0.0566,
      "step": 8175
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6984804342346674,
      "learning_rate": 2.614299986991603e-05,
      "loss": 0.1375,
      "step": 8176
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5693327813617604,
      "learning_rate": 2.6138044520800976e-05,
      "loss": 0.0918,
      "step": 8177
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6573684086282413,
      "learning_rate": 2.61330891268803e-05,
      "loss": 0.0789,
      "step": 8178
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5120917402184506,
      "learning_rate": 2.61281336883491e-05,
      "loss": 0.0829,
      "step": 8179
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7172390993225536,
      "learning_rate": 2.6123178205402455e-05,
      "loss": 0.1547,
      "step": 8180
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7611355387165898,
      "learning_rate": 2.6118222678235487e-05,
      "loss": 0.1258,
      "step": 8181
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7073343823583551,
      "learning_rate": 2.6113267107043286e-05,
      "loss": 0.1401,
      "step": 8182
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5669771397949238,
      "learning_rate": 2.6108311492020963e-05,
      "loss": 0.1548,
      "step": 8183
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9579589364677933,
      "learning_rate": 2.6103355833363623e-05,
      "loss": 0.2272,
      "step": 8184
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.44835254650767,
      "learning_rate": 2.6098400131266377e-05,
      "loss": 0.6066,
      "step": 8185
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5565973120441069,
      "learning_rate": 2.609344438592432e-05,
      "loss": 0.0822,
      "step": 8186
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4431653961024233,
      "learning_rate": 2.6088488597532586e-05,
      "loss": 0.0565,
      "step": 8187
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4636774715699485,
      "learning_rate": 2.6083532766286268e-05,
      "loss": 0.0473,
      "step": 8188
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6524579272849588,
      "learning_rate": 2.607857689238049e-05,
      "loss": 0.1332,
      "step": 8189
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0894592136367958,
      "learning_rate": 2.6073620976010365e-05,
      "loss": 0.3103,
      "step": 8190
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6124985277376954,
      "learning_rate": 2.606866501737102e-05,
      "loss": 0.107,
      "step": 8191
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5771444529054438,
      "learning_rate": 2.6063709016657557e-05,
      "loss": 0.0965,
      "step": 8192
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3518134747017557,
      "learning_rate": 2.6058752974065115e-05,
      "loss": 0.0426,
      "step": 8193
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.978033973655687,
      "learning_rate": 2.6053796889788806e-05,
      "loss": 0.2269,
      "step": 8194
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6091430600986026,
      "learning_rate": 2.6048840764023762e-05,
      "loss": 0.0771,
      "step": 8195
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7999882129258296,
      "learning_rate": 2.60438845969651e-05,
      "loss": 0.2399,
      "step": 8196
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9820786573001928,
      "learning_rate": 2.603892838880796e-05,
      "loss": 0.2201,
      "step": 8197
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5376090737965682,
      "learning_rate": 2.6033972139747458e-05,
      "loss": 0.0451,
      "step": 8198
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.690335190736016,
      "learning_rate": 2.6029015849978737e-05,
      "loss": 0.9482,
      "step": 8199
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5539401531485789,
      "learning_rate": 2.602405951969692e-05,
      "loss": 0.0765,
      "step": 8200
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9002052328010847,
      "learning_rate": 2.6019103149097153e-05,
      "loss": 0.3329,
      "step": 8201
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.226183601812087,
      "learning_rate": 2.6014146738374556e-05,
      "loss": 0.3537,
      "step": 8202
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4132483258986244,
      "learning_rate": 2.6009190287724283e-05,
      "loss": 0.0296,
      "step": 8203
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7461631371260801,
      "learning_rate": 2.6004233797341456e-05,
      "loss": 0.2575,
      "step": 8204
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5796122223976659,
      "learning_rate": 2.5999277267421235e-05,
      "loss": 0.0873,
      "step": 8205
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5738151956433671,
      "learning_rate": 2.5994320698158742e-05,
      "loss": 0.1328,
      "step": 8206
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8595541867781898,
      "learning_rate": 2.5989364089749135e-05,
      "loss": 0.2199,
      "step": 8207
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6835613543673028,
      "learning_rate": 2.5984407442387555e-05,
      "loss": 0.19,
      "step": 8208
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.44763449417680984,
      "learning_rate": 2.5979450756269153e-05,
      "loss": 0.0808,
      "step": 8209
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6830304069526527,
      "learning_rate": 2.597449403158907e-05,
      "loss": 0.0797,
      "step": 8210
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5904915394518992,
      "learning_rate": 2.5969537268542455e-05,
      "loss": 0.119,
      "step": 8211
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8750437544355005,
      "learning_rate": 2.5964580467324466e-05,
      "loss": 0.1797,
      "step": 8212
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6633698253649543,
      "learning_rate": 2.5959623628130246e-05,
      "loss": 0.1274,
      "step": 8213
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5164286345725181,
      "learning_rate": 2.5954666751154965e-05,
      "loss": 0.1161,
      "step": 8214
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7091990155185789,
      "learning_rate": 2.5949709836593765e-05,
      "loss": 0.1287,
      "step": 8215
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8698733556295398,
      "learning_rate": 2.594475288464181e-05,
      "loss": 0.2104,
      "step": 8216
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2387614999885823,
      "learning_rate": 2.5939795895494255e-05,
      "loss": 0.2891,
      "step": 8217
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5901233439508922,
      "learning_rate": 2.5934838869346267e-05,
      "loss": 0.3349,
      "step": 8218
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6680682435170466,
      "learning_rate": 2.5929881806392996e-05,
      "loss": 0.1129,
      "step": 8219
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9553287534664315,
      "learning_rate": 2.5924924706829616e-05,
      "loss": 0.1623,
      "step": 8220
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9100995299062227,
      "learning_rate": 2.5919967570851284e-05,
      "loss": 0.302,
      "step": 8221
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1714534509276198,
      "learning_rate": 2.591501039865318e-05,
      "loss": 0.4012,
      "step": 8222
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8042815105160646,
      "learning_rate": 2.5910053190430444e-05,
      "loss": 0.1398,
      "step": 8223
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6187954362673249,
      "learning_rate": 2.5905095946378277e-05,
      "loss": 0.0738,
      "step": 8224
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5955594075559797,
      "learning_rate": 2.590013866669182e-05,
      "loss": 0.0717,
      "step": 8225
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7385711926081118,
      "learning_rate": 2.5895181351566273e-05,
      "loss": 0.1139,
      "step": 8226
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6151590961711266,
      "learning_rate": 2.5890224001196784e-05,
      "loss": 0.0626,
      "step": 8227
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.48530043136000045,
      "learning_rate": 2.588526661577854e-05,
      "loss": 0.049,
      "step": 8228
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1892263652477835,
      "learning_rate": 2.5880309195506712e-05,
      "loss": 0.1864,
      "step": 8229
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9494578131878726,
      "learning_rate": 2.5875351740576493e-05,
      "loss": 0.1433,
      "step": 8230
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.151409266909581,
      "learning_rate": 2.5870394251183032e-05,
      "loss": 0.2359,
      "step": 8231
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5865227738736257,
      "learning_rate": 2.586543672752154e-05,
      "loss": 0.0972,
      "step": 8232
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6025998191769003,
      "learning_rate": 2.586047916978717e-05,
      "loss": 0.1093,
      "step": 8233
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7345210280302159,
      "learning_rate": 2.5855521578175136e-05,
      "loss": 0.117,
      "step": 8234
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7709667526428514,
      "learning_rate": 2.585056395288059e-05,
      "loss": 0.1433,
      "step": 8235
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5671831740239481,
      "learning_rate": 2.5845606294098735e-05,
      "loss": 0.0637,
      "step": 8236
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.685434557944196,
      "learning_rate": 2.5840648602024753e-05,
      "loss": 0.1625,
      "step": 8237
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1767666304370326,
      "learning_rate": 2.583569087685384e-05,
      "loss": 0.2314,
      "step": 8238
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.900025257162457,
      "learning_rate": 2.583073311878117e-05,
      "loss": 0.5551,
      "step": 8239
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5561710361659431,
      "learning_rate": 2.5825775328001943e-05,
      "loss": 0.1492,
      "step": 8240
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4037382066688917,
      "learning_rate": 2.582081750471136e-05,
      "loss": 0.0728,
      "step": 8241
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7187853899991498,
      "learning_rate": 2.581585964910459e-05,
      "loss": 0.1488,
      "step": 8242
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9073958935183529,
      "learning_rate": 2.581090176137685e-05,
      "loss": 0.1538,
      "step": 8243
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5536646099588133,
      "learning_rate": 2.5805943841723322e-05,
      "loss": 0.1492,
      "step": 8244
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.523927761204976,
      "learning_rate": 2.580098589033921e-05,
      "loss": 0.0882,
      "step": 8245
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.5696169998119873,
      "learning_rate": 2.5796027907419707e-05,
      "loss": 0.4668,
      "step": 8246
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7554765637783369,
      "learning_rate": 2.5791069893160013e-05,
      "loss": 0.1443,
      "step": 8247
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1186781705140145,
      "learning_rate": 2.5786111847755335e-05,
      "loss": 0.3175,
      "step": 8248
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7165332096648206,
      "learning_rate": 2.578115377140087e-05,
      "loss": 0.2145,
      "step": 8249
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6522633858993612,
      "learning_rate": 2.5776195664291818e-05,
      "loss": 0.1357,
      "step": 8250
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.653733376234683,
      "learning_rate": 2.577123752662338e-05,
      "loss": 0.1365,
      "step": 8251
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4443654314448494,
      "learning_rate": 2.5766279358590785e-05,
      "loss": 0.0539,
      "step": 8252
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5417941702271235,
      "learning_rate": 2.5761321160389206e-05,
      "loss": 0.087,
      "step": 8253
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8152289901514241,
      "learning_rate": 2.575636293221388e-05,
      "loss": 0.1908,
      "step": 8254
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7678729072315004,
      "learning_rate": 2.5751404674259994e-05,
      "loss": 0.1166,
      "step": 8255
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7241188197916566,
      "learning_rate": 2.5746446386722772e-05,
      "loss": 0.1384,
      "step": 8256
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6692274393624104,
      "learning_rate": 2.5741488069797414e-05,
      "loss": 0.1072,
      "step": 8257
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1363988722234324,
      "learning_rate": 2.573652972367915e-05,
      "loss": 0.1446,
      "step": 8258
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3115103156412244,
      "learning_rate": 2.5731571348563172e-05,
      "loss": 0.8131,
      "step": 8259
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8546240012282452,
      "learning_rate": 2.5726612944644717e-05,
      "loss": 0.0776,
      "step": 8260
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0506232596575973,
      "learning_rate": 2.5721654512118982e-05,
      "loss": 0.2347,
      "step": 8261
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4527753291734182,
      "learning_rate": 2.5716696051181193e-05,
      "loss": 0.3865,
      "step": 8262
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6995474764044506,
      "learning_rate": 2.5711737562026565e-05,
      "loss": 0.1237,
      "step": 8263
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7189655954763468,
      "learning_rate": 2.5706779044850325e-05,
      "loss": 0.1545,
      "step": 8264
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3253993585500357,
      "learning_rate": 2.5701820499847684e-05,
      "loss": 0.2808,
      "step": 8265
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6651262266360033,
      "learning_rate": 2.5696861927213867e-05,
      "loss": 0.1376,
      "step": 8266
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1283339417926403,
      "learning_rate": 2.5691903327144096e-05,
      "loss": 0.2433,
      "step": 8267
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8181381178926144,
      "learning_rate": 2.56869446998336e-05,
      "loss": 0.1314,
      "step": 8268
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6098786415062278,
      "learning_rate": 2.568198604547759e-05,
      "loss": 0.0769,
      "step": 8269
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5733093586839426,
      "learning_rate": 2.5677027364271305e-05,
      "loss": 0.1754,
      "step": 8270
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.45078266414511897,
      "learning_rate": 2.5672068656409965e-05,
      "loss": 0.0604,
      "step": 8271
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5465695040828966,
      "learning_rate": 2.5667109922088806e-05,
      "loss": 0.1083,
      "step": 8272
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.387327155097249,
      "learning_rate": 2.5662151161503046e-05,
      "loss": 0.1832,
      "step": 8273
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6371012961689878,
      "learning_rate": 2.5657192374847926e-05,
      "loss": 0.0771,
      "step": 8274
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7046926556951569,
      "learning_rate": 2.5652233562318667e-05,
      "loss": 0.1041,
      "step": 8275
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7789967941982265,
      "learning_rate": 2.5647274724110503e-05,
      "loss": 0.1873,
      "step": 8276
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5647356957594342,
      "learning_rate": 2.5642315860418674e-05,
      "loss": 0.1145,
      "step": 8277
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.630256654588233,
      "learning_rate": 2.5637356971438413e-05,
      "loss": 0.1205,
      "step": 8278
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9172934879602027,
      "learning_rate": 2.5632398057364944e-05,
      "loss": 0.2746,
      "step": 8279
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.395855972480386,
      "learning_rate": 2.5627439118393513e-05,
      "loss": 0.4744,
      "step": 8280
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6854427782820217,
      "learning_rate": 2.5622480154719353e-05,
      "loss": 0.1229,
      "step": 8281
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8746711614381618,
      "learning_rate": 2.5617521166537707e-05,
      "loss": 0.326,
      "step": 8282
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6863822400020624,
      "learning_rate": 2.5612562154043803e-05,
      "loss": 0.0767,
      "step": 8283
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2510529487737978,
      "learning_rate": 2.5607603117432892e-05,
      "loss": 0.1614,
      "step": 8284
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7070373828594845,
      "learning_rate": 2.5602644056900205e-05,
      "loss": 0.0941,
      "step": 8285
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6466082892384367,
      "learning_rate": 2.5597684972641e-05,
      "loss": 0.0795,
      "step": 8286
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1336119539671428,
      "learning_rate": 2.5592725864850504e-05,
      "loss": 0.3085,
      "step": 8287
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6223452253682228,
      "learning_rate": 2.5587766733723967e-05,
      "loss": 0.1446,
      "step": 8288
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3807188247810236,
      "learning_rate": 2.5582807579456626e-05,
      "loss": 0.4047,
      "step": 8289
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.482214671283421,
      "learning_rate": 2.557784840224374e-05,
      "loss": 0.7144,
      "step": 8290
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8224477960849346,
      "learning_rate": 2.557288920228055e-05,
      "loss": 0.1323,
      "step": 8291
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6755827705111256,
      "learning_rate": 2.5567929979762294e-05,
      "loss": 0.0913,
      "step": 8292
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9133744728473701,
      "learning_rate": 2.5562970734884234e-05,
      "loss": 0.2926,
      "step": 8293
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4867843330644651,
      "learning_rate": 2.5558011467841615e-05,
      "loss": 0.0682,
      "step": 8294
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5723555820024833,
      "learning_rate": 2.555305217882967e-05,
      "loss": 0.0738,
      "step": 8295
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6420541984430156,
      "learning_rate": 2.5548092868043683e-05,
      "loss": 0.1392,
      "step": 8296
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.48243007204298677,
      "learning_rate": 2.5543133535678875e-05,
      "loss": 0.0978,
      "step": 8297
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5242507811102155,
      "learning_rate": 2.5538174181930515e-05,
      "loss": 0.0874,
      "step": 8298
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7510952667211418,
      "learning_rate": 2.553321480699385e-05,
      "loss": 0.1765,
      "step": 8299
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.671222769750262,
      "learning_rate": 2.552825541106414e-05,
      "loss": 0.3119,
      "step": 8300
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8369906838303194,
      "learning_rate": 2.552329599433663e-05,
      "loss": 0.1214,
      "step": 8301
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6490989135746299,
      "learning_rate": 2.5518336557006584e-05,
      "loss": 0.1593,
      "step": 8302
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3336445665263925,
      "learning_rate": 2.5513377099269252e-05,
      "loss": 0.1022,
      "step": 8303
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8980751384926956,
      "learning_rate": 2.55084176213199e-05,
      "loss": 0.1025,
      "step": 8304
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7851646989687484,
      "learning_rate": 2.5503458123353787e-05,
      "loss": 0.149,
      "step": 8305
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.00973336578806,
      "learning_rate": 2.5498498605566163e-05,
      "loss": 0.3212,
      "step": 8306
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.454390381885136,
      "learning_rate": 2.549353906815229e-05,
      "loss": 0.077,
      "step": 8307
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.145787551786984,
      "learning_rate": 2.548857951130743e-05,
      "loss": 0.3552,
      "step": 8308
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.69387690524657,
      "learning_rate": 2.5483619935226848e-05,
      "loss": 0.244,
      "step": 8309
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6511454743451521,
      "learning_rate": 2.5478660340105796e-05,
      "loss": 0.1293,
      "step": 8310
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5700561327892761,
      "learning_rate": 2.547370072613955e-05,
      "loss": 0.1126,
      "step": 8311
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7798852855203159,
      "learning_rate": 2.546874109352337e-05,
      "loss": 0.1167,
      "step": 8312
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4784659514994707,
      "learning_rate": 2.5463781442452505e-05,
      "loss": 0.0813,
      "step": 8313
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8051323507070169,
      "learning_rate": 2.5458821773122243e-05,
      "loss": 0.1749,
      "step": 8314
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5378850407371083,
      "learning_rate": 2.545386208572783e-05,
      "loss": 0.121,
      "step": 8315
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.551070180809176,
      "learning_rate": 2.5448902380464545e-05,
      "loss": 0.1147,
      "step": 8316
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.018925906052375,
      "learning_rate": 2.5443942657527652e-05,
      "loss": 0.2312,
      "step": 8317
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6326878637861476,
      "learning_rate": 2.543898291711242e-05,
      "loss": 0.1092,
      "step": 8318
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9792760369998612,
      "learning_rate": 2.543402315941411e-05,
      "loss": 0.2536,
      "step": 8319
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9625790793548803,
      "learning_rate": 2.542906338462801e-05,
      "loss": 0.2473,
      "step": 8320
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.818229894345902,
      "learning_rate": 2.5424103592949366e-05,
      "loss": 0.1736,
      "step": 8321
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5303819880251766,
      "learning_rate": 2.5419143784573462e-05,
      "loss": 0.0567,
      "step": 8322
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5823883697835246,
      "learning_rate": 2.5414183959695564e-05,
      "loss": 0.1575,
      "step": 8323
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6183833094263134,
      "learning_rate": 2.5409224118510954e-05,
      "loss": 0.0937,
      "step": 8324
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4100877426078808,
      "learning_rate": 2.540426426121489e-05,
      "loss": 0.494,
      "step": 8325
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7953490081087647,
      "learning_rate": 2.539930438800266e-05,
      "loss": 0.1373,
      "step": 8326
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4345846941774912,
      "learning_rate": 2.539434449906952e-05,
      "loss": 0.0503,
      "step": 8327
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3263811792549642,
      "learning_rate": 2.5389384594610767e-05,
      "loss": 0.3891,
      "step": 8328
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.937587206589527,
      "learning_rate": 2.5384424674821648e-05,
      "loss": 0.2624,
      "step": 8329
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0941431635653265,
      "learning_rate": 2.5379464739897467e-05,
      "loss": 0.2044,
      "step": 8330
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6584978673353393,
      "learning_rate": 2.537450479003348e-05,
      "loss": 0.1376,
      "step": 8331
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4730038704997669,
      "learning_rate": 2.536954482542498e-05,
      "loss": 0.0649,
      "step": 8332
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9728250933365128,
      "learning_rate": 2.5364584846267227e-05,
      "loss": 0.1178,
      "step": 8333
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.066535942565,
      "learning_rate": 2.5359624852755508e-05,
      "loss": 0.5457,
      "step": 8334
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4035570547737915,
      "learning_rate": 2.53546648450851e-05,
      "loss": 0.0663,
      "step": 8335
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7019941422912628,
      "learning_rate": 2.5349704823451286e-05,
      "loss": 0.1405,
      "step": 8336
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.614216247013466,
      "learning_rate": 2.534474478804934e-05,
      "loss": 0.1147,
      "step": 8337
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7079883273998764,
      "learning_rate": 2.533978473907455e-05,
      "loss": 0.1403,
      "step": 8338
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.709051043093676,
      "learning_rate": 2.5334824676722186e-05,
      "loss": 0.1082,
      "step": 8339
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5572171526297645,
      "learning_rate": 2.5329864601187537e-05,
      "loss": 0.0991,
      "step": 8340
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6520991388681351,
      "learning_rate": 2.5324904512665888e-05,
      "loss": 0.1501,
      "step": 8341
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.610849865381872,
      "learning_rate": 2.5319944411352503e-05,
      "loss": 0.0791,
      "step": 8342
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.49554620681581163,
      "learning_rate": 2.5314984297442684e-05,
      "loss": 0.0746,
      "step": 8343
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5952299340830643,
      "learning_rate": 2.5310024171131706e-05,
      "loss": 0.1136,
      "step": 8344
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7253794994633592,
      "learning_rate": 2.5305064032614855e-05,
      "loss": 0.1667,
      "step": 8345
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8836772915878576,
      "learning_rate": 2.5300103882087413e-05,
      "loss": 0.3101,
      "step": 8346
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8891735698146465,
      "learning_rate": 2.529514371974467e-05,
      "loss": 0.2328,
      "step": 8347
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6245261987666616,
      "learning_rate": 2.5290183545781903e-05,
      "loss": 0.1934,
      "step": 8348
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5092889785833641,
      "learning_rate": 2.5285223360394405e-05,
      "loss": 0.0768,
      "step": 8349
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8672301939511866,
      "learning_rate": 2.5280263163777457e-05,
      "loss": 0.2915,
      "step": 8350
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9747223201700027,
      "learning_rate": 2.5275302956126346e-05,
      "loss": 0.0896,
      "step": 8351
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5317019073154013,
      "learning_rate": 2.527034273763636e-05,
      "loss": 0.101,
      "step": 8352
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8208115639757769,
      "learning_rate": 2.5265382508502794e-05,
      "loss": 0.2073,
      "step": 8353
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6127607491190206,
      "learning_rate": 2.526042226892092e-05,
      "loss": 0.4208,
      "step": 8354
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.02460417826708,
      "learning_rate": 2.525546201908604e-05,
      "loss": 0.2565,
      "step": 8355
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6167554504965884,
      "learning_rate": 2.525050175919343e-05,
      "loss": 0.1186,
      "step": 8356
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5371813670804974,
      "learning_rate": 2.5245541489438395e-05,
      "loss": 0.0705,
      "step": 8357
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4128832389620592,
      "learning_rate": 2.5240581210016207e-05,
      "loss": 0.057,
      "step": 8358
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1618096349387494,
      "learning_rate": 2.523562092112217e-05,
      "loss": 0.2836,
      "step": 8359
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7019510412272811,
      "learning_rate": 2.523066062295157e-05,
      "loss": 0.1173,
      "step": 8360
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8948607113915402,
      "learning_rate": 2.522570031569969e-05,
      "loss": 0.3596,
      "step": 8361
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5264893505551478,
      "learning_rate": 2.5220739999561833e-05,
      "loss": 0.0692,
      "step": 8362
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.718020413128106,
      "learning_rate": 2.5215779674733282e-05,
      "loss": 0.1378,
      "step": 8363
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5410785463030203,
      "learning_rate": 2.5210819341409326e-05,
      "loss": 0.0907,
      "step": 8364
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.105214411820408,
      "learning_rate": 2.520585899978527e-05,
      "loss": 0.2346,
      "step": 8365
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8780966026466304,
      "learning_rate": 2.5200898650056387e-05,
      "loss": 0.2787,
      "step": 8366
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4038130949789698,
      "learning_rate": 2.519593829241799e-05,
      "loss": 0.0687,
      "step": 8367
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7573271519873324,
      "learning_rate": 2.5190977927065357e-05,
      "loss": 0.173,
      "step": 8368
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0581781739883727,
      "learning_rate": 2.5186017554193785e-05,
      "loss": 0.3478,
      "step": 8369
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5536012913766086,
      "learning_rate": 2.5181057173998573e-05,
      "loss": 0.0894,
      "step": 8370
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8825061182657757,
      "learning_rate": 2.5176096786675006e-05,
      "loss": 0.2152,
      "step": 8371
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6674702316630514,
      "learning_rate": 2.5171136392418393e-05,
      "loss": 0.1659,
      "step": 8372
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.654007146269283,
      "learning_rate": 2.5166175991424002e-05,
      "loss": 0.0748,
      "step": 8373
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.45472916656173207,
      "learning_rate": 2.5161215583887153e-05,
      "loss": 0.0388,
      "step": 8374
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.38834613353395,
      "learning_rate": 2.5156255170003133e-05,
      "loss": 0.0632,
      "step": 8375
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5558431551613724,
      "learning_rate": 2.5151294749967225e-05,
      "loss": 0.1016,
      "step": 8376
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7958803909420372,
      "learning_rate": 2.514633432397474e-05,
      "loss": 0.1082,
      "step": 8377
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6149946264690525,
      "learning_rate": 2.5141373892220965e-05,
      "loss": 0.1327,
      "step": 8378
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.772547487261228,
      "learning_rate": 2.5136413454901204e-05,
      "loss": 0.495,
      "step": 8379
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.443426766316063,
      "learning_rate": 2.5131453012210742e-05,
      "loss": 0.3852,
      "step": 8380
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2071897462083647,
      "learning_rate": 2.512649256434489e-05,
      "loss": 0.2963,
      "step": 8381
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7980241653291132,
      "learning_rate": 2.512153211149892e-05,
      "loss": 0.1808,
      "step": 8382
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7670373305175686,
      "learning_rate": 2.5116571653868154e-05,
      "loss": 0.0825,
      "step": 8383
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7247196520941591,
      "learning_rate": 2.5111611191647878e-05,
      "loss": 0.1079,
      "step": 8384
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.948658394644469,
      "learning_rate": 2.5106650725033387e-05,
      "loss": 0.1464,
      "step": 8385
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6305794514135643,
      "learning_rate": 2.5101690254219978e-05,
      "loss": 0.267,
      "step": 8386
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5574069933072432,
      "learning_rate": 2.5096729779402956e-05,
      "loss": 0.0562,
      "step": 8387
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4624130865310188,
      "learning_rate": 2.5091769300777606e-05,
      "loss": 0.0874,
      "step": 8388
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6355729047185791,
      "learning_rate": 2.5086808818539244e-05,
      "loss": 0.1308,
      "step": 8389
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7727815652939135,
      "learning_rate": 2.5081848332883144e-05,
      "loss": 0.1522,
      "step": 8390
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6899165278223414,
      "learning_rate": 2.5076887844004626e-05,
      "loss": 0.1676,
      "step": 8391
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4135514658551728,
      "learning_rate": 2.5071927352098972e-05,
      "loss": 0.0836,
      "step": 8392
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.46588197837967293,
      "learning_rate": 2.5066966857361497e-05,
      "loss": 0.0868,
      "step": 8393
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5849159308495594,
      "learning_rate": 2.506200635998748e-05,
      "loss": 0.0658,
      "step": 8394
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7104021783078444,
      "learning_rate": 2.5057045860172228e-05,
      "loss": 0.1454,
      "step": 8395
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9736130150635053,
      "learning_rate": 2.5052085358111045e-05,
      "loss": 0.2566,
      "step": 8396
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1288849633539577,
      "learning_rate": 2.5047124853999222e-05,
      "loss": 0.1978,
      "step": 8397
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5504181565626224,
      "learning_rate": 2.5042164348032056e-05,
      "loss": 0.086,
      "step": 8398
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.45772305483991665,
      "learning_rate": 2.5037203840404856e-05,
      "loss": 0.0663,
      "step": 8399
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7002945262176121,
      "learning_rate": 2.5032243331312904e-05,
      "loss": 0.1035,
      "step": 8400
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9345554135686672,
      "learning_rate": 2.502728282095152e-05,
      "loss": 0.2102,
      "step": 8401
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7624984909122878,
      "learning_rate": 2.5022322309515982e-05,
      "loss": 0.2278,
      "step": 8402
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8101637133212723,
      "learning_rate": 2.5017361797201605e-05,
      "loss": 0.2329,
      "step": 8403
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5384424436152129,
      "learning_rate": 2.5012401284203678e-05,
      "loss": 0.0852,
      "step": 8404
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5153012559459117,
      "learning_rate": 2.5007440770717505e-05,
      "loss": 0.0866,
      "step": 8405
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5139279360435688,
      "learning_rate": 2.5002480256938387e-05,
      "loss": 0.0971,
      "step": 8406
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6329766134424022,
      "learning_rate": 2.4997519743061616e-05,
      "loss": 0.1765,
      "step": 8407
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9019936725781216,
      "learning_rate": 2.4992559229282498e-05,
      "loss": 0.1968,
      "step": 8408
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8100454380961066,
      "learning_rate": 2.4987598715796324e-05,
      "loss": 0.2616,
      "step": 8409
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5906196893405499,
      "learning_rate": 2.49826382027984e-05,
      "loss": 0.1405,
      "step": 8410
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8478457892868118,
      "learning_rate": 2.4977677690484023e-05,
      "loss": 0.1311,
      "step": 8411
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0195814137879196,
      "learning_rate": 2.4972717179048493e-05,
      "loss": 0.2201,
      "step": 8412
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4777850099264109,
      "learning_rate": 2.49677566686871e-05,
      "loss": 0.0432,
      "step": 8413
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.543009166471289,
      "learning_rate": 2.4962796159595153e-05,
      "loss": 0.1112,
      "step": 8414
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5320707259396131,
      "learning_rate": 2.495783565196795e-05,
      "loss": 0.1165,
      "step": 8415
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4446058817711498,
      "learning_rate": 2.495287514600079e-05,
      "loss": 0.0591,
      "step": 8416
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.934375239343742,
      "learning_rate": 2.4947914641888957e-05,
      "loss": 0.5191,
      "step": 8417
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5088207609104702,
      "learning_rate": 2.4942954139827774e-05,
      "loss": 0.0749,
      "step": 8418
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8384094344771365,
      "learning_rate": 2.4937993640012526e-05,
      "loss": 0.1205,
      "step": 8419
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6842491663025394,
      "learning_rate": 2.4933033142638516e-05,
      "loss": 0.1322,
      "step": 8420
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5940758313789657,
      "learning_rate": 2.4928072647901023e-05,
      "loss": 0.116,
      "step": 8421
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8734653359650998,
      "learning_rate": 2.4923112155995376e-05,
      "loss": 0.1993,
      "step": 8422
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5946175667974462,
      "learning_rate": 2.4918151667116858e-05,
      "loss": 0.1359,
      "step": 8423
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8453557907121598,
      "learning_rate": 2.491319118146077e-05,
      "loss": 0.2322,
      "step": 8424
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.510167507465622,
      "learning_rate": 2.490823069922239e-05,
      "loss": 0.0835,
      "step": 8425
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6279246524200348,
      "learning_rate": 2.490327022059705e-05,
      "loss": 0.1112,
      "step": 8426
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5654329387943638,
      "learning_rate": 2.4898309745780028e-05,
      "loss": 0.08,
      "step": 8427
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8324114232166487,
      "learning_rate": 2.4893349274966622e-05,
      "loss": 0.2697,
      "step": 8428
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.46587640771762223,
      "learning_rate": 2.4888388808352125e-05,
      "loss": 0.0707,
      "step": 8429
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.39424437635596155,
      "learning_rate": 2.4883428346131848e-05,
      "loss": 0.0424,
      "step": 8430
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6182674049728132,
      "learning_rate": 2.4878467888501084e-05,
      "loss": 0.0747,
      "step": 8431
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.588393507259414,
      "learning_rate": 2.4873507435655124e-05,
      "loss": 0.0878,
      "step": 8432
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.635045126265123,
      "learning_rate": 2.4868546987789257e-05,
      "loss": 0.1303,
      "step": 8433
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3882791710433624,
      "learning_rate": 2.48635865450988e-05,
      "loss": 0.2947,
      "step": 8434
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.837333528180871,
      "learning_rate": 2.4858626107779038e-05,
      "loss": 0.3308,
      "step": 8435
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0815054213583668,
      "learning_rate": 2.485366567602527e-05,
      "loss": 0.3244,
      "step": 8436
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.872006886461692,
      "learning_rate": 2.484870525003278e-05,
      "loss": 0.1716,
      "step": 8437
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5582064720200476,
      "learning_rate": 2.4843744829996873e-05,
      "loss": 0.0687,
      "step": 8438
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6691134645319416,
      "learning_rate": 2.483878441611285e-05,
      "loss": 0.1185,
      "step": 8439
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7621909205155457,
      "learning_rate": 2.4833824008576e-05,
      "loss": 0.2511,
      "step": 8440
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9084160651841853,
      "learning_rate": 2.4828863607581616e-05,
      "loss": 0.1474,
      "step": 8441
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.536167844065945,
      "learning_rate": 2.4823903213324997e-05,
      "loss": 0.1087,
      "step": 8442
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3802519756371467,
      "learning_rate": 2.4818942826001433e-05,
      "loss": 0.0444,
      "step": 8443
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.2083930869897763,
      "learning_rate": 2.4813982445806217e-05,
      "loss": 0.4941,
      "step": 8444
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3798161764436334,
      "learning_rate": 2.4809022072934645e-05,
      "loss": 0.3096,
      "step": 8445
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6165362540923996,
      "learning_rate": 2.4804061707582015e-05,
      "loss": 0.1259,
      "step": 8446
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7293239115059841,
      "learning_rate": 2.479910134994362e-05,
      "loss": 0.2406,
      "step": 8447
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6192146995586628,
      "learning_rate": 2.4794141000214742e-05,
      "loss": 0.1168,
      "step": 8448
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6666071266882164,
      "learning_rate": 2.4789180658590673e-05,
      "loss": 0.13,
      "step": 8449
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8430217384270043,
      "learning_rate": 2.4784220325266724e-05,
      "loss": 0.1554,
      "step": 8450
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5360950379080553,
      "learning_rate": 2.4779260000438176e-05,
      "loss": 0.1199,
      "step": 8451
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6430697430771021,
      "learning_rate": 2.477429968430032e-05,
      "loss": 0.1416,
      "step": 8452
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3985613209965458,
      "learning_rate": 2.4769339377048433e-05,
      "loss": 0.0901,
      "step": 8453
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.589663922475676,
      "learning_rate": 2.4764379078877835e-05,
      "loss": 0.0959,
      "step": 8454
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.40472979942087556,
      "learning_rate": 2.47594187899838e-05,
      "loss": 0.0949,
      "step": 8455
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6168493338216152,
      "learning_rate": 2.4754458510561618e-05,
      "loss": 0.1288,
      "step": 8456
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7158487004311607,
      "learning_rate": 2.474949824080657e-05,
      "loss": 0.1494,
      "step": 8457
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.631077161627049,
      "learning_rate": 2.4744537980913966e-05,
      "loss": 0.1081,
      "step": 8458
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5006054086563296,
      "learning_rate": 2.473957773107909e-05,
      "loss": 0.0781,
      "step": 8459
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5777131131463225,
      "learning_rate": 2.473461749149722e-05,
      "loss": 0.1612,
      "step": 8460
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6755010377268612,
      "learning_rate": 2.472965726236364e-05,
      "loss": 0.1133,
      "step": 8461
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.550293429682713,
      "learning_rate": 2.4724697043873656e-05,
      "loss": 0.1027,
      "step": 8462
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5744179723800265,
      "learning_rate": 2.4719736836222552e-05,
      "loss": 0.0931,
      "step": 8463
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.577581624552997,
      "learning_rate": 2.4714776639605604e-05,
      "loss": 0.0888,
      "step": 8464
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5244656487201018,
      "learning_rate": 2.47098164542181e-05,
      "loss": 0.0511,
      "step": 8465
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0891720329597234,
      "learning_rate": 2.4704856280255332e-05,
      "loss": 0.4279,
      "step": 8466
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5973026495986186,
      "learning_rate": 2.469989611791259e-05,
      "loss": 0.1355,
      "step": 8467
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5359780158220152,
      "learning_rate": 2.4694935967385154e-05,
      "loss": 0.0756,
      "step": 8468
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.45859350223110207,
      "learning_rate": 2.46899758288683e-05,
      "loss": 0.1013,
      "step": 8469
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6803579673738424,
      "learning_rate": 2.468501570255732e-05,
      "loss": 0.0981,
      "step": 8470
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5306761276797487,
      "learning_rate": 2.4680055588647503e-05,
      "loss": 0.0627,
      "step": 8471
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6710378301663643,
      "learning_rate": 2.4675095487334128e-05,
      "loss": 0.0934,
      "step": 8472
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2417594167008492,
      "learning_rate": 2.4670135398812465e-05,
      "loss": 0.2949,
      "step": 8473
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7254444694549758,
      "learning_rate": 2.466517532327782e-05,
      "loss": 0.1616,
      "step": 8474
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8386267463935084,
      "learning_rate": 2.4660215260925453e-05,
      "loss": 0.2519,
      "step": 8475
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.618642057946041,
      "learning_rate": 2.465525521195066e-05,
      "loss": 0.112,
      "step": 8476
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.43138831299855357,
      "learning_rate": 2.4650295176548717e-05,
      "loss": 0.0421,
      "step": 8477
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9049496403512183,
      "learning_rate": 2.4645335154914904e-05,
      "loss": 0.1749,
      "step": 8478
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6228565677252561,
      "learning_rate": 2.46403751472445e-05,
      "loss": 0.1139,
      "step": 8479
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.420473393030871,
      "learning_rate": 2.4635415153732776e-05,
      "loss": 0.0602,
      "step": 8480
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5784164366754957,
      "learning_rate": 2.4630455174575026e-05,
      "loss": 0.1219,
      "step": 8481
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8302246431448823,
      "learning_rate": 2.4625495209966525e-05,
      "loss": 0.1569,
      "step": 8482
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9973876681682969,
      "learning_rate": 2.4620535260102542e-05,
      "loss": 0.4332,
      "step": 8483
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5307629863026958,
      "learning_rate": 2.4615575325178347e-05,
      "loss": 0.0843,
      "step": 8484
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6575243520586469,
      "learning_rate": 2.461061540538924e-05,
      "loss": 0.1219,
      "step": 8485
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9843445776912312,
      "learning_rate": 2.460565550093048e-05,
      "loss": 0.1409,
      "step": 8486
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6626136507330408,
      "learning_rate": 2.460069561199735e-05,
      "loss": 0.117,
      "step": 8487
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8893940917184894,
      "learning_rate": 2.459573573878511e-05,
      "loss": 0.1562,
      "step": 8488
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7859651047631663,
      "learning_rate": 2.459077588148905e-05,
      "loss": 0.11,
      "step": 8489
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4957604993549662,
      "learning_rate": 2.458581604030444e-05,
      "loss": 0.0962,
      "step": 8490
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5733084452034422,
      "learning_rate": 2.4580856215426547e-05,
      "loss": 0.0692,
      "step": 8491
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5739139923713282,
      "learning_rate": 2.4575896407050636e-05,
      "loss": 0.1299,
      "step": 8492
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4569433124931,
      "learning_rate": 2.4570936615371998e-05,
      "loss": 0.1467,
      "step": 8493
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5282219795812629,
      "learning_rate": 2.456597684058589e-05,
      "loss": 0.0827,
      "step": 8494
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8664672657689554,
      "learning_rate": 2.456101708288759e-05,
      "loss": 0.2023,
      "step": 8495
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7696290367003609,
      "learning_rate": 2.4556057342472347e-05,
      "loss": 0.157,
      "step": 8496
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.899437193327432,
      "learning_rate": 2.4551097619535457e-05,
      "loss": 0.2653,
      "step": 8497
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8676309735703228,
      "learning_rate": 2.4546137914272178e-05,
      "loss": 0.1816,
      "step": 8498
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5543902668448761,
      "learning_rate": 2.454117822687777e-05,
      "loss": 0.0696,
      "step": 8499
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6307320733270015,
      "learning_rate": 2.45362185575475e-05,
      "loss": 0.0558,
      "step": 8500
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5359261087319719,
      "learning_rate": 2.453125890647664e-05,
      "loss": 0.0788,
      "step": 8501
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.40902659525909124,
      "learning_rate": 2.4526299273860456e-05,
      "loss": 0.0812,
      "step": 8502
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6235426230615017,
      "learning_rate": 2.452133965989421e-05,
      "loss": 0.0938,
      "step": 8503
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1396785670793568,
      "learning_rate": 2.4516380064773158e-05,
      "loss": 0.2119,
      "step": 8504
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8275367759764367,
      "learning_rate": 2.4511420488692578e-05,
      "loss": 0.0797,
      "step": 8505
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6163965505404687,
      "learning_rate": 2.4506460931847712e-05,
      "loss": 0.0669,
      "step": 8506
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5781220525306751,
      "learning_rate": 2.4501501394433847e-05,
      "loss": 0.0793,
      "step": 8507
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6394718814883866,
      "learning_rate": 2.4496541876646216e-05,
      "loss": 0.0949,
      "step": 8508
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6623755913708701,
      "learning_rate": 2.44915823786801e-05,
      "loss": 0.0512,
      "step": 8509
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6387563951239469,
      "learning_rate": 2.448662290073075e-05,
      "loss": 0.0927,
      "step": 8510
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.47038486930550594,
      "learning_rate": 2.4481663442993426e-05,
      "loss": 0.0882,
      "step": 8511
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6740427982978238,
      "learning_rate": 2.4476704005663374e-05,
      "loss": 0.2162,
      "step": 8512
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5615597424518768,
      "learning_rate": 2.4471744588935867e-05,
      "loss": 0.1036,
      "step": 8513
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.48909346361834377,
      "learning_rate": 2.4466785193006157e-05,
      "loss": 0.0736,
      "step": 8514
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6261007873700025,
      "learning_rate": 2.4461825818069494e-05,
      "loss": 0.1108,
      "step": 8515
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7522338457497727,
      "learning_rate": 2.4456866464321127e-05,
      "loss": 0.1593,
      "step": 8516
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8322599630474375,
      "learning_rate": 2.4451907131956327e-05,
      "loss": 0.1538,
      "step": 8517
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.669888998957724,
      "learning_rate": 2.444694782117033e-05,
      "loss": 0.1962,
      "step": 8518
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7250817643860986,
      "learning_rate": 2.44419885321584e-05,
      "loss": 0.1998,
      "step": 8519
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5895503953906014,
      "learning_rate": 2.443702926511577e-05,
      "loss": 0.0745,
      "step": 8520
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0165646694087673,
      "learning_rate": 2.4432070020237708e-05,
      "loss": 0.2198,
      "step": 8521
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3390301221824916,
      "learning_rate": 2.442711079771946e-05,
      "loss": 0.6443,
      "step": 8522
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.936448427307795,
      "learning_rate": 2.4422151597756268e-05,
      "loss": 0.1375,
      "step": 8523
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5062509083563209,
      "learning_rate": 2.4417192420543373e-05,
      "loss": 0.1365,
      "step": 8524
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.42852849462365133,
      "learning_rate": 2.4412233266276042e-05,
      "loss": 0.0443,
      "step": 8525
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1728673468574276,
      "learning_rate": 2.4407274135149506e-05,
      "loss": 0.3837,
      "step": 8526
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.399525562225674,
      "learning_rate": 2.440231502735901e-05,
      "loss": 0.0519,
      "step": 8527
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8948873787792458,
      "learning_rate": 2.4397355943099794e-05,
      "loss": 0.1054,
      "step": 8528
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8689201800723555,
      "learning_rate": 2.4392396882567114e-05,
      "loss": 0.2928,
      "step": 8529
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7269787999341609,
      "learning_rate": 2.4387437845956203e-05,
      "loss": 0.12,
      "step": 8530
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7324701242142422,
      "learning_rate": 2.438247883346231e-05,
      "loss": 0.1545,
      "step": 8531
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7463940988967062,
      "learning_rate": 2.437751984528065e-05,
      "loss": 0.1357,
      "step": 8532
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0207883829090456,
      "learning_rate": 2.437256088160649e-05,
      "loss": 0.214,
      "step": 8533
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6903426273840682,
      "learning_rate": 2.4367601942635062e-05,
      "loss": 0.0847,
      "step": 8534
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5586025287822641,
      "learning_rate": 2.4362643028561596e-05,
      "loss": 0.1247,
      "step": 8535
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6015616276428079,
      "learning_rate": 2.4357684139581328e-05,
      "loss": 0.1098,
      "step": 8536
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.49887798622432744,
      "learning_rate": 2.43527252758895e-05,
      "loss": 0.0604,
      "step": 8537
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7180197065851757,
      "learning_rate": 2.4347766437681336e-05,
      "loss": 0.1762,
      "step": 8538
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7432292537267986,
      "learning_rate": 2.434280762515208e-05,
      "loss": 0.1359,
      "step": 8539
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.42266080002161494,
      "learning_rate": 2.4337848838496953e-05,
      "loss": 0.1025,
      "step": 8540
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6092307107307398,
      "learning_rate": 2.43328900779112e-05,
      "loss": 0.121,
      "step": 8541
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9194498709179741,
      "learning_rate": 2.432793134359004e-05,
      "loss": 0.1926,
      "step": 8542
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4613754488855518,
      "learning_rate": 2.43229726357287e-05,
      "loss": 0.0667,
      "step": 8543
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4297673887829559,
      "learning_rate": 2.4318013954522413e-05,
      "loss": 0.0982,
      "step": 8544
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6384376084273372,
      "learning_rate": 2.431305530016641e-05,
      "loss": 0.1084,
      "step": 8545
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5954303510938862,
      "learning_rate": 2.4308096672855913e-05,
      "loss": 0.0783,
      "step": 8546
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7965546689449997,
      "learning_rate": 2.4303138072786143e-05,
      "loss": 0.1371,
      "step": 8547
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.42997004856994114,
      "learning_rate": 2.4298179500152318e-05,
      "loss": 0.0802,
      "step": 8548
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4325760510890893,
      "learning_rate": 2.4293220955149677e-05,
      "loss": 0.4971,
      "step": 8549
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8528932494092084,
      "learning_rate": 2.4288262437973437e-05,
      "loss": 0.1584,
      "step": 8550
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5826527936629916,
      "learning_rate": 2.4283303948818816e-05,
      "loss": 0.0896,
      "step": 8551
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8245422176009222,
      "learning_rate": 2.427834548788102e-05,
      "loss": 0.139,
      "step": 8552
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5322120110077734,
      "learning_rate": 2.427338705535529e-05,
      "loss": 0.0752,
      "step": 8553
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7667084999693493,
      "learning_rate": 2.426842865143683e-05,
      "loss": 0.1256,
      "step": 8554
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5327515563765198,
      "learning_rate": 2.4263470276320864e-05,
      "loss": 0.0464,
      "step": 8555
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.443248272629009,
      "learning_rate": 2.4258511930202585e-05,
      "loss": 0.0728,
      "step": 8556
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8167562751798564,
      "learning_rate": 2.4253553613277237e-05,
      "loss": 0.2211,
      "step": 8557
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7116859934731754,
      "learning_rate": 2.4248595325740015e-05,
      "loss": 0.1145,
      "step": 8558
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.348109521981023,
      "learning_rate": 2.4243637067786133e-05,
      "loss": 0.2603,
      "step": 8559
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6287220587383993,
      "learning_rate": 2.4238678839610793e-05,
      "loss": 0.12,
      "step": 8560
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5762694918657401,
      "learning_rate": 2.4233720641409224e-05,
      "loss": 0.0864,
      "step": 8561
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.983561428720592,
      "learning_rate": 2.422876247337662e-05,
      "loss": 0.3672,
      "step": 8562
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9116804482989272,
      "learning_rate": 2.4223804335708194e-05,
      "loss": 0.3475,
      "step": 8563
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5963021820288038,
      "learning_rate": 2.4218846228599134e-05,
      "loss": 0.1384,
      "step": 8564
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4690711706842603,
      "learning_rate": 2.421388815224467e-05,
      "loss": 0.1121,
      "step": 8565
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5116654293969283,
      "learning_rate": 2.4208930106839993e-05,
      "loss": 0.1499,
      "step": 8566
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8237004155872507,
      "learning_rate": 2.4203972092580305e-05,
      "loss": 0.1677,
      "step": 8567
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5994394226854323,
      "learning_rate": 2.4199014109660797e-05,
      "loss": 0.1861,
      "step": 8568
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6695862262837903,
      "learning_rate": 2.419405615827668e-05,
      "loss": 0.128,
      "step": 8569
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4971841746812693,
      "learning_rate": 2.4189098238623157e-05,
      "loss": 0.0985,
      "step": 8570
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6932334415413371,
      "learning_rate": 2.4184140350895417e-05,
      "loss": 0.2279,
      "step": 8571
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.075355489390726,
      "learning_rate": 2.4179182495288646e-05,
      "loss": 0.3852,
      "step": 8572
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5819430435981484,
      "learning_rate": 2.417422467199806e-05,
      "loss": 0.1592,
      "step": 8573
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.42789158083390066,
      "learning_rate": 2.4169266881218834e-05,
      "loss": 0.062,
      "step": 8574
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6490034509422876,
      "learning_rate": 2.4164309123146166e-05,
      "loss": 0.1078,
      "step": 8575
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5835335210634486,
      "learning_rate": 2.4159351397975246e-05,
      "loss": 0.1145,
      "step": 8576
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.896819012418824,
      "learning_rate": 2.4154393705901267e-05,
      "loss": 0.4839,
      "step": 8577
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6410487087976486,
      "learning_rate": 2.4149436047119418e-05,
      "loss": 0.1694,
      "step": 8578
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5309516226745788,
      "learning_rate": 2.414447842182488e-05,
      "loss": 0.075,
      "step": 8579
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.555556497335285,
      "learning_rate": 2.4139520830212824e-05,
      "loss": 0.1039,
      "step": 8580
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5264780135495079,
      "learning_rate": 2.4134563272478468e-05,
      "loss": 0.0866,
      "step": 8581
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5136540922672944,
      "learning_rate": 2.412960574881697e-05,
      "loss": 0.0886,
      "step": 8582
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8612480461702616,
      "learning_rate": 2.4124648259423522e-05,
      "loss": 0.188,
      "step": 8583
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5548317455648046,
      "learning_rate": 2.4119690804493284e-05,
      "loss": 0.0753,
      "step": 8584
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7644757987838167,
      "learning_rate": 2.4114733384221465e-05,
      "loss": 0.1833,
      "step": 8585
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8654037661609156,
      "learning_rate": 2.4109775998803226e-05,
      "loss": 0.1282,
      "step": 8586
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.45359547561398583,
      "learning_rate": 2.4104818648433743e-05,
      "loss": 0.0887,
      "step": 8587
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5172576079744637,
      "learning_rate": 2.4099861333308176e-05,
      "loss": 0.2489,
      "step": 8588
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7430747254626199,
      "learning_rate": 2.4094904053621732e-05,
      "loss": 0.2072,
      "step": 8589
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5837339956865254,
      "learning_rate": 2.408994680956956e-05,
      "loss": 0.147,
      "step": 8590
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6094579085204404,
      "learning_rate": 2.4084989601346837e-05,
      "loss": 0.1149,
      "step": 8591
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5117598949046618,
      "learning_rate": 2.4080032429148715e-05,
      "loss": 0.0739,
      "step": 8592
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.519418182660246,
      "learning_rate": 2.407507529317039e-05,
      "loss": 0.0885,
      "step": 8593
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6871125168860284,
      "learning_rate": 2.407011819360701e-05,
      "loss": 0.1652,
      "step": 8594
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1062481369772468,
      "learning_rate": 2.4065161130653742e-05,
      "loss": 0.2304,
      "step": 8595
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0853165916394576,
      "learning_rate": 2.4060204104505747e-05,
      "loss": 0.2717,
      "step": 8596
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5694823459730781,
      "learning_rate": 2.405524711535819e-05,
      "loss": 0.1165,
      "step": 8597
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8632513055510025,
      "learning_rate": 2.405029016340624e-05,
      "loss": 0.1595,
      "step": 8598
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9234260127808973,
      "learning_rate": 2.4045333248845044e-05,
      "loss": 0.1213,
      "step": 8599
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.706895602726811,
      "learning_rate": 2.4040376371869753e-05,
      "loss": 0.1436,
      "step": 8600
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.41497357182305106,
      "learning_rate": 2.403541953267554e-05,
      "loss": 0.0606,
      "step": 8601
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6041119746626358,
      "learning_rate": 2.4030462731457554e-05,
      "loss": 0.0922,
      "step": 8602
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7193624023904226,
      "learning_rate": 2.402550596841094e-05,
      "loss": 0.161,
      "step": 8603
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7656630152167494,
      "learning_rate": 2.4020549243730856e-05,
      "loss": 0.1503,
      "step": 8604
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0510298778672127,
      "learning_rate": 2.401559255761245e-05,
      "loss": 0.3967,
      "step": 8605
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6613490717310452,
      "learning_rate": 2.4010635910250868e-05,
      "loss": 0.1341,
      "step": 8606
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5454978730275932,
      "learning_rate": 2.4005679301841257e-05,
      "loss": 0.1207,
      "step": 8607
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.832083018494014,
      "learning_rate": 2.4000722732578775e-05,
      "loss": 0.1734,
      "step": 8608
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6114518966006564,
      "learning_rate": 2.3995766202658547e-05,
      "loss": 0.0963,
      "step": 8609
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6324260216659121,
      "learning_rate": 2.399080971227573e-05,
      "loss": 0.0611,
      "step": 8610
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6093660310469409,
      "learning_rate": 2.3985853261625447e-05,
      "loss": 0.0871,
      "step": 8611
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6714173866896858,
      "learning_rate": 2.3980896850902856e-05,
      "loss": 0.1268,
      "step": 8612
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8242719765248918,
      "learning_rate": 2.3975940480303085e-05,
      "loss": 0.2089,
      "step": 8613
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8881975557848746,
      "learning_rate": 2.3970984150021276e-05,
      "loss": 0.2991,
      "step": 8614
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6936863491525599,
      "learning_rate": 2.3966027860252544e-05,
      "loss": 0.17,
      "step": 8615
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6093923640120433,
      "learning_rate": 2.396107161119205e-05,
      "loss": 0.1127,
      "step": 8616
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7395677517962069,
      "learning_rate": 2.3956115403034906e-05,
      "loss": 0.1207,
      "step": 8617
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5618127902801133,
      "learning_rate": 2.395115923597625e-05,
      "loss": 0.1434,
      "step": 8618
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5421731997332959,
      "learning_rate": 2.3946203110211197e-05,
      "loss": 0.0649,
      "step": 8619
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.45886976351448794,
      "learning_rate": 2.394124702593489e-05,
      "loss": 0.0746,
      "step": 8620
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3837726817753286,
      "learning_rate": 2.393629098334245e-05,
      "loss": 0.5561,
      "step": 8621
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.428984263095936,
      "learning_rate": 2.3931334982628996e-05,
      "loss": 0.0576,
      "step": 8622
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7918524672112547,
      "learning_rate": 2.3926379023989638e-05,
      "loss": 0.1655,
      "step": 8623
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0146059447860905,
      "learning_rate": 2.3921423107619518e-05,
      "loss": 0.2001,
      "step": 8624
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5880422927348614,
      "learning_rate": 2.391646723371374e-05,
      "loss": 0.0844,
      "step": 8625
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1606150951164387,
      "learning_rate": 2.391151140246743e-05,
      "loss": 0.3907,
      "step": 8626
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.693316385379176,
      "learning_rate": 2.390655561407568e-05,
      "loss": 0.1442,
      "step": 8627
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.765433543048732,
      "learning_rate": 2.3901599868733632e-05,
      "loss": 0.1462,
      "step": 8628
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5115138239703533,
      "learning_rate": 2.3896644166636383e-05,
      "loss": 0.0851,
      "step": 8629
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5874423964512681,
      "learning_rate": 2.3891688507979046e-05,
      "loss": 0.0578,
      "step": 8630
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6782384624432649,
      "learning_rate": 2.388673289295672e-05,
      "loss": 0.1598,
      "step": 8631
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7577933625723847,
      "learning_rate": 2.3881777321764516e-05,
      "loss": 0.1046,
      "step": 8632
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5595575655819498,
      "learning_rate": 2.387682179459755e-05,
      "loss": 0.0801,
      "step": 8633
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5120926785411469,
      "learning_rate": 2.3871866311650914e-05,
      "loss": 0.0829,
      "step": 8634
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7504009526585078,
      "learning_rate": 2.3866910873119703e-05,
      "loss": 0.1551,
      "step": 8635
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5930463592779807,
      "learning_rate": 2.3861955479199027e-05,
      "loss": 0.1148,
      "step": 8636
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6899836345790383,
      "learning_rate": 2.3857000130083975e-05,
      "loss": 0.0827,
      "step": 8637
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4417388059636222,
      "learning_rate": 2.3852044825969655e-05,
      "loss": 0.0461,
      "step": 8638
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6880323070432149,
      "learning_rate": 2.3847089567051144e-05,
      "loss": 0.1076,
      "step": 8639
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7018651108091014,
      "learning_rate": 2.3842134353523552e-05,
      "loss": 0.1026,
      "step": 8640
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7583642330320146,
      "learning_rate": 2.383717918558196e-05,
      "loss": 0.1906,
      "step": 8641
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7292346476343394,
      "learning_rate": 2.3832224063421454e-05,
      "loss": 0.1892,
      "step": 8642
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5112069241295021,
      "learning_rate": 2.3827268987237114e-05,
      "loss": 0.0937,
      "step": 8643
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1073266749848334,
      "learning_rate": 2.3822313957224042e-05,
      "loss": 0.1073,
      "step": 8644
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6179652446169012,
      "learning_rate": 2.381735897357732e-05,
      "loss": 0.107,
      "step": 8645
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2102841795586028,
      "learning_rate": 2.381240403649202e-05,
      "loss": 0.3569,
      "step": 8646
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7239637487163129,
      "learning_rate": 2.380744914616321e-05,
      "loss": 0.1864,
      "step": 8647
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.412266470767118,
      "learning_rate": 2.3802494302785996e-05,
      "loss": 0.0439,
      "step": 8648
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7288534700746767,
      "learning_rate": 2.3797539506555437e-05,
      "loss": 0.7166,
      "step": 8649
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.079758544156146,
      "learning_rate": 2.3792584757666614e-05,
      "loss": 0.2156,
      "step": 8650
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6435846666233827,
      "learning_rate": 2.378763005631458e-05,
      "loss": 0.2165,
      "step": 8651
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5855764815597256,
      "learning_rate": 2.3782675402694432e-05,
      "loss": 0.0982,
      "step": 8652
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8914756153460568,
      "learning_rate": 2.3777720797001224e-05,
      "loss": 0.2742,
      "step": 8653
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.43684385547708204,
      "learning_rate": 2.3772766239430028e-05,
      "loss": 0.0658,
      "step": 8654
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5352849760805137,
      "learning_rate": 2.376781173017589e-05,
      "loss": 0.0597,
      "step": 8655
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6971569220485483,
      "learning_rate": 2.37628572694339e-05,
      "loss": 0.4707,
      "step": 8656
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7686891736751938,
      "learning_rate": 2.3757902857399103e-05,
      "loss": 0.1681,
      "step": 8657
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6466341214075426,
      "learning_rate": 2.3752948494266563e-05,
      "loss": 0.133,
      "step": 8658
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7514866339871132,
      "learning_rate": 2.3747994180231325e-05,
      "loss": 0.1537,
      "step": 8659
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5002998343228288,
      "learning_rate": 2.3743039915488462e-05,
      "loss": 0.0635,
      "step": 8660
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6755706531430578,
      "learning_rate": 2.373808570023302e-05,
      "loss": 0.1057,
      "step": 8661
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6650091863298516,
      "learning_rate": 2.3733131534660047e-05,
      "loss": 0.0917,
      "step": 8662
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.670817427700493,
      "learning_rate": 2.3728177418964584e-05,
      "loss": 0.1184,
      "step": 8663
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5044123308222256,
      "learning_rate": 2.3723223353341698e-05,
      "loss": 0.0704,
      "step": 8664
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.394209702853852,
      "learning_rate": 2.3718269337986422e-05,
      "loss": 0.3888,
      "step": 8665
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7073188388024022,
      "learning_rate": 2.37133153730938e-05,
      "loss": 0.194,
      "step": 8666
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6042142823496321,
      "learning_rate": 2.370836145885887e-05,
      "loss": 0.1077,
      "step": 8667
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.451716035467913,
      "learning_rate": 2.370340759547668e-05,
      "loss": 0.0644,
      "step": 8668
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.41359707310943244,
      "learning_rate": 2.369845378314226e-05,
      "loss": 0.0453,
      "step": 8669
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6289333690593476,
      "learning_rate": 2.369350002205065e-05,
      "loss": 0.1555,
      "step": 8670
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6593786463481347,
      "learning_rate": 2.3688546312396875e-05,
      "loss": 0.1545,
      "step": 8671
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5322103158005196,
      "learning_rate": 2.3683592654375977e-05,
      "loss": 0.9205,
      "step": 8672
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7190110735009281,
      "learning_rate": 2.3678639048182982e-05,
      "loss": 0.098,
      "step": 8673
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8336065891827205,
      "learning_rate": 2.3673685494012907e-05,
      "loss": 0.1587,
      "step": 8674
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.49622774597967373,
      "learning_rate": 2.3668731992060784e-05,
      "loss": 0.0837,
      "step": 8675
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7039840532755558,
      "learning_rate": 2.3663778542521643e-05,
      "loss": 0.1234,
      "step": 8676
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9740676821740639,
      "learning_rate": 2.36588251455905e-05,
      "loss": 0.2204,
      "step": 8677
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.45943912163646106,
      "learning_rate": 2.365387180146237e-05,
      "loss": 0.0618,
      "step": 8678
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5647078089380406,
      "learning_rate": 2.364891851033226e-05,
      "loss": 0.0802,
      "step": 8679
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7469481321264754,
      "learning_rate": 2.3643965272395208e-05,
      "loss": 0.1996,
      "step": 8680
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7078207315595458,
      "learning_rate": 2.3639012087846215e-05,
      "loss": 0.2522,
      "step": 8681
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7500230318931457,
      "learning_rate": 2.3634058956880296e-05,
      "loss": 0.1,
      "step": 8682
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9275714207252195,
      "learning_rate": 2.362910587969244e-05,
      "loss": 0.3129,
      "step": 8683
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6898048796514085,
      "learning_rate": 2.3624152856477678e-05,
      "loss": 0.2627,
      "step": 8684
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5162783472935469,
      "learning_rate": 2.3619199887431004e-05,
      "loss": 0.1041,
      "step": 8685
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.645362098596266,
      "learning_rate": 2.3614246972747423e-05,
      "loss": 0.1807,
      "step": 8686
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9458234930454874,
      "learning_rate": 2.360929411262192e-05,
      "loss": 0.2231,
      "step": 8687
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.079391206177659,
      "learning_rate": 2.3604341307249515e-05,
      "loss": 0.2056,
      "step": 8688
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5889674022769057,
      "learning_rate": 2.3599388556825192e-05,
      "loss": 0.1024,
      "step": 8689
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5197147641336818,
      "learning_rate": 2.3594435861543944e-05,
      "loss": 0.1293,
      "step": 8690
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6106617880611425,
      "learning_rate": 2.3589483221600755e-05,
      "loss": 0.1257,
      "step": 8691
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0026829567772646,
      "learning_rate": 2.3584530637190634e-05,
      "loss": 0.3416,
      "step": 8692
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8355507293805096,
      "learning_rate": 2.3579578108508553e-05,
      "loss": 0.1436,
      "step": 8693
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5394135239118047,
      "learning_rate": 2.3574625635749504e-05,
      "loss": 0.1734,
      "step": 8694
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8833376926686534,
      "learning_rate": 2.3569673219108452e-05,
      "loss": 0.1769,
      "step": 8695
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5807453737293284,
      "learning_rate": 2.35647208587804e-05,
      "loss": 0.0892,
      "step": 8696
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.37813297555284475,
      "learning_rate": 2.3559768554960317e-05,
      "loss": 0.0586,
      "step": 8697
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5488528561011474,
      "learning_rate": 2.355481630784318e-05,
      "loss": 0.0873,
      "step": 8698
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7331122841080008,
      "learning_rate": 2.3549864117623953e-05,
      "loss": 0.1125,
      "step": 8699
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0930276674058583,
      "learning_rate": 2.3544911984497618e-05,
      "loss": 0.2418,
      "step": 8700
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5487382499557689,
      "learning_rate": 2.3539959908659144e-05,
      "loss": 0.101,
      "step": 8701
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0354924647935628,
      "learning_rate": 2.3535007890303497e-05,
      "loss": 0.3865,
      "step": 8702
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7375469109022235,
      "learning_rate": 2.3530055929625633e-05,
      "loss": 0.1336,
      "step": 8703
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6291111614403831,
      "learning_rate": 2.3525104026820523e-05,
      "loss": 0.1375,
      "step": 8704
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5085079685556774,
      "learning_rate": 2.3520152182083123e-05,
      "loss": 0.0745,
      "step": 8705
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1766071532764517,
      "learning_rate": 2.3515200395608396e-05,
      "loss": 0.2676,
      "step": 8706
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0630023720747506,
      "learning_rate": 2.351024866759129e-05,
      "loss": 0.5318,
      "step": 8707
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.43637938102782947,
      "learning_rate": 2.3505296998226766e-05,
      "loss": 0.0439,
      "step": 8708
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.57494404171094,
      "learning_rate": 2.3500345387709768e-05,
      "loss": 0.0886,
      "step": 8709
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.656392916455014,
      "learning_rate": 2.349539383623525e-05,
      "loss": 0.1327,
      "step": 8710
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4052509065518268,
      "learning_rate": 2.3490442343998138e-05,
      "loss": 0.047,
      "step": 8711
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6349683840530023,
      "learning_rate": 2.3485490911193413e-05,
      "loss": 0.2406,
      "step": 8712
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4352000974039625,
      "learning_rate": 2.348053953801599e-05,
      "loss": 0.0626,
      "step": 8713
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4262028133629148,
      "learning_rate": 2.3475588224660815e-05,
      "loss": 0.0452,
      "step": 8714
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7840222011629409,
      "learning_rate": 2.3470636971322815e-05,
      "loss": 0.2222,
      "step": 8715
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8791629534795019,
      "learning_rate": 2.346568577819694e-05,
      "loss": 0.2507,
      "step": 8716
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8104361591977122,
      "learning_rate": 2.346073464547812e-05,
      "loss": 0.1592,
      "step": 8717
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6118428680360914,
      "learning_rate": 2.345578357336128e-05,
      "loss": 0.1267,
      "step": 8718
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2906661548370595,
      "learning_rate": 2.3450832562041336e-05,
      "loss": 0.4234,
      "step": 8719
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8056594307359909,
      "learning_rate": 2.3445881611713235e-05,
      "loss": 0.2095,
      "step": 8720
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6535027686349979,
      "learning_rate": 2.3440930722571888e-05,
      "loss": 0.1093,
      "step": 8721
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6592519463074036,
      "learning_rate": 2.3435979894812216e-05,
      "loss": 0.1774,
      "step": 8722
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7851374754302416,
      "learning_rate": 2.3431029128629125e-05,
      "loss": 0.1935,
      "step": 8723
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6208879398943163,
      "learning_rate": 2.3426078424217555e-05,
      "loss": 0.1659,
      "step": 8724
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5845657045281805,
      "learning_rate": 2.3421127781772405e-05,
      "loss": 0.1334,
      "step": 8725
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1786466941684144,
      "learning_rate": 2.3416177201488584e-05,
      "loss": 0.3135,
      "step": 8726
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5724102700691054,
      "learning_rate": 2.341122668356099e-05,
      "loss": 0.1394,
      "step": 8727
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.45946078475888913,
      "learning_rate": 2.3406276228184555e-05,
      "loss": 0.0749,
      "step": 8728
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7235356570636052,
      "learning_rate": 2.340132583555416e-05,
      "loss": 0.1374,
      "step": 8729
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7770825794827024,
      "learning_rate": 2.3396375505864722e-05,
      "loss": 0.1064,
      "step": 8730
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8127741101140424,
      "learning_rate": 2.339142523931111e-05,
      "loss": 0.1813,
      "step": 8731
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8645093807865479,
      "learning_rate": 2.3386475036088254e-05,
      "loss": 0.2325,
      "step": 8732
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3992918631721389,
      "learning_rate": 2.3381524896391033e-05,
      "loss": 0.0341,
      "step": 8733
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.49592715428637824,
      "learning_rate": 2.3376574820414335e-05,
      "loss": 0.0973,
      "step": 8734
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8754098245601728,
      "learning_rate": 2.3371624808353046e-05,
      "loss": 0.1548,
      "step": 8735
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8358839125038046,
      "learning_rate": 2.336667486040206e-05,
      "loss": 0.1687,
      "step": 8736
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7324221078503406,
      "learning_rate": 2.336172497675625e-05,
      "loss": 0.0533,
      "step": 8737
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5070027097462892,
      "learning_rate": 2.3356775157610505e-05,
      "loss": 0.064,
      "step": 8738
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.990153764157854,
      "learning_rate": 2.3351825403159698e-05,
      "loss": 0.2766,
      "step": 8739
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5766171313157213,
      "learning_rate": 2.334687571359871e-05,
      "loss": 0.1163,
      "step": 8740
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4808591846201932,
      "learning_rate": 2.334192608912241e-05,
      "loss": 0.0622,
      "step": 8741
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9865172200803096,
      "learning_rate": 2.3336976529925656e-05,
      "loss": 0.1528,
      "step": 8742
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4398641125532392,
      "learning_rate": 2.3332027036203337e-05,
      "loss": 0.0952,
      "step": 8743
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7364955904508401,
      "learning_rate": 2.3327077608150313e-05,
      "loss": 0.1147,
      "step": 8744
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6021176516890084,
      "learning_rate": 2.3322128245961442e-05,
      "loss": 0.1851,
      "step": 8745
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3661217172260547,
      "learning_rate": 2.3317178949831574e-05,
      "loss": 0.0793,
      "step": 8746
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.711868229437665,
      "learning_rate": 2.3312229719955585e-05,
      "loss": 0.166,
      "step": 8747
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.723352792689333,
      "learning_rate": 2.3307280556528322e-05,
      "loss": 0.1761,
      "step": 8748
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6004344925635808,
      "learning_rate": 2.330233145974464e-05,
      "loss": 0.1867,
      "step": 8749
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.147322871400086,
      "learning_rate": 2.329738242979937e-05,
      "loss": 0.403,
      "step": 8750
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8193549624480757,
      "learning_rate": 2.3292433466887387e-05,
      "loss": 0.2186,
      "step": 8751
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6785134716592887,
      "learning_rate": 2.3287484571203522e-05,
      "loss": 0.148,
      "step": 8752
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8410935905282542,
      "learning_rate": 2.3282535742942615e-05,
      "loss": 0.2966,
      "step": 8753
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0778151843642116,
      "learning_rate": 2.3277586982299498e-05,
      "loss": 0.1733,
      "step": 8754
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5975506963322809,
      "learning_rate": 2.3272638289469026e-05,
      "loss": 0.0887,
      "step": 8755
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6156615965180415,
      "learning_rate": 2.3267689664646023e-05,
      "loss": 0.1344,
      "step": 8756
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5771359017598789,
      "learning_rate": 2.326274110802532e-05,
      "loss": 0.0757,
      "step": 8757
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5957209137311668,
      "learning_rate": 2.325779261980173e-05,
      "loss": 0.0789,
      "step": 8758
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5254328805451925,
      "learning_rate": 2.3252844200170107e-05,
      "loss": 0.0852,
      "step": 8759
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6929414895139387,
      "learning_rate": 2.324789584932526e-05,
      "loss": 0.1224,
      "step": 8760
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5408910854063036,
      "learning_rate": 2.3242947567462007e-05,
      "loss": 0.0951,
      "step": 8761
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4307823942109995,
      "learning_rate": 2.323799935477516e-05,
      "loss": 0.0601,
      "step": 8762
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.712309965343077,
      "learning_rate": 2.323305121145955e-05,
      "loss": 0.1511,
      "step": 8763
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7298121962350592,
      "learning_rate": 2.3228103137709983e-05,
      "loss": 0.1742,
      "step": 8764
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8163702526511644,
      "learning_rate": 2.3223155133721263e-05,
      "loss": 0.1434,
      "step": 8765
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4571822572216857,
      "learning_rate": 2.321820719968819e-05,
      "loss": 0.0797,
      "step": 8766
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.590239178164799,
      "learning_rate": 2.3213259335805586e-05,
      "loss": 0.1177,
      "step": 8767
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5145688164445599,
      "learning_rate": 2.320831154226824e-05,
      "loss": 0.118,
      "step": 8768
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9630515110741547,
      "learning_rate": 2.3203363819270952e-05,
      "loss": 0.2317,
      "step": 8769
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6580863553562769,
      "learning_rate": 2.3198416167008516e-05,
      "loss": 0.1187,
      "step": 8770
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7743225718845842,
      "learning_rate": 2.319346858567573e-05,
      "loss": 0.1914,
      "step": 8771
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8119187820445327,
      "learning_rate": 2.318852107546738e-05,
      "loss": 0.1584,
      "step": 8772
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6481866260681833,
      "learning_rate": 2.3183573636578252e-05,
      "loss": 0.0967,
      "step": 8773
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5192115137354842,
      "learning_rate": 2.317862626920313e-05,
      "loss": 0.0501,
      "step": 8774
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6764697429339033,
      "learning_rate": 2.31736789735368e-05,
      "loss": 0.1443,
      "step": 8775
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6156877115733388,
      "learning_rate": 2.316873174977404e-05,
      "loss": 0.0755,
      "step": 8776
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5542152324101187,
      "learning_rate": 2.316378459810962e-05,
      "loss": 0.1551,
      "step": 8777
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.529820878491784,
      "learning_rate": 2.3158837518738312e-05,
      "loss": 0.1677,
      "step": 8778
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.772664140576744,
      "learning_rate": 2.3153890511854896e-05,
      "loss": 0.2147,
      "step": 8779
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5156997088896998,
      "learning_rate": 2.3148943577654138e-05,
      "loss": 0.1251,
      "step": 8780
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6744103663686905,
      "learning_rate": 2.3143996716330796e-05,
      "loss": 0.1246,
      "step": 8781
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5670660313601978,
      "learning_rate": 2.313904992807962e-05,
      "loss": 0.1924,
      "step": 8782
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6116709593298286,
      "learning_rate": 2.3134103213095394e-05,
      "loss": 0.0774,
      "step": 8783
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6751228560906974,
      "learning_rate": 2.3129156571572864e-05,
      "loss": 0.1173,
      "step": 8784
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6586501918502782,
      "learning_rate": 2.3124210003706777e-05,
      "loss": 0.0766,
      "step": 8785
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.701800338823129,
      "learning_rate": 2.311926350969188e-05,
      "loss": 0.1893,
      "step": 8786
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4234885431237017,
      "learning_rate": 2.3114317089722934e-05,
      "loss": 0.0519,
      "step": 8787
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7801437785387297,
      "learning_rate": 2.3109370743994676e-05,
      "loss": 0.1968,
      "step": 8788
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5913658014874735,
      "learning_rate": 2.3104424472701848e-05,
      "loss": 0.07,
      "step": 8789
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.874019590356178,
      "learning_rate": 2.309947827603918e-05,
      "loss": 0.3593,
      "step": 8790
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6647423369535235,
      "learning_rate": 2.3094532154201418e-05,
      "loss": 0.1598,
      "step": 8791
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8572607625603759,
      "learning_rate": 2.3089586107383295e-05,
      "loss": 0.3424,
      "step": 8792
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.575876440271669,
      "learning_rate": 2.3084640135779532e-05,
      "loss": 0.1225,
      "step": 8793
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6436456875099069,
      "learning_rate": 2.307969423958485e-05,
      "loss": 0.202,
      "step": 8794
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5959299834322788,
      "learning_rate": 2.3074748418993998e-05,
      "loss": 0.0444,
      "step": 8795
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.48461623105068047,
      "learning_rate": 2.3069802674201674e-05,
      "loss": 0.0536,
      "step": 8796
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0470854000469823,
      "learning_rate": 2.3064857005402606e-05,
      "loss": 0.2512,
      "step": 8797
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5374734354500498,
      "learning_rate": 2.3059911412791498e-05,
      "loss": 0.1013,
      "step": 8798
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5524818508386317,
      "learning_rate": 2.3054965896563066e-05,
      "loss": 0.1116,
      "step": 8799
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8605314196572287,
      "learning_rate": 2.3050020456912028e-05,
      "loss": 0.2083,
      "step": 8800
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9954771154628693,
      "learning_rate": 2.3045075094033084e-05,
      "loss": 0.3418,
      "step": 8801
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.67258402875815,
      "learning_rate": 2.3040129808120924e-05,
      "loss": 0.1374,
      "step": 8802
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.49224915474848446,
      "learning_rate": 2.303518459937027e-05,
      "loss": 0.0831,
      "step": 8803
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9346784317871344,
      "learning_rate": 2.3030239467975802e-05,
      "loss": 0.2581,
      "step": 8804
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5135522984381798,
      "learning_rate": 2.302529441413221e-05,
      "loss": 0.0867,
      "step": 8805
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5248114048228695,
      "learning_rate": 2.30203494380342e-05,
      "loss": 0.1013,
      "step": 8806
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5632364446913215,
      "learning_rate": 2.301540453987645e-05,
      "loss": 0.1206,
      "step": 8807
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.667582416535721,
      "learning_rate": 2.301045971985365e-05,
      "loss": 0.1178,
      "step": 8808
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5985328539971844,
      "learning_rate": 2.3005514978160475e-05,
      "loss": 0.1082,
      "step": 8809
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3874280434850879,
      "learning_rate": 2.3000570314991593e-05,
      "loss": 0.4983,
      "step": 8810
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.780471833710151,
      "learning_rate": 2.2995625730541704e-05,
      "loss": 0.1414,
      "step": 8811
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6707861540147243,
      "learning_rate": 2.299068122500546e-05,
      "loss": 0.1453,
      "step": 8812
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5434664747039006,
      "learning_rate": 2.298573679857754e-05,
      "loss": 0.0578,
      "step": 8813
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5863979987324136,
      "learning_rate": 2.29807924514526e-05,
      "loss": 0.0651,
      "step": 8814
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7167815484493768,
      "learning_rate": 2.2975848183825312e-05,
      "loss": 0.1932,
      "step": 8815
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.689742313001637,
      "learning_rate": 2.2970903995890337e-05,
      "loss": 0.1039,
      "step": 8816
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2782153561143283,
      "learning_rate": 2.296595988784232e-05,
      "loss": 0.6472,
      "step": 8817
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5123223060426961,
      "learning_rate": 2.2961015859875915e-05,
      "loss": 0.0403,
      "step": 8818
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.687739007405965,
      "learning_rate": 2.2956071912185786e-05,
      "loss": 0.114,
      "step": 8819
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9668565069674552,
      "learning_rate": 2.2951128044966568e-05,
      "loss": 0.2453,
      "step": 8820
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3643903832397645,
      "learning_rate": 2.294618425841291e-05,
      "loss": 0.0386,
      "step": 8821
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4333168847428142,
      "learning_rate": 2.294124055271944e-05,
      "loss": 0.0422,
      "step": 8822
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8570793453598056,
      "learning_rate": 2.2936296928080815e-05,
      "loss": 0.5649,
      "step": 8823
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6632649030757803,
      "learning_rate": 2.2931353384691655e-05,
      "loss": 0.1612,
      "step": 8824
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5902407445059726,
      "learning_rate": 2.2926409922746596e-05,
      "loss": 0.0714,
      "step": 8825
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7378729971317856,
      "learning_rate": 2.2921466542440256e-05,
      "loss": 0.1829,
      "step": 8826
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6524906597927594,
      "learning_rate": 2.2916523243967277e-05,
      "loss": 0.1379,
      "step": 8827
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6364457960074154,
      "learning_rate": 2.2911580027522274e-05,
      "loss": 0.16,
      "step": 8828
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6481865063687633,
      "learning_rate": 2.290663689329986e-05,
      "loss": 0.0893,
      "step": 8829
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.019823806161871,
      "learning_rate": 2.2901693841494645e-05,
      "loss": 0.0737,
      "step": 8830
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5851396438890184,
      "learning_rate": 2.289675087230125e-05,
      "loss": 0.0985,
      "step": 8831
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4768504927447885,
      "learning_rate": 2.289180798591428e-05,
      "loss": 0.07,
      "step": 8832
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6347713191328387,
      "learning_rate": 2.288686518252835e-05,
      "loss": 0.1382,
      "step": 8833
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7193128849423698,
      "learning_rate": 2.288192246233804e-05,
      "loss": 0.126,
      "step": 8834
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7613829144084912,
      "learning_rate": 2.2876979825537966e-05,
      "loss": 0.1708,
      "step": 8835
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7120649126449092,
      "learning_rate": 2.2872037272322717e-05,
      "loss": 0.3339,
      "step": 8836
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6426418827968824,
      "learning_rate": 2.2867094802886888e-05,
      "loss": 0.1078,
      "step": 8837
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8704217708857269,
      "learning_rate": 2.2862152417425058e-05,
      "loss": 0.0708,
      "step": 8838
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0646743894475494,
      "learning_rate": 2.2857210116131828e-05,
      "loss": 0.1763,
      "step": 8839
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.43276265227507804,
      "learning_rate": 2.2852267899201772e-05,
      "loss": 0.1117,
      "step": 8840
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5453410670972001,
      "learning_rate": 2.2847325766829463e-05,
      "loss": 0.0737,
      "step": 8841
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9555441718315719,
      "learning_rate": 2.2842383719209475e-05,
      "loss": 0.3221,
      "step": 8842
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6627982811804545,
      "learning_rate": 2.2837441756536396e-05,
      "loss": 0.1712,
      "step": 8843
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9663985796156193,
      "learning_rate": 2.2832499879004784e-05,
      "loss": 0.2627,
      "step": 8844
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6640081750969729,
      "learning_rate": 2.2827558086809205e-05,
      "loss": 0.1344,
      "step": 8845
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5883123919884644,
      "learning_rate": 2.2822616380144214e-05,
      "loss": 0.1245,
      "step": 8846
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6429774941460609,
      "learning_rate": 2.281767475920438e-05,
      "loss": 0.1025,
      "step": 8847
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0485131219924468,
      "learning_rate": 2.2812733224184264e-05,
      "loss": 0.2894,
      "step": 8848
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4604401522797839,
      "learning_rate": 2.2807791775278402e-05,
      "loss": 0.0841,
      "step": 8849
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7622714738340559,
      "learning_rate": 2.280285041268134e-05,
      "loss": 0.1132,
      "step": 8850
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6071379932308155,
      "learning_rate": 2.2797909136587643e-05,
      "loss": 0.1092,
      "step": 8851
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5944558396557904,
      "learning_rate": 2.279296794719184e-05,
      "loss": 0.1606,
      "step": 8852
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.47565476969628784,
      "learning_rate": 2.2788026844688472e-05,
      "loss": 0.0959,
      "step": 8853
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5178472504210749,
      "learning_rate": 2.278308582927206e-05,
      "loss": 0.0714,
      "step": 8854
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0639628754371777,
      "learning_rate": 2.277814490113716e-05,
      "loss": 0.5141,
      "step": 8855
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9975155631123116,
      "learning_rate": 2.277320406047829e-05,
      "loss": 0.462,
      "step": 8856
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5127174896320463,
      "learning_rate": 2.2768263307489967e-05,
      "loss": 0.0861,
      "step": 8857
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5241909053449006,
      "learning_rate": 2.276332264236671e-05,
      "loss": 0.0958,
      "step": 8858
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3351280037180975,
      "learning_rate": 2.275838206530305e-05,
      "loss": 0.3243,
      "step": 8859
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5978404784229978,
      "learning_rate": 2.27534415764935e-05,
      "loss": 0.1039,
      "step": 8860
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5882634020191657,
      "learning_rate": 2.2748501176132563e-05,
      "loss": 0.1327,
      "step": 8861
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6818279455155014,
      "learning_rate": 2.274356086441474e-05,
      "loss": 0.1862,
      "step": 8862
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6604690308902776,
      "learning_rate": 2.273862064153455e-05,
      "loss": 0.1336,
      "step": 8863
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5358958900100115,
      "learning_rate": 2.2733680507686487e-05,
      "loss": 0.1544,
      "step": 8864
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5813163194743692,
      "learning_rate": 2.2728740463065046e-05,
      "loss": 0.0784,
      "step": 8865
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4467020091340732,
      "learning_rate": 2.2723800507864713e-05,
      "loss": 0.0718,
      "step": 8866
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8658628904249692,
      "learning_rate": 2.271886064227999e-05,
      "loss": 0.1422,
      "step": 8867
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5314502158851756,
      "learning_rate": 2.271392086650536e-05,
      "loss": 0.0785,
      "step": 8868
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.48990256679110283,
      "learning_rate": 2.270898118073531e-05,
      "loss": 0.0681,
      "step": 8869
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5566581104029015,
      "learning_rate": 2.2704041585164303e-05,
      "loss": 0.2273,
      "step": 8870
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9326208463287191,
      "learning_rate": 2.269910207998683e-05,
      "loss": 0.1894,
      "step": 8871
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4643870050541454,
      "learning_rate": 2.2694162665397357e-05,
      "loss": 0.0991,
      "step": 8872
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4710038912981735,
      "learning_rate": 2.268922334159034e-05,
      "loss": 0.0676,
      "step": 8873
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7373920945951434,
      "learning_rate": 2.268428410876027e-05,
      "loss": 0.137,
      "step": 8874
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.809563464274458,
      "learning_rate": 2.2679344967101594e-05,
      "loss": 0.5292,
      "step": 8875
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4684566306509342,
      "learning_rate": 2.2674405916808765e-05,
      "loss": 0.0798,
      "step": 8876
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7559273139281798,
      "learning_rate": 2.2669466958076237e-05,
      "loss": 0.1032,
      "step": 8877
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6656723484852258,
      "learning_rate": 2.266452809109847e-05,
      "loss": 0.2539,
      "step": 8878
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6597400274308132,
      "learning_rate": 2.2659589316069913e-05,
      "loss": 0.1237,
      "step": 8879
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5061018930601049,
      "learning_rate": 2.2654650633184997e-05,
      "loss": 0.0452,
      "step": 8880
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8997652964361116,
      "learning_rate": 2.264971204263816e-05,
      "loss": 0.1807,
      "step": 8881
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5180321183638469,
      "learning_rate": 2.264477354462385e-05,
      "loss": 0.0718,
      "step": 8882
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7162674750358233,
      "learning_rate": 2.263983513933649e-05,
      "loss": 0.4918,
      "step": 8883
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4495579342253155,
      "learning_rate": 2.2634896826970523e-05,
      "loss": 0.0671,
      "step": 8884
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.2464755021688485,
      "learning_rate": 2.2629958607720346e-05,
      "loss": 0.4822,
      "step": 8885
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.806805767203541,
      "learning_rate": 2.2625020481780406e-05,
      "loss": 0.2585,
      "step": 8886
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5351518658648642,
      "learning_rate": 2.2620082449345115e-05,
      "loss": 0.0748,
      "step": 8887
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7452047769325328,
      "learning_rate": 2.2615144510608886e-05,
      "loss": 0.1288,
      "step": 8888
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.757384193624074,
      "learning_rate": 2.261020666576611e-05,
      "loss": 0.3097,
      "step": 8889
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9231898495642613,
      "learning_rate": 2.2605268915011225e-05,
      "loss": 0.2976,
      "step": 8890
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6184426845508472,
      "learning_rate": 2.260033125853862e-05,
      "loss": 0.1431,
      "step": 8891
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.753386622965946,
      "learning_rate": 2.259539369654269e-05,
      "loss": 0.1634,
      "step": 8892
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.629466013256264,
      "learning_rate": 2.259045622921782e-05,
      "loss": 0.4091,
      "step": 8893
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6321667444767419,
      "learning_rate": 2.2585518856758432e-05,
      "loss": 0.1126,
      "step": 8894
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.41163422573719416,
      "learning_rate": 2.2580581579358894e-05,
      "loss": 0.0839,
      "step": 8895
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9376374614336636,
      "learning_rate": 2.2575644397213595e-05,
      "loss": 0.3213,
      "step": 8896
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5750772819961522,
      "learning_rate": 2.2570707310516905e-05,
      "loss": 0.1128,
      "step": 8897
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6334426763105532,
      "learning_rate": 2.256577031946322e-05,
      "loss": 0.0939,
      "step": 8898
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.529316139396474,
      "learning_rate": 2.256083342424689e-05,
      "loss": 0.3117,
      "step": 8899
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5351608020050492,
      "learning_rate": 2.2555896625062306e-05,
      "loss": 0.1035,
      "step": 8900
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5825845291230685,
      "learning_rate": 2.2550959922103816e-05,
      "loss": 0.0908,
      "step": 8901
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8280136432805469,
      "learning_rate": 2.2546023315565794e-05,
      "loss": 0.1364,
      "step": 8902
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5274420638224647,
      "learning_rate": 2.2541086805642596e-05,
      "loss": 0.0961,
      "step": 8903
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6333520954258676,
      "learning_rate": 2.2536150392528565e-05,
      "loss": 0.1038,
      "step": 8904
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6338540628436422,
      "learning_rate": 2.2531214076418063e-05,
      "loss": 0.138,
      "step": 8905
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2092568164491289,
      "learning_rate": 2.252627785750543e-05,
      "loss": 0.3181,
      "step": 8906
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6657311058078356,
      "learning_rate": 2.2521341735985016e-05,
      "loss": 0.131,
      "step": 8907
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6764199203713485,
      "learning_rate": 2.2516405712051153e-05,
      "loss": 0.107,
      "step": 8908
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.395565494972912,
      "learning_rate": 2.2511469785898166e-05,
      "loss": 0.3981,
      "step": 8909
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.137088491644469,
      "learning_rate": 2.2506533957720406e-05,
      "loss": 0.2676,
      "step": 8910
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9141292591185327,
      "learning_rate": 2.2501598227712197e-05,
      "loss": 0.1888,
      "step": 8911
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5342887271169284,
      "learning_rate": 2.249666259606785e-05,
      "loss": 0.0946,
      "step": 8912
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7108669855276873,
      "learning_rate": 2.249172706298168e-05,
      "loss": 0.1574,
      "step": 8913
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5443358714448491,
      "learning_rate": 2.2486791628648028e-05,
      "loss": 0.1145,
      "step": 8914
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.724924039556289,
      "learning_rate": 2.248185629326119e-05,
      "loss": 0.1521,
      "step": 8915
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8053298707386531,
      "learning_rate": 2.247692105701547e-05,
      "loss": 0.0955,
      "step": 8916
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4652882081810469,
      "learning_rate": 2.247198592010517e-05,
      "loss": 0.0725,
      "step": 8917
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9150248522007293,
      "learning_rate": 2.2467050882724605e-05,
      "loss": 0.1534,
      "step": 8918
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6416987070785332,
      "learning_rate": 2.2462115945068064e-05,
      "loss": 0.1872,
      "step": 8919
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4839995889237652,
      "learning_rate": 2.2457181107329832e-05,
      "loss": 0.0901,
      "step": 8920
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2224744265487313,
      "learning_rate": 2.2452246369704197e-05,
      "loss": 0.4807,
      "step": 8921
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5903525384832776,
      "learning_rate": 2.2447311732385457e-05,
      "loss": 0.1451,
      "step": 8922
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8439612060466056,
      "learning_rate": 2.244237719556788e-05,
      "loss": 0.1771,
      "step": 8923
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5764081240785789,
      "learning_rate": 2.2437442759445754e-05,
      "loss": 0.1383,
      "step": 8924
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6451295912297098,
      "learning_rate": 2.2432508424213325e-05,
      "loss": 0.1056,
      "step": 8925
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3280125325977807,
      "learning_rate": 2.2427574190064894e-05,
      "loss": 0.3597,
      "step": 8926
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7128675746846987,
      "learning_rate": 2.242264005719471e-05,
      "loss": 0.1369,
      "step": 8927
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8890380180550106,
      "learning_rate": 2.2417706025797032e-05,
      "loss": 0.1814,
      "step": 8928
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9950293521738777,
      "learning_rate": 2.2412772096066116e-05,
      "loss": 0.2414,
      "step": 8929
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9186308750492908,
      "learning_rate": 2.2407838268196215e-05,
      "loss": 0.1798,
      "step": 8930
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7583524517811082,
      "learning_rate": 2.2402904542381586e-05,
      "loss": 0.1004,
      "step": 8931
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6995947301322757,
      "learning_rate": 2.2397970918816465e-05,
      "loss": 0.1316,
      "step": 8932
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1531665110517935,
      "learning_rate": 2.239303739769509e-05,
      "loss": 0.2355,
      "step": 8933
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5156860579690993,
      "learning_rate": 2.23881039792117e-05,
      "loss": 0.0978,
      "step": 8934
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4773358819792393,
      "learning_rate": 2.2383170663560536e-05,
      "loss": 0.2456,
      "step": 8935
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7680618771588952,
      "learning_rate": 2.237823745093581e-05,
      "loss": 0.1602,
      "step": 8936
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5016960748923247,
      "learning_rate": 2.2373304341531757e-05,
      "loss": 0.0624,
      "step": 8937
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.43550766559864756,
      "learning_rate": 2.2368371335542596e-05,
      "loss": 0.0774,
      "step": 8938
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4878990776866296,
      "learning_rate": 2.2363438433162543e-05,
      "loss": 0.0912,
      "step": 8939
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7425191628866766,
      "learning_rate": 2.235850563458581e-05,
      "loss": 0.2104,
      "step": 8940
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.44311381512252207,
      "learning_rate": 2.235357294000659e-05,
      "loss": 0.0758,
      "step": 8941
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7940307389923058,
      "learning_rate": 2.234864034961911e-05,
      "loss": 0.0884,
      "step": 8942
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6891196689525406,
      "learning_rate": 2.2343707863617564e-05,
      "loss": 0.1331,
      "step": 8943
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7692428198796833,
      "learning_rate": 2.2338775482196143e-05,
      "loss": 0.2762,
      "step": 8944
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5951238909327358,
      "learning_rate": 2.2333843205549025e-05,
      "loss": 0.1217,
      "step": 8945
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3163216610831328,
      "learning_rate": 2.232891103387042e-05,
      "loss": 0.0488,
      "step": 8946
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6335343557987119,
      "learning_rate": 2.2323978967354504e-05,
      "loss": 0.1764,
      "step": 8947
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9528288049566338,
      "learning_rate": 2.2319047006195453e-05,
      "loss": 0.2303,
      "step": 8948
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7031054382105139,
      "learning_rate": 2.2314115150587433e-05,
      "loss": 0.1522,
      "step": 8949
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.49998149649173695,
      "learning_rate": 2.2309183400724636e-05,
      "loss": 0.0906,
      "step": 8950
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5163140375438674,
      "learning_rate": 2.2304251756801216e-05,
      "loss": 0.1257,
      "step": 8951
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6596867273212642,
      "learning_rate": 2.2299320219011336e-05,
      "loss": 0.1766,
      "step": 8952
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6276567801215504,
      "learning_rate": 2.2294388787549145e-05,
      "loss": 0.1317,
      "step": 8953
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9782519186613532,
      "learning_rate": 2.2289457462608816e-05,
      "loss": 0.1794,
      "step": 8954
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7800012056165895,
      "learning_rate": 2.2284526244384487e-05,
      "loss": 0.2135,
      "step": 8955
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.503436384595306,
      "learning_rate": 2.227959513307031e-05,
      "loss": 0.3926,
      "step": 8956
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8833198251905835,
      "learning_rate": 2.227466412886041e-05,
      "loss": 0.2118,
      "step": 8957
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5565357369745639,
      "learning_rate": 2.2269733231948946e-05,
      "loss": 0.1021,
      "step": 8958
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.346777011542814,
      "learning_rate": 2.2264802442530042e-05,
      "loss": 0.2754,
      "step": 8959
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7261897440798951,
      "learning_rate": 2.2259871760797828e-05,
      "loss": 0.2254,
      "step": 8960
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0909559382424432,
      "learning_rate": 2.2254941186946414e-05,
      "loss": 0.2466,
      "step": 8961
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4364150686816007,
      "learning_rate": 2.2250010721169946e-05,
      "loss": 0.0971,
      "step": 8962
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.34273375381122595,
      "learning_rate": 2.224508036366252e-05,
      "loss": 0.0723,
      "step": 8963
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7249757350328412,
      "learning_rate": 2.224015011461826e-05,
      "loss": 0.1134,
      "step": 8964
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.845317237223534,
      "learning_rate": 2.2235219974231264e-05,
      "loss": 0.3837,
      "step": 8965
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2322666814050858,
      "learning_rate": 2.2230289942695638e-05,
      "loss": 0.4091,
      "step": 8966
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9499665950396162,
      "learning_rate": 2.2225360020205483e-05,
      "loss": 0.169,
      "step": 8967
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8240236465406787,
      "learning_rate": 2.2220430206954896e-05,
      "loss": 0.1741,
      "step": 8968
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5771343833768,
      "learning_rate": 2.2215500503137958e-05,
      "loss": 0.1247,
      "step": 8969
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.602861741860221,
      "learning_rate": 2.2210570908948767e-05,
      "loss": 0.0829,
      "step": 8970
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5614951617551129,
      "learning_rate": 2.2205641424581396e-05,
      "loss": 0.0646,
      "step": 8971
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5379805661801036,
      "learning_rate": 2.2200712050229923e-05,
      "loss": 0.1078,
      "step": 8972
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4002052768436152,
      "learning_rate": 2.2195782786088418e-05,
      "loss": 0.0533,
      "step": 8973
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.741889138344351,
      "learning_rate": 2.2190853632350962e-05,
      "loss": 0.2261,
      "step": 8974
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.43355174377641986,
      "learning_rate": 2.218592458921161e-05,
      "loss": 0.0577,
      "step": 8975
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0492242754757808,
      "learning_rate": 2.2180995656864428e-05,
      "loss": 0.2643,
      "step": 8976
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.46653474046516485,
      "learning_rate": 2.2176066835503456e-05,
      "loss": 0.1136,
      "step": 8977
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5682622790372954,
      "learning_rate": 2.2171138125322766e-05,
      "loss": 0.1032,
      "step": 8978
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.765890671993709,
      "learning_rate": 2.2166209526516394e-05,
      "loss": 0.1032,
      "step": 8979
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4929327100190302,
      "learning_rate": 2.2161281039278384e-05,
      "loss": 0.0975,
      "step": 8980
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.732096136127355,
      "learning_rate": 2.2156352663802765e-05,
      "loss": 0.1154,
      "step": 8981
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5538408114230126,
      "learning_rate": 2.2151424400283588e-05,
      "loss": 0.0719,
      "step": 8982
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8714392322022941,
      "learning_rate": 2.2146496248914876e-05,
      "loss": 0.2651,
      "step": 8983
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.419631962153018,
      "learning_rate": 2.2141568209890652e-05,
      "loss": 0.6945,
      "step": 8984
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5399662063653893,
      "learning_rate": 2.2136640283404925e-05,
      "loss": 0.0851,
      "step": 8985
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4916053909536884,
      "learning_rate": 2.213171246965173e-05,
      "loss": 0.0804,
      "step": 8986
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.37306555435421773,
      "learning_rate": 2.2126784768825073e-05,
      "loss": 0.0507,
      "step": 8987
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6718977751751831,
      "learning_rate": 2.212185718111896e-05,
      "loss": 0.0865,
      "step": 8988
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0250834282776826,
      "learning_rate": 2.211692970672738e-05,
      "loss": 0.2778,
      "step": 8989
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7905861640107752,
      "learning_rate": 2.2112002345844355e-05,
      "loss": 0.1898,
      "step": 8990
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.572688527245844,
      "learning_rate": 2.2107075098663867e-05,
      "loss": 0.0789,
      "step": 8991
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7693549831426831,
      "learning_rate": 2.2102147965379904e-05,
      "loss": 0.1645,
      "step": 8992
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.40253063720676907,
      "learning_rate": 2.2097220946186445e-05,
      "loss": 0.0758,
      "step": 8993
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8728779839407853,
      "learning_rate": 2.2092294041277486e-05,
      "loss": 0.1866,
      "step": 8994
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4061703559510639,
      "learning_rate": 2.2087367250846996e-05,
      "loss": 0.0723,
      "step": 8995
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1753910861962562,
      "learning_rate": 2.2082440575088944e-05,
      "loss": 0.3109,
      "step": 8996
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6687414980774826,
      "learning_rate": 2.2077514014197294e-05,
      "loss": 0.1602,
      "step": 8997
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5479066149140019,
      "learning_rate": 2.207258756836601e-05,
      "loss": 0.1288,
      "step": 8998
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.76392665912714,
      "learning_rate": 2.2067661237789056e-05,
      "loss": 0.1023,
      "step": 8999
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6072879354513581,
      "learning_rate": 2.206273502266038e-05,
      "loss": 0.1014,
      "step": 9000
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2340415929144026,
      "learning_rate": 2.2057808923173927e-05,
      "loss": 0.2833,
      "step": 9001
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7681291296628919,
      "learning_rate": 2.2052882939523654e-05,
      "loss": 0.2725,
      "step": 9002
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5078076151977897,
      "learning_rate": 2.2047957071903487e-05,
      "loss": 0.0775,
      "step": 9003
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5231757933055354,
      "learning_rate": 2.2043031320507364e-05,
      "loss": 0.0905,
      "step": 9004
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8891000177400262,
      "learning_rate": 2.2038105685529214e-05,
      "loss": 0.1751,
      "step": 9005
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.395285290083332,
      "learning_rate": 2.203318016716297e-05,
      "loss": 0.1122,
      "step": 9006
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0493910230423793,
      "learning_rate": 2.2028254765602555e-05,
      "loss": 0.2221,
      "step": 9007
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5869092549724926,
      "learning_rate": 2.2023329481041865e-05,
      "loss": 0.0898,
      "step": 9008
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.441714895450545,
      "learning_rate": 2.201840431367484e-05,
      "loss": 0.0742,
      "step": 9009
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7658886625762291,
      "learning_rate": 2.2013479263695368e-05,
      "loss": 0.1863,
      "step": 9010
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6647181732123228,
      "learning_rate": 2.2008554331297366e-05,
      "loss": 0.1199,
      "step": 9011
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2715916958074067,
      "learning_rate": 2.2003629516674705e-05,
      "loss": 0.3032,
      "step": 9012
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9132728779540801,
      "learning_rate": 2.199870482002131e-05,
      "loss": 0.2334,
      "step": 9013
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7240037338148043,
      "learning_rate": 2.1993780241531063e-05,
      "loss": 0.139,
      "step": 9014
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.42910935366496655,
      "learning_rate": 2.1988855781397837e-05,
      "loss": 0.0714,
      "step": 9015
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5601635980918261,
      "learning_rate": 2.1983931439815508e-05,
      "loss": 0.118,
      "step": 9016
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4763711954600016,
      "learning_rate": 2.197900721697797e-05,
      "loss": 0.0874,
      "step": 9017
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5488656442694074,
      "learning_rate": 2.1974083113079082e-05,
      "loss": 0.106,
      "step": 9018
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6918433938828392,
      "learning_rate": 2.1969159128312712e-05,
      "loss": 0.1117,
      "step": 9019
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5119966110712438,
      "learning_rate": 2.196423526287271e-05,
      "loss": 0.1183,
      "step": 9020
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5455523908234707,
      "learning_rate": 2.1959311516952947e-05,
      "loss": 0.1024,
      "step": 9021
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1114990680594985,
      "learning_rate": 2.1954387890747273e-05,
      "loss": 0.5026,
      "step": 9022
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.49393319357567717,
      "learning_rate": 2.194946438444953e-05,
      "loss": 0.1435,
      "step": 9023
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6199442436262397,
      "learning_rate": 2.194454099825355e-05,
      "loss": 0.1817,
      "step": 9024
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5455929702578454,
      "learning_rate": 2.193961773235319e-05,
      "loss": 0.0991,
      "step": 9025
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6348675881912552,
      "learning_rate": 2.1934694586942274e-05,
      "loss": 0.1768,
      "step": 9026
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4787219089874145,
      "learning_rate": 2.1929771562214628e-05,
      "loss": 0.0673,
      "step": 9027
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.588706175168422,
      "learning_rate": 2.192484865836407e-05,
      "loss": 0.0767,
      "step": 9028
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4694237997261966,
      "learning_rate": 2.1919925875584427e-05,
      "loss": 0.0767,
      "step": 9029
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.7301726842568845,
      "learning_rate": 2.1915003214069513e-05,
      "loss": 0.4934,
      "step": 9030
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7582494915670168,
      "learning_rate": 2.1910080674013133e-05,
      "loss": 0.0832,
      "step": 9031
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7091209593342761,
      "learning_rate": 2.190515825560909e-05,
      "loss": 0.0891,
      "step": 9032
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.43274549342274765,
      "learning_rate": 2.190023595905118e-05,
      "loss": 0.0765,
      "step": 9033
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5229197741853774,
      "learning_rate": 2.189531378453321e-05,
      "loss": 0.0856,
      "step": 9034
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.49584098534649956,
      "learning_rate": 2.1890391732248956e-05,
      "loss": 0.1219,
      "step": 9035
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2470642627871191,
      "learning_rate": 2.1885469802392204e-05,
      "loss": 0.4931,
      "step": 9036
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7765692242377225,
      "learning_rate": 2.1880547995156747e-05,
      "loss": 0.1959,
      "step": 9037
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.34945958237186353,
      "learning_rate": 2.1875626310736353e-05,
      "loss": 0.0471,
      "step": 9038
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8407355163343222,
      "learning_rate": 2.1870704749324784e-05,
      "loss": 0.1777,
      "step": 9039
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5963855029637463,
      "learning_rate": 2.1865783311115805e-05,
      "loss": 0.1563,
      "step": 9040
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.48359425214392143,
      "learning_rate": 2.1860861996303195e-05,
      "loss": 0.0764,
      "step": 9041
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.535302055785791,
      "learning_rate": 2.1855940805080694e-05,
      "loss": 0.0934,
      "step": 9042
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8097087274745031,
      "learning_rate": 2.1851019737642057e-05,
      "loss": 0.237,
      "step": 9043
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.43371113275005696,
      "learning_rate": 2.184609879418102e-05,
      "loss": 0.0455,
      "step": 9044
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5473067761856009,
      "learning_rate": 2.1841177974891348e-05,
      "loss": 0.0852,
      "step": 9045
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4999449889001032,
      "learning_rate": 2.183625727996676e-05,
      "loss": 0.106,
      "step": 9046
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8287951235087688,
      "learning_rate": 2.183133670960099e-05,
      "loss": 0.1175,
      "step": 9047
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6107530604723093,
      "learning_rate": 2.1826416263987757e-05,
      "loss": 0.1512,
      "step": 9048
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5894144519309271,
      "learning_rate": 2.1821495943320797e-05,
      "loss": 0.1344,
      "step": 9049
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5237057121045602,
      "learning_rate": 2.181657574779382e-05,
      "loss": 0.0896,
      "step": 9050
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5275141067397998,
      "learning_rate": 2.181165567760054e-05,
      "loss": 0.1134,
      "step": 9051
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6848575927709638,
      "learning_rate": 2.1806735732934646e-05,
      "loss": 0.1563,
      "step": 9052
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7009894128010611,
      "learning_rate": 2.180181591398987e-05,
      "loss": 0.1738,
      "step": 9053
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.526303360381945,
      "learning_rate": 2.1796896220959888e-05,
      "loss": 0.5742,
      "step": 9054
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8094555465412551,
      "learning_rate": 2.1791976654038403e-05,
      "loss": 0.1833,
      "step": 9055
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8853089861081584,
      "learning_rate": 2.1787057213419085e-05,
      "loss": 0.2212,
      "step": 9056
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8736832858562253,
      "learning_rate": 2.1782137899295636e-05,
      "loss": 0.1551,
      "step": 9057
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.520898414715857,
      "learning_rate": 2.1777218711861723e-05,
      "loss": 0.1016,
      "step": 9058
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5066657993447735,
      "learning_rate": 2.177229965131102e-05,
      "loss": 0.0782,
      "step": 9059
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8153369955760219,
      "learning_rate": 2.1767380717837187e-05,
      "loss": 0.1824,
      "step": 9060
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4605244781236428,
      "learning_rate": 2.1762461911633894e-05,
      "loss": 0.0849,
      "step": 9061
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4564906355472096,
      "learning_rate": 2.17575432328948e-05,
      "loss": 0.109,
      "step": 9062
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6235758123577461,
      "learning_rate": 2.1752624681813555e-05,
      "loss": 0.1396,
      "step": 9063
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0446289404878786,
      "learning_rate": 2.1747706258583795e-05,
      "loss": 0.2222,
      "step": 9064
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.48464242750502895,
      "learning_rate": 2.1742787963399176e-05,
      "loss": 0.1173,
      "step": 9065
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5836896239690583,
      "learning_rate": 2.1737869796453326e-05,
      "loss": 0.0778,
      "step": 9066
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4798745597686666,
      "learning_rate": 2.1732951757939883e-05,
      "loss": 0.0784,
      "step": 9067
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5638266745182091,
      "learning_rate": 2.1728033848052466e-05,
      "loss": 0.1665,
      "step": 9068
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5148568216416478,
      "learning_rate": 2.1723116066984706e-05,
      "loss": 0.1124,
      "step": 9069
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.47609133950372934,
      "learning_rate": 2.1718198414930215e-05,
      "loss": 0.08,
      "step": 9070
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5553126196406135,
      "learning_rate": 2.1713280892082604e-05,
      "loss": 0.0763,
      "step": 9071
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5532354500617906,
      "learning_rate": 2.170836349863547e-05,
      "loss": 0.086,
      "step": 9072
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.450983411588947,
      "learning_rate": 2.1703446234782438e-05,
      "loss": 0.0662,
      "step": 9073
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5871222069176317,
      "learning_rate": 2.1698529100717085e-05,
      "loss": 0.1822,
      "step": 9074
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7936747640236717,
      "learning_rate": 2.1693612096633013e-05,
      "loss": 0.2487,
      "step": 9075
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8185237144934735,
      "learning_rate": 2.1688695222723787e-05,
      "loss": 0.151,
      "step": 9076
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4179490614385906,
      "learning_rate": 2.1683778479183015e-05,
      "loss": 0.0554,
      "step": 9077
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.47716084957573895,
      "learning_rate": 2.167886186620426e-05,
      "loss": 0.1006,
      "step": 9078
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5858475055033445,
      "learning_rate": 2.1673945383981098e-05,
      "loss": 0.1163,
      "step": 9079
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4166818757354061,
      "learning_rate": 2.1669029032707074e-05,
      "loss": 0.0652,
      "step": 9080
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6893260341711178,
      "learning_rate": 2.1664112812575777e-05,
      "loss": 0.1297,
      "step": 9081
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7881836326264857,
      "learning_rate": 2.165919672378075e-05,
      "loss": 0.1178,
      "step": 9082
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5483962556501486,
      "learning_rate": 2.1654280766515543e-05,
      "loss": 0.087,
      "step": 9083
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4947801010946872,
      "learning_rate": 2.1649364940973686e-05,
      "loss": 0.0554,
      "step": 9084
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5011362216348005,
      "learning_rate": 2.1644449247348745e-05,
      "loss": 0.0714,
      "step": 9085
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7009877840047043,
      "learning_rate": 2.163953368583424e-05,
      "loss": 0.1823,
      "step": 9086
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7085593239056209,
      "learning_rate": 2.1634618256623706e-05,
      "loss": 0.1324,
      "step": 9087
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4205662448834807,
      "learning_rate": 2.162970295991065e-05,
      "loss": 0.0485,
      "step": 9088
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8943082052609275,
      "learning_rate": 2.1624787795888615e-05,
      "loss": 0.281,
      "step": 9089
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7925424541021172,
      "learning_rate": 2.1619872764751102e-05,
      "loss": 0.1278,
      "step": 9090
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7483149506101663,
      "learning_rate": 2.1614957866691628e-05,
      "loss": 0.1458,
      "step": 9091
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.49977318718232805,
      "learning_rate": 2.1610043101903667e-05,
      "loss": 0.0872,
      "step": 9092
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5243941246436629,
      "learning_rate": 2.1605128470580756e-05,
      "loss": 0.1242,
      "step": 9093
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9209038074313435,
      "learning_rate": 2.160021397291637e-05,
      "loss": 0.2421,
      "step": 9094
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6561325968877045,
      "learning_rate": 2.1595299609103995e-05,
      "loss": 0.0976,
      "step": 9095
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.36705278475203207,
      "learning_rate": 2.1590385379337107e-05,
      "loss": 0.0558,
      "step": 9096
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.633024429114126,
      "learning_rate": 2.15854712838092e-05,
      "loss": 0.1452,
      "step": 9097
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.200858926156274,
      "learning_rate": 2.158055732271373e-05,
      "loss": 0.229,
      "step": 9098
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.32149992208708,
      "learning_rate": 2.1575643496244167e-05,
      "loss": 0.2979,
      "step": 9099
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7008103556688753,
      "learning_rate": 2.1570729804593977e-05,
      "loss": 0.1105,
      "step": 9100
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6939730774514521,
      "learning_rate": 2.1565816247956617e-05,
      "loss": 0.1362,
      "step": 9101
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5746864218038285,
      "learning_rate": 2.156090282652553e-05,
      "loss": 0.1169,
      "step": 9102
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5836910453839113,
      "learning_rate": 2.1555989540494154e-05,
      "loss": 0.1228,
      "step": 9103
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6180169987919595,
      "learning_rate": 2.1551076390055943e-05,
      "loss": 0.0939,
      "step": 9104
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.657227185706698,
      "learning_rate": 2.1546163375404334e-05,
      "loss": 0.1077,
      "step": 9105
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5763576133791535,
      "learning_rate": 2.154125049673274e-05,
      "loss": 0.0954,
      "step": 9106
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3508127291887562,
      "learning_rate": 2.1536337754234596e-05,
      "loss": 0.4668,
      "step": 9107
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8824149833209028,
      "learning_rate": 2.153142514810331e-05,
      "loss": 0.1445,
      "step": 9108
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5444569634938757,
      "learning_rate": 2.152651267853231e-05,
      "loss": 0.0962,
      "step": 9109
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.097558561747826,
      "learning_rate": 2.1521600345714997e-05,
      "loss": 0.2526,
      "step": 9110
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5889781386714735,
      "learning_rate": 2.1516688149844767e-05,
      "loss": 0.0712,
      "step": 9111
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0682664579704642,
      "learning_rate": 2.1511776091115016e-05,
      "loss": 0.2055,
      "step": 9112
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9093326156617811,
      "learning_rate": 2.1506864169719145e-05,
      "loss": 0.219,
      "step": 9113
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3906127563712245,
      "learning_rate": 2.150195238585054e-05,
      "loss": 0.5783,
      "step": 9114
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.48682390418292243,
      "learning_rate": 2.1497040739702574e-05,
      "loss": 0.0658,
      "step": 9115
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9901890991546785,
      "learning_rate": 2.1492129231468612e-05,
      "loss": 0.232,
      "step": 9116
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5585867998689,
      "learning_rate": 2.148721786134205e-05,
      "loss": 0.0733,
      "step": 9117
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5566259300015542,
      "learning_rate": 2.1482306629516237e-05,
      "loss": 0.1236,
      "step": 9118
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4059807979088803,
      "learning_rate": 2.1477395536184536e-05,
      "loss": 0.0897,
      "step": 9119
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6680123431692515,
      "learning_rate": 2.1472484581540283e-05,
      "loss": 0.209,
      "step": 9120
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4396700088224096,
      "learning_rate": 2.1467573765776853e-05,
      "loss": 0.0745,
      "step": 9121
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5371062356787708,
      "learning_rate": 2.1462663089087574e-05,
      "loss": 0.131,
      "step": 9122
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8297787854633302,
      "learning_rate": 2.1457752551665784e-05,
      "loss": 0.1449,
      "step": 9123
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.698821352084891,
      "learning_rate": 2.1452842153704805e-05,
      "loss": 0.2269,
      "step": 9124
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.712807341009339,
      "learning_rate": 2.1447931895397983e-05,
      "loss": 0.1187,
      "step": 9125
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2907163665168302,
      "learning_rate": 2.1443021776938625e-05,
      "loss": 0.467,
      "step": 9126
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5913391690125941,
      "learning_rate": 2.1438111798520054e-05,
      "loss": 0.1095,
      "step": 9127
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6626532139409105,
      "learning_rate": 2.1433201960335567e-05,
      "loss": 0.1269,
      "step": 9128
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5716812454878307,
      "learning_rate": 2.1428292262578476e-05,
      "loss": 0.0791,
      "step": 9129
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9230812620167825,
      "learning_rate": 2.142338270544208e-05,
      "loss": 0.2447,
      "step": 9130
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6887031405586166,
      "learning_rate": 2.141847328911967e-05,
      "loss": 0.1466,
      "step": 9131
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5315733310957804,
      "learning_rate": 2.1413564013804532e-05,
      "loss": 0.1058,
      "step": 9132
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.39978391225174525,
      "learning_rate": 2.1408654879689955e-05,
      "loss": 0.0568,
      "step": 9133
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5895103600279289,
      "learning_rate": 2.1403745886969202e-05,
      "loss": 0.0977,
      "step": 9134
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.113035421035165,
      "learning_rate": 2.1398837035835555e-05,
      "loss": 0.402,
      "step": 9135
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5387294903878876,
      "learning_rate": 2.139392832648227e-05,
      "loss": 0.1309,
      "step": 9136
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5847728882855675,
      "learning_rate": 2.138901975910262e-05,
      "loss": 0.1262,
      "step": 9137
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.567627716666419,
      "learning_rate": 2.1384111333889845e-05,
      "loss": 0.1618,
      "step": 9138
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6349561083010193,
      "learning_rate": 2.137920305103719e-05,
      "loss": 0.1113,
      "step": 9139
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7678210010028231,
      "learning_rate": 2.1374294910737912e-05,
      "loss": 0.1455,
      "step": 9140
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.47462338098539314,
      "learning_rate": 2.136938691318525e-05,
      "loss": 0.029,
      "step": 9141
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7050758125097666,
      "learning_rate": 2.136447905857242e-05,
      "loss": 0.1129,
      "step": 9142
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6098192625746272,
      "learning_rate": 2.135957134709265e-05,
      "loss": 0.1133,
      "step": 9143
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8007746598324621,
      "learning_rate": 2.1354663778939172e-05,
      "loss": 0.2188,
      "step": 9144
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5920697274574978,
      "learning_rate": 2.1349756354305194e-05,
      "loss": 0.1137,
      "step": 9145
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3952988204340516,
      "learning_rate": 2.1344849073383923e-05,
      "loss": 0.0707,
      "step": 9146
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.874747727573324,
      "learning_rate": 2.1339941936368558e-05,
      "loss": 0.2585,
      "step": 9147
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.48874009357039155,
      "learning_rate": 2.1335034943452308e-05,
      "loss": 0.0845,
      "step": 9148
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8888328886908194,
      "learning_rate": 2.133012809482836e-05,
      "loss": 0.1713,
      "step": 9149
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0203619240736836,
      "learning_rate": 2.13252213906899e-05,
      "loss": 0.2132,
      "step": 9150
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3923549656798816,
      "learning_rate": 2.1320314831230096e-05,
      "loss": 0.0861,
      "step": 9151
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.129378395693772,
      "learning_rate": 2.1315408416642145e-05,
      "loss": 0.2549,
      "step": 9152
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4933698607331563,
      "learning_rate": 2.1310502147119205e-05,
      "loss": 0.1226,
      "step": 9153
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7386688578842117,
      "learning_rate": 2.130559602285444e-05,
      "loss": 0.1339,
      "step": 9154
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.47139486735206876,
      "learning_rate": 2.1300690044040994e-05,
      "loss": 0.0955,
      "step": 9155
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5907124388704036,
      "learning_rate": 2.1295784210872047e-05,
      "loss": 0.1211,
      "step": 9156
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8365435289857384,
      "learning_rate": 2.1290878523540727e-05,
      "loss": 0.2221,
      "step": 9157
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3978729369684134,
      "learning_rate": 2.128597298224018e-05,
      "loss": 0.0532,
      "step": 9158
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4236738725946729,
      "learning_rate": 2.1281067587163532e-05,
      "loss": 0.061,
      "step": 9159
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.44368026697735663,
      "learning_rate": 2.1276162338503918e-05,
      "loss": 0.0998,
      "step": 9160
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5497042711235861,
      "learning_rate": 2.127125723645447e-05,
      "loss": 0.1509,
      "step": 9161
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.45421863272354074,
      "learning_rate": 2.1266352281208297e-05,
      "loss": 0.083,
      "step": 9162
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9627711458903274,
      "learning_rate": 2.12614474729585e-05,
      "loss": 0.1302,
      "step": 9163
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2869958094349234,
      "learning_rate": 2.1256542811898204e-05,
      "loss": 0.2848,
      "step": 9164
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7273411660657201,
      "learning_rate": 2.12516382982205e-05,
      "loss": 0.1656,
      "step": 9165
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6716623382004054,
      "learning_rate": 2.1246733932118475e-05,
      "loss": 0.1604,
      "step": 9166
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6959748565657357,
      "learning_rate": 2.124182971378523e-05,
      "loss": 0.1415,
      "step": 9167
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.45579631275180416,
      "learning_rate": 2.1236925643413845e-05,
      "loss": 0.0889,
      "step": 9168
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.538047221098144,
      "learning_rate": 2.1232021721197394e-05,
      "loss": 0.084,
      "step": 9169
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0785913921193289,
      "learning_rate": 2.1227117947328948e-05,
      "loss": 0.2934,
      "step": 9170
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5236913409959225,
      "learning_rate": 2.122221432200156e-05,
      "loss": 0.0861,
      "step": 9171
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.49506307753589246,
      "learning_rate": 2.1217310845408313e-05,
      "loss": 0.1072,
      "step": 9172
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7127561423779096,
      "learning_rate": 2.121240751774225e-05,
      "loss": 0.1649,
      "step": 9173
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.976273245065477,
      "learning_rate": 2.120750433919642e-05,
      "loss": 0.2606,
      "step": 9174
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9197111671788685,
      "learning_rate": 2.120260130996385e-05,
      "loss": 0.1137,
      "step": 9175
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8018674912418103,
      "learning_rate": 2.11976984302376e-05,
      "loss": 0.1669,
      "step": 9176
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.36308955041151314,
      "learning_rate": 2.1192795700210683e-05,
      "loss": 0.0497,
      "step": 9177
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7117019082450796,
      "learning_rate": 2.118789312007613e-05,
      "loss": 0.1353,
      "step": 9178
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6231050973488539,
      "learning_rate": 2.1182990690026947e-05,
      "loss": 0.1796,
      "step": 9179
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5337921866231468,
      "learning_rate": 2.1178088410256166e-05,
      "loss": 0.0971,
      "step": 9180
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9312261559995478,
      "learning_rate": 2.1173186280956787e-05,
      "loss": 0.2792,
      "step": 9181
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8421340100794755,
      "learning_rate": 2.11682843023218e-05,
      "loss": 0.2452,
      "step": 9182
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4970053408431212,
      "learning_rate": 2.11633824745442e-05,
      "loss": 0.1281,
      "step": 9183
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4710936860652035,
      "learning_rate": 2.1158480797816992e-05,
      "loss": 0.0664,
      "step": 9184
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5119682857106153,
      "learning_rate": 2.1153579272333147e-05,
      "loss": 0.0648,
      "step": 9185
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7933333070057441,
      "learning_rate": 2.1148677898285644e-05,
      "loss": 0.0862,
      "step": 9186
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5847988978731291,
      "learning_rate": 2.114377667586744e-05,
      "loss": 0.1379,
      "step": 9187
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7019078617962842,
      "learning_rate": 2.1138875605271523e-05,
      "loss": 0.0842,
      "step": 9188
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1354209146768852,
      "learning_rate": 2.113397468669084e-05,
      "loss": 0.2702,
      "step": 9189
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8044629669556852,
      "learning_rate": 2.1129073920318346e-05,
      "loss": 0.1893,
      "step": 9190
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7700539846294457,
      "learning_rate": 2.1124173306346978e-05,
      "loss": 0.1161,
      "step": 9191
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3924832491777002,
      "learning_rate": 2.1119272844969686e-05,
      "loss": 0.4138,
      "step": 9192
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.43905024070377646,
      "learning_rate": 2.111437253637941e-05,
      "loss": 0.0599,
      "step": 9193
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4883218449817167,
      "learning_rate": 2.110947238076907e-05,
      "loss": 0.0909,
      "step": 9194
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.46215305762692144,
      "learning_rate": 2.1104572378331587e-05,
      "loss": 0.0834,
      "step": 9195
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6718645256688527,
      "learning_rate": 2.1099672529259888e-05,
      "loss": 0.1213,
      "step": 9196
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7169178220058893,
      "learning_rate": 2.1094772833746866e-05,
      "loss": 0.1909,
      "step": 9197
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.592929018834346,
      "learning_rate": 2.1089873291985446e-05,
      "loss": 0.1291,
      "step": 9198
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7645450935944822,
      "learning_rate": 2.1084973904168513e-05,
      "loss": 0.236,
      "step": 9199
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5980113884262984,
      "learning_rate": 2.1080074670488964e-05,
      "loss": 0.1266,
      "step": 9200
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5682739918624851,
      "learning_rate": 2.1075175591139685e-05,
      "loss": 0.0738,
      "step": 9201
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0979550688834705,
      "learning_rate": 2.107027666631356e-05,
      "loss": 0.3278,
      "step": 9202
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1901401388834028,
      "learning_rate": 2.1065377896203444e-05,
      "loss": 0.1574,
      "step": 9203
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6635582326997479,
      "learning_rate": 2.1060479281002234e-05,
      "loss": 0.1287,
      "step": 9204
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.449610742264756,
      "learning_rate": 2.1055580820902774e-05,
      "loss": 0.0525,
      "step": 9205
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5949976377071163,
      "learning_rate": 2.105068251609793e-05,
      "loss": 0.143,
      "step": 9206
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7787428801456326,
      "learning_rate": 2.104578436678053e-05,
      "loss": 0.1987,
      "step": 9207
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4508667015882324,
      "learning_rate": 2.1040886373143443e-05,
      "loss": 0.0734,
      "step": 9208
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5353992709068048,
      "learning_rate": 2.10359885353795e-05,
      "loss": 0.1089,
      "step": 9209
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.329377868875133,
      "learning_rate": 2.1031090853681528e-05,
      "loss": 0.4287,
      "step": 9210
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5026427839177525,
      "learning_rate": 2.1026193328242343e-05,
      "loss": 0.0693,
      "step": 9211
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8629841768482757,
      "learning_rate": 2.1021295959254785e-05,
      "loss": 0.1686,
      "step": 9212
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6362085424190999,
      "learning_rate": 2.101639874691166e-05,
      "loss": 0.2226,
      "step": 9213
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8439666358149817,
      "learning_rate": 2.1011501691405768e-05,
      "loss": 0.1844,
      "step": 9214
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5750702954807299,
      "learning_rate": 2.10066047929299e-05,
      "loss": 0.1304,
      "step": 9215
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.47045421687860217,
      "learning_rate": 2.1001708051676878e-05,
      "loss": 0.0957,
      "step": 9216
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5993451012909043,
      "learning_rate": 2.0996811467839474e-05,
      "loss": 0.1138,
      "step": 9217
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5978329953406629,
      "learning_rate": 2.0991915041610472e-05,
      "loss": 0.1502,
      "step": 9218
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5989495344626999,
      "learning_rate": 2.0987018773182638e-05,
      "loss": 0.0814,
      "step": 9219
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.36276629293598395,
      "learning_rate": 2.0982122662748756e-05,
      "loss": 0.0597,
      "step": 9220
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.530465131412269,
      "learning_rate": 2.097722671050159e-05,
      "loss": 0.0887,
      "step": 9221
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8665168499048641,
      "learning_rate": 2.097233091663389e-05,
      "loss": 0.2283,
      "step": 9222
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3977198447495741,
      "learning_rate": 2.0967435281338396e-05,
      "loss": 0.0568,
      "step": 9223
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4500630485476429,
      "learning_rate": 2.0962539804807875e-05,
      "loss": 0.0439,
      "step": 9224
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6350413902153287,
      "learning_rate": 2.095764448723506e-05,
      "loss": 0.0939,
      "step": 9225
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3834452881666244,
      "learning_rate": 2.0952749328812675e-05,
      "loss": 0.0604,
      "step": 9226
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4511926485588576,
      "learning_rate": 2.0947854329733447e-05,
      "loss": 0.0639,
      "step": 9227
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7869259259032179,
      "learning_rate": 2.0942959490190096e-05,
      "loss": 0.1529,
      "step": 9228
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6089598855579603,
      "learning_rate": 2.093806481037534e-05,
      "loss": 0.1255,
      "step": 9229
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6465239085307155,
      "learning_rate": 2.0933170290481888e-05,
      "loss": 0.131,
      "step": 9230
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8940714248961881,
      "learning_rate": 2.092827593070243e-05,
      "loss": 0.2974,
      "step": 9231
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4413242603229435,
      "learning_rate": 2.0923381731229667e-05,
      "loss": 0.0518,
      "step": 9232
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.45159359023525886,
      "learning_rate": 2.0918487692256295e-05,
      "loss": 0.075,
      "step": 9233
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8172523061803045,
      "learning_rate": 2.0913593813974976e-05,
      "loss": 0.1431,
      "step": 9234
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0192588132084757,
      "learning_rate": 2.09087000965784e-05,
      "loss": 0.2178,
      "step": 9235
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8929691369567273,
      "learning_rate": 2.090380654025924e-05,
      "loss": 0.2363,
      "step": 9236
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5546907307424677,
      "learning_rate": 2.089891314521015e-05,
      "loss": 0.1117,
      "step": 9237
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.709922625827422,
      "learning_rate": 2.089401991162379e-05,
      "loss": 0.0802,
      "step": 9238
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6169481316324817,
      "learning_rate": 2.0889126839692798e-05,
      "loss": 0.1841,
      "step": 9239
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6071394421159291,
      "learning_rate": 2.0884233929609836e-05,
      "loss": 0.1014,
      "step": 9240
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8192255524400835,
      "learning_rate": 2.0879341181567537e-05,
      "loss": 0.1209,
      "step": 9241
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7105279188329058,
      "learning_rate": 2.087444859575853e-05,
      "loss": 0.115,
      "step": 9242
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4465446968948694,
      "learning_rate": 2.0869556172375424e-05,
      "loss": 0.0786,
      "step": 9243
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6286921875787357,
      "learning_rate": 2.0864663911610864e-05,
      "loss": 0.6494,
      "step": 9244
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6441580913347458,
      "learning_rate": 2.085977181365745e-05,
      "loss": 0.1184,
      "step": 9245
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.540437105877476,
      "learning_rate": 2.0854879878707787e-05,
      "loss": 0.1205,
      "step": 9246
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4885610132143339,
      "learning_rate": 2.084998810695446e-05,
      "loss": 0.1699,
      "step": 9247
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6181770556553181,
      "learning_rate": 2.0845096498590092e-05,
      "loss": 0.1429,
      "step": 9248
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5090192062202032,
      "learning_rate": 2.084020505380725e-05,
      "loss": 0.118,
      "step": 9249
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7256943233471645,
      "learning_rate": 2.0835313772798515e-05,
      "loss": 0.1436,
      "step": 9250
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.66862192161293,
      "learning_rate": 2.0830422655756448e-05,
      "loss": 0.1057,
      "step": 9251
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5240183594260293,
      "learning_rate": 2.0825531702873646e-05,
      "loss": 0.0656,
      "step": 9252
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6958071978005723,
      "learning_rate": 2.082064091434265e-05,
      "loss": 0.2134,
      "step": 9253
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7966548596907366,
      "learning_rate": 2.0815750290356018e-05,
      "loss": 0.1144,
      "step": 9254
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8907509883613978,
      "learning_rate": 2.0810859831106284e-05,
      "loss": 0.1644,
      "step": 9255
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.49284879396065406,
      "learning_rate": 2.0805969536786008e-05,
      "loss": 0.1098,
      "step": 9256
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.691368541053029,
      "learning_rate": 2.080107940758772e-05,
      "loss": 0.1144,
      "step": 9257
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4985677797597736,
      "learning_rate": 2.0796189443703944e-05,
      "loss": 0.1149,
      "step": 9258
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.49072149640675294,
      "learning_rate": 2.0791299645327202e-05,
      "loss": 0.1053,
      "step": 9259
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.682342094181518,
      "learning_rate": 2.0786410012650006e-05,
      "loss": 0.2349,
      "step": 9260
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9895514803657566,
      "learning_rate": 2.078152054586487e-05,
      "loss": 0.2321,
      "step": 9261
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4552635016398177,
      "learning_rate": 2.0776631245164296e-05,
      "loss": 0.0649,
      "step": 9262
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6942706355996587,
      "learning_rate": 2.077174211074077e-05,
      "loss": 0.1006,
      "step": 9263
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.512960774356453,
      "learning_rate": 2.0766853142786794e-05,
      "loss": 0.0922,
      "step": 9264
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.704220360339468,
      "learning_rate": 2.0761964341494838e-05,
      "loss": 0.2125,
      "step": 9265
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6393430958071158,
      "learning_rate": 2.0757075707057388e-05,
      "loss": 0.2139,
      "step": 9266
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4539535348982505,
      "learning_rate": 2.0752187239666904e-05,
      "loss": 0.0703,
      "step": 9267
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.40516972158655806,
      "learning_rate": 2.074729893951586e-05,
      "loss": 0.0775,
      "step": 9268
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6520451611406558,
      "learning_rate": 2.0742410806796697e-05,
      "loss": 0.1016,
      "step": 9269
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.463903872447476,
      "learning_rate": 2.0737522841701877e-05,
      "loss": 0.07,
      "step": 9270
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1039141926855893,
      "learning_rate": 2.0732635044423826e-05,
      "loss": 0.3249,
      "step": 9271
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.539890716890167,
      "learning_rate": 2.0727747415155005e-05,
      "loss": 0.0796,
      "step": 9272
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9656706177124474,
      "learning_rate": 2.0722859954087828e-05,
      "loss": 0.2416,
      "step": 9273
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5408729345864159,
      "learning_rate": 2.0717972661414707e-05,
      "loss": 0.0814,
      "step": 9274
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6907708018369969,
      "learning_rate": 2.0713085537328085e-05,
      "loss": 0.0824,
      "step": 9275
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5883025457715463,
      "learning_rate": 2.0708198582020357e-05,
      "loss": 0.0899,
      "step": 9276
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6677142046332877,
      "learning_rate": 2.070331179568393e-05,
      "loss": 0.1977,
      "step": 9277
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.44494017694032423,
      "learning_rate": 2.0698425178511177e-05,
      "loss": 0.0521,
      "step": 9278
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6479861153576579,
      "learning_rate": 2.0693538730694525e-05,
      "loss": 0.1331,
      "step": 9279
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6321200435468494,
      "learning_rate": 2.0688652452426338e-05,
      "loss": 0.1917,
      "step": 9280
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4952751318590434,
      "learning_rate": 2.0683766343898996e-05,
      "loss": 0.1243,
      "step": 9281
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5572977258549406,
      "learning_rate": 2.067888040530485e-05,
      "loss": 0.084,
      "step": 9282
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.997194241435408,
      "learning_rate": 2.0673994636836296e-05,
      "loss": 0.3728,
      "step": 9283
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2087202326666415,
      "learning_rate": 2.066910903868567e-05,
      "loss": 0.338,
      "step": 9284
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8733479481698397,
      "learning_rate": 2.0664223611045327e-05,
      "loss": 0.1233,
      "step": 9285
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7422232463235525,
      "learning_rate": 2.0659338354107597e-05,
      "loss": 0.1407,
      "step": 9286
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6803246802686272,
      "learning_rate": 2.065445326806484e-05,
      "loss": 0.1426,
      "step": 9287
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5240718454404815,
      "learning_rate": 2.0649568353109366e-05,
      "loss": 0.0534,
      "step": 9288
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5540481439846794,
      "learning_rate": 2.0644683609433506e-05,
      "loss": 0.1195,
      "step": 9289
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6235013584542883,
      "learning_rate": 2.063979903722957e-05,
      "loss": 0.0967,
      "step": 9290
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4448000929010861,
      "learning_rate": 2.0634914636689868e-05,
      "loss": 0.0602,
      "step": 9291
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7741765609039928,
      "learning_rate": 2.0630030408006714e-05,
      "loss": 0.1232,
      "step": 9292
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6411987930336962,
      "learning_rate": 2.0625146351372393e-05,
      "loss": 0.1259,
      "step": 9293
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5444813758531337,
      "learning_rate": 2.062026246697919e-05,
      "loss": 0.0985,
      "step": 9294
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5503056967376689,
      "learning_rate": 2.0615378755019398e-05,
      "loss": 0.1976,
      "step": 9295
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7497285447243406,
      "learning_rate": 2.061049521568529e-05,
      "loss": 0.1956,
      "step": 9296
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.35536921703960983,
      "learning_rate": 2.0605611849169122e-05,
      "loss": 0.0537,
      "step": 9297
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7005280911057532,
      "learning_rate": 2.0600728655663164e-05,
      "loss": 0.2125,
      "step": 9298
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5807002370833786,
      "learning_rate": 2.0595845635359678e-05,
      "loss": 0.112,
      "step": 9299
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5162555247977306,
      "learning_rate": 2.0590962788450908e-05,
      "loss": 0.0873,
      "step": 9300
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5938361368035624,
      "learning_rate": 2.058608011512909e-05,
      "loss": 0.1626,
      "step": 9301
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4463763585516101,
      "learning_rate": 2.058119761558645e-05,
      "loss": 0.0517,
      "step": 9302
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7323349225011019,
      "learning_rate": 2.057631529001524e-05,
      "loss": 0.2063,
      "step": 9303
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4741078777201378,
      "learning_rate": 2.0571433138607666e-05,
      "loss": 0.105,
      "step": 9304
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7935054308836246,
      "learning_rate": 2.0566551161555946e-05,
      "loss": 0.13,
      "step": 9305
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5398079816117968,
      "learning_rate": 2.0561669359052267e-05,
      "loss": 0.0912,
      "step": 9306
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5382720416548592,
      "learning_rate": 2.0556787731288864e-05,
      "loss": 0.117,
      "step": 9307
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5715034895640922,
      "learning_rate": 2.055190627845791e-05,
      "loss": 0.0979,
      "step": 9308
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6730798943481424,
      "learning_rate": 2.054702500075159e-05,
      "loss": 0.1,
      "step": 9309
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5186565229489528,
      "learning_rate": 2.054214389836208e-05,
      "loss": 0.1048,
      "step": 9310
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6248694199454934,
      "learning_rate": 2.053726297148157e-05,
      "loss": 0.2784,
      "step": 9311
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.080700748463774,
      "learning_rate": 2.053238222030221e-05,
      "loss": 0.1932,
      "step": 9312
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3627736730676061,
      "learning_rate": 2.0527501645016174e-05,
      "loss": 0.048,
      "step": 9313
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9713818362043445,
      "learning_rate": 2.0522621245815585e-05,
      "loss": 0.3098,
      "step": 9314
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6660974767029525,
      "learning_rate": 2.051774102289262e-05,
      "loss": 0.1396,
      "step": 9315
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5129343745052887,
      "learning_rate": 2.05128609764394e-05,
      "loss": 0.1324,
      "step": 9316
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6796230236572,
      "learning_rate": 2.0507981106648064e-05,
      "loss": 0.1186,
      "step": 9317
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8209824595925523,
      "learning_rate": 2.050310141371072e-05,
      "loss": 0.1435,
      "step": 9318
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7995600415451333,
      "learning_rate": 2.04982218978195e-05,
      "loss": 0.2095,
      "step": 9319
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.094379933579757,
      "learning_rate": 2.0493342559166517e-05,
      "loss": 0.3851,
      "step": 9320
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7214973085848588,
      "learning_rate": 2.0488463397943865e-05,
      "loss": 0.1786,
      "step": 9321
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7208942267546296,
      "learning_rate": 2.0483584414343633e-05,
      "loss": 0.1836,
      "step": 9322
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5568774507654433,
      "learning_rate": 2.047870560855793e-05,
      "loss": 0.1281,
      "step": 9323
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3881799497647407,
      "learning_rate": 2.0473826980778828e-05,
      "loss": 0.0506,
      "step": 9324
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6588017511797126,
      "learning_rate": 2.0468948531198398e-05,
      "loss": 0.1076,
      "step": 9325
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8858027868684166,
      "learning_rate": 2.0464070260008707e-05,
      "loss": 0.1433,
      "step": 9326
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.48157450133298635,
      "learning_rate": 2.0459192167401832e-05,
      "loss": 0.0916,
      "step": 9327
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.594579998202979,
      "learning_rate": 2.0454314253569802e-05,
      "loss": 0.0863,
      "step": 9328
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5466830878153293,
      "learning_rate": 2.0449436518704687e-05,
      "loss": 0.0889,
      "step": 9329
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2572771113241579,
      "learning_rate": 2.0444558962998513e-05,
      "loss": 0.4,
      "step": 9330
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4925457504650767,
      "learning_rate": 2.043968158664332e-05,
      "loss": 0.0713,
      "step": 9331
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9553607465975454,
      "learning_rate": 2.0434804389831136e-05,
      "loss": 0.1714,
      "step": 9332
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0439180949060878,
      "learning_rate": 2.0429927372753965e-05,
      "loss": 0.1925,
      "step": 9333
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.461887431271028,
      "learning_rate": 2.042505053560383e-05,
      "loss": 0.0574,
      "step": 9334
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5538572877135349,
      "learning_rate": 2.0420173878572742e-05,
      "loss": 0.0858,
      "step": 9335
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6372732341698237,
      "learning_rate": 2.0415297401852684e-05,
      "loss": 0.2324,
      "step": 9336
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6880764280987773,
      "learning_rate": 2.0410421105635662e-05,
      "loss": 0.1219,
      "step": 9337
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5645457909703003,
      "learning_rate": 2.040554499011363e-05,
      "loss": 0.0819,
      "step": 9338
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6417471499033502,
      "learning_rate": 2.0400669055478603e-05,
      "loss": 0.0831,
      "step": 9339
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4736479737201135,
      "learning_rate": 2.039579330192253e-05,
      "loss": 0.0636,
      "step": 9340
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6488213393749527,
      "learning_rate": 2.039091772963737e-05,
      "loss": 0.1426,
      "step": 9341
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6318119450092754,
      "learning_rate": 2.0386042338815075e-05,
      "loss": 0.0871,
      "step": 9342
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.48813734204724285,
      "learning_rate": 2.038116712964761e-05,
      "loss": 0.0885,
      "step": 9343
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4874562718682942,
      "learning_rate": 2.0376292102326904e-05,
      "loss": 0.0762,
      "step": 9344
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5866390882935395,
      "learning_rate": 2.0371417257044893e-05,
      "loss": 0.2082,
      "step": 9345
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.30945507430993896,
      "learning_rate": 2.0366542593993493e-05,
      "loss": 0.0368,
      "step": 9346
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.32623376285302796,
      "learning_rate": 2.036166811336464e-05,
      "loss": 0.0594,
      "step": 9347
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5732062279192615,
      "learning_rate": 2.0356793815350238e-05,
      "loss": 0.1576,
      "step": 9348
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6858742947475325,
      "learning_rate": 2.035191970014219e-05,
      "loss": 0.1686,
      "step": 9349
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6521386764215861,
      "learning_rate": 2.034704576793239e-05,
      "loss": 0.121,
      "step": 9350
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8715598743440703,
      "learning_rate": 2.0342172018912738e-05,
      "loss": 0.1889,
      "step": 9351
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8190302379749732,
      "learning_rate": 2.033729845327511e-05,
      "loss": 0.2496,
      "step": 9352
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2901227523548038,
      "learning_rate": 2.0332425071211385e-05,
      "loss": 0.4074,
      "step": 9353
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.619974497849394,
      "learning_rate": 2.032755187291342e-05,
      "loss": 0.2202,
      "step": 9354
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9307449879426551,
      "learning_rate": 2.0322678858573098e-05,
      "loss": 0.1443,
      "step": 9355
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6734368192924859,
      "learning_rate": 2.031780602838226e-05,
      "loss": 0.0932,
      "step": 9356
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6960853851687753,
      "learning_rate": 2.0312933382532758e-05,
      "loss": 0.2745,
      "step": 9357
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4926829378234963,
      "learning_rate": 2.0308060921216415e-05,
      "loss": 0.0674,
      "step": 9358
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.43929962301555264,
      "learning_rate": 2.0303188644625083e-05,
      "loss": 0.1341,
      "step": 9359
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5895786680063964,
      "learning_rate": 2.0298316552950584e-05,
      "loss": 0.1205,
      "step": 9360
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4803464619514664,
      "learning_rate": 2.0293444646384728e-05,
      "loss": 0.1017,
      "step": 9361
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7096369090189667,
      "learning_rate": 2.028857292511933e-05,
      "loss": 0.1826,
      "step": 9362
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.722196939538057,
      "learning_rate": 2.0283701389346195e-05,
      "loss": 0.5924,
      "step": 9363
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5566789803557252,
      "learning_rate": 2.0278830039257113e-05,
      "loss": 0.0914,
      "step": 9364
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5982062274710525,
      "learning_rate": 2.0273958875043874e-05,
      "loss": 0.1355,
      "step": 9365
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5665146962379727,
      "learning_rate": 2.026908789689826e-05,
      "loss": 0.1326,
      "step": 9366
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5814498612114117,
      "learning_rate": 2.0264217105012053e-05,
      "loss": 0.1379,
      "step": 9367
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4367765347363532,
      "learning_rate": 2.0259346499577014e-05,
      "loss": 0.0583,
      "step": 9368
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5400852495649223,
      "learning_rate": 2.0254476080784898e-05,
      "loss": 0.2446,
      "step": 9369
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0424253515580524,
      "learning_rate": 2.0249605848827448e-05,
      "loss": 0.0862,
      "step": 9370
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2980089615667902,
      "learning_rate": 2.0244735803896434e-05,
      "loss": 0.1188,
      "step": 9371
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6301628472909179,
      "learning_rate": 2.0239865946183578e-05,
      "loss": 0.0636,
      "step": 9372
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6483906607039015,
      "learning_rate": 2.023499627588061e-05,
      "loss": 0.1033,
      "step": 9373
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5161493951101004,
      "learning_rate": 2.0230126793179242e-05,
      "loss": 0.082,
      "step": 9374
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8429089408179701,
      "learning_rate": 2.0225257498271217e-05,
      "loss": 0.2671,
      "step": 9375
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.40418200597638043,
      "learning_rate": 2.0220388391348223e-05,
      "loss": 0.0756,
      "step": 9376
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5383232833921273,
      "learning_rate": 2.0215519472601964e-05,
      "loss": 0.0992,
      "step": 9377
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.46439996354115703,
      "learning_rate": 2.021065074222412e-05,
      "loss": 0.0579,
      "step": 9378
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6690112725355469,
      "learning_rate": 2.02057822004064e-05,
      "loss": 0.1682,
      "step": 9379
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6315871836992593,
      "learning_rate": 2.0200913847340468e-05,
      "loss": 0.0936,
      "step": 9380
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6379661758232796,
      "learning_rate": 2.0196045683218003e-05,
      "loss": 0.1715,
      "step": 9381
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5909122460795582,
      "learning_rate": 2.0191177708230645e-05,
      "loss": 0.1336,
      "step": 9382
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5810995205872387,
      "learning_rate": 2.0186309922570083e-05,
      "loss": 0.1766,
      "step": 9383
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7490085285745245,
      "learning_rate": 2.0181442326427946e-05,
      "loss": 0.1673,
      "step": 9384
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9630516866444376,
      "learning_rate": 2.017657491999588e-05,
      "loss": 0.2185,
      "step": 9385
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7265643887286501,
      "learning_rate": 2.0171707703465502e-05,
      "loss": 0.1343,
      "step": 9386
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6556388806894421,
      "learning_rate": 2.016684067702847e-05,
      "loss": 0.1031,
      "step": 9387
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7659509858104486,
      "learning_rate": 2.0161973840876377e-05,
      "loss": 0.107,
      "step": 9388
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3502408933648973,
      "learning_rate": 2.0157107195200843e-05,
      "loss": 0.41,
      "step": 9389
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0791966006753904,
      "learning_rate": 2.0152240740193462e-05,
      "loss": 0.4029,
      "step": 9390
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.037575520542271,
      "learning_rate": 2.0147374476045844e-05,
      "loss": 0.3373,
      "step": 9391
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1571861375684895,
      "learning_rate": 2.0142508402949573e-05,
      "loss": 0.5791,
      "step": 9392
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7234982174690902,
      "learning_rate": 2.0137642521096228e-05,
      "loss": 0.0738,
      "step": 9393
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6313545685729016,
      "learning_rate": 2.0132776830677374e-05,
      "loss": 0.214,
      "step": 9394
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6776802407229091,
      "learning_rate": 2.0127911331884593e-05,
      "loss": 0.149,
      "step": 9395
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.028070998875869,
      "learning_rate": 2.0123046024909426e-05,
      "loss": 0.2181,
      "step": 9396
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6901210564019521,
      "learning_rate": 2.011818090994344e-05,
      "loss": 0.1825,
      "step": 9397
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.47309670073002086,
      "learning_rate": 2.011331598717816e-05,
      "loss": 0.1146,
      "step": 9398
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.4969939141457527,
      "learning_rate": 2.0108451256805145e-05,
      "loss": 0.0929,
      "step": 9399
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2054621848777682,
      "learning_rate": 2.0103586719015905e-05,
      "loss": 0.4287,
      "step": 9400
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.38511940805431466,
      "learning_rate": 2.0098722374001965e-05,
      "loss": 0.086,
      "step": 9401
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4531297830578548,
      "learning_rate": 2.0093858221954825e-05,
      "loss": 0.0781,
      "step": 9402
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.49648289461078354,
      "learning_rate": 2.0088994263066018e-05,
      "loss": 0.0972,
      "step": 9403
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4809286948479434,
      "learning_rate": 2.0084130497527023e-05,
      "loss": 0.0791,
      "step": 9404
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5444551527702843,
      "learning_rate": 2.0079266925529323e-05,
      "loss": 0.1652,
      "step": 9405
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6128371817387931,
      "learning_rate": 2.0074403547264427e-05,
      "loss": 0.073,
      "step": 9406
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0957958073922331,
      "learning_rate": 2.0069540362923786e-05,
      "loss": 0.3624,
      "step": 9407
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5620995643612997,
      "learning_rate": 2.006467737269888e-05,
      "loss": 0.0667,
      "step": 9408
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9466621161966299,
      "learning_rate": 2.005981457678115e-05,
      "loss": 0.325,
      "step": 9409
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6739444768768981,
      "learning_rate": 2.0054951975362067e-05,
      "loss": 0.1669,
      "step": 9410
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7337099575464975,
      "learning_rate": 2.0050089568633075e-05,
      "loss": 0.1902,
      "step": 9411
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4483642091296387,
      "learning_rate": 2.00452273567856e-05,
      "loss": 0.0502,
      "step": 9412
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.32744636515178993,
      "learning_rate": 2.0040365340011066e-05,
      "loss": 0.0417,
      "step": 9413
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8011581942274778,
      "learning_rate": 2.0035503518500915e-05,
      "loss": 0.1486,
      "step": 9414
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4558899208814832,
      "learning_rate": 2.0030641892446547e-05,
      "loss": 0.1211,
      "step": 9415
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7197999781484847,
      "learning_rate": 2.002578046203937e-05,
      "loss": 0.2616,
      "step": 9416
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6161219340983417,
      "learning_rate": 2.0020919227470768e-05,
      "loss": 0.131,
      "step": 9417
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.44355242809191536,
      "learning_rate": 2.0016058188932157e-05,
      "loss": 0.0605,
      "step": 9418
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3736566910653725,
      "learning_rate": 2.0011197346614906e-05,
      "loss": 0.0622,
      "step": 9419
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5541459688838275,
      "learning_rate": 2.0006336700710395e-05,
      "loss": 0.0755,
      "step": 9420
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4763200871305161,
      "learning_rate": 2.0001476251409976e-05,
      "loss": 0.0914,
      "step": 9421
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.41973061506192194,
      "learning_rate": 1.999661599890502e-05,
      "loss": 0.0384,
      "step": 9422
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9405139489046482,
      "learning_rate": 1.9991755943386886e-05,
      "loss": 0.1645,
      "step": 9423
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5599881731876155,
      "learning_rate": 1.9986896085046908e-05,
      "loss": 0.1054,
      "step": 9424
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6495148488323265,
      "learning_rate": 1.998203642407642e-05,
      "loss": 0.1266,
      "step": 9425
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6453501516323016,
      "learning_rate": 1.997717696066676e-05,
      "loss": 0.1188,
      "step": 9426
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7711734101123611,
      "learning_rate": 1.9972317695009234e-05,
      "loss": 0.123,
      "step": 9427
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6248939693498404,
      "learning_rate": 1.9967458627295172e-05,
      "loss": 0.1035,
      "step": 9428
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.911123966235811,
      "learning_rate": 1.9962599757715862e-05,
      "loss": 0.1892,
      "step": 9429
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.45734448090454466,
      "learning_rate": 1.995774108646262e-05,
      "loss": 0.0742,
      "step": 9430
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5692073524825795,
      "learning_rate": 1.995288261372672e-05,
      "loss": 0.0735,
      "step": 9431
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7350540505893279,
      "learning_rate": 1.994802433969945e-05,
      "loss": 0.2248,
      "step": 9432
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6011240226374885,
      "learning_rate": 1.9943166264572075e-05,
      "loss": 0.1115,
      "step": 9433
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.521193374421901,
      "learning_rate": 1.9938308388535877e-05,
      "loss": 0.1228,
      "step": 9434
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6091633284396794,
      "learning_rate": 1.9933450711782102e-05,
      "loss": 0.1211,
      "step": 9435
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8058948735125012,
      "learning_rate": 1.992859323450201e-05,
      "loss": 0.1866,
      "step": 9436
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7290173247219923,
      "learning_rate": 1.992373595688682e-05,
      "loss": 0.179,
      "step": 9437
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8330573458076126,
      "learning_rate": 1.9918878879127798e-05,
      "loss": 0.1057,
      "step": 9438
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.851875696423219,
      "learning_rate": 1.9914022001416154e-05,
      "loss": 0.3174,
      "step": 9439
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3952491414637259,
      "learning_rate": 1.990916532394311e-05,
      "loss": 0.0759,
      "step": 9440
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6996066490276228,
      "learning_rate": 1.9904308846899865e-05,
      "loss": 0.1542,
      "step": 9441
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0314388779917607,
      "learning_rate": 1.9899452570477642e-05,
      "loss": 0.0828,
      "step": 9442
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3157598825695813,
      "learning_rate": 1.989459649486763e-05,
      "loss": 0.2332,
      "step": 9443
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4354518162479927,
      "learning_rate": 1.988974062026101e-05,
      "loss": 0.0668,
      "step": 9444
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5383994292851723,
      "learning_rate": 1.9884884946848953e-05,
      "loss": 0.0574,
      "step": 9445
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6541186653848265,
      "learning_rate": 1.9880029474822654e-05,
      "loss": 0.584,
      "step": 9446
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0094566147928843,
      "learning_rate": 1.9875174204373263e-05,
      "loss": 0.2678,
      "step": 9447
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6040057814975175,
      "learning_rate": 1.9870319135691938e-05,
      "loss": 0.1703,
      "step": 9448
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.046447473769545,
      "learning_rate": 1.986546426896981e-05,
      "loss": 0.7945,
      "step": 9449
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7525077030805176,
      "learning_rate": 1.9860609604398048e-05,
      "loss": 0.1266,
      "step": 9450
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5280775388126003,
      "learning_rate": 1.9855755142167766e-05,
      "loss": 0.0775,
      "step": 9451
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.46016422614361624,
      "learning_rate": 1.9850900882470095e-05,
      "loss": 0.0604,
      "step": 9452
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6142334310328785,
      "learning_rate": 1.9846046825496134e-05,
      "loss": 0.0715,
      "step": 9453
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.45277591812615153,
      "learning_rate": 1.9841192971437016e-05,
      "loss": 0.0559,
      "step": 9454
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9020261134068362,
      "learning_rate": 1.9836339320483825e-05,
      "loss": 0.26,
      "step": 9455
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0777517938607186,
      "learning_rate": 1.9831485872827656e-05,
      "loss": 0.1804,
      "step": 9456
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5680214255856814,
      "learning_rate": 1.9826632628659585e-05,
      "loss": 0.1557,
      "step": 9457
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6802788608766833,
      "learning_rate": 1.9821779588170703e-05,
      "loss": 0.1132,
      "step": 9458
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9617821778200832,
      "learning_rate": 1.9816926751552064e-05,
      "loss": 0.244,
      "step": 9459
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5341119602729713,
      "learning_rate": 1.9812074118994738e-05,
      "loss": 0.0846,
      "step": 9460
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9822729687541191,
      "learning_rate": 1.9807221690689767e-05,
      "loss": 0.3082,
      "step": 9461
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6881876068741855,
      "learning_rate": 1.9802369466828206e-05,
      "loss": 0.1348,
      "step": 9462
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7138272246963004,
      "learning_rate": 1.9797517447601084e-05,
      "loss": 0.1113,
      "step": 9463
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4667654811307797,
      "learning_rate": 1.9792665633199424e-05,
      "loss": 0.0667,
      "step": 9464
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.157957425242136,
      "learning_rate": 1.9787814023814245e-05,
      "loss": 0.3262,
      "step": 9465
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4584101945100761,
      "learning_rate": 1.9782962619636574e-05,
      "loss": 0.0969,
      "step": 9466
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9322322173953949,
      "learning_rate": 1.97781114208574e-05,
      "loss": 0.4463,
      "step": 9467
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6972153972679702,
      "learning_rate": 1.9773260427667724e-05,
      "loss": 0.0599,
      "step": 9468
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5339458988748051,
      "learning_rate": 1.976840964025852e-05,
      "loss": 0.1145,
      "step": 9469
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9072944105033053,
      "learning_rate": 1.976355905882079e-05,
      "loss": 0.2056,
      "step": 9470
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.836199811312785,
      "learning_rate": 1.975870868354549e-05,
      "loss": 0.2561,
      "step": 9471
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6999184471563363,
      "learning_rate": 1.9753858514623584e-05,
      "loss": 0.1087,
      "step": 9472
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5354331454530152,
      "learning_rate": 1.974900855224602e-05,
      "loss": 0.0635,
      "step": 9473
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9751514695824524,
      "learning_rate": 1.974415879660376e-05,
      "loss": 0.3761,
      "step": 9474
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3020813614357325,
      "learning_rate": 1.973930924788774e-05,
      "loss": 0.2965,
      "step": 9475
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8060871449382334,
      "learning_rate": 1.9734459906288885e-05,
      "loss": 0.2062,
      "step": 9476
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4129294650365138,
      "learning_rate": 1.9729610771998102e-05,
      "loss": 0.074,
      "step": 9477
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7697217240870632,
      "learning_rate": 1.9724761845206336e-05,
      "loss": 0.1383,
      "step": 9478
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4234553166433022,
      "learning_rate": 1.9719913126104472e-05,
      "loss": 0.0463,
      "step": 9479
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.674461540897985,
      "learning_rate": 1.9715064614883417e-05,
      "loss": 0.2038,
      "step": 9480
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7108389896602663,
      "learning_rate": 1.971021631173404e-05,
      "loss": 0.0895,
      "step": 9481
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6270939600077098,
      "learning_rate": 1.970536821684725e-05,
      "loss": 0.0897,
      "step": 9482
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9021829808151018,
      "learning_rate": 1.970052033041391e-05,
      "loss": 0.1969,
      "step": 9483
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.43844229609071056,
      "learning_rate": 1.969567265262488e-05,
      "loss": 0.1037,
      "step": 9484
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5869838353704698,
      "learning_rate": 1.969082518367101e-05,
      "loss": 0.1021,
      "step": 9485
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5277386246784959,
      "learning_rate": 1.968597792374317e-05,
      "loss": 0.0924,
      "step": 9486
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4956843797468223,
      "learning_rate": 1.9681130873032182e-05,
      "loss": 0.0637,
      "step": 9487
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5956993403916526,
      "learning_rate": 1.9676284031728888e-05,
      "loss": 0.1528,
      "step": 9488
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6118519495376726,
      "learning_rate": 1.96714374000241e-05,
      "loss": 0.1452,
      "step": 9489
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6227805166200536,
      "learning_rate": 1.966659097810864e-05,
      "loss": 0.1431,
      "step": 9490
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7081821668380944,
      "learning_rate": 1.9661744766173324e-05,
      "loss": 0.1277,
      "step": 9491
    },
    {
      "epoch": 0.58,
      "grad_norm": 10.457259670248062,
      "learning_rate": 1.965689876440894e-05,
      "loss": 0.8768,
      "step": 9492
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6214233484561619,
      "learning_rate": 1.9652052973006275e-05,
      "loss": 0.1121,
      "step": 9493
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6386590290556292,
      "learning_rate": 1.9647207392156125e-05,
      "loss": 0.097,
      "step": 9494
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5313463308860445,
      "learning_rate": 1.964236202204925e-05,
      "loss": 0.0913,
      "step": 9495
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4594159338679362,
      "learning_rate": 1.9637516862876424e-05,
      "loss": 0.0797,
      "step": 9496
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.49784927764750775,
      "learning_rate": 1.96326719148284e-05,
      "loss": 0.1322,
      "step": 9497
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8080166982613319,
      "learning_rate": 1.9627827178095934e-05,
      "loss": 0.1505,
      "step": 9498
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8552358823889044,
      "learning_rate": 1.9622982652869762e-05,
      "loss": 0.2241,
      "step": 9499
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5647317754860117,
      "learning_rate": 1.9618138339340614e-05,
      "loss": 0.1421,
      "step": 9500
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4642103335065294,
      "learning_rate": 1.9613294237699208e-05,
      "loss": 0.0715,
      "step": 9501
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5447424019541182,
      "learning_rate": 1.960845034813628e-05,
      "loss": 0.1075,
      "step": 9502
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.441217988362636,
      "learning_rate": 1.9603606670842528e-05,
      "loss": 0.0622,
      "step": 9503
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5041687099898007,
      "learning_rate": 1.9598763206008643e-05,
      "loss": 0.0924,
      "step": 9504
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4421518087807241,
      "learning_rate": 1.9593919953825314e-05,
      "loss": 0.0589,
      "step": 9505
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9191951452644566,
      "learning_rate": 1.958907691448324e-05,
      "loss": 0.1999,
      "step": 9506
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9188619657710685,
      "learning_rate": 1.9584234088173088e-05,
      "loss": 0.2397,
      "step": 9507
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8949940471249526,
      "learning_rate": 1.9579391475085523e-05,
      "loss": 0.2819,
      "step": 9508
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4836613804256446,
      "learning_rate": 1.9574549075411185e-05,
      "loss": 0.0632,
      "step": 9509
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6411233694359438,
      "learning_rate": 1.9569706889340752e-05,
      "loss": 0.1547,
      "step": 9510
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6489418563114783,
      "learning_rate": 1.956486491706485e-05,
      "loss": 0.1711,
      "step": 9511
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9197715568416541,
      "learning_rate": 1.956002315877411e-05,
      "loss": 0.1693,
      "step": 9512
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7940484727856043,
      "learning_rate": 1.955518161465915e-05,
      "loss": 0.1219,
      "step": 9513
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.45034129019116415,
      "learning_rate": 1.95503402849106e-05,
      "loss": 0.0778,
      "step": 9514
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.47121305521194445,
      "learning_rate": 1.9545499169719055e-05,
      "loss": 0.0861,
      "step": 9515
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.47032777770724304,
      "learning_rate": 1.954065826927512e-05,
      "loss": 0.0891,
      "step": 9516
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.038121207348975,
      "learning_rate": 1.953581758376937e-05,
      "loss": 0.1933,
      "step": 9517
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.40384564188346694,
      "learning_rate": 1.9530977113392413e-05,
      "loss": 0.0617,
      "step": 9518
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9244493456449622,
      "learning_rate": 1.9526136858334804e-05,
      "loss": 0.1172,
      "step": 9519
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2872260679893088,
      "learning_rate": 1.952129681878711e-05,
      "loss": 0.5161,
      "step": 9520
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.48851577692598375,
      "learning_rate": 1.9516456994939874e-05,
      "loss": 0.0594,
      "step": 9521
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9274776359308241,
      "learning_rate": 1.951161738698367e-05,
      "loss": 0.1869,
      "step": 9522
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7293128182879418,
      "learning_rate": 1.950677799510902e-05,
      "loss": 0.1398,
      "step": 9523
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5039434502403639,
      "learning_rate": 1.9501938819506458e-05,
      "loss": 0.0611,
      "step": 9524
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5484863352599735,
      "learning_rate": 1.9497099860366497e-05,
      "loss": 0.1105,
      "step": 9525
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4781429156215902,
      "learning_rate": 1.9492261117879667e-05,
      "loss": 0.0704,
      "step": 9526
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5280477699696505,
      "learning_rate": 1.948742259223646e-05,
      "loss": 0.1069,
      "step": 9527
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5764789390760702,
      "learning_rate": 1.9482584283627378e-05,
      "loss": 0.0959,
      "step": 9528
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7445194969123312,
      "learning_rate": 1.94777461922429e-05,
      "loss": 0.2653,
      "step": 9529
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6822992240790428,
      "learning_rate": 1.947290831827352e-05,
      "loss": 0.2049,
      "step": 9530
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6731709273814902,
      "learning_rate": 1.9468070661909704e-05,
      "loss": 0.1418,
      "step": 9531
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5514742813987726,
      "learning_rate": 1.9463233223341904e-05,
      "loss": 0.138,
      "step": 9532
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.41594638228018815,
      "learning_rate": 1.9458396002760578e-05,
      "loss": 0.0977,
      "step": 9533
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4382740457452983,
      "learning_rate": 1.9453559000356178e-05,
      "loss": 0.0809,
      "step": 9534
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3181113180149961,
      "learning_rate": 1.9448722216319136e-05,
      "loss": 0.3802,
      "step": 9535
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5111761462671851,
      "learning_rate": 1.9443885650839877e-05,
      "loss": 0.0812,
      "step": 9536
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4819000325523108,
      "learning_rate": 1.9439049304108815e-05,
      "loss": 0.0879,
      "step": 9537
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5404730148554515,
      "learning_rate": 1.9434213176316372e-05,
      "loss": 0.1406,
      "step": 9538
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4353974740497328,
      "learning_rate": 1.9429377267652953e-05,
      "loss": 0.062,
      "step": 9539
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8059701232386918,
      "learning_rate": 1.9424541578308928e-05,
      "loss": 0.1892,
      "step": 9540
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4718314633778359,
      "learning_rate": 1.9419706108474705e-05,
      "loss": 0.0988,
      "step": 9541
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8584843601966088,
      "learning_rate": 1.9414870858340655e-05,
      "loss": 0.1381,
      "step": 9542
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9279554052682742,
      "learning_rate": 1.9410035828097144e-05,
      "loss": 0.1689,
      "step": 9543
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8114720577669416,
      "learning_rate": 1.9405201017934514e-05,
      "loss": 0.1398,
      "step": 9544
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7462091861748437,
      "learning_rate": 1.940036642804314e-05,
      "loss": 0.0891,
      "step": 9545
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9277089645596408,
      "learning_rate": 1.9395532058613357e-05,
      "loss": 0.188,
      "step": 9546
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4806673685296639,
      "learning_rate": 1.9390697909835493e-05,
      "loss": 0.0678,
      "step": 9547
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.33123139534645196,
      "learning_rate": 1.938586398189986e-05,
      "loss": 0.0548,
      "step": 9548
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5235050173990868,
      "learning_rate": 1.9381030274996793e-05,
      "loss": 0.1094,
      "step": 9549
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6498613528043926,
      "learning_rate": 1.9376196789316598e-05,
      "loss": 0.0839,
      "step": 9550
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5225287980173424,
      "learning_rate": 1.937136352504956e-05,
      "loss": 0.0618,
      "step": 9551
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8719381510670462,
      "learning_rate": 1.9366530482385972e-05,
      "loss": 0.1634,
      "step": 9552
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5038363701804972,
      "learning_rate": 1.9361697661516116e-05,
      "loss": 0.0945,
      "step": 9553
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6192461139452681,
      "learning_rate": 1.9356865062630265e-05,
      "loss": 0.1587,
      "step": 9554
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5044288428242586,
      "learning_rate": 1.9352032685918686e-05,
      "loss": 0.0601,
      "step": 9555
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4349167669383382,
      "learning_rate": 1.934720053157162e-05,
      "loss": 0.0448,
      "step": 9556
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5876166490724436,
      "learning_rate": 1.9342368599779326e-05,
      "loss": 0.1321,
      "step": 9557
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6376055370587227,
      "learning_rate": 1.9337536890732032e-05,
      "loss": 0.169,
      "step": 9558
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0806330811973193,
      "learning_rate": 1.933270540461997e-05,
      "loss": 0.2794,
      "step": 9559
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.43820109069538704,
      "learning_rate": 1.9327874141633352e-05,
      "loss": 0.0404,
      "step": 9560
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8517653595815802,
      "learning_rate": 1.9323043101962403e-05,
      "loss": 0.1615,
      "step": 9561
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4857343748577381,
      "learning_rate": 1.9318212285797316e-05,
      "loss": 0.0877,
      "step": 9562
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6349626863861956,
      "learning_rate": 1.931338169332828e-05,
      "loss": 0.1232,
      "step": 9563
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4787151752925708,
      "learning_rate": 1.9308551324745476e-05,
      "loss": 0.0879,
      "step": 9564
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4715938034486372,
      "learning_rate": 1.930372118023909e-05,
      "loss": 0.0669,
      "step": 9565
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5033540077033979,
      "learning_rate": 1.929889125999929e-05,
      "loss": 0.0692,
      "step": 9566
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5232201475035059,
      "learning_rate": 1.929406156421622e-05,
      "loss": 0.0628,
      "step": 9567
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5731110819349998,
      "learning_rate": 1.928923209308003e-05,
      "loss": 0.052,
      "step": 9568
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.42930637118006143,
      "learning_rate": 1.9284402846780874e-05,
      "loss": 0.1055,
      "step": 9569
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6246749409865006,
      "learning_rate": 1.9279573825508875e-05,
      "loss": 0.0797,
      "step": 9570
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4181231774799302,
      "learning_rate": 1.927474502945415e-05,
      "loss": 0.0719,
      "step": 9571
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8269754133572138,
      "learning_rate": 1.9269916458806814e-05,
      "loss": 0.1585,
      "step": 9572
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.631247767921383,
      "learning_rate": 1.9265088113756976e-05,
      "loss": 0.1346,
      "step": 9573
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7548245497821685,
      "learning_rate": 1.9260259994494733e-05,
      "loss": 0.2967,
      "step": 9574
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8993526193040055,
      "learning_rate": 1.925543210121017e-05,
      "loss": 0.322,
      "step": 9575
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7370368959620461,
      "learning_rate": 1.9250604434093346e-05,
      "loss": 0.178,
      "step": 9576
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5285578584282662,
      "learning_rate": 1.924577699333436e-05,
      "loss": 0.0726,
      "step": 9577
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6627408447618903,
      "learning_rate": 1.9240949779123258e-05,
      "loss": 0.0893,
      "step": 9578
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8182075150118372,
      "learning_rate": 1.9236122791650088e-05,
      "loss": 0.2242,
      "step": 9579
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8391465646311507,
      "learning_rate": 1.9231296031104885e-05,
      "loss": 0.2455,
      "step": 9580
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6609998192631502,
      "learning_rate": 1.9226469497677703e-05,
      "loss": 0.1222,
      "step": 9581
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.906001318995141,
      "learning_rate": 1.9221643191558554e-05,
      "loss": 0.0798,
      "step": 9582
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5078637258811499,
      "learning_rate": 1.9216817112937456e-05,
      "loss": 0.0991,
      "step": 9583
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5639204883186062,
      "learning_rate": 1.9211991262004398e-05,
      "loss": 0.0758,
      "step": 9584
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.586510396714714,
      "learning_rate": 1.9207165638949403e-05,
      "loss": 0.1272,
      "step": 9585
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.49630379326024854,
      "learning_rate": 1.9202340243962453e-05,
      "loss": 0.1434,
      "step": 9586
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6313227176946817,
      "learning_rate": 1.919751507723352e-05,
      "loss": 0.1599,
      "step": 9587
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6949207777598227,
      "learning_rate": 1.919269013895257e-05,
      "loss": 0.0935,
      "step": 9588
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3376668399723695,
      "learning_rate": 1.9187865429309573e-05,
      "loss": 0.0513,
      "step": 9589
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5778591810070406,
      "learning_rate": 1.9183040948494486e-05,
      "loss": 0.1022,
      "step": 9590
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5257951328919003,
      "learning_rate": 1.9178216696697243e-05,
      "loss": 0.1701,
      "step": 9591
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9174771663195389,
      "learning_rate": 1.9173392674107776e-05,
      "loss": 0.1467,
      "step": 9592
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.950159495631583,
      "learning_rate": 1.916856888091602e-05,
      "loss": 0.1381,
      "step": 9593
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7187387019209953,
      "learning_rate": 1.9163745317311892e-05,
      "loss": 0.1232,
      "step": 9594
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.512446829675759,
      "learning_rate": 1.915892198348528e-05,
      "loss": 0.093,
      "step": 9595
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5121740098477378,
      "learning_rate": 1.9154098879626102e-05,
      "loss": 0.0926,
      "step": 9596
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.984377314493626,
      "learning_rate": 1.9149276005924244e-05,
      "loss": 0.3286,
      "step": 9597
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4241991553028629,
      "learning_rate": 1.914445336256958e-05,
      "loss": 0.085,
      "step": 9598
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8001043467582716,
      "learning_rate": 1.9139630949751987e-05,
      "loss": 0.1185,
      "step": 9599
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6055995044913406,
      "learning_rate": 1.9134808767661312e-05,
      "loss": 0.0969,
      "step": 9600
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7152497648131275,
      "learning_rate": 1.912998681648743e-05,
      "loss": 0.1589,
      "step": 9601
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8032370876999879,
      "learning_rate": 1.9125165096420175e-05,
      "loss": 0.1045,
      "step": 9602
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4500929670078539,
      "learning_rate": 1.9120343607649376e-05,
      "loss": 0.1202,
      "step": 9603
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.49807717942685575,
      "learning_rate": 1.911552235036486e-05,
      "loss": 0.07,
      "step": 9604
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.603326141520422,
      "learning_rate": 1.911070132475645e-05,
      "loss": 0.1306,
      "step": 9605
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4479348188364587,
      "learning_rate": 1.910588053101395e-05,
      "loss": 0.097,
      "step": 9606
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.40767137194007685,
      "learning_rate": 1.910105996932716e-05,
      "loss": 0.0872,
      "step": 9607
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.48607112519304796,
      "learning_rate": 1.9096239639885858e-05,
      "loss": 0.0797,
      "step": 9608
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.605925621749307,
      "learning_rate": 1.9091419542879842e-05,
      "loss": 0.14,
      "step": 9609
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8887826757269819,
      "learning_rate": 1.908659967849887e-05,
      "loss": 0.5381,
      "step": 9610
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4665065089965373,
      "learning_rate": 1.908178004693271e-05,
      "loss": 0.0911,
      "step": 9611
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5201615626327933,
      "learning_rate": 1.9076960648371098e-05,
      "loss": 0.1071,
      "step": 9612
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5792355171552203,
      "learning_rate": 1.9072141483003803e-05,
      "loss": 0.1147,
      "step": 9613
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.40990436444619655,
      "learning_rate": 1.9067322551020543e-05,
      "loss": 0.075,
      "step": 9614
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4685446567185926,
      "learning_rate": 1.906250385261105e-05,
      "loss": 0.0624,
      "step": 9615
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7150863542125301,
      "learning_rate": 1.905768538796502e-05,
      "loss": 0.1528,
      "step": 9616
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9713178078362847,
      "learning_rate": 1.905286715727219e-05,
      "loss": 0.2443,
      "step": 9617
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6930122959909356,
      "learning_rate": 1.904804916072224e-05,
      "loss": 0.1766,
      "step": 9618
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5789621065696551,
      "learning_rate": 1.904323139850486e-05,
      "loss": 0.0954,
      "step": 9619
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5852260622517788,
      "learning_rate": 1.9038413870809724e-05,
      "loss": 0.1227,
      "step": 9620
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6006872052918594,
      "learning_rate": 1.9033596577826506e-05,
      "loss": 0.142,
      "step": 9621
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8710459161316887,
      "learning_rate": 1.9028779519744874e-05,
      "loss": 0.195,
      "step": 9622
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.42253447038761105,
      "learning_rate": 1.902396269675447e-05,
      "loss": 0.0604,
      "step": 9623
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7053877794015503,
      "learning_rate": 1.9019146109044933e-05,
      "loss": 0.1611,
      "step": 9624
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.84457451922106,
      "learning_rate": 1.9014329756805905e-05,
      "loss": 0.1356,
      "step": 9625
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.48614278988356163,
      "learning_rate": 1.9009513640227e-05,
      "loss": 0.0721,
      "step": 9626
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5063096288947256,
      "learning_rate": 1.900469775949784e-05,
      "loss": 0.0787,
      "step": 9627
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.720593786802628,
      "learning_rate": 1.899988211480802e-05,
      "loss": 0.1479,
      "step": 9628
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5813196249883061,
      "learning_rate": 1.899506670634715e-05,
      "loss": 0.0912,
      "step": 9629
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5894242473876692,
      "learning_rate": 1.8990251534304805e-05,
      "loss": 0.0916,
      "step": 9630
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4939911940182065,
      "learning_rate": 1.8985436598870564e-05,
      "loss": 0.0829,
      "step": 9631
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5371941460728415,
      "learning_rate": 1.8980621900233982e-05,
      "loss": 0.0852,
      "step": 9632
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0051809145845536,
      "learning_rate": 1.8975807438584642e-05,
      "loss": 0.5097,
      "step": 9633
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.38893035964518363,
      "learning_rate": 1.8970993214112083e-05,
      "loss": 0.0531,
      "step": 9634
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8322779096039916,
      "learning_rate": 1.896617922700584e-05,
      "loss": 0.2451,
      "step": 9635
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8166831278643106,
      "learning_rate": 1.8961365477455436e-05,
      "loss": 0.1809,
      "step": 9636
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.523615279817746,
      "learning_rate": 1.8956551965650408e-05,
      "loss": 0.1131,
      "step": 9637
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5876500335235983,
      "learning_rate": 1.8951738691780262e-05,
      "loss": 0.1052,
      "step": 9638
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8867931768371866,
      "learning_rate": 1.8946925656034497e-05,
      "loss": 0.2925,
      "step": 9639
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.675653446856496,
      "learning_rate": 1.89421128586026e-05,
      "loss": 0.1818,
      "step": 9640
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5533655329745836,
      "learning_rate": 1.893730029967407e-05,
      "loss": 0.1433,
      "step": 9641
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4934963177602694,
      "learning_rate": 1.893248797943837e-05,
      "loss": 0.3887,
      "step": 9642
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3660702228897026,
      "learning_rate": 1.892767589808497e-05,
      "loss": 0.3752,
      "step": 9643
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6732132708505292,
      "learning_rate": 1.892286405580331e-05,
      "loss": 0.1082,
      "step": 9644
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6650942088731981,
      "learning_rate": 1.8918052452782858e-05,
      "loss": 0.0853,
      "step": 9645
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1433709977965472,
      "learning_rate": 1.891324108921304e-05,
      "loss": 0.3187,
      "step": 9646
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5182103527457228,
      "learning_rate": 1.890842996528328e-05,
      "loss": 0.0695,
      "step": 9647
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7004744447208173,
      "learning_rate": 1.890361908118299e-05,
      "loss": 0.1509,
      "step": 9648
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5903107781400294,
      "learning_rate": 1.88988084371016e-05,
      "loss": 0.1459,
      "step": 9649
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5013476205667906,
      "learning_rate": 1.889399803322849e-05,
      "loss": 0.0879,
      "step": 9650
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4890292291237498,
      "learning_rate": 1.8889187869753054e-05,
      "loss": 0.1117,
      "step": 9651
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4941322157459119,
      "learning_rate": 1.8884377946864656e-05,
      "loss": 0.0789,
      "step": 9652
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5365394022738323,
      "learning_rate": 1.8879568264752698e-05,
      "loss": 0.0648,
      "step": 9653
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.41817294257212867,
      "learning_rate": 1.8874758823606523e-05,
      "loss": 0.064,
      "step": 9654
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0805705512262713,
      "learning_rate": 1.886994962361548e-05,
      "loss": 0.2627,
      "step": 9655
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9088081071150733,
      "learning_rate": 1.886514066496891e-05,
      "loss": 0.0902,
      "step": 9656
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.769236423075173,
      "learning_rate": 1.8860331947856154e-05,
      "loss": 0.1949,
      "step": 9657
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4056446712493376,
      "learning_rate": 1.8855523472466523e-05,
      "loss": 0.0646,
      "step": 9658
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5755069211158887,
      "learning_rate": 1.8850715238989343e-05,
      "loss": 0.1295,
      "step": 9659
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5001390292282363,
      "learning_rate": 1.8845907247613908e-05,
      "loss": 0.1044,
      "step": 9660
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.693445898293957,
      "learning_rate": 1.8841099498529517e-05,
      "loss": 0.1266,
      "step": 9661
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4100462747888619,
      "learning_rate": 1.8836291991925454e-05,
      "loss": 0.0659,
      "step": 9662
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5518442716368003,
      "learning_rate": 1.8831484727990988e-05,
      "loss": 0.0905,
      "step": 9663
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8049363932587931,
      "learning_rate": 1.882667770691539e-05,
      "loss": 0.128,
      "step": 9664
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5214143485005404,
      "learning_rate": 1.8821870928887916e-05,
      "loss": 0.0767,
      "step": 9665
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8103172389807877,
      "learning_rate": 1.8817064394097815e-05,
      "loss": 0.2019,
      "step": 9666
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4754969202602225,
      "learning_rate": 1.8812258102734317e-05,
      "loss": 0.0906,
      "step": 9667
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5873481523327405,
      "learning_rate": 1.8807452054986642e-05,
      "loss": 0.1117,
      "step": 9668
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5127413825755115,
      "learning_rate": 1.8802646251044033e-05,
      "loss": 0.1022,
      "step": 9669
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4208558990797869,
      "learning_rate": 1.8797840691095677e-05,
      "loss": 0.0896,
      "step": 9670
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.676178135934046,
      "learning_rate": 1.879303537533077e-05,
      "loss": 0.4113,
      "step": 9671
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5135281520202158,
      "learning_rate": 1.8788230303938518e-05,
      "loss": 0.0982,
      "step": 9672
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9219632018602191,
      "learning_rate": 1.878342547710809e-05,
      "loss": 0.1288,
      "step": 9673
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5152931402690517,
      "learning_rate": 1.8778620895028658e-05,
      "loss": 0.1192,
      "step": 9674
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.951000307742036,
      "learning_rate": 1.877381655788937e-05,
      "loss": 0.1485,
      "step": 9675
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.41662355138410756,
      "learning_rate": 1.8769012465879396e-05,
      "loss": 0.0889,
      "step": 9676
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5842368745174554,
      "learning_rate": 1.876420861918787e-05,
      "loss": 0.0986,
      "step": 9677
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5996395521545965,
      "learning_rate": 1.875940501800391e-05,
      "loss": 0.1409,
      "step": 9678
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5220912055798377,
      "learning_rate": 1.8754601662516646e-05,
      "loss": 0.0843,
      "step": 9679
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.457220169871511,
      "learning_rate": 1.87497985529152e-05,
      "loss": 0.0574,
      "step": 9680
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7538168757691636,
      "learning_rate": 1.8744995689388658e-05,
      "loss": 0.2527,
      "step": 9681
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7187592757122134,
      "learning_rate": 1.874019307212612e-05,
      "loss": 0.198,
      "step": 9682
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7562920397171619,
      "learning_rate": 1.8735390701316663e-05,
      "loss": 0.1265,
      "step": 9683
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6992668722966866,
      "learning_rate": 1.8730588577149372e-05,
      "loss": 0.2834,
      "step": 9684
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3881520302789588,
      "learning_rate": 1.8725786699813296e-05,
      "loss": 0.0769,
      "step": 9685
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5306511616899071,
      "learning_rate": 1.87209850694975e-05,
      "loss": 0.0694,
      "step": 9686
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4970737083061608,
      "learning_rate": 1.8716183686391013e-05,
      "loss": 0.0513,
      "step": 9687
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4116623086125638,
      "learning_rate": 1.8711382550682888e-05,
      "loss": 0.0922,
      "step": 9688
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3771503194732925,
      "learning_rate": 1.8706581662562127e-05,
      "loss": 0.0582,
      "step": 9689
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3308043296651753,
      "learning_rate": 1.8701781022217768e-05,
      "loss": 0.1183,
      "step": 9690
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7343667082330753,
      "learning_rate": 1.8696980629838795e-05,
      "loss": 0.137,
      "step": 9691
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7943314900633641,
      "learning_rate": 1.869218048561422e-05,
      "loss": 0.1732,
      "step": 9692
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6691978766269968,
      "learning_rate": 1.8687380589733017e-05,
      "loss": 0.1331,
      "step": 9693
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0085571511102236,
      "learning_rate": 1.868258094238416e-05,
      "loss": 0.5083,
      "step": 9694
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6928023758989904,
      "learning_rate": 1.8677781543756616e-05,
      "loss": 0.1397,
      "step": 9695
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.47870563534160493,
      "learning_rate": 1.867298239403935e-05,
      "loss": 0.0799,
      "step": 9696
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5104182143530853,
      "learning_rate": 1.8668183493421305e-05,
      "loss": 0.1389,
      "step": 9697
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.529869936825453,
      "learning_rate": 1.8663384842091412e-05,
      "loss": 0.0978,
      "step": 9698
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6607840854238224,
      "learning_rate": 1.865858644023859e-05,
      "loss": 0.0778,
      "step": 9699
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6054077761481341,
      "learning_rate": 1.865378828805177e-05,
      "loss": 0.1328,
      "step": 9700
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4307882531962224,
      "learning_rate": 1.8648990385719858e-05,
      "loss": 0.058,
      "step": 9701
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.746206538715446,
      "learning_rate": 1.8644192733431746e-05,
      "loss": 0.2437,
      "step": 9702
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6887384661036191,
      "learning_rate": 1.8639395331376307e-05,
      "loss": 0.1641,
      "step": 9703
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5348427837109221,
      "learning_rate": 1.8634598179742445e-05,
      "loss": 0.087,
      "step": 9704
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.607294080951613,
      "learning_rate": 1.8629801278719015e-05,
      "loss": 0.1131,
      "step": 9705
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6573481800598462,
      "learning_rate": 1.8625004628494875e-05,
      "loss": 0.1033,
      "step": 9706
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8767070203426117,
      "learning_rate": 1.8620208229258863e-05,
      "loss": 0.3446,
      "step": 9707
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1194352417387206,
      "learning_rate": 1.861541208119983e-05,
      "loss": 0.3815,
      "step": 9708
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5329964261637996,
      "learning_rate": 1.8610616184506603e-05,
      "loss": 0.0677,
      "step": 9709
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6465010896458938,
      "learning_rate": 1.8605820539367994e-05,
      "loss": 0.1496,
      "step": 9710
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5327426721320389,
      "learning_rate": 1.86010251459728e-05,
      "loss": 0.1014,
      "step": 9711
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.775150358810828,
      "learning_rate": 1.8596230004509846e-05,
      "loss": 0.2125,
      "step": 9712
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5902992498383113,
      "learning_rate": 1.8591435115167907e-05,
      "loss": 0.4241,
      "step": 9713
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5158598271831417,
      "learning_rate": 1.858664047813576e-05,
      "loss": 0.0964,
      "step": 9714
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.48662370644885855,
      "learning_rate": 1.8581846093602163e-05,
      "loss": 0.0905,
      "step": 9715
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0172360436913352,
      "learning_rate": 1.8577051961755894e-05,
      "loss": 0.3268,
      "step": 9716
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7043961439927672,
      "learning_rate": 1.8572258082785694e-05,
      "loss": 0.1048,
      "step": 9717
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6211865884960114,
      "learning_rate": 1.8567464456880295e-05,
      "loss": 0.1061,
      "step": 9718
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5426876148000654,
      "learning_rate": 1.856267108422843e-05,
      "loss": 0.0916,
      "step": 9719
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3917540850531722,
      "learning_rate": 1.8557877965018817e-05,
      "loss": 0.0494,
      "step": 9720
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2020184703044763,
      "learning_rate": 1.855308509944017e-05,
      "loss": 0.2835,
      "step": 9721
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.603319541811854,
      "learning_rate": 1.8548292487681186e-05,
      "loss": 0.0991,
      "step": 9722
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7369895198909897,
      "learning_rate": 1.854350012993054e-05,
      "loss": 0.3167,
      "step": 9723
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7934520009054805,
      "learning_rate": 1.8538708026376932e-05,
      "loss": 0.0998,
      "step": 9724
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8995187565317628,
      "learning_rate": 1.8533916177209016e-05,
      "loss": 0.2937,
      "step": 9725
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6225399885539747,
      "learning_rate": 1.852912458261545e-05,
      "loss": 0.1109,
      "step": 9726
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7596914361452136,
      "learning_rate": 1.8524333242784886e-05,
      "loss": 0.1553,
      "step": 9727
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4983927164699056,
      "learning_rate": 1.851954215790597e-05,
      "loss": 0.0806,
      "step": 9728
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6824106876919993,
      "learning_rate": 1.8514751328167324e-05,
      "loss": 0.1974,
      "step": 9729
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9514409055046815,
      "learning_rate": 1.8509960753757562e-05,
      "loss": 0.1984,
      "step": 9730
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5110286170284855,
      "learning_rate": 1.8505170434865287e-05,
      "loss": 0.0691,
      "step": 9731
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4572255086866484,
      "learning_rate": 1.850038037167912e-05,
      "loss": 0.0778,
      "step": 9732
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4904652096990042,
      "learning_rate": 1.8495590564387637e-05,
      "loss": 0.0649,
      "step": 9733
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6946497767060967,
      "learning_rate": 1.849080101317941e-05,
      "loss": 0.1538,
      "step": 9734
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7831667836996248,
      "learning_rate": 1.8486011718243003e-05,
      "loss": 0.1491,
      "step": 9735
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5469339462179187,
      "learning_rate": 1.8481222679766995e-05,
      "loss": 0.1095,
      "step": 9736
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.44172175453520895,
      "learning_rate": 1.8476433897939924e-05,
      "loss": 0.0884,
      "step": 9737
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1723569957767488,
      "learning_rate": 1.8471645372950322e-05,
      "loss": 0.106,
      "step": 9738
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7137182566285412,
      "learning_rate": 1.846685710498671e-05,
      "loss": 0.1916,
      "step": 9739
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1774107257967608,
      "learning_rate": 1.846206909423763e-05,
      "loss": 0.3037,
      "step": 9740
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6907295686927999,
      "learning_rate": 1.8457281340891572e-05,
      "loss": 0.1661,
      "step": 9741
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9224841918642277,
      "learning_rate": 1.8452493845137038e-05,
      "loss": 0.1863,
      "step": 9742
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4079815182308121,
      "learning_rate": 1.8447706607162503e-05,
      "loss": 0.0755,
      "step": 9743
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4961882873736118,
      "learning_rate": 1.8442919627156463e-05,
      "loss": 0.0577,
      "step": 9744
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3759145023269415,
      "learning_rate": 1.843813290530738e-05,
      "loss": 0.0477,
      "step": 9745
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6884245260114906,
      "learning_rate": 1.8433346441803706e-05,
      "loss": 0.1775,
      "step": 9746
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6684673552989312,
      "learning_rate": 1.8428560236833877e-05,
      "loss": 0.1556,
      "step": 9747
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5049428187685809,
      "learning_rate": 1.8423774290586355e-05,
      "loss": 0.1044,
      "step": 9748
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6357339406340332,
      "learning_rate": 1.8418988603249548e-05,
      "loss": 0.1241,
      "step": 9749
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.47165161717864335,
      "learning_rate": 1.841420317501188e-05,
      "loss": 0.1019,
      "step": 9750
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8880355550906641,
      "learning_rate": 1.840941800606174e-05,
      "loss": 0.2326,
      "step": 9751
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0090003415314837,
      "learning_rate": 1.8404633096587547e-05,
      "loss": 0.1979,
      "step": 9752
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6742761687020274,
      "learning_rate": 1.8399848446777673e-05,
      "loss": 0.1549,
      "step": 9753
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2458949899833467,
      "learning_rate": 1.83950640568205e-05,
      "loss": 0.5219,
      "step": 9754
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.44444880212788385,
      "learning_rate": 1.839027992690438e-05,
      "loss": 0.0722,
      "step": 9755
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6081511555831287,
      "learning_rate": 1.8385496057217687e-05,
      "loss": 0.1058,
      "step": 9756
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6939245842437826,
      "learning_rate": 1.8380712447948744e-05,
      "loss": 0.1327,
      "step": 9757
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5246705371592444,
      "learning_rate": 1.8375929099285906e-05,
      "loss": 0.0694,
      "step": 9758
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8471055768587445,
      "learning_rate": 1.837114601141748e-05,
      "loss": 0.2931,
      "step": 9759
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6172001104861804,
      "learning_rate": 1.836636318453179e-05,
      "loss": 0.0977,
      "step": 9760
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7473878450251541,
      "learning_rate": 1.8361580618817134e-05,
      "loss": 0.15,
      "step": 9761
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2067153957264245,
      "learning_rate": 1.8356798314461803e-05,
      "loss": 0.2679,
      "step": 9762
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.742180822828984,
      "learning_rate": 1.835201627165408e-05,
      "loss": 0.1156,
      "step": 9763
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1403551345475518,
      "learning_rate": 1.8347234490582255e-05,
      "loss": 0.181,
      "step": 9764
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5923176628485228,
      "learning_rate": 1.8342452971434567e-05,
      "loss": 0.0494,
      "step": 9765
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.49665000157279476,
      "learning_rate": 1.8337671714399283e-05,
      "loss": 0.108,
      "step": 9766
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6017029266627323,
      "learning_rate": 1.8332890719664625e-05,
      "loss": 0.0936,
      "step": 9767
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.931357726278549,
      "learning_rate": 1.8328109987418846e-05,
      "loss": 0.2035,
      "step": 9768
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7449387546057976,
      "learning_rate": 1.8323329517850164e-05,
      "loss": 0.1018,
      "step": 9769
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.191607189903413,
      "learning_rate": 1.8318549311146777e-05,
      "loss": 0.3683,
      "step": 9770
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4316876297336934,
      "learning_rate": 1.8313769367496885e-05,
      "loss": 0.049,
      "step": 9771
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.41705690165801707,
      "learning_rate": 1.8308989687088695e-05,
      "loss": 0.0799,
      "step": 9772
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5525603480632003,
      "learning_rate": 1.8304210270110378e-05,
      "loss": 0.0724,
      "step": 9773
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4839265256978146,
      "learning_rate": 1.82994311167501e-05,
      "loss": 0.062,
      "step": 9774
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5018977823621482,
      "learning_rate": 1.829465222719601e-05,
      "loss": 0.0752,
      "step": 9775
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7136671439297032,
      "learning_rate": 1.828987360163628e-05,
      "loss": 0.1175,
      "step": 9776
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5206680127288029,
      "learning_rate": 1.828509524025903e-05,
      "loss": 0.083,
      "step": 9777
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.48912701477946813,
      "learning_rate": 1.8280317143252402e-05,
      "loss": 0.0807,
      "step": 9778
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6351288407086493,
      "learning_rate": 1.8275539310804484e-05,
      "loss": 0.2793,
      "step": 9779
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4611661014214448,
      "learning_rate": 1.827076174310342e-05,
      "loss": 0.0866,
      "step": 9780
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8242824233697035,
      "learning_rate": 1.8265984440337285e-05,
      "loss": 0.1995,
      "step": 9781
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6553976706542568,
      "learning_rate": 1.826120740269417e-05,
      "loss": 0.3233,
      "step": 9782
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7965604649236497,
      "learning_rate": 1.825643063036214e-05,
      "loss": 0.2532,
      "step": 9783
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5682833314276324,
      "learning_rate": 1.8251654123529273e-05,
      "loss": 0.0652,
      "step": 9784
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8253684651255583,
      "learning_rate": 1.8246877882383627e-05,
      "loss": 0.1698,
      "step": 9785
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5802921622424551,
      "learning_rate": 1.8242101907113234e-05,
      "loss": 0.1034,
      "step": 9786
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5220036525773311,
      "learning_rate": 1.8237326197906128e-05,
      "loss": 0.0832,
      "step": 9787
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.36082000443411955,
      "learning_rate": 1.8232550754950338e-05,
      "loss": 0.0493,
      "step": 9788
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6230644158330061,
      "learning_rate": 1.8227775578433877e-05,
      "loss": 0.1019,
      "step": 9789
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4818594597047954,
      "learning_rate": 1.822300066854475e-05,
      "loss": 0.0504,
      "step": 9790
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9300597690062692,
      "learning_rate": 1.8218226025470934e-05,
      "loss": 0.1759,
      "step": 9791
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4217558662622757,
      "learning_rate": 1.8213451649400425e-05,
      "loss": 0.0585,
      "step": 9792
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.482352374144942,
      "learning_rate": 1.820867754052119e-05,
      "loss": 0.0739,
      "step": 9793
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4849238871375223,
      "learning_rate": 1.820390369902118e-05,
      "loss": 0.0695,
      "step": 9794
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.83742897950201,
      "learning_rate": 1.8199130125088354e-05,
      "loss": 0.1258,
      "step": 9795
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5296467636617804,
      "learning_rate": 1.8194356818910653e-05,
      "loss": 0.0725,
      "step": 9796
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3926048914318702,
      "learning_rate": 1.8189583780676006e-05,
      "loss": 0.0651,
      "step": 9797
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.30233239910794474,
      "learning_rate": 1.8184811010572322e-05,
      "loss": 0.0404,
      "step": 9798
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8071368877706855,
      "learning_rate": 1.81800385087875e-05,
      "loss": 0.2736,
      "step": 9799
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4302928609495141,
      "learning_rate": 1.8175266275509466e-05,
      "loss": 0.0785,
      "step": 9800
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.47731983104642184,
      "learning_rate": 1.8170494310926085e-05,
      "loss": 0.0664,
      "step": 9801
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5562223348879374,
      "learning_rate": 1.816572261522524e-05,
      "loss": 0.0531,
      "step": 9802
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6307219811181088,
      "learning_rate": 1.8160951188594783e-05,
      "loss": 0.1133,
      "step": 9803
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4996049999428547,
      "learning_rate": 1.815618003122259e-05,
      "loss": 0.0848,
      "step": 9804
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4829872410547301,
      "learning_rate": 1.8151409143296492e-05,
      "loss": 0.0652,
      "step": 9805
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3620450870764802,
      "learning_rate": 1.8146638525004313e-05,
      "loss": 0.3738,
      "step": 9806
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5310966701494023,
      "learning_rate": 1.81418681765339e-05,
      "loss": 0.0812,
      "step": 9807
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7649178612047315,
      "learning_rate": 1.8137098098073053e-05,
      "loss": 0.2975,
      "step": 9808
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8143574611049056,
      "learning_rate": 1.813232828980957e-05,
      "loss": 0.2502,
      "step": 9809
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0966771697233229,
      "learning_rate": 1.8127558751931233e-05,
      "loss": 0.2526,
      "step": 9810
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8000568067970095,
      "learning_rate": 1.8122789484625847e-05,
      "loss": 0.6514,
      "step": 9811
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.378872252780618,
      "learning_rate": 1.8118020488081163e-05,
      "loss": 0.0519,
      "step": 9812
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.282618084725414,
      "learning_rate": 1.811325176248495e-05,
      "loss": 0.4197,
      "step": 9813
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6932896871485514,
      "learning_rate": 1.8108483308024935e-05,
      "loss": 0.117,
      "step": 9814
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.42996215819846123,
      "learning_rate": 1.8103715124888884e-05,
      "loss": 0.0994,
      "step": 9815
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5433861363525637,
      "learning_rate": 1.809894721326451e-05,
      "loss": 0.094,
      "step": 9816
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6806830298462834,
      "learning_rate": 1.8094179573339532e-05,
      "loss": 0.1382,
      "step": 9817
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.060466394679788,
      "learning_rate": 1.8089412205301646e-05,
      "loss": 0.3042,
      "step": 9818
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6363965739862348,
      "learning_rate": 1.808464510933856e-05,
      "loss": 0.0675,
      "step": 9819
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5309073676424094,
      "learning_rate": 1.8079878285637947e-05,
      "loss": 0.121,
      "step": 9820
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5294412890698114,
      "learning_rate": 1.807511173438749e-05,
      "loss": 0.1045,
      "step": 9821
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0475496227886525,
      "learning_rate": 1.8070345455774844e-05,
      "loss": 0.1942,
      "step": 9822
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5921237841584941,
      "learning_rate": 1.8065579449987667e-05,
      "loss": 0.1334,
      "step": 9823
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6967012924486286,
      "learning_rate": 1.80608137172136e-05,
      "loss": 0.1477,
      "step": 9824
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6287869185845231,
      "learning_rate": 1.805604825764026e-05,
      "loss": 0.1952,
      "step": 9825
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9004490346545315,
      "learning_rate": 1.8051283071455277e-05,
      "loss": 0.301,
      "step": 9826
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5030026994620341,
      "learning_rate": 1.8046518158846267e-05,
      "loss": 0.1074,
      "step": 9827
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5593639011096223,
      "learning_rate": 1.804175352000082e-05,
      "loss": 0.1271,
      "step": 9828
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5719056035003619,
      "learning_rate": 1.8036989155106525e-05,
      "loss": 0.0718,
      "step": 9829
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0132393803295028,
      "learning_rate": 1.8032225064350945e-05,
      "loss": 0.3666,
      "step": 9830
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4028945577447058,
      "learning_rate": 1.8027461247921668e-05,
      "loss": 0.0717,
      "step": 9831
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6558826424704459,
      "learning_rate": 1.8022697706006242e-05,
      "loss": 0.1175,
      "step": 9832
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7786474698476223,
      "learning_rate": 1.8017934438792203e-05,
      "loss": 0.2192,
      "step": 9833
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7216665042247113,
      "learning_rate": 1.801317144646708e-05,
      "loss": 0.1079,
      "step": 9834
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5671350294724835,
      "learning_rate": 1.8008408729218417e-05,
      "loss": 0.1204,
      "step": 9835
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7269644656585252,
      "learning_rate": 1.800364628723371e-05,
      "loss": 0.1037,
      "step": 9836
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5620996744524716,
      "learning_rate": 1.7998884120700465e-05,
      "loss": 0.0849,
      "step": 9837
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9763099424346282,
      "learning_rate": 1.799412222980616e-05,
      "loss": 0.2271,
      "step": 9838
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5215271322537552,
      "learning_rate": 1.7989360614738293e-05,
      "loss": 0.0966,
      "step": 9839
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6252448972766923,
      "learning_rate": 1.7984599275684323e-05,
      "loss": 0.1134,
      "step": 9840
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4424086371887948,
      "learning_rate": 1.7979838212831706e-05,
      "loss": 0.0944,
      "step": 9841
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4459320374159907,
      "learning_rate": 1.7975077426367882e-05,
      "loss": 0.0814,
      "step": 9842
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.520046571311663,
      "learning_rate": 1.79703169164803e-05,
      "loss": 0.1017,
      "step": 9843
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5128458090979574,
      "learning_rate": 1.7965556683356382e-05,
      "loss": 0.0618,
      "step": 9844
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8646998270381872,
      "learning_rate": 1.796079672718354e-05,
      "loss": 0.2119,
      "step": 9845
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7022254338584472,
      "learning_rate": 1.7956037048149164e-05,
      "loss": 0.1384,
      "step": 9846
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7957240058514494,
      "learning_rate": 1.7951277646440667e-05,
      "loss": 0.1269,
      "step": 9847
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7866220108314662,
      "learning_rate": 1.7946518522245422e-05,
      "loss": 0.1583,
      "step": 9848
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5641555110351786,
      "learning_rate": 1.7941759675750796e-05,
      "loss": 0.0754,
      "step": 9849
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6227592456770867,
      "learning_rate": 1.793700110714415e-05,
      "loss": 0.1181,
      "step": 9850
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9606784332882202,
      "learning_rate": 1.7932242816612825e-05,
      "loss": 0.2662,
      "step": 9851
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6104315191121625,
      "learning_rate": 1.7927484804344178e-05,
      "loss": 0.18,
      "step": 9852
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3615704938641848,
      "learning_rate": 1.792272707052552e-05,
      "loss": 0.2784,
      "step": 9853
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.652469610492481,
      "learning_rate": 1.7917969615344163e-05,
      "loss": 0.1573,
      "step": 9854
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5750961347291401,
      "learning_rate": 1.791321243898742e-05,
      "loss": 0.1028,
      "step": 9855
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5746542182579957,
      "learning_rate": 1.7908455541642584e-05,
      "loss": 0.1026,
      "step": 9856
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5996050045371848,
      "learning_rate": 1.7903698923496936e-05,
      "loss": 0.0737,
      "step": 9857
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5806265254494982,
      "learning_rate": 1.7898942584737744e-05,
      "loss": 0.1406,
      "step": 9858
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1065144053730764,
      "learning_rate": 1.7894186525552275e-05,
      "loss": 0.3434,
      "step": 9859
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.94139740810056,
      "learning_rate": 1.7889430746127776e-05,
      "loss": 0.2711,
      "step": 9860
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5889967556864077,
      "learning_rate": 1.7884675246651485e-05,
      "loss": 0.0995,
      "step": 9861
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7164736728572793,
      "learning_rate": 1.7879920027310622e-05,
      "loss": 0.2845,
      "step": 9862
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4816181484641424,
      "learning_rate": 1.7875165088292413e-05,
      "loss": 0.6198,
      "step": 9863
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5666798697798205,
      "learning_rate": 1.787041042978407e-05,
      "loss": 0.0853,
      "step": 9864
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.62380843113532,
      "learning_rate": 1.7865656051972766e-05,
      "loss": 0.1144,
      "step": 9865
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.795315239148576,
      "learning_rate": 1.7860901955045694e-05,
      "loss": 0.2444,
      "step": 9866
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9291171722149747,
      "learning_rate": 1.785614813919004e-05,
      "loss": 0.1593,
      "step": 9867
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7858958389074484,
      "learning_rate": 1.785139460459295e-05,
      "loss": 0.1316,
      "step": 9868
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8667990674882579,
      "learning_rate": 1.7846641351441578e-05,
      "loss": 0.2788,
      "step": 9869
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.48227583810664426,
      "learning_rate": 1.7841888379923052e-05,
      "loss": 0.0875,
      "step": 9870
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.42767106069352223,
      "learning_rate": 1.783713569022452e-05,
      "loss": 0.0564,
      "step": 9871
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3495983996303849,
      "learning_rate": 1.7832383282533094e-05,
      "loss": 0.0422,
      "step": 9872
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.034905282858618,
      "learning_rate": 1.782763115703587e-05,
      "loss": 0.409,
      "step": 9873
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9316952208864967,
      "learning_rate": 1.782287931391994e-05,
      "loss": 0.2578,
      "step": 9874
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3909250729684214,
      "learning_rate": 1.7818127753372403e-05,
      "loss": 0.0653,
      "step": 9875
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7889679387094966,
      "learning_rate": 1.7813376475580325e-05,
      "loss": 0.2267,
      "step": 9876
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.510645150711117,
      "learning_rate": 1.7808625480730767e-05,
      "loss": 0.1018,
      "step": 9877
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5541562407027721,
      "learning_rate": 1.7803874769010763e-05,
      "loss": 0.1677,
      "step": 9878
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.38813852524085346,
      "learning_rate": 1.779912434060738e-05,
      "loss": 0.0553,
      "step": 9879
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7895666349640605,
      "learning_rate": 1.7794374195707636e-05,
      "loss": 0.1323,
      "step": 9880
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4657987558725811,
      "learning_rate": 1.7789624334498538e-05,
      "loss": 0.078,
      "step": 9881
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7137782443112726,
      "learning_rate": 1.778487475716709e-05,
      "loss": 0.1429,
      "step": 9882
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8546598506871287,
      "learning_rate": 1.7780125463900305e-05,
      "loss": 0.3058,
      "step": 9883
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4514565547908307,
      "learning_rate": 1.7775376454885153e-05,
      "loss": 0.3793,
      "step": 9884
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6872916092075715,
      "learning_rate": 1.7770627730308608e-05,
      "loss": 0.1807,
      "step": 9885
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6683212041499157,
      "learning_rate": 1.7765879290357624e-05,
      "loss": 0.1273,
      "step": 9886
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8894218552562069,
      "learning_rate": 1.7761131135219164e-05,
      "loss": 0.2089,
      "step": 9887
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.43002145356329946,
      "learning_rate": 1.7756383265080155e-05,
      "loss": 0.0579,
      "step": 9888
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8046940010572033,
      "learning_rate": 1.775163568012753e-05,
      "loss": 0.2184,
      "step": 9889
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5901255474719324,
      "learning_rate": 1.7746888380548198e-05,
      "loss": 0.1264,
      "step": 9890
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7111260815782996,
      "learning_rate": 1.7742141366529076e-05,
      "loss": 0.1889,
      "step": 9891
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8352287335364332,
      "learning_rate": 1.773739463825705e-05,
      "loss": 0.0785,
      "step": 9892
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5127090543072589,
      "learning_rate": 1.7732648195918993e-05,
      "loss": 0.1201,
      "step": 9893
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.43791723658845166,
      "learning_rate": 1.772790203970179e-05,
      "loss": 0.0784,
      "step": 9894
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8625610862763795,
      "learning_rate": 1.7723156169792297e-05,
      "loss": 0.2012,
      "step": 9895
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5235198898063634,
      "learning_rate": 1.7718410586377362e-05,
      "loss": 0.0811,
      "step": 9896
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6161814546781142,
      "learning_rate": 1.771366528964382e-05,
      "loss": 0.2044,
      "step": 9897
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5765503650898176,
      "learning_rate": 1.770892027977849e-05,
      "loss": 0.0921,
      "step": 9898
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.41705933626019986,
      "learning_rate": 1.77041755569682e-05,
      "loss": 0.0566,
      "step": 9899
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5347042047347768,
      "learning_rate": 1.769943112139975e-05,
      "loss": 0.1142,
      "step": 9900
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.597552039100281,
      "learning_rate": 1.769468697325993e-05,
      "loss": 0.1049,
      "step": 9901
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8449602558603151,
      "learning_rate": 1.7689943112735507e-05,
      "loss": 0.2689,
      "step": 9902
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.695739275227711,
      "learning_rate": 1.7685199540013274e-05,
      "loss": 0.128,
      "step": 9903
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6095395817214707,
      "learning_rate": 1.7680456255279976e-05,
      "loss": 0.0821,
      "step": 9904
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5263679326753405,
      "learning_rate": 1.767571325872236e-05,
      "loss": 0.0941,
      "step": 9905
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.49769219657001906,
      "learning_rate": 1.7670970550527155e-05,
      "loss": 0.1518,
      "step": 9906
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5146171302063246,
      "learning_rate": 1.76662281308811e-05,
      "loss": 0.229,
      "step": 9907
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.46623664139980325,
      "learning_rate": 1.76614859999709e-05,
      "loss": 0.1037,
      "step": 9908
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6512400359593005,
      "learning_rate": 1.7656744157983253e-05,
      "loss": 0.1126,
      "step": 9909
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4871091279634654,
      "learning_rate": 1.7652002605104845e-05,
      "loss": 0.0975,
      "step": 9910
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6691240103596998,
      "learning_rate": 1.764726134152237e-05,
      "loss": 0.1415,
      "step": 9911
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.427732177011704,
      "learning_rate": 1.7642520367422483e-05,
      "loss": 0.1933,
      "step": 9912
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4133021586791952,
      "learning_rate": 1.7637779682991846e-05,
      "loss": 0.0574,
      "step": 9913
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6076174936245524,
      "learning_rate": 1.7633039288417086e-05,
      "loss": 0.0797,
      "step": 9914
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5989602602272563,
      "learning_rate": 1.762829918388486e-05,
      "loss": 0.0862,
      "step": 9915
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5531942926203842,
      "learning_rate": 1.7623559369581776e-05,
      "loss": 0.086,
      "step": 9916
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.42605625635055494,
      "learning_rate": 1.7618819845694452e-05,
      "loss": 0.0637,
      "step": 9917
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5001900428508594,
      "learning_rate": 1.7614080612409466e-05,
      "loss": 0.0659,
      "step": 9918
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5844163354864472,
      "learning_rate": 1.7609341669913428e-05,
      "loss": 0.0997,
      "step": 9919
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5331668074362647,
      "learning_rate": 1.7604603018392906e-05,
      "loss": 0.1326,
      "step": 9920
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6237112961239971,
      "learning_rate": 1.7599864658034464e-05,
      "loss": 0.0964,
      "step": 9921
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6648866896136766,
      "learning_rate": 1.7595126589024653e-05,
      "loss": 0.0682,
      "step": 9922
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8180978776585668,
      "learning_rate": 1.759038881155002e-05,
      "loss": 0.1993,
      "step": 9923
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.45518906816057875,
      "learning_rate": 1.758565132579708e-05,
      "loss": 0.1005,
      "step": 9924
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4594402020607689,
      "learning_rate": 1.7580914131952363e-05,
      "loss": 0.0874,
      "step": 9925
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6886978954903024,
      "learning_rate": 1.7576177230202376e-05,
      "loss": 0.2428,
      "step": 9926
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4745195895795862,
      "learning_rate": 1.7571440620733613e-05,
      "loss": 0.0685,
      "step": 9927
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7408041763187978,
      "learning_rate": 1.756670430373256e-05,
      "loss": 0.4387,
      "step": 9928
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.605450467539774,
      "learning_rate": 1.756196827938568e-05,
      "loss": 0.1266,
      "step": 9929
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.64905645783588,
      "learning_rate": 1.7557232547879436e-05,
      "loss": 0.1265,
      "step": 9930
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3622483673442494,
      "learning_rate": 1.755249710940029e-05,
      "loss": 0.0588,
      "step": 9931
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.49988357835288055,
      "learning_rate": 1.7547761964134664e-05,
      "loss": 0.0708,
      "step": 9932
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.41685239290638665,
      "learning_rate": 1.7543027112268994e-05,
      "loss": 0.0658,
      "step": 9933
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8901053195394524,
      "learning_rate": 1.7538292553989684e-05,
      "loss": 0.1517,
      "step": 9934
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6520755379451638,
      "learning_rate": 1.753355828948315e-05,
      "loss": 0.1038,
      "step": 9935
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.595616451218645,
      "learning_rate": 1.752882431893578e-05,
      "loss": 0.1368,
      "step": 9936
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6816807443842143,
      "learning_rate": 1.7524090642533935e-05,
      "loss": 0.1212,
      "step": 9937
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7555703422994342,
      "learning_rate": 1.7519357260464013e-05,
      "loss": 0.1446,
      "step": 9938
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6052627039268704,
      "learning_rate": 1.751462417291236e-05,
      "loss": 0.1217,
      "step": 9939
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3889781849457242,
      "learning_rate": 1.750989138006531e-05,
      "loss": 0.0545,
      "step": 9940
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7072558319611147,
      "learning_rate": 1.7505158882109203e-05,
      "loss": 0.1626,
      "step": 9941
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5305844608339046,
      "learning_rate": 1.7500426679230368e-05,
      "loss": 0.2144,
      "step": 9942
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6898348171233892,
      "learning_rate": 1.749569477161511e-05,
      "loss": 0.1188,
      "step": 9943
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6136396027287551,
      "learning_rate": 1.7490963159449725e-05,
      "loss": 0.1895,
      "step": 9944
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8809241085228672,
      "learning_rate": 1.748623184292049e-05,
      "loss": 0.1441,
      "step": 9945
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5390030167643571,
      "learning_rate": 1.7481500822213708e-05,
      "loss": 0.1381,
      "step": 9946
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6651192875495526,
      "learning_rate": 1.7476770097515627e-05,
      "loss": 0.1467,
      "step": 9947
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6406877034936692,
      "learning_rate": 1.7472039669012495e-05,
      "loss": 0.1827,
      "step": 9948
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5246751003781195,
      "learning_rate": 1.7467309536890556e-05,
      "loss": 0.1194,
      "step": 9949
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6200729625868631,
      "learning_rate": 1.7462579701336033e-05,
      "loss": 0.1051,
      "step": 9950
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6703937691858857,
      "learning_rate": 1.745785016253516e-05,
      "loss": 0.1164,
      "step": 9951
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8739680156051696,
      "learning_rate": 1.7453120920674126e-05,
      "loss": 0.1513,
      "step": 9952
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.39880069409204105,
      "learning_rate": 1.744839197593913e-05,
      "loss": 0.0685,
      "step": 9953
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6198660616612706,
      "learning_rate": 1.7443663328516358e-05,
      "loss": 0.115,
      "step": 9954
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5641560582709378,
      "learning_rate": 1.7438934978591977e-05,
      "loss": 0.1131,
      "step": 9955
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0256178431807714,
      "learning_rate": 1.7434206926352134e-05,
      "loss": 0.3206,
      "step": 9956
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.46237024347624156,
      "learning_rate": 1.742947917198299e-05,
      "loss": 0.0835,
      "step": 9957
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5319666308866433,
      "learning_rate": 1.7424751715670682e-05,
      "loss": 0.1327,
      "step": 9958
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.618368656023886,
      "learning_rate": 1.7420024557601327e-05,
      "loss": 0.0794,
      "step": 9959
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7595754493382685,
      "learning_rate": 1.7415297697961038e-05,
      "loss": 0.2495,
      "step": 9960
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6706145852104699,
      "learning_rate": 1.7410571136935906e-05,
      "loss": 0.1363,
      "step": 9961
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5273776863696046,
      "learning_rate": 1.7405844874712035e-05,
      "loss": 0.0784,
      "step": 9962
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7833236526335335,
      "learning_rate": 1.7401118911475496e-05,
      "loss": 0.101,
      "step": 9963
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5190475744748357,
      "learning_rate": 1.739639324741235e-05,
      "loss": 0.1025,
      "step": 9964
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7542504859698939,
      "learning_rate": 1.739166788270864e-05,
      "loss": 0.1617,
      "step": 9965
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5701974104109336,
      "learning_rate": 1.7386942817550427e-05,
      "loss": 0.0814,
      "step": 9966
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5659857800343412,
      "learning_rate": 1.7382218052123735e-05,
      "loss": 0.0959,
      "step": 9967
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.409489942136514,
      "learning_rate": 1.737749358661458e-05,
      "loss": 0.099,
      "step": 9968
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1836282188412677,
      "learning_rate": 1.7372769421208952e-05,
      "loss": 0.2694,
      "step": 9969
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7875883599338277,
      "learning_rate": 1.7368045556092865e-05,
      "loss": 0.1883,
      "step": 9970
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5508303257219335,
      "learning_rate": 1.7363321991452296e-05,
      "loss": 0.1229,
      "step": 9971
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7447070075297928,
      "learning_rate": 1.7358598727473218e-05,
      "loss": 0.1224,
      "step": 9972
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6626076174907395,
      "learning_rate": 1.735387576434157e-05,
      "loss": 0.1005,
      "step": 9973
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.553523217389945,
      "learning_rate": 1.7349153102243322e-05,
      "loss": 0.1274,
      "step": 9974
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.0641226071336467,
      "learning_rate": 1.7344430741364403e-05,
      "loss": 0.4293,
      "step": 9975
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.356441924609528,
      "learning_rate": 1.7339708681890733e-05,
      "loss": 0.835,
      "step": 9976
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6601418626938672,
      "learning_rate": 1.7334986924008206e-05,
      "loss": 0.0975,
      "step": 9977
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3652425987645462,
      "learning_rate": 1.7330265467902752e-05,
      "loss": 0.0497,
      "step": 9978
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6079861077544512,
      "learning_rate": 1.732554431376024e-05,
      "loss": 0.1476,
      "step": 9979
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4006246239805062,
      "learning_rate": 1.7320823461766552e-05,
      "loss": 0.0411,
      "step": 9980
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5427710111471381,
      "learning_rate": 1.731610291210754e-05,
      "loss": 0.1611,
      "step": 9981
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8885795395879412,
      "learning_rate": 1.7311382664969066e-05,
      "loss": 0.1552,
      "step": 9982
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4747073005334948,
      "learning_rate": 1.730666272053697e-05,
      "loss": 0.1368,
      "step": 9983
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.402392816922503,
      "learning_rate": 1.730194307899708e-05,
      "loss": 0.0353,
      "step": 9984
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6827200368821063,
      "learning_rate": 1.7297223740535197e-05,
      "loss": 0.1265,
      "step": 9985
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.36947422925471407,
      "learning_rate": 1.7292504705337144e-05,
      "loss": 0.0881,
      "step": 9986
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7605245118239943,
      "learning_rate": 1.72877859735887e-05,
      "loss": 0.2062,
      "step": 9987
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5958537461475849,
      "learning_rate": 1.7283067545475652e-05,
      "loss": 0.1009,
      "step": 9988
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5030237356544209,
      "learning_rate": 1.727834942118376e-05,
      "loss": 0.0943,
      "step": 9989
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.377012247728662,
      "learning_rate": 1.7273631600898795e-05,
      "loss": 0.0588,
      "step": 9990
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9806889042212715,
      "learning_rate": 1.7268914084806485e-05,
      "loss": 0.3472,
      "step": 9991
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4326226665768643,
      "learning_rate": 1.726419687309257e-05,
      "loss": 0.0673,
      "step": 9992
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.46107771441897855,
      "learning_rate": 1.725947996594276e-05,
      "loss": 0.0598,
      "step": 9993
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5115783148272305,
      "learning_rate": 1.7254763363542782e-05,
      "loss": 0.0597,
      "step": 9994
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5160994064096752,
      "learning_rate": 1.725004706607832e-05,
      "loss": 0.0858,
      "step": 9995
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5237351085280181,
      "learning_rate": 1.724533107373506e-05,
      "loss": 0.1032,
      "step": 9996
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6452700035028067,
      "learning_rate": 1.724061538669866e-05,
      "loss": 0.2299,
      "step": 9997
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.702625038203272,
      "learning_rate": 1.7235900005154804e-05,
      "loss": 0.2269,
      "step": 9998
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6015777623352605,
      "learning_rate": 1.7231184929289128e-05,
      "loss": 0.1189,
      "step": 9999
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5308499158298251,
      "learning_rate": 1.7226470159287273e-05,
      "loss": 0.0877,
      "step": 10000
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6867840571098724,
      "learning_rate": 1.7221755695334847e-05,
      "loss": 0.1372,
      "step": 10001
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5361622416099714,
      "learning_rate": 1.7217041537617484e-05,
      "loss": 0.1103,
      "step": 10002
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7170217474687244,
      "learning_rate": 1.7212327686320773e-05,
      "loss": 0.2682,
      "step": 10003
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5269265020910819,
      "learning_rate": 1.7207614141630305e-05,
      "loss": 0.0721,
      "step": 10004
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.34457035226738963,
      "learning_rate": 1.720290090373164e-05,
      "loss": 0.0484,
      "step": 10005
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6929044492296338,
      "learning_rate": 1.7198187972810366e-05,
      "loss": 0.1447,
      "step": 10006
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.138414518382434,
      "learning_rate": 1.7193475349052017e-05,
      "loss": 0.9051,
      "step": 10007
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0264787862395062,
      "learning_rate": 1.7188763032642145e-05,
      "loss": 0.3017,
      "step": 10008
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5230834625558368,
      "learning_rate": 1.718405102376626e-05,
      "loss": 0.1053,
      "step": 10009
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.36151956871687996,
      "learning_rate": 1.7179339322609893e-05,
      "loss": 0.04,
      "step": 10010
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.698953755283077,
      "learning_rate": 1.7174627929358546e-05,
      "loss": 0.2325,
      "step": 10011
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5127972774109033,
      "learning_rate": 1.7169916844197705e-05,
      "loss": 0.08,
      "step": 10012
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5925351933931835,
      "learning_rate": 1.716520606731284e-05,
      "loss": 0.0985,
      "step": 10013
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4720671053476523,
      "learning_rate": 1.7160495598889437e-05,
      "loss": 0.1045,
      "step": 10014
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7145867230547787,
      "learning_rate": 1.715578543911294e-05,
      "loss": 0.1372,
      "step": 10015
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6847438941151572,
      "learning_rate": 1.715107558816879e-05,
      "loss": 0.1123,
      "step": 10016
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6443352782737856,
      "learning_rate": 1.7146366046242418e-05,
      "loss": 0.0771,
      "step": 10017
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4065072314624762,
      "learning_rate": 1.714165681351924e-05,
      "loss": 0.0807,
      "step": 10018
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.48274027624089716,
      "learning_rate": 1.7136947890184675e-05,
      "loss": 0.0745,
      "step": 10019
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6753868662516881,
      "learning_rate": 1.71322392764241e-05,
      "loss": 0.197,
      "step": 10020
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8938953857728326,
      "learning_rate": 1.71275309724229e-05,
      "loss": 0.2461,
      "step": 10021
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.47882448153826207,
      "learning_rate": 1.7122822978366458e-05,
      "loss": 0.0607,
      "step": 10022
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6717247171117475,
      "learning_rate": 1.7118115294440114e-05,
      "loss": 0.139,
      "step": 10023
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5847849951973733,
      "learning_rate": 1.711340792082922e-05,
      "loss": 0.1045,
      "step": 10024
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6451415684835556,
      "learning_rate": 1.7108700857719102e-05,
      "loss": 0.1194,
      "step": 10025
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5564840063321391,
      "learning_rate": 1.7103994105295097e-05,
      "loss": 0.1218,
      "step": 10026
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5811419694716631,
      "learning_rate": 1.7099287663742504e-05,
      "loss": 0.0746,
      "step": 10027
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6170680422006356,
      "learning_rate": 1.7094581533246616e-05,
      "loss": 0.1183,
      "step": 10028
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5918738955408929,
      "learning_rate": 1.7089875713992706e-05,
      "loss": 0.0924,
      "step": 10029
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.49072318159183614,
      "learning_rate": 1.708517020616607e-05,
      "loss": 0.0621,
      "step": 10030
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5917887450123949,
      "learning_rate": 1.7080465009951957e-05,
      "loss": 0.0935,
      "step": 10031
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3812926971864803,
      "learning_rate": 1.7075760125535613e-05,
      "loss": 0.0418,
      "step": 10032
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8650047977697268,
      "learning_rate": 1.7071055553102255e-05,
      "loss": 0.4796,
      "step": 10033
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6343681030942401,
      "learning_rate": 1.7066351292837138e-05,
      "loss": 0.1245,
      "step": 10034
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5895878002967248,
      "learning_rate": 1.7061647344925458e-05,
      "loss": 0.0978,
      "step": 10035
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4760182193267342,
      "learning_rate": 1.705694370955241e-05,
      "loss": 0.0738,
      "step": 10036
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.41347922981513446,
      "learning_rate": 1.7052240386903168e-05,
      "loss": 0.0589,
      "step": 10037
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0462096268487595,
      "learning_rate": 1.7047537377162928e-05,
      "loss": 0.2337,
      "step": 10038
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.46012609450172687,
      "learning_rate": 1.7042834680516843e-05,
      "loss": 0.1348,
      "step": 10039
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5594617258953365,
      "learning_rate": 1.7038132297150057e-05,
      "loss": 0.0692,
      "step": 10040
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5622551802796408,
      "learning_rate": 1.70334302272477e-05,
      "loss": 0.0887,
      "step": 10041
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4926007098765357,
      "learning_rate": 1.7028728470994915e-05,
      "loss": 0.3756,
      "step": 10042
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7280974376890318,
      "learning_rate": 1.7024027028576803e-05,
      "loss": 0.1726,
      "step": 10043
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3610468677398264,
      "learning_rate": 1.701932590017846e-05,
      "loss": 0.0469,
      "step": 10044
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6741142164004722,
      "learning_rate": 1.701462508598497e-05,
      "loss": 0.1658,
      "step": 10045
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6183607083272616,
      "learning_rate": 1.7009924586181423e-05,
      "loss": 0.1013,
      "step": 10046
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8401423822730504,
      "learning_rate": 1.7005224400952868e-05,
      "loss": 0.2051,
      "step": 10047
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0704627987183242,
      "learning_rate": 1.700052453048436e-05,
      "loss": 0.1618,
      "step": 10048
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.47766341901085274,
      "learning_rate": 1.699582497496093e-05,
      "loss": 0.0661,
      "step": 10049
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4344536609779949,
      "learning_rate": 1.6991125734567605e-05,
      "loss": 0.0498,
      "step": 10050
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7490196624151676,
      "learning_rate": 1.698642680948941e-05,
      "loss": 0.1591,
      "step": 10051
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4415392034507967,
      "learning_rate": 1.698172819991133e-05,
      "loss": 0.0462,
      "step": 10052
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0349707041296587,
      "learning_rate": 1.697702990601835e-05,
      "loss": 0.2735,
      "step": 10053
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5495072641173253,
      "learning_rate": 1.697233192799546e-05,
      "loss": 0.1998,
      "step": 10054
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5487520680154804,
      "learning_rate": 1.696763426602761e-05,
      "loss": 0.0723,
      "step": 10055
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4387226964474124,
      "learning_rate": 1.696293692029976e-05,
      "loss": 0.096,
      "step": 10056
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8575865087432696,
      "learning_rate": 1.695823989099684e-05,
      "loss": 0.3283,
      "step": 10057
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4462308030462065,
      "learning_rate": 1.6953543178303787e-05,
      "loss": 0.0892,
      "step": 10058
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.013312259474142,
      "learning_rate": 1.69488467824055e-05,
      "loss": 0.224,
      "step": 10059
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5978978795430235,
      "learning_rate": 1.694415070348689e-05,
      "loss": 0.0943,
      "step": 10060
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8667457976012641,
      "learning_rate": 1.6939454941732823e-05,
      "loss": 0.177,
      "step": 10061
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.743433469321669,
      "learning_rate": 1.6934759497328206e-05,
      "loss": 0.1646,
      "step": 10062
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.47549675938295105,
      "learning_rate": 1.6930064370457887e-05,
      "loss": 0.0665,
      "step": 10063
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5153540337121248,
      "learning_rate": 1.692536956130672e-05,
      "loss": 0.121,
      "step": 10064
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7877824965428731,
      "learning_rate": 1.692067507005953e-05,
      "loss": 0.1753,
      "step": 10065
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8559099782272378,
      "learning_rate": 1.691598089690116e-05,
      "loss": 0.2377,
      "step": 10066
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6935699137912483,
      "learning_rate": 1.6911287042016415e-05,
      "loss": 0.1166,
      "step": 10067
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0192742727576563,
      "learning_rate": 1.6906593505590103e-05,
      "loss": 0.152,
      "step": 10068
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4986963456206054,
      "learning_rate": 1.6901900287806988e-05,
      "loss": 0.1065,
      "step": 10069
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7642751721686797,
      "learning_rate": 1.6897207388851874e-05,
      "loss": 0.1259,
      "step": 10070
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5479394344804361,
      "learning_rate": 1.6892514808909512e-05,
      "loss": 0.127,
      "step": 10071
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4195054787692936,
      "learning_rate": 1.6887822548164644e-05,
      "loss": 0.3298,
      "step": 10072
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5250212685905697,
      "learning_rate": 1.6883130606802023e-05,
      "loss": 0.0994,
      "step": 10073
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7406101846571014,
      "learning_rate": 1.687843898500637e-05,
      "loss": 0.1804,
      "step": 10074
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8194354204026663,
      "learning_rate": 1.6873747682962394e-05,
      "loss": 0.1958,
      "step": 10075
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6233939009011403,
      "learning_rate": 1.6869056700854787e-05,
      "loss": 0.1282,
      "step": 10076
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5398164942053569,
      "learning_rate": 1.6864366038868252e-05,
      "loss": 0.08,
      "step": 10077
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7705332809828235,
      "learning_rate": 1.6859675697187458e-05,
      "loss": 0.1697,
      "step": 10078
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8549646186362718,
      "learning_rate": 1.6854985675997066e-05,
      "loss": 0.1725,
      "step": 10079
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.639473472897914,
      "learning_rate": 1.685029597548172e-05,
      "loss": 0.1324,
      "step": 10080
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5578757493749129,
      "learning_rate": 1.6845606595826062e-05,
      "loss": 0.1056,
      "step": 10081
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6559877975456312,
      "learning_rate": 1.6840917537214722e-05,
      "loss": 0.1488,
      "step": 10082
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5994499169809355,
      "learning_rate": 1.6836228799832304e-05,
      "loss": 0.1284,
      "step": 10083
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4947397054562182,
      "learning_rate": 1.6831540383863405e-05,
      "loss": 0.0715,
      "step": 10084
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6060165873792168,
      "learning_rate": 1.6826852289492624e-05,
      "loss": 0.1664,
      "step": 10085
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.563557491844756,
      "learning_rate": 1.682216451690452e-05,
      "loss": 0.1066,
      "step": 10086
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8591029971525457,
      "learning_rate": 1.681747706628366e-05,
      "loss": 0.1899,
      "step": 10087
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6180995142170388,
      "learning_rate": 1.6812789937814587e-05,
      "loss": 0.1609,
      "step": 10088
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5638679045959782,
      "learning_rate": 1.680810313168185e-05,
      "loss": 0.1076,
      "step": 10089
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.642289367013954,
      "learning_rate": 1.6803416648069963e-05,
      "loss": 0.1101,
      "step": 10090
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9723285865044577,
      "learning_rate": 1.679873048716344e-05,
      "loss": 0.2198,
      "step": 10091
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.41355908561905974,
      "learning_rate": 1.6794044649146758e-05,
      "loss": 0.0637,
      "step": 10092
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.695762907213066,
      "learning_rate": 1.6789359134204436e-05,
      "loss": 0.1083,
      "step": 10093
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5954606034838479,
      "learning_rate": 1.6784673942520925e-05,
      "loss": 0.0839,
      "step": 10094
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5536623335992862,
      "learning_rate": 1.677998907428069e-05,
      "loss": 0.0969,
      "step": 10095
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7903957865508506,
      "learning_rate": 1.6775304529668163e-05,
      "loss": 0.1976,
      "step": 10096
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7551007642789312,
      "learning_rate": 1.6770620308867806e-05,
      "loss": 0.1992,
      "step": 10097
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8879231291306999,
      "learning_rate": 1.6765936412064026e-05,
      "loss": 0.1558,
      "step": 10098
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.497407391209954,
      "learning_rate": 1.676125283944123e-05,
      "loss": 0.0897,
      "step": 10099
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4084181183525142,
      "learning_rate": 1.67565695911838e-05,
      "loss": 0.0549,
      "step": 10100
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3062267537633704,
      "learning_rate": 1.6751886667476145e-05,
      "loss": 0.0534,
      "step": 10101
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.573791666207521,
      "learning_rate": 1.6747204068502623e-05,
      "loss": 0.1551,
      "step": 10102
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.653958321824781,
      "learning_rate": 1.6742521794447592e-05,
      "loss": 0.1858,
      "step": 10103
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.683744247295065,
      "learning_rate": 1.6737839845495383e-05,
      "loss": 0.1299,
      "step": 10104
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.47202045737755166,
      "learning_rate": 1.6733158221830356e-05,
      "loss": 0.176,
      "step": 10105
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6640162708374382,
      "learning_rate": 1.672847692363681e-05,
      "loss": 0.1087,
      "step": 10106
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7168166181632911,
      "learning_rate": 1.672379595109906e-05,
      "loss": 0.1261,
      "step": 10107
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0582062582326692,
      "learning_rate": 1.671911530440138e-05,
      "loss": 0.2806,
      "step": 10108
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.36473476919403625,
      "learning_rate": 1.6714434983728076e-05,
      "loss": 0.0688,
      "step": 10109
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.590451576032268,
      "learning_rate": 1.670975498926341e-05,
      "loss": 0.1215,
      "step": 10110
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.501930532730136,
      "learning_rate": 1.6705075321191626e-05,
      "loss": 0.1173,
      "step": 10111
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4803198841424378,
      "learning_rate": 1.6700395979696962e-05,
      "loss": 0.0773,
      "step": 10112
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5745357884556236,
      "learning_rate": 1.6695716964963668e-05,
      "loss": 0.095,
      "step": 10113
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3306041825960078,
      "learning_rate": 1.6691038277175947e-05,
      "loss": 0.0448,
      "step": 10114
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.43316558539477623,
      "learning_rate": 1.668635991651801e-05,
      "loss": 0.0605,
      "step": 10115
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5927070670387448,
      "learning_rate": 1.668168188317403e-05,
      "loss": 0.1271,
      "step": 10116
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4655778488967715,
      "learning_rate": 1.6677004177328198e-05,
      "loss": 0.2664,
      "step": 10117
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1315292394863727,
      "learning_rate": 1.6672326799164677e-05,
      "loss": 0.2398,
      "step": 10118
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0176041020348352,
      "learning_rate": 1.666764974886762e-05,
      "loss": 0.3999,
      "step": 10119
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4001655535443899,
      "learning_rate": 1.666297302662116e-05,
      "loss": 0.061,
      "step": 10120
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5449605721822866,
      "learning_rate": 1.665829663260943e-05,
      "loss": 0.1105,
      "step": 10121
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6822469501190491,
      "learning_rate": 1.665362056701654e-05,
      "loss": 0.1001,
      "step": 10122
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6361065632468196,
      "learning_rate": 1.6648944830026592e-05,
      "loss": 0.131,
      "step": 10123
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6447340472560072,
      "learning_rate": 1.6644269421823666e-05,
      "loss": 0.1023,
      "step": 10124
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3496101825155563,
      "learning_rate": 1.6639594342591843e-05,
      "loss": 0.0405,
      "step": 10125
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9962275860031022,
      "learning_rate": 1.6634919592515187e-05,
      "loss": 0.2837,
      "step": 10126
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5962327934965173,
      "learning_rate": 1.663024517177774e-05,
      "loss": 0.2031,
      "step": 10127
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5897463013574709,
      "learning_rate": 1.662557108056353e-05,
      "loss": 0.2068,
      "step": 10128
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8005207773241232,
      "learning_rate": 1.6620897319056597e-05,
      "loss": 0.1406,
      "step": 10129
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3826334936224156,
      "learning_rate": 1.661622388744094e-05,
      "loss": 0.0811,
      "step": 10130
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6193731936603826,
      "learning_rate": 1.661155078590056e-05,
      "loss": 0.1552,
      "step": 10131
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3100700153981523,
      "learning_rate": 1.6606878014619424e-05,
      "loss": 0.0511,
      "step": 10132
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.42969665410566077,
      "learning_rate": 1.6602205573781533e-05,
      "loss": 0.0486,
      "step": 10133
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6680458545100215,
      "learning_rate": 1.659753346357082e-05,
      "loss": 0.1508,
      "step": 10134
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5443341718778865,
      "learning_rate": 1.659286168417124e-05,
      "loss": 0.1755,
      "step": 10135
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5389389936062009,
      "learning_rate": 1.658819023576671e-05,
      "loss": 0.0857,
      "step": 10136
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4840099006476106,
      "learning_rate": 1.658351911854117e-05,
      "loss": 0.0685,
      "step": 10137
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9689745444246658,
      "learning_rate": 1.6578848332678508e-05,
      "loss": 0.0916,
      "step": 10138
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9055558047748008,
      "learning_rate": 1.6574177878362628e-05,
      "loss": 0.2295,
      "step": 10139
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8497231479993768,
      "learning_rate": 1.6569507755777392e-05,
      "loss": 0.1895,
      "step": 10140
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6335219160720117,
      "learning_rate": 1.656483796510669e-05,
      "loss": 0.1471,
      "step": 10141
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0322648403126855,
      "learning_rate": 1.6560168506534356e-05,
      "loss": 0.3606,
      "step": 10142
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5345849974560297,
      "learning_rate": 1.6555499380244245e-05,
      "loss": 0.1139,
      "step": 10143
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.32806101040827323,
      "learning_rate": 1.655083058642016e-05,
      "loss": 0.0471,
      "step": 10144
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3811071842365567,
      "learning_rate": 1.654616212524594e-05,
      "loss": 0.0512,
      "step": 10145
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7164962896672052,
      "learning_rate": 1.6541493996905376e-05,
      "loss": 0.1576,
      "step": 10146
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6062455168756332,
      "learning_rate": 1.653682620158226e-05,
      "loss": 0.0753,
      "step": 10147
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5869994697070996,
      "learning_rate": 1.653215873946035e-05,
      "loss": 0.1271,
      "step": 10148
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5696013486809506,
      "learning_rate": 1.6527491610723426e-05,
      "loss": 0.0943,
      "step": 10149
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1937044481822618,
      "learning_rate": 1.652282481555523e-05,
      "loss": 0.3048,
      "step": 10150
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.569826848013804,
      "learning_rate": 1.6518158354139497e-05,
      "loss": 0.1012,
      "step": 10151
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5776527545185901,
      "learning_rate": 1.6513492226659944e-05,
      "loss": 0.1171,
      "step": 10152
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4270631920224279,
      "learning_rate": 1.6508826433300288e-05,
      "loss": 0.1131,
      "step": 10153
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6806146005277602,
      "learning_rate": 1.6504160974244222e-05,
      "loss": 0.1774,
      "step": 10154
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.47634002345342863,
      "learning_rate": 1.649949584967542e-05,
      "loss": 0.1036,
      "step": 10155
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.42352286686698637,
      "learning_rate": 1.649483105977756e-05,
      "loss": 0.0543,
      "step": 10156
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8105940657177668,
      "learning_rate": 1.6490166604734304e-05,
      "loss": 0.2114,
      "step": 10157
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.42000958441041697,
      "learning_rate": 1.648550248472928e-05,
      "loss": 0.0403,
      "step": 10158
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6475466691558868,
      "learning_rate": 1.6480838699946134e-05,
      "loss": 0.197,
      "step": 10159
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6913482788689402,
      "learning_rate": 1.647617525056846e-05,
      "loss": 0.2193,
      "step": 10160
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9006646448771076,
      "learning_rate": 1.6471512136779886e-05,
      "loss": 0.2463,
      "step": 10161
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1321260086987306,
      "learning_rate": 1.646684935876399e-05,
      "loss": 0.2859,
      "step": 10162
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5606505420085968,
      "learning_rate": 1.6462186916704353e-05,
      "loss": 0.0967,
      "step": 10163
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2444933976906587,
      "learning_rate": 1.6457524810784523e-05,
      "loss": 0.249,
      "step": 10164
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5586985666247498,
      "learning_rate": 1.6452863041188073e-05,
      "loss": 0.0639,
      "step": 10165
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.768873177168927,
      "learning_rate": 1.644820160809853e-05,
      "loss": 0.1598,
      "step": 10166
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6030139351910739,
      "learning_rate": 1.6443540511699423e-05,
      "loss": 0.0961,
      "step": 10167
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5736430902200145,
      "learning_rate": 1.6438879752174248e-05,
      "loss": 0.1155,
      "step": 10168
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6361530067317164,
      "learning_rate": 1.643421932970652e-05,
      "loss": 0.103,
      "step": 10169
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.32814776309045696,
      "learning_rate": 1.6429559244479713e-05,
      "loss": 0.0351,
      "step": 10170
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5045957630443895,
      "learning_rate": 1.6424899496677304e-05,
      "loss": 0.119,
      "step": 10171
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5303400177278575,
      "learning_rate": 1.6420240086482737e-05,
      "loss": 0.1336,
      "step": 10172
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4532478205530236,
      "learning_rate": 1.6415581014079477e-05,
      "loss": 0.0589,
      "step": 10173
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9822782003895871,
      "learning_rate": 1.641092227965095e-05,
      "loss": 0.2656,
      "step": 10174
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5205687102654147,
      "learning_rate": 1.640626388338056e-05,
      "loss": 0.0821,
      "step": 10175
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7397590770352783,
      "learning_rate": 1.6401605825451713e-05,
      "loss": 0.2145,
      "step": 10176
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7556129802600028,
      "learning_rate": 1.6396948106047817e-05,
      "loss": 0.0853,
      "step": 10177
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5529354728341384,
      "learning_rate": 1.6392290725352238e-05,
      "loss": 0.1007,
      "step": 10178
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.073884307142476,
      "learning_rate": 1.6387633683548344e-05,
      "loss": 0.254,
      "step": 10179
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2065259225752667,
      "learning_rate": 1.638297698081948e-05,
      "loss": 0.3649,
      "step": 10180
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9799269269054878,
      "learning_rate": 1.6378320617348986e-05,
      "loss": 0.2492,
      "step": 10181
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6443184832030695,
      "learning_rate": 1.6373664593320196e-05,
      "loss": 0.0554,
      "step": 10182
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5481596212428416,
      "learning_rate": 1.6369008908916413e-05,
      "loss": 0.1484,
      "step": 10183
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3960598668153841,
      "learning_rate": 1.6364353564320926e-05,
      "loss": 0.0617,
      "step": 10184
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6483140312875056,
      "learning_rate": 1.6359698559717034e-05,
      "loss": 0.0881,
      "step": 10185
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6214690034942459,
      "learning_rate": 1.6355043895287997e-05,
      "loss": 0.1236,
      "step": 10186
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8462401769838213,
      "learning_rate": 1.6350389571217083e-05,
      "loss": 0.1699,
      "step": 10187
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4321090887382906,
      "learning_rate": 1.6345735587687526e-05,
      "loss": 0.5308,
      "step": 10188
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5592799291829023,
      "learning_rate": 1.634108194488257e-05,
      "loss": 0.1512,
      "step": 10189
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5168315775093569,
      "learning_rate": 1.633642864298542e-05,
      "loss": 0.094,
      "step": 10190
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5459182806477096,
      "learning_rate": 1.6331775682179282e-05,
      "loss": 0.1166,
      "step": 10191
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.44390606591803844,
      "learning_rate": 1.6327123062647337e-05,
      "loss": 0.0468,
      "step": 10192
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5078640740562365,
      "learning_rate": 1.6322470784572785e-05,
      "loss": 0.047,
      "step": 10193
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.737054204955068,
      "learning_rate": 1.6317818848138777e-05,
      "loss": 0.3722,
      "step": 10194
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5510710537122311,
      "learning_rate": 1.6313167253528465e-05,
      "loss": 0.1177,
      "step": 10195
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3164762345526828,
      "learning_rate": 1.630851600092497e-05,
      "loss": 0.2794,
      "step": 10196
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9014288867770344,
      "learning_rate": 1.630386509051144e-05,
      "loss": 0.1357,
      "step": 10197
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7097214973901517,
      "learning_rate": 1.6299214522470972e-05,
      "loss": 0.0786,
      "step": 10198
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5522594161264512,
      "learning_rate": 1.629456429698667e-05,
      "loss": 0.1722,
      "step": 10199
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6095271258011596,
      "learning_rate": 1.6289914414241598e-05,
      "loss": 0.0617,
      "step": 10200
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3617626327430688,
      "learning_rate": 1.6285264874418843e-05,
      "loss": 0.0581,
      "step": 10201
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.535113034859032,
      "learning_rate": 1.628061567770146e-05,
      "loss": 0.0899,
      "step": 10202
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.49144300088646103,
      "learning_rate": 1.6275966824272476e-05,
      "loss": 0.0765,
      "step": 10203
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1043737945191308,
      "learning_rate": 1.6271318314314943e-05,
      "loss": 0.1529,
      "step": 10204
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.701776235409146,
      "learning_rate": 1.6266670148011858e-05,
      "loss": 0.1842,
      "step": 10205
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3699508320787575,
      "learning_rate": 1.6262022325546232e-05,
      "loss": 0.3203,
      "step": 10206
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2750703290525756,
      "learning_rate": 1.625737484710104e-05,
      "loss": 0.4384,
      "step": 10207
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2219774225733837,
      "learning_rate": 1.625272771285927e-05,
      "loss": 0.3247,
      "step": 10208
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8271791935779204,
      "learning_rate": 1.6248080923003888e-05,
      "loss": 0.1249,
      "step": 10209
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5849241074428165,
      "learning_rate": 1.6243434477717828e-05,
      "loss": 0.0774,
      "step": 10210
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4338994903984868,
      "learning_rate": 1.6238788377184024e-05,
      "loss": 0.0575,
      "step": 10211
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1267356115445233,
      "learning_rate": 1.6234142621585397e-05,
      "loss": 0.3997,
      "step": 10212
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5239311061094641,
      "learning_rate": 1.6229497211104866e-05,
      "loss": 0.076,
      "step": 10213
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.634685563912565,
      "learning_rate": 1.622485214592531e-05,
      "loss": 0.1595,
      "step": 10214
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7385614659143157,
      "learning_rate": 1.6220207426229615e-05,
      "loss": 0.1866,
      "step": 10215
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.626031311613563,
      "learning_rate": 1.6215563052200648e-05,
      "loss": 0.1525,
      "step": 10216
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8996202070406897,
      "learning_rate": 1.6210919024021258e-05,
      "loss": 0.1712,
      "step": 10217
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5114978184355916,
      "learning_rate": 1.6206275341874284e-05,
      "loss": 0.0645,
      "step": 10218
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7358777514128192,
      "learning_rate": 1.620163200594255e-05,
      "loss": 0.0905,
      "step": 10219
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3832142586052993,
      "learning_rate": 1.6196989016408876e-05,
      "loss": 0.0729,
      "step": 10220
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7212574670104043,
      "learning_rate": 1.619234637345605e-05,
      "loss": 0.2317,
      "step": 10221
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5536665319777018,
      "learning_rate": 1.6187704077266862e-05,
      "loss": 0.1768,
      "step": 10222
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4460208083100457,
      "learning_rate": 1.618306212802407e-05,
      "loss": 0.0461,
      "step": 10223
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5599764616800186,
      "learning_rate": 1.617842052591045e-05,
      "loss": 0.152,
      "step": 10224
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2350709033703644,
      "learning_rate": 1.6173779271108738e-05,
      "loss": 0.265,
      "step": 10225
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6133892262317641,
      "learning_rate": 1.6169138363801662e-05,
      "loss": 0.1526,
      "step": 10226
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7597809419869194,
      "learning_rate": 1.6164497804171924e-05,
      "loss": 0.1742,
      "step": 10227
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.789879000731792,
      "learning_rate": 1.615985759240225e-05,
      "loss": 0.232,
      "step": 10228
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5555795170364568,
      "learning_rate": 1.615521772867532e-05,
      "loss": 0.0771,
      "step": 10229
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4831054651055323,
      "learning_rate": 1.6150578213173806e-05,
      "loss": 0.0715,
      "step": 10230
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4078190440849974,
      "learning_rate": 1.614593904608036e-05,
      "loss": 0.0926,
      "step": 10231
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6007274676288982,
      "learning_rate": 1.614130022757765e-05,
      "loss": 0.1559,
      "step": 10232
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5211370636622308,
      "learning_rate": 1.6136661757848297e-05,
      "loss": 0.1774,
      "step": 10233
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5200525323139742,
      "learning_rate": 1.6132023637074923e-05,
      "loss": 0.098,
      "step": 10234
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7700187176090889,
      "learning_rate": 1.6127385865440125e-05,
      "loss": 0.216,
      "step": 10235
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6282236873121962,
      "learning_rate": 1.612274844312651e-05,
      "loss": 0.1068,
      "step": 10236
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3989021214046004,
      "learning_rate": 1.6118111370316656e-05,
      "loss": 0.0604,
      "step": 10237
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6739096799459446,
      "learning_rate": 1.6113474647193124e-05,
      "loss": 0.1168,
      "step": 10238
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.659409648823629,
      "learning_rate": 1.6108838273938452e-05,
      "loss": 0.4566,
      "step": 10239
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7613884594142436,
      "learning_rate": 1.61042022507352e-05,
      "loss": 0.1064,
      "step": 10240
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1213341841509432,
      "learning_rate": 1.6099566577765877e-05,
      "loss": 0.269,
      "step": 10241
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6167392053814139,
      "learning_rate": 1.6094931255213e-05,
      "loss": 0.1945,
      "step": 10242
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3350901629407994,
      "learning_rate": 1.609029628325905e-05,
      "loss": 0.044,
      "step": 10243
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.37667927080801794,
      "learning_rate": 1.6085661662086533e-05,
      "loss": 0.0417,
      "step": 10244
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5286786498169105,
      "learning_rate": 1.60810273918779e-05,
      "loss": 0.1168,
      "step": 10245
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.489706760986049,
      "learning_rate": 1.6076393472815616e-05,
      "loss": 0.1407,
      "step": 10246
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6815289677543125,
      "learning_rate": 1.607175990508211e-05,
      "loss": 0.1576,
      "step": 10247
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4705640586552243,
      "learning_rate": 1.606712668885982e-05,
      "loss": 0.0583,
      "step": 10248
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7197424452187038,
      "learning_rate": 1.6062493824331148e-05,
      "loss": 0.1365,
      "step": 10249
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5000977220957514,
      "learning_rate": 1.6057861311678503e-05,
      "loss": 0.0695,
      "step": 10250
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5376036834385272,
      "learning_rate": 1.6053229151084264e-05,
      "loss": 0.0699,
      "step": 10251
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0572889522189333,
      "learning_rate": 1.604859734273081e-05,
      "loss": 0.2789,
      "step": 10252
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2516231584520519,
      "learning_rate": 1.6043965886800495e-05,
      "loss": 0.5412,
      "step": 10253
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5298804028067443,
      "learning_rate": 1.6039334783475654e-05,
      "loss": 0.1006,
      "step": 10254
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5022720090581385,
      "learning_rate": 1.6034704032938623e-05,
      "loss": 0.0929,
      "step": 10255
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.490725577853123,
      "learning_rate": 1.603007363537173e-05,
      "loss": 0.0772,
      "step": 10256
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4309266072872054,
      "learning_rate": 1.6025443590957264e-05,
      "loss": 0.046,
      "step": 10257
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5975237308288845,
      "learning_rate": 1.6020813899877517e-05,
      "loss": 0.1229,
      "step": 10258
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5751387927667716,
      "learning_rate": 1.6016184562314752e-05,
      "loss": 0.0873,
      "step": 10259
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5755500932593447,
      "learning_rate": 1.6011555578451246e-05,
      "loss": 0.1173,
      "step": 10260
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4742495903739415,
      "learning_rate": 1.6006926948469238e-05,
      "loss": 0.0721,
      "step": 10261
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5589178070795507,
      "learning_rate": 1.6002298672550962e-05,
      "loss": 0.0979,
      "step": 10262
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6597466288295386,
      "learning_rate": 1.599767075087863e-05,
      "loss": 0.1246,
      "step": 10263
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4815648947038681,
      "learning_rate": 1.599304318363446e-05,
      "loss": 0.1422,
      "step": 10264
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3576461304039084,
      "learning_rate": 1.5988415971000632e-05,
      "loss": 0.0651,
      "step": 10265
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8574834492795406,
      "learning_rate": 1.598378911315933e-05,
      "loss": 0.2365,
      "step": 10266
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.89079364631298,
      "learning_rate": 1.59791626102927e-05,
      "loss": 0.1961,
      "step": 10267
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8529278982867345,
      "learning_rate": 1.5974536462582913e-05,
      "loss": 0.2385,
      "step": 10268
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.641694843454442,
      "learning_rate": 1.5969910670212097e-05,
      "loss": 0.1055,
      "step": 10269
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7005540027715362,
      "learning_rate": 1.5965285233362364e-05,
      "loss": 0.1614,
      "step": 10270
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.39497074531523085,
      "learning_rate": 1.5960660152215817e-05,
      "loss": 0.0781,
      "step": 10271
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5323220185632407,
      "learning_rate": 1.595603542695457e-05,
      "loss": 0.0936,
      "step": 10272
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.48031702394724957,
      "learning_rate": 1.595141105776069e-05,
      "loss": 0.0686,
      "step": 10273
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4398005199610184,
      "learning_rate": 1.594678704481624e-05,
      "loss": 0.0572,
      "step": 10274
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5567475679728179,
      "learning_rate": 1.5942163388303268e-05,
      "loss": 0.0874,
      "step": 10275
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7818307477552958,
      "learning_rate": 1.593754008840382e-05,
      "loss": 0.1116,
      "step": 10276
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6296670231119414,
      "learning_rate": 1.5932917145299914e-05,
      "loss": 0.1325,
      "step": 10277
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7663121082973179,
      "learning_rate": 1.5928294559173555e-05,
      "loss": 0.3202,
      "step": 10278
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6046493826395127,
      "learning_rate": 1.592367233020674e-05,
      "loss": 0.1545,
      "step": 10279
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4836819813342499,
      "learning_rate": 1.591905045858145e-05,
      "loss": 0.0758,
      "step": 10280
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6500642874827003,
      "learning_rate": 1.5914428944479654e-05,
      "loss": 0.0794,
      "step": 10281
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6971609113842796,
      "learning_rate": 1.5909807788083304e-05,
      "loss": 0.2143,
      "step": 10282
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6489575973696385,
      "learning_rate": 1.5905186989574335e-05,
      "loss": 0.1693,
      "step": 10283
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4838947462810283,
      "learning_rate": 1.590056654913467e-05,
      "loss": 0.17,
      "step": 10284
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4539997537965952,
      "learning_rate": 1.5895946466946225e-05,
      "loss": 0.1091,
      "step": 10285
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.80727578593078,
      "learning_rate": 1.5891326743190885e-05,
      "loss": 0.189,
      "step": 10286
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5846642808525783,
      "learning_rate": 1.588670737805054e-05,
      "loss": 0.1198,
      "step": 10287
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5971672015455842,
      "learning_rate": 1.588208837170706e-05,
      "loss": 0.0987,
      "step": 10288
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1101028402068447,
      "learning_rate": 1.58774697243423e-05,
      "loss": 0.264,
      "step": 10289
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7461117348176186,
      "learning_rate": 1.5872851436138092e-05,
      "loss": 0.1034,
      "step": 10290
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4679187447730988,
      "learning_rate": 1.5868233507276254e-05,
      "loss": 0.0974,
      "step": 10291
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6760094584994815,
      "learning_rate": 1.586361593793862e-05,
      "loss": 0.108,
      "step": 10292
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0363800449624363,
      "learning_rate": 1.585899872830697e-05,
      "loss": 0.2967,
      "step": 10293
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4156745226324129,
      "learning_rate": 1.5854381878563095e-05,
      "loss": 0.0617,
      "step": 10294
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7014477056729578,
      "learning_rate": 1.584976538888875e-05,
      "loss": 0.1437,
      "step": 10295
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.45080913849326243,
      "learning_rate": 1.5845149259465703e-05,
      "loss": 0.102,
      "step": 10296
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5868776824027823,
      "learning_rate": 1.5840533490475694e-05,
      "loss": 0.0811,
      "step": 10297
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.40599157751332365,
      "learning_rate": 1.583591808210045e-05,
      "loss": 0.0489,
      "step": 10298
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8748750691305784,
      "learning_rate": 1.5831303034521665e-05,
      "loss": 0.251,
      "step": 10299
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0629047469192576,
      "learning_rate": 1.5826688347921063e-05,
      "loss": 0.3266,
      "step": 10300
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4461877717706861,
      "learning_rate": 1.5822074022480313e-05,
      "loss": 0.0608,
      "step": 10301
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3929374274163595,
      "learning_rate": 1.5817460058381088e-05,
      "loss": 0.0637,
      "step": 10302
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.39362402614897407,
      "learning_rate": 1.5812846455805032e-05,
      "loss": 0.0649,
      "step": 10303
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5644678472185867,
      "learning_rate": 1.58082332149338e-05,
      "loss": 0.0948,
      "step": 10304
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8490294513861895,
      "learning_rate": 1.5803620335949015e-05,
      "loss": 0.2105,
      "step": 10305
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6860386148012179,
      "learning_rate": 1.5799007819032293e-05,
      "loss": 0.1704,
      "step": 10306
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9008892773924474,
      "learning_rate": 1.5794395664365212e-05,
      "loss": 0.3778,
      "step": 10307
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1305488364441536,
      "learning_rate": 1.5789783872129388e-05,
      "loss": 0.3877,
      "step": 10308
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6195434610869595,
      "learning_rate": 1.578517244250637e-05,
      "loss": 0.132,
      "step": 10309
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5251021921344046,
      "learning_rate": 1.5780561375677716e-05,
      "loss": 0.0414,
      "step": 10310
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5559204580269912,
      "learning_rate": 1.5775950671824957e-05,
      "loss": 0.0875,
      "step": 10311
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.41403117752543606,
      "learning_rate": 1.577134033112964e-05,
      "loss": 0.0496,
      "step": 10312
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6087577336981997,
      "learning_rate": 1.5766730353773273e-05,
      "loss": 0.1655,
      "step": 10313
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6999666026224041,
      "learning_rate": 1.5762120739937348e-05,
      "loss": 0.2184,
      "step": 10314
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7796408018622153,
      "learning_rate": 1.575751148980334e-05,
      "loss": 0.2888,
      "step": 10315
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4257163904230467,
      "learning_rate": 1.575290260355274e-05,
      "loss": 0.1362,
      "step": 10316
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.42594818940551177,
      "learning_rate": 1.574829408136698e-05,
      "loss": 0.069,
      "step": 10317
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5542382277314598,
      "learning_rate": 1.574368592342752e-05,
      "loss": 0.1111,
      "step": 10318
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5156900065193444,
      "learning_rate": 1.573907812991578e-05,
      "loss": 0.0844,
      "step": 10319
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6615985106735899,
      "learning_rate": 1.5734470701013172e-05,
      "loss": 0.1137,
      "step": 10320
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.24885337524458034,
      "learning_rate": 1.5729863636901087e-05,
      "loss": 0.0353,
      "step": 10321
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7774903479769354,
      "learning_rate": 1.572525693776092e-05,
      "loss": 0.1677,
      "step": 10322
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5721192974351024,
      "learning_rate": 1.5720650603774026e-05,
      "loss": 0.1003,
      "step": 10323
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.015969332568155,
      "learning_rate": 1.5716044635121774e-05,
      "loss": 0.2896,
      "step": 10324
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7305134518583045,
      "learning_rate": 1.5711439031985502e-05,
      "loss": 0.2963,
      "step": 10325
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4222658882391111,
      "learning_rate": 1.570683379454653e-05,
      "loss": 0.0458,
      "step": 10326
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5290967990253523,
      "learning_rate": 1.570222892298616e-05,
      "loss": 0.0621,
      "step": 10327
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7023181236987351,
      "learning_rate": 1.5697624417485713e-05,
      "loss": 0.2267,
      "step": 10328
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5671018779905028,
      "learning_rate": 1.5693020278226462e-05,
      "loss": 0.0875,
      "step": 10329
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4880890725987787,
      "learning_rate": 1.568841650538967e-05,
      "loss": 0.0627,
      "step": 10330
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6655309229280566,
      "learning_rate": 1.568381309915658e-05,
      "loss": 0.1592,
      "step": 10331
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6421520243459081,
      "learning_rate": 1.567921005970846e-05,
      "loss": 0.1129,
      "step": 10332
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6249065732687614,
      "learning_rate": 1.5674607387226516e-05,
      "loss": 0.0863,
      "step": 10333
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.45443109196385995,
      "learning_rate": 1.5670005081891966e-05,
      "loss": 0.0588,
      "step": 10334
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6909205617443036,
      "learning_rate": 1.5665403143885994e-05,
      "loss": 0.1086,
      "step": 10335
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6061756880208585,
      "learning_rate": 1.5660801573389792e-05,
      "loss": 0.1226,
      "step": 10336
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.46931475826169455,
      "learning_rate": 1.5656200370584535e-05,
      "loss": 0.0899,
      "step": 10337
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7288833420839882,
      "learning_rate": 1.5651599535651352e-05,
      "loss": 0.115,
      "step": 10338
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.44115847277409725,
      "learning_rate": 1.56469990687714e-05,
      "loss": 0.0705,
      "step": 10339
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4588200739211291,
      "learning_rate": 1.5642398970125804e-05,
      "loss": 0.0665,
      "step": 10340
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5892757006764329,
      "learning_rate": 1.5637799239895664e-05,
      "loss": 0.0956,
      "step": 10341
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6620409142579453,
      "learning_rate": 1.5633199878262075e-05,
      "loss": 0.1725,
      "step": 10342
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.526069040594371,
      "learning_rate": 1.562860088540612e-05,
      "loss": 0.07,
      "step": 10343
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6260642340616006,
      "learning_rate": 1.5624002261508866e-05,
      "loss": 0.2583,
      "step": 10344
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4782021277203158,
      "learning_rate": 1.5619404006751363e-05,
      "loss": 0.1023,
      "step": 10345
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1490396426381313,
      "learning_rate": 1.5614806121314645e-05,
      "loss": 0.2813,
      "step": 10346
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8988043470975481,
      "learning_rate": 1.561020860537974e-05,
      "loss": 0.2122,
      "step": 10347
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3908851215079702,
      "learning_rate": 1.5605611459127645e-05,
      "loss": 0.0356,
      "step": 10348
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4869439751652919,
      "learning_rate": 1.5601014682739367e-05,
      "loss": 0.1015,
      "step": 10349
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6904579386305583,
      "learning_rate": 1.5596418276395868e-05,
      "loss": 0.2408,
      "step": 10350
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.621918009694122,
      "learning_rate": 1.559182224027813e-05,
      "loss": 0.098,
      "step": 10351
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3598137003892676,
      "learning_rate": 1.5587226574567093e-05,
      "loss": 0.0445,
      "step": 10352
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4881467799573032,
      "learning_rate": 1.5582631279443688e-05,
      "loss": 0.068,
      "step": 10353
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6630526861172727,
      "learning_rate": 1.5578036355088833e-05,
      "loss": 0.1288,
      "step": 10354
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.45092992385429576,
      "learning_rate": 1.5573441801683445e-05,
      "loss": 0.0589,
      "step": 10355
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2255235058723306,
      "learning_rate": 1.5568847619408412e-05,
      "loss": 0.2723,
      "step": 10356
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6759968837371865,
      "learning_rate": 1.5564253808444607e-05,
      "loss": 0.1567,
      "step": 10357
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9017071195888826,
      "learning_rate": 1.5559660368972878e-05,
      "loss": 0.2814,
      "step": 10358
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6451724361340744,
      "learning_rate": 1.55550673011741e-05,
      "loss": 0.0637,
      "step": 10359
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5774853113612728,
      "learning_rate": 1.555047460522909e-05,
      "loss": 0.0836,
      "step": 10360
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6120885781348135,
      "learning_rate": 1.554588228131867e-05,
      "loss": 0.1976,
      "step": 10361
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3078866831001964,
      "learning_rate": 1.5541290329623626e-05,
      "loss": 0.0443,
      "step": 10362
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7717355351641594,
      "learning_rate": 1.5536698750324774e-05,
      "loss": 0.2142,
      "step": 10363
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5054077598216892,
      "learning_rate": 1.5532107543602868e-05,
      "loss": 0.1145,
      "step": 10364
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.475746260109421,
      "learning_rate": 1.5527516709638678e-05,
      "loss": 0.0856,
      "step": 10365
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4748658650091046,
      "learning_rate": 1.5522926248612933e-05,
      "loss": 0.0957,
      "step": 10366
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.45453946527716643,
      "learning_rate": 1.5518336160706387e-05,
      "loss": 0.091,
      "step": 10367
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9492420976921859,
      "learning_rate": 1.551374644609974e-05,
      "loss": 0.2128,
      "step": 10368
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4323447317382456,
      "learning_rate": 1.5509157104973694e-05,
      "loss": 0.087,
      "step": 10369
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.41476784660022564,
      "learning_rate": 1.550456813750892e-05,
      "loss": 0.0788,
      "step": 10370
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0521383682876582,
      "learning_rate": 1.5499979543886117e-05,
      "loss": 0.2356,
      "step": 10371
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4702869748681311,
      "learning_rate": 1.549539132428593e-05,
      "loss": 0.0811,
      "step": 10372
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.45513634384903373,
      "learning_rate": 1.549080347888899e-05,
      "loss": 0.0725,
      "step": 10373
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6484452550638262,
      "learning_rate": 1.548621600787593e-05,
      "loss": 0.1196,
      "step": 10374
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7637258040853298,
      "learning_rate": 1.5481628911427374e-05,
      "loss": 0.2323,
      "step": 10375
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.49140175386585755,
      "learning_rate": 1.5477042189723907e-05,
      "loss": 0.054,
      "step": 10376
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.444593626884875,
      "learning_rate": 1.5472455842946117e-05,
      "loss": 0.0724,
      "step": 10377
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5189611012002977,
      "learning_rate": 1.546786987127456e-05,
      "loss": 0.0882,
      "step": 10378
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9413233984040443,
      "learning_rate": 1.54632842748898e-05,
      "loss": 0.0737,
      "step": 10379
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2893630859889125,
      "learning_rate": 1.5458699053972376e-05,
      "loss": 0.3094,
      "step": 10380
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7040566209669417,
      "learning_rate": 1.545411420870281e-05,
      "loss": 0.4246,
      "step": 10381
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5695798215150345,
      "learning_rate": 1.54495297392616e-05,
      "loss": 0.1178,
      "step": 10382
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3214809335132294,
      "learning_rate": 1.5444945645829258e-05,
      "loss": 0.4998,
      "step": 10383
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8518667894899068,
      "learning_rate": 1.544036192858625e-05,
      "loss": 0.2338,
      "step": 10384
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7501333517595461,
      "learning_rate": 1.543577858771304e-05,
      "loss": 0.1674,
      "step": 10385
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.38273487473705964,
      "learning_rate": 1.543119562339008e-05,
      "loss": 0.0459,
      "step": 10386
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.44056444499401365,
      "learning_rate": 1.542661303579782e-05,
      "loss": 0.0462,
      "step": 10387
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.477962480244464,
      "learning_rate": 1.5422030825116655e-05,
      "loss": 0.2468,
      "step": 10388
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5832504631140303,
      "learning_rate": 1.5417448991527007e-05,
      "loss": 0.1277,
      "step": 10389
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3267433830035749,
      "learning_rate": 1.5412867535209246e-05,
      "loss": 0.3052,
      "step": 10390
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5797412806099695,
      "learning_rate": 1.5408286456343772e-05,
      "loss": 0.0728,
      "step": 10391
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6632960416034281,
      "learning_rate": 1.5403705755110936e-05,
      "loss": 0.0832,
      "step": 10392
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7418670758471652,
      "learning_rate": 1.539912543169108e-05,
      "loss": 0.1606,
      "step": 10393
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4457209127859081,
      "learning_rate": 1.539454548626453e-05,
      "loss": 0.0818,
      "step": 10394
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.46230781248437713,
      "learning_rate": 1.5389965919011616e-05,
      "loss": 0.0613,
      "step": 10395
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5550949826240396,
      "learning_rate": 1.538538673011263e-05,
      "loss": 0.1536,
      "step": 10396
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5016663059175552,
      "learning_rate": 1.538080791974786e-05,
      "loss": 0.098,
      "step": 10397
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2678732819511012,
      "learning_rate": 1.5376229488097565e-05,
      "loss": 0.5132,
      "step": 10398
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5337884827162854,
      "learning_rate": 1.5371651435342026e-05,
      "loss": 0.1185,
      "step": 10399
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7755270817537357,
      "learning_rate": 1.5367073761661466e-05,
      "loss": 0.1572,
      "step": 10400
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.084648064229226,
      "learning_rate": 1.5362496467236116e-05,
      "loss": 0.1837,
      "step": 10401
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7128543676802953,
      "learning_rate": 1.535791955224618e-05,
      "loss": 0.1685,
      "step": 10402
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7072182050969077,
      "learning_rate": 1.5353343016871868e-05,
      "loss": 0.117,
      "step": 10403
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0175700276550002,
      "learning_rate": 1.5348766861293358e-05,
      "loss": 0.2475,
      "step": 10404
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7690320583090795,
      "learning_rate": 1.534419108569081e-05,
      "loss": 0.2342,
      "step": 10405
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.36293233276216197,
      "learning_rate": 1.533961569024437e-05,
      "loss": 0.0889,
      "step": 10406
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7860658093307387,
      "learning_rate": 1.5335040675134195e-05,
      "loss": 0.1156,
      "step": 10407
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7623114736522408,
      "learning_rate": 1.5330466040540395e-05,
      "loss": 0.1055,
      "step": 10408
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8233922117971542,
      "learning_rate": 1.5325891786643076e-05,
      "loss": 0.1417,
      "step": 10409
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5326254909684087,
      "learning_rate": 1.5321317913622326e-05,
      "loss": 0.1314,
      "step": 10410
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4246511406266864,
      "learning_rate": 1.5316744421658225e-05,
      "loss": 0.4961,
      "step": 10411
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0953249679199386,
      "learning_rate": 1.5312171310930846e-05,
      "loss": 0.7898,
      "step": 10412
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5259639190408235,
      "learning_rate": 1.5307598581620218e-05,
      "loss": 0.0565,
      "step": 10413
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4918041336438407,
      "learning_rate": 1.530302623390638e-05,
      "loss": 0.1226,
      "step": 10414
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.534690603244939,
      "learning_rate": 1.5298454267969353e-05,
      "loss": 0.1007,
      "step": 10415
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4295447092169274,
      "learning_rate": 1.529388268398913e-05,
      "loss": 0.0545,
      "step": 10416
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7467711139964384,
      "learning_rate": 1.5289311482145705e-05,
      "loss": 0.1113,
      "step": 10417
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.46829517545391947,
      "learning_rate": 1.528474066261904e-05,
      "loss": 0.0787,
      "step": 10418
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.41251930198038184,
      "learning_rate": 1.5280170225589107e-05,
      "loss": 0.06,
      "step": 10419
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5372077570985521,
      "learning_rate": 1.527560017123584e-05,
      "loss": 0.1502,
      "step": 10420
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5071307631020854,
      "learning_rate": 1.5271030499739155e-05,
      "loss": 0.0927,
      "step": 10421
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5487474057282519,
      "learning_rate": 1.5266461211278966e-05,
      "loss": 0.0964,
      "step": 10422
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8426389946246083,
      "learning_rate": 1.5261892306035184e-05,
      "loss": 0.1657,
      "step": 10423
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.11179239074707,
      "learning_rate": 1.5257323784187683e-05,
      "loss": 0.0793,
      "step": 10424
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.49499055139792025,
      "learning_rate": 1.5252755645916329e-05,
      "loss": 0.0892,
      "step": 10425
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8688605259165464,
      "learning_rate": 1.5248187891400955e-05,
      "loss": 0.1334,
      "step": 10426
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6941322787111868,
      "learning_rate": 1.5243620520821427e-05,
      "loss": 0.1222,
      "step": 10427
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5516613285956546,
      "learning_rate": 1.5239053534357551e-05,
      "loss": 0.1292,
      "step": 10428
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7570546767974521,
      "learning_rate": 1.523448693218913e-05,
      "loss": 0.2245,
      "step": 10429
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.44409778651925724,
      "learning_rate": 1.5229920714495948e-05,
      "loss": 0.0582,
      "step": 10430
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5097343793743295,
      "learning_rate": 1.5225354881457802e-05,
      "loss": 0.0698,
      "step": 10431
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.43507783263065497,
      "learning_rate": 1.5220789433254437e-05,
      "loss": 0.0789,
      "step": 10432
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6411530388072467,
      "learning_rate": 1.5216224370065602e-05,
      "loss": 0.0933,
      "step": 10433
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5400914594867908,
      "learning_rate": 1.5211659692071016e-05,
      "loss": 0.1006,
      "step": 10434
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8803812241266017,
      "learning_rate": 1.5207095399450411e-05,
      "loss": 0.1866,
      "step": 10435
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6793528738425871,
      "learning_rate": 1.5202531492383476e-05,
      "loss": 0.0964,
      "step": 10436
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5393228494628054,
      "learning_rate": 1.5197967971049903e-05,
      "loss": 0.061,
      "step": 10437
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7969163883068036,
      "learning_rate": 1.5193404835629343e-05,
      "loss": 0.1455,
      "step": 10438
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5849885100659523,
      "learning_rate": 1.5188842086301474e-05,
      "loss": 0.1363,
      "step": 10439
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.48094676371668876,
      "learning_rate": 1.5184279723245923e-05,
      "loss": 0.0899,
      "step": 10440
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6054870185475999,
      "learning_rate": 1.5179717746642314e-05,
      "loss": 0.1326,
      "step": 10441
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7409506080843955,
      "learning_rate": 1.5175156156670244e-05,
      "loss": 0.1682,
      "step": 10442
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6413759066098642,
      "learning_rate": 1.517059495350933e-05,
      "loss": 0.1399,
      "step": 10443
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.44803752107324213,
      "learning_rate": 1.5166034137339137e-05,
      "loss": 0.055,
      "step": 10444
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6384075118048086,
      "learning_rate": 1.5161473708339225e-05,
      "loss": 0.1303,
      "step": 10445
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5224840269118856,
      "learning_rate": 1.5156913666689137e-05,
      "loss": 0.1036,
      "step": 10446
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7192368918099532,
      "learning_rate": 1.5152354012568424e-05,
      "loss": 0.1086,
      "step": 10447
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.48282374223460994,
      "learning_rate": 1.5147794746156583e-05,
      "loss": 0.1116,
      "step": 10448
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6364416124720592,
      "learning_rate": 1.5143235867633126e-05,
      "loss": 0.1034,
      "step": 10449
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4582887573582241,
      "learning_rate": 1.5138677377177536e-05,
      "loss": 0.054,
      "step": 10450
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4146114057356003,
      "learning_rate": 1.5134119274969289e-05,
      "loss": 0.0668,
      "step": 10451
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.49366428866005857,
      "learning_rate": 1.5129561561187835e-05,
      "loss": 0.0804,
      "step": 10452
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4460678639764748,
      "learning_rate": 1.5125004236012613e-05,
      "loss": 0.0655,
      "step": 10453
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5211460516729072,
      "learning_rate": 1.5120447299623054e-05,
      "loss": 0.0685,
      "step": 10454
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.571311767639386,
      "learning_rate": 1.5115890752198564e-05,
      "loss": 0.0913,
      "step": 10455
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5185981961786736,
      "learning_rate": 1.5111334593918542e-05,
      "loss": 0.0755,
      "step": 10456
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.48732294824428196,
      "learning_rate": 1.5106778824962364e-05,
      "loss": 0.0969,
      "step": 10457
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5533980257851564,
      "learning_rate": 1.5102223445509383e-05,
      "loss": 0.1003,
      "step": 10458
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3431209348871432,
      "learning_rate": 1.509766845573897e-05,
      "loss": 0.0313,
      "step": 10459
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.662471738780321,
      "learning_rate": 1.5093113855830445e-05,
      "loss": 0.608,
      "step": 10460
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8685619637957221,
      "learning_rate": 1.5088559645963129e-05,
      "loss": 0.2663,
      "step": 10461
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.41087803212341134,
      "learning_rate": 1.508400582631631e-05,
      "loss": 0.0536,
      "step": 10462
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9828294654129766,
      "learning_rate": 1.5079452397069302e-05,
      "loss": 0.4227,
      "step": 10463
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3855911818380958,
      "learning_rate": 1.5074899358401362e-05,
      "loss": 0.0674,
      "step": 10464
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6017288130796471,
      "learning_rate": 1.5070346710491746e-05,
      "loss": 0.1075,
      "step": 10465
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8197750922469021,
      "learning_rate": 1.5065794453519688e-05,
      "loss": 0.1463,
      "step": 10466
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4850712696527702,
      "learning_rate": 1.506124258766443e-05,
      "loss": 0.074,
      "step": 10467
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5640250916590431,
      "learning_rate": 1.5056691113105177e-05,
      "loss": 0.1058,
      "step": 10468
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6057860677731813,
      "learning_rate": 1.5052140030021108e-05,
      "loss": 0.0974,
      "step": 10469
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0770109931947838,
      "learning_rate": 1.5047589338591429e-05,
      "loss": 0.4148,
      "step": 10470
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.47094476593979406,
      "learning_rate": 1.5043039038995288e-05,
      "loss": 0.0841,
      "step": 10471
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5208162863148101,
      "learning_rate": 1.5038489131411834e-05,
      "loss": 0.0513,
      "step": 10472
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.42216221025011164,
      "learning_rate": 1.5033939616020193e-05,
      "loss": 0.0628,
      "step": 10473
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7609466105256605,
      "learning_rate": 1.5029390492999504e-05,
      "loss": 0.1836,
      "step": 10474
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.39617119292004316,
      "learning_rate": 1.5024841762528857e-05,
      "loss": 0.077,
      "step": 10475
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3449529535466081,
      "learning_rate": 1.5020293424787335e-05,
      "loss": 0.0594,
      "step": 10476
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9661023040530939,
      "learning_rate": 1.5015745479954014e-05,
      "loss": 0.0843,
      "step": 10477
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6361241425363103,
      "learning_rate": 1.5011197928207948e-05,
      "loss": 0.0854,
      "step": 10478
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9564803190753388,
      "learning_rate": 1.5006650769728175e-05,
      "loss": 0.2053,
      "step": 10479
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5502423999975328,
      "learning_rate": 1.5002104004693729e-05,
      "loss": 0.0537,
      "step": 10480
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.47272015670604406,
      "learning_rate": 1.499755763328361e-05,
      "loss": 0.0883,
      "step": 10481
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6563383491232884,
      "learning_rate": 1.4993011655676819e-05,
      "loss": 0.0965,
      "step": 10482
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8942603251832092,
      "learning_rate": 1.4988466072052332e-05,
      "loss": 0.2538,
      "step": 10483
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5780347379312241,
      "learning_rate": 1.4983920882589104e-05,
      "loss": 0.1758,
      "step": 10484
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.40146825590000784,
      "learning_rate": 1.4979376087466088e-05,
      "loss": 0.0593,
      "step": 10485
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5432194147075592,
      "learning_rate": 1.4974831686862223e-05,
      "loss": 0.0854,
      "step": 10486
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6383342758465241,
      "learning_rate": 1.4970287680956421e-05,
      "loss": 0.2396,
      "step": 10487
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.473675618542961,
      "learning_rate": 1.496574406992758e-05,
      "loss": 0.0559,
      "step": 10488
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6804124653152619,
      "learning_rate": 1.4961200853954577e-05,
      "loss": 0.1448,
      "step": 10489
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.43705804601313797,
      "learning_rate": 1.4956658033216297e-05,
      "loss": 0.045,
      "step": 10490
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.34193279243304625,
      "learning_rate": 1.4952115607891592e-05,
      "loss": 0.0446,
      "step": 10491
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6266405562692385,
      "learning_rate": 1.4947573578159296e-05,
      "loss": 0.1715,
      "step": 10492
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5393358840996385,
      "learning_rate": 1.4943031944198225e-05,
      "loss": 0.0728,
      "step": 10493
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9695434800720155,
      "learning_rate": 1.49384907061872e-05,
      "loss": 0.2285,
      "step": 10494
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7011051381318946,
      "learning_rate": 1.4933949864305008e-05,
      "loss": 0.1322,
      "step": 10495
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.46042694512353066,
      "learning_rate": 1.4929409418730422e-05,
      "loss": 0.0943,
      "step": 10496
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.527912709638617,
      "learning_rate": 1.4924869369642197e-05,
      "loss": 0.1112,
      "step": 10497
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7931182563660717,
      "learning_rate": 1.4920329717219095e-05,
      "loss": 0.2345,
      "step": 10498
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8573797907147712,
      "learning_rate": 1.4915790461639839e-05,
      "loss": 0.2466,
      "step": 10499
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7031770112708852,
      "learning_rate": 1.4911251603083135e-05,
      "loss": 0.1622,
      "step": 10500
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.57284670978756,
      "learning_rate": 1.4906713141727677e-05,
      "loss": 0.1112,
      "step": 10501
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.848849240252035,
      "learning_rate": 1.4902175077752167e-05,
      "loss": 0.1437,
      "step": 10502
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3471124274576136,
      "learning_rate": 1.4897637411335258e-05,
      "loss": 0.0511,
      "step": 10503
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.753234197282699,
      "learning_rate": 1.4893100142655608e-05,
      "loss": 0.1685,
      "step": 10504
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5750295338720437,
      "learning_rate": 1.4888563271891837e-05,
      "loss": 0.0866,
      "step": 10505
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6726020505220142,
      "learning_rate": 1.4884026799222589e-05,
      "loss": 0.2562,
      "step": 10506
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4052994634779538,
      "learning_rate": 1.4879490724826454e-05,
      "loss": 0.0583,
      "step": 10507
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6658735500468161,
      "learning_rate": 1.4874955048882022e-05,
      "loss": 0.1362,
      "step": 10508
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6833558947089222,
      "learning_rate": 1.4870419771567863e-05,
      "loss": 0.0918,
      "step": 10509
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.48950019698726904,
      "learning_rate": 1.4865884893062531e-05,
      "loss": 0.1252,
      "step": 10510
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3468244640085107,
      "learning_rate": 1.4861350413544583e-05,
      "loss": 0.043,
      "step": 10511
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5771737654741375,
      "learning_rate": 1.4856816333192536e-05,
      "loss": 0.1374,
      "step": 10512
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3530138074936125,
      "learning_rate": 1.4852282652184896e-05,
      "loss": 0.041,
      "step": 10513
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2009701823993977,
      "learning_rate": 1.4847749370700162e-05,
      "loss": 0.1856,
      "step": 10514
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.46712562757818,
      "learning_rate": 1.4843216488916812e-05,
      "loss": 0.0812,
      "step": 10515
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4677063132708539,
      "learning_rate": 1.4838684007013304e-05,
      "loss": 0.0581,
      "step": 10516
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.39779273868777637,
      "learning_rate": 1.4834151925168088e-05,
      "loss": 0.0591,
      "step": 10517
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5294403682162827,
      "learning_rate": 1.48296202435596e-05,
      "loss": 0.1109,
      "step": 10518
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5117785110181081,
      "learning_rate": 1.4825088962366255e-05,
      "loss": 0.0666,
      "step": 10519
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8393165055982186,
      "learning_rate": 1.4820558081766447e-05,
      "loss": 0.1189,
      "step": 10520
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5474521757529942,
      "learning_rate": 1.4816027601938554e-05,
      "loss": 0.1122,
      "step": 10521
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.44747077548944586,
      "learning_rate": 1.4811497523060963e-05,
      "loss": 0.0674,
      "step": 10522
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8870122147989614,
      "learning_rate": 1.4806967845312017e-05,
      "loss": 0.1928,
      "step": 10523
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.49789817506164785,
      "learning_rate": 1.480243856887005e-05,
      "loss": 0.0956,
      "step": 10524
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4336221440256675,
      "learning_rate": 1.4797909693913376e-05,
      "loss": 0.0976,
      "step": 10525
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5539814739264535,
      "learning_rate": 1.479338122062032e-05,
      "loss": 0.1626,
      "step": 10526
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7365164755046435,
      "learning_rate": 1.4788853149169158e-05,
      "loss": 0.0958,
      "step": 10527
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.40078849037475534,
      "learning_rate": 1.4784325479738167e-05,
      "loss": 0.0455,
      "step": 10528
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4390379095604643,
      "learning_rate": 1.477979821250559e-05,
      "loss": 0.0478,
      "step": 10529
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4750575994342722,
      "learning_rate": 1.4775271347649699e-05,
      "loss": 0.0893,
      "step": 10530
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4104580434474347,
      "learning_rate": 1.4770744885348697e-05,
      "loss": 0.1291,
      "step": 10531
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6200483722720338,
      "learning_rate": 1.4766218825780803e-05,
      "loss": 0.1338,
      "step": 10532
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8662327101088836,
      "learning_rate": 1.4761693169124196e-05,
      "loss": 0.2044,
      "step": 10533
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.45541773570190075,
      "learning_rate": 1.4757167915557077e-05,
      "loss": 0.0998,
      "step": 10534
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3968877778028369,
      "learning_rate": 1.4752643065257598e-05,
      "loss": 0.0497,
      "step": 10535
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5246984524790628,
      "learning_rate": 1.4748118618403906e-05,
      "loss": 0.0777,
      "step": 10536
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4315600299609956,
      "learning_rate": 1.4743594575174119e-05,
      "loss": 0.0627,
      "step": 10537
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5787700931317901,
      "learning_rate": 1.4739070935746378e-05,
      "loss": 0.0894,
      "step": 10538
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4452990615269683,
      "learning_rate": 1.4734547700298765e-05,
      "loss": 0.0638,
      "step": 10539
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.721981307254958,
      "learning_rate": 1.4730024869009368e-05,
      "loss": 0.2789,
      "step": 10540
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9542826538864112,
      "learning_rate": 1.4725502442056243e-05,
      "loss": 0.309,
      "step": 10541
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3730412613747816,
      "learning_rate": 1.4720980419617456e-05,
      "loss": 0.0567,
      "step": 10542
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9160798615992481,
      "learning_rate": 1.4716458801871038e-05,
      "loss": 0.2821,
      "step": 10543
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7279648753066346,
      "learning_rate": 1.4711937588995009e-05,
      "loss": 0.3061,
      "step": 10544
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5355587415854894,
      "learning_rate": 1.4707416781167363e-05,
      "loss": 0.069,
      "step": 10545
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5939713450793349,
      "learning_rate": 1.47028963785661e-05,
      "loss": 0.0869,
      "step": 10546
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2230292980951121,
      "learning_rate": 1.4698376381369184e-05,
      "loss": 0.0269,
      "step": 10547
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4354945517092014,
      "learning_rate": 1.4693856789754577e-05,
      "loss": 0.0509,
      "step": 10548
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7812161621194015,
      "learning_rate": 1.4689337603900208e-05,
      "loss": 0.1285,
      "step": 10549
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8003403389849882,
      "learning_rate": 1.4684818823984011e-05,
      "loss": 0.0968,
      "step": 10550
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6565450307529971,
      "learning_rate": 1.4680300450183892e-05,
      "loss": 0.146,
      "step": 10551
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7607220991410079,
      "learning_rate": 1.4675782482677739e-05,
      "loss": 0.1761,
      "step": 10552
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9218804584853526,
      "learning_rate": 1.4671264921643419e-05,
      "loss": 0.239,
      "step": 10553
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3886336069173643,
      "learning_rate": 1.466674776725881e-05,
      "loss": 0.0511,
      "step": 10554
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.41788118566464155,
      "learning_rate": 1.466223101970175e-05,
      "loss": 0.0831,
      "step": 10555
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.40313453570885466,
      "learning_rate": 1.4657714679150059e-05,
      "loss": 0.0734,
      "step": 10556
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.32270919111757285,
      "learning_rate": 1.465319874578155e-05,
      "loss": 0.0489,
      "step": 10557
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8266981312023172,
      "learning_rate": 1.4648683219774022e-05,
      "loss": 0.2466,
      "step": 10558
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.42222026909099486,
      "learning_rate": 1.4644168101305261e-05,
      "loss": 0.0508,
      "step": 10559
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5248656982950426,
      "learning_rate": 1.4639653390553024e-05,
      "loss": 0.4284,
      "step": 10560
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3798554032959133,
      "learning_rate": 1.4635139087695045e-05,
      "loss": 0.0892,
      "step": 10561
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6915599103512895,
      "learning_rate": 1.463062519290908e-05,
      "loss": 0.158,
      "step": 10562
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.533199695893515,
      "learning_rate": 1.4626111706372834e-05,
      "loss": 0.1233,
      "step": 10563
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6703178713996018,
      "learning_rate": 1.4621598628264002e-05,
      "loss": 0.1262,
      "step": 10564
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0313773489605236,
      "learning_rate": 1.461708595876026e-05,
      "loss": 0.144,
      "step": 10565
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5594849503386095,
      "learning_rate": 1.4612573698039295e-05,
      "loss": 0.118,
      "step": 10566
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3929911813674982,
      "learning_rate": 1.460806184627875e-05,
      "loss": 0.0865,
      "step": 10567
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.44392129356361676,
      "learning_rate": 1.4603550403656258e-05,
      "loss": 0.0938,
      "step": 10568
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3253203109965888,
      "learning_rate": 1.4599039370349427e-05,
      "loss": 0.2742,
      "step": 10569
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4616786547178226,
      "learning_rate": 1.459452874653588e-05,
      "loss": 0.086,
      "step": 10570
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5783276957325872,
      "learning_rate": 1.4590018532393196e-05,
      "loss": 0.1162,
      "step": 10571
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1426562511412646,
      "learning_rate": 1.4585508728098935e-05,
      "loss": 0.067,
      "step": 10572
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.370633209242572,
      "learning_rate": 1.4580999333830659e-05,
      "loss": 0.4515,
      "step": 10573
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.43026156667159454,
      "learning_rate": 1.4576490349765904e-05,
      "loss": 0.0476,
      "step": 10574
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.476031657301939,
      "learning_rate": 1.4571981776082208e-05,
      "loss": 0.0713,
      "step": 10575
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4039865748213031,
      "learning_rate": 1.456747361295705e-05,
      "loss": 0.0415,
      "step": 10576
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.47500960872627646,
      "learning_rate": 1.4562965860567934e-05,
      "loss": 0.1204,
      "step": 10577
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7962190124565124,
      "learning_rate": 1.4558458519092332e-05,
      "loss": 0.2367,
      "step": 10578
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5013803895106149,
      "learning_rate": 1.4553951588707711e-05,
      "loss": 0.1357,
      "step": 10579
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.37505842423978797,
      "learning_rate": 1.4549445069591495e-05,
      "loss": 0.0412,
      "step": 10580
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.6365571287137497,
      "learning_rate": 1.4544938961921112e-05,
      "loss": 0.7414,
      "step": 10581
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.009012690631883,
      "learning_rate": 1.4540433265873977e-05,
      "loss": 0.1344,
      "step": 10582
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5975622864745636,
      "learning_rate": 1.453592798162749e-05,
      "loss": 0.0944,
      "step": 10583
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9454736977228689,
      "learning_rate": 1.453142310935901e-05,
      "loss": 0.2383,
      "step": 10584
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4166371993804458,
      "learning_rate": 1.4526918649245901e-05,
      "loss": 0.0671,
      "step": 10585
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6694904583679538,
      "learning_rate": 1.4522414601465512e-05,
      "loss": 0.0937,
      "step": 10586
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.643609992824538,
      "learning_rate": 1.451791096619518e-05,
      "loss": 0.1835,
      "step": 10587
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7379532170746294,
      "learning_rate": 1.4513407743612194e-05,
      "loss": 0.1313,
      "step": 10588
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.38879176708873436,
      "learning_rate": 1.4508904933893864e-05,
      "loss": 0.0526,
      "step": 10589
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.465056538550285,
      "learning_rate": 1.450440253721746e-05,
      "loss": 0.0764,
      "step": 10590
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4849425720795053,
      "learning_rate": 1.4499900553760264e-05,
      "loss": 0.0839,
      "step": 10591
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.39238293967339183,
      "learning_rate": 1.4495398983699499e-05,
      "loss": 0.0571,
      "step": 10592
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5475688666742543,
      "learning_rate": 1.4490897827212402e-05,
      "loss": 0.0733,
      "step": 10593
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5033113729286333,
      "learning_rate": 1.4486397084476189e-05,
      "loss": 0.097,
      "step": 10594
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6804589737506117,
      "learning_rate": 1.4481896755668067e-05,
      "loss": 0.1368,
      "step": 10595
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7031406195969074,
      "learning_rate": 1.4477396840965202e-05,
      "loss": 0.1781,
      "step": 10596
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6374237635180086,
      "learning_rate": 1.4472897340544762e-05,
      "loss": 0.1383,
      "step": 10597
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6937155342278519,
      "learning_rate": 1.4468398254583909e-05,
      "loss": 0.1828,
      "step": 10598
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7745884880391922,
      "learning_rate": 1.4463899583259752e-05,
      "loss": 0.2306,
      "step": 10599
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5391256503614662,
      "learning_rate": 1.4459401326749423e-05,
      "loss": 0.1147,
      "step": 10600
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7317113586402596,
      "learning_rate": 1.4454903485230014e-05,
      "loss": 0.1856,
      "step": 10601
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.155769507827389,
      "learning_rate": 1.4450406058878623e-05,
      "loss": 0.2584,
      "step": 10602
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5175941205388301,
      "learning_rate": 1.4445909047872298e-05,
      "loss": 0.0714,
      "step": 10603
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8483538494669639,
      "learning_rate": 1.4441412452388098e-05,
      "loss": 0.1763,
      "step": 10604
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6352409858435757,
      "learning_rate": 1.4436916272603054e-05,
      "loss": 0.103,
      "step": 10605
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6135871365731467,
      "learning_rate": 1.44324205086942e-05,
      "loss": 0.1093,
      "step": 10606
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9251994671287648,
      "learning_rate": 1.4427925160838512e-05,
      "loss": 0.3167,
      "step": 10607
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.49251637850580715,
      "learning_rate": 1.4423430229212986e-05,
      "loss": 0.1011,
      "step": 10608
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1450217759386367,
      "learning_rate": 1.4418935713994594e-05,
      "loss": 0.3131,
      "step": 10609
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4511695131932563,
      "learning_rate": 1.4414441615360298e-05,
      "loss": 0.3515,
      "step": 10610
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5785473703404008,
      "learning_rate": 1.4409947933487006e-05,
      "loss": 0.1657,
      "step": 10611
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7390797930021917,
      "learning_rate": 1.440545466855166e-05,
      "loss": 0.122,
      "step": 10612
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5318494799023014,
      "learning_rate": 1.4400961820731151e-05,
      "loss": 0.1008,
      "step": 10613
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4115253835090744,
      "learning_rate": 1.4396469390202385e-05,
      "loss": 0.0507,
      "step": 10614
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7004880792422039,
      "learning_rate": 1.4391977377142207e-05,
      "loss": 0.1361,
      "step": 10615
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7469176805693772,
      "learning_rate": 1.4387485781727483e-05,
      "loss": 0.2185,
      "step": 10616
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.35515344057373627,
      "learning_rate": 1.4382994604135046e-05,
      "loss": 0.037,
      "step": 10617
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3778199334076588,
      "learning_rate": 1.4378503844541735e-05,
      "loss": 0.044,
      "step": 10618
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1388010060044578,
      "learning_rate": 1.4374013503124329e-05,
      "loss": 0.447,
      "step": 10619
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9009998566218105,
      "learning_rate": 1.4369523580059624e-05,
      "loss": 0.1928,
      "step": 10620
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5237328266219554,
      "learning_rate": 1.4365034075524397e-05,
      "loss": 0.1352,
      "step": 10621
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4147610811130791,
      "learning_rate": 1.436054498969541e-05,
      "loss": 0.0518,
      "step": 10622
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3956390294295373,
      "learning_rate": 1.4356056322749378e-05,
      "loss": 0.0631,
      "step": 10623
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5168207776409556,
      "learning_rate": 1.4351568074863043e-05,
      "loss": 0.0968,
      "step": 10624
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5998580893101702,
      "learning_rate": 1.4347080246213102e-05,
      "loss": 0.1758,
      "step": 10625
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5906581188571203,
      "learning_rate": 1.4342592836976254e-05,
      "loss": 0.1507,
      "step": 10626
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4971499614597977,
      "learning_rate": 1.4338105847329156e-05,
      "loss": 0.1306,
      "step": 10627
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.71470841665414,
      "learning_rate": 1.433361927744847e-05,
      "loss": 0.117,
      "step": 10628
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6174264862378428,
      "learning_rate": 1.4329133127510849e-05,
      "loss": 0.1128,
      "step": 10629
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5430421588828884,
      "learning_rate": 1.4324647397692894e-05,
      "loss": 0.0872,
      "step": 10630
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9279143507292801,
      "learning_rate": 1.4320162088171224e-05,
      "loss": 0.102,
      "step": 10631
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.38479637829814983,
      "learning_rate": 1.4315677199122424e-05,
      "loss": 0.0712,
      "step": 10632
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4921862464574526,
      "learning_rate": 1.4311192730723078e-05,
      "loss": 0.0943,
      "step": 10633
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6919248214349991,
      "learning_rate": 1.4306708683149727e-05,
      "loss": 0.1714,
      "step": 10634
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9535466421013346,
      "learning_rate": 1.4302225056578928e-05,
      "loss": 0.2532,
      "step": 10635
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5579305274609225,
      "learning_rate": 1.4297741851187177e-05,
      "loss": 0.1636,
      "step": 10636
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.40805866508145366,
      "learning_rate": 1.4293259067151016e-05,
      "loss": 0.1145,
      "step": 10637
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4811521816017951,
      "learning_rate": 1.4288776704646911e-05,
      "loss": 0.0458,
      "step": 10638
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.62751866197441,
      "learning_rate": 1.4284294763851356e-05,
      "loss": 0.1106,
      "step": 10639
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5645267188668941,
      "learning_rate": 1.427981324494077e-05,
      "loss": 0.1259,
      "step": 10640
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8045390235761727,
      "learning_rate": 1.4275332148091647e-05,
      "loss": 0.3276,
      "step": 10641
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1110376267502533,
      "learning_rate": 1.4270851473480367e-05,
      "loss": 0.1641,
      "step": 10642
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6082714413188122,
      "learning_rate": 1.4266371221283366e-05,
      "loss": 0.1075,
      "step": 10643
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0142633434125625,
      "learning_rate": 1.4261891391677002e-05,
      "loss": 0.3054,
      "step": 10644
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.42952693628745114,
      "learning_rate": 1.4257411984837693e-05,
      "loss": 0.0769,
      "step": 10645
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.53980832156975,
      "learning_rate": 1.4252933000941759e-05,
      "loss": 0.205,
      "step": 10646
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.47014061154885123,
      "learning_rate": 1.4248454440165565e-05,
      "loss": 0.0988,
      "step": 10647
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6029932541255763,
      "learning_rate": 1.4243976302685408e-05,
      "loss": 0.1511,
      "step": 10648
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4140465499542974,
      "learning_rate": 1.423949858867763e-05,
      "loss": 0.0789,
      "step": 10649
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.64089733416497,
      "learning_rate": 1.4235021298318496e-05,
      "loss": 0.1529,
      "step": 10650
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5360353221904173,
      "learning_rate": 1.4230544431784299e-05,
      "loss": 0.0714,
      "step": 10651
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5782829508953143,
      "learning_rate": 1.4226067989251262e-05,
      "loss": 0.0944,
      "step": 10652
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4110828103294017,
      "learning_rate": 1.4221591970895673e-05,
      "loss": 0.0429,
      "step": 10653
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5228841735799626,
      "learning_rate": 1.4217116376893722e-05,
      "loss": 0.0858,
      "step": 10654
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4321423051863105,
      "learning_rate": 1.4212641207421637e-05,
      "loss": 0.5024,
      "step": 10655
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5975151810421226,
      "learning_rate": 1.4208166462655587e-05,
      "loss": 0.0925,
      "step": 10656
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2101294564390805,
      "learning_rate": 1.420369214277176e-05,
      "loss": 0.2552,
      "step": 10657
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.501825598726217,
      "learning_rate": 1.419921824794631e-05,
      "loss": 0.0758,
      "step": 10658
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8857804806270329,
      "learning_rate": 1.419474477835539e-05,
      "loss": 0.3452,
      "step": 10659
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.506751561714185,
      "learning_rate": 1.4190271734175098e-05,
      "loss": 0.0865,
      "step": 10660
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5780278851691487,
      "learning_rate": 1.4185799115581558e-05,
      "loss": 0.0952,
      "step": 10661
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.45300780991996453,
      "learning_rate": 1.4181326922750856e-05,
      "loss": 0.0764,
      "step": 10662
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.49399892784572774,
      "learning_rate": 1.4176855155859075e-05,
      "loss": 0.0574,
      "step": 10663
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4409749393653793,
      "learning_rate": 1.4172383815082258e-05,
      "loss": 0.072,
      "step": 10664
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5023923304905676,
      "learning_rate": 1.4167912900596448e-05,
      "loss": 0.071,
      "step": 10665
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6017257946887499,
      "learning_rate": 1.4163442412577677e-05,
      "loss": 0.1442,
      "step": 10666
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.39173325002174675,
      "learning_rate": 1.4158972351201938e-05,
      "loss": 0.0817,
      "step": 10667
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.37957560444646055,
      "learning_rate": 1.4154502716645224e-05,
      "loss": 0.0882,
      "step": 10668
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5493156355802056,
      "learning_rate": 1.4150033509083514e-05,
      "loss": 0.0768,
      "step": 10669
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5530765588228752,
      "learning_rate": 1.4145564728692767e-05,
      "loss": 0.1631,
      "step": 10670
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.49187647024211373,
      "learning_rate": 1.414109637564891e-05,
      "loss": 0.1106,
      "step": 10671
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.42737690888237284,
      "learning_rate": 1.4136628450127868e-05,
      "loss": 0.0349,
      "step": 10672
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.41237113262689484,
      "learning_rate": 1.4132160952305552e-05,
      "loss": 0.0522,
      "step": 10673
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.48604751844790256,
      "learning_rate": 1.4127693882357854e-05,
      "loss": 0.0748,
      "step": 10674
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4147151981408134,
      "learning_rate": 1.4123227240460632e-05,
      "loss": 0.0733,
      "step": 10675
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5160237355012609,
      "learning_rate": 1.4118761026789747e-05,
      "loss": 0.082,
      "step": 10676
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5010606319224362,
      "learning_rate": 1.4114295241521041e-05,
      "loss": 0.0851,
      "step": 10677
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6499311108942287,
      "learning_rate": 1.4109829884830342e-05,
      "loss": 0.1222,
      "step": 10678
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6124624731082777,
      "learning_rate": 1.4105364956893436e-05,
      "loss": 0.1136,
      "step": 10679
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3966215953651677,
      "learning_rate": 1.4100900457886119e-05,
      "loss": 0.6239,
      "step": 10680
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8806323224494963,
      "learning_rate": 1.4096436387984163e-05,
      "loss": 0.2445,
      "step": 10681
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7259557205683717,
      "learning_rate": 1.409197274736333e-05,
      "loss": 0.1731,
      "step": 10682
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6828522065512612,
      "learning_rate": 1.4087509536199333e-05,
      "loss": 0.1115,
      "step": 10683
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5024299884921187,
      "learning_rate": 1.408304675466791e-05,
      "loss": 0.1168,
      "step": 10684
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6148029229228243,
      "learning_rate": 1.4078584402944761e-05,
      "loss": 0.1277,
      "step": 10685
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.46495029432637724,
      "learning_rate": 1.4074122481205582e-05,
      "loss": 0.0575,
      "step": 10686
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6478776165687703,
      "learning_rate": 1.4069660989626022e-05,
      "loss": 0.2379,
      "step": 10687
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5706677413310826,
      "learning_rate": 1.4065199928381738e-05,
      "loss": 0.1563,
      "step": 10688
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9683904708247271,
      "learning_rate": 1.4060739297648373e-05,
      "loss": 0.3762,
      "step": 10689
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.47477882232803154,
      "learning_rate": 1.4056279097601549e-05,
      "loss": 0.0583,
      "step": 10690
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8029472347047597,
      "learning_rate": 1.4051819328416854e-05,
      "loss": 0.2073,
      "step": 10691
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6396697755903573,
      "learning_rate": 1.4047359990269876e-05,
      "loss": 0.0856,
      "step": 10692
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5430664735543546,
      "learning_rate": 1.4042901083336182e-05,
      "loss": 0.122,
      "step": 10693
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5231213943114141,
      "learning_rate": 1.4038442607791338e-05,
      "loss": 0.0661,
      "step": 10694
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7331339093073367,
      "learning_rate": 1.4033984563810853e-05,
      "loss": 0.1191,
      "step": 10695
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6565800528182637,
      "learning_rate": 1.4029526951570257e-05,
      "loss": 0.1375,
      "step": 10696
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7319788671154545,
      "learning_rate": 1.4025069771245056e-05,
      "loss": 0.1922,
      "step": 10697
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.43194950677185073,
      "learning_rate": 1.4020613023010715e-05,
      "loss": 0.0663,
      "step": 10698
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8676293460457464,
      "learning_rate": 1.4016156707042705e-05,
      "loss": 0.2478,
      "step": 10699
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6056965929341102,
      "learning_rate": 1.4011700823516477e-05,
      "loss": 0.1105,
      "step": 10700
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5567626210107972,
      "learning_rate": 1.4007245372607474e-05,
      "loss": 0.1176,
      "step": 10701
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4960571744698416,
      "learning_rate": 1.4002790354491091e-05,
      "loss": 0.069,
      "step": 10702
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0961373124039104,
      "learning_rate": 1.3998335769342742e-05,
      "loss": 0.272,
      "step": 10703
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7392860179507862,
      "learning_rate": 1.3993881617337776e-05,
      "loss": 0.105,
      "step": 10704
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.37429431346209335,
      "learning_rate": 1.3989427898651602e-05,
      "loss": 0.0491,
      "step": 10705
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.36492577755135563,
      "learning_rate": 1.3984974613459529e-05,
      "loss": 0.0508,
      "step": 10706
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5939443347924682,
      "learning_rate": 1.398052176193691e-05,
      "loss": 0.1211,
      "step": 10707
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7491028840423721,
      "learning_rate": 1.397606934425903e-05,
      "loss": 0.2136,
      "step": 10708
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7962451107583718,
      "learning_rate": 1.397161736060122e-05,
      "loss": 0.176,
      "step": 10709
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7864239469048903,
      "learning_rate": 1.3967165811138727e-05,
      "loss": 0.2461,
      "step": 10710
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5944719037283399,
      "learning_rate": 1.396271469604683e-05,
      "loss": 0.1914,
      "step": 10711
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8620078282337303,
      "learning_rate": 1.395826401550075e-05,
      "loss": 0.191,
      "step": 10712
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.46617564618297447,
      "learning_rate": 1.3953813769675746e-05,
      "loss": 0.0589,
      "step": 10713
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4969575816338618,
      "learning_rate": 1.3949363958747002e-05,
      "loss": 0.0689,
      "step": 10714
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4614428653435008,
      "learning_rate": 1.3944914582889728e-05,
      "loss": 0.0654,
      "step": 10715
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9834142243872815,
      "learning_rate": 1.394046564227907e-05,
      "loss": 0.3568,
      "step": 10716
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4823454718609259,
      "learning_rate": 1.393601713709023e-05,
      "loss": 0.0674,
      "step": 10717
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5325886790322759,
      "learning_rate": 1.3931569067498306e-05,
      "loss": 0.0833,
      "step": 10718
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8769812942709095,
      "learning_rate": 1.3927121433678458e-05,
      "loss": 0.221,
      "step": 10719
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.40915244095016956,
      "learning_rate": 1.392267423580575e-05,
      "loss": 0.0644,
      "step": 10720
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3975128291683733,
      "learning_rate": 1.3918227474055318e-05,
      "loss": 0.075,
      "step": 10721
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7234705246863556,
      "learning_rate": 1.3913781148602203e-05,
      "loss": 0.1071,
      "step": 10722
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.37218386597270003,
      "learning_rate": 1.3909335259621478e-05,
      "loss": 0.0573,
      "step": 10723
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6253701060258433,
      "learning_rate": 1.3904889807288164e-05,
      "loss": 0.1049,
      "step": 10724
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.45516450626520677,
      "learning_rate": 1.390044479177729e-05,
      "loss": 0.0705,
      "step": 10725
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0308627547021054,
      "learning_rate": 1.3896000213263855e-05,
      "loss": 0.3351,
      "step": 10726
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3427544537256011,
      "learning_rate": 1.3891556071922862e-05,
      "loss": 0.0839,
      "step": 10727
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6081538854575711,
      "learning_rate": 1.3887112367929261e-05,
      "loss": 0.1692,
      "step": 10728
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5526576447215017,
      "learning_rate": 1.3882669101458007e-05,
      "loss": 0.0703,
      "step": 10729
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0747890671816296,
      "learning_rate": 1.387822627268404e-05,
      "loss": 0.5348,
      "step": 10730
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.40540719319587193,
      "learning_rate": 1.3873783881782282e-05,
      "loss": 0.0799,
      "step": 10731
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0406001083375418,
      "learning_rate": 1.3869341928927621e-05,
      "loss": 0.1699,
      "step": 10732
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6566576600570806,
      "learning_rate": 1.386490041429494e-05,
      "loss": 0.1098,
      "step": 10733
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5421675969980132,
      "learning_rate": 1.3860459338059123e-05,
      "loss": 0.0601,
      "step": 10734
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4002485335704645,
      "learning_rate": 1.3856018700394985e-05,
      "loss": 0.0645,
      "step": 10735
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6550379630479766,
      "learning_rate": 1.3851578501477397e-05,
      "loss": 0.1578,
      "step": 10736
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6462417144576899,
      "learning_rate": 1.3847138741481145e-05,
      "loss": 0.1989,
      "step": 10737
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5959262311329558,
      "learning_rate": 1.384269942058104e-05,
      "loss": 0.1004,
      "step": 10738
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5307847943067172,
      "learning_rate": 1.383826053895184e-05,
      "loss": 0.1053,
      "step": 10739
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9523517334935109,
      "learning_rate": 1.3833822096768337e-05,
      "loss": 0.2517,
      "step": 10740
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4004274261791333,
      "learning_rate": 1.3829384094205252e-05,
      "loss": 0.0692,
      "step": 10741
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2530041661664444,
      "learning_rate": 1.3824946531437332e-05,
      "loss": 0.6204,
      "step": 10742
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7120717535049323,
      "learning_rate": 1.3820509408639254e-05,
      "loss": 0.2728,
      "step": 10743
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6015488581704702,
      "learning_rate": 1.3816072725985757e-05,
      "loss": 0.1317,
      "step": 10744
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5256253701543441,
      "learning_rate": 1.381163648365148e-05,
      "loss": 0.1509,
      "step": 10745
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.43945842087111214,
      "learning_rate": 1.38072006818111e-05,
      "loss": 0.0703,
      "step": 10746
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5078378403491716,
      "learning_rate": 1.3802765320639232e-05,
      "loss": 0.0756,
      "step": 10747
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6755656110991812,
      "learning_rate": 1.3798330400310539e-05,
      "loss": 0.1126,
      "step": 10748
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.025482539444896,
      "learning_rate": 1.3793895920999598e-05,
      "loss": 0.1902,
      "step": 10749
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6784411386344771,
      "learning_rate": 1.3789461882881014e-05,
      "loss": 0.1581,
      "step": 10750
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.39166175598402037,
      "learning_rate": 1.3785028286129331e-05,
      "loss": 0.1075,
      "step": 10751
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4290938850967423,
      "learning_rate": 1.3780595130919138e-05,
      "loss": 0.07,
      "step": 10752
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7901191351021758,
      "learning_rate": 1.3776162417424948e-05,
      "loss": 0.1617,
      "step": 10753
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.504218639923265,
      "learning_rate": 1.3771730145821299e-05,
      "loss": 0.1567,
      "step": 10754
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.41873179075084027,
      "learning_rate": 1.3767298316282667e-05,
      "loss": 0.0561,
      "step": 10755
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7185740845264194,
      "learning_rate": 1.3762866928983553e-05,
      "loss": 0.1098,
      "step": 10756
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4958189702087297,
      "learning_rate": 1.3758435984098422e-05,
      "loss": 0.126,
      "step": 10757
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6525380762842505,
      "learning_rate": 1.3754005481801735e-05,
      "loss": 0.2139,
      "step": 10758
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4162677047647912,
      "learning_rate": 1.37495754222679e-05,
      "loss": 0.0774,
      "step": 10759
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7822918926084085,
      "learning_rate": 1.3745145805671342e-05,
      "loss": 0.3009,
      "step": 10760
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8647436814217211,
      "learning_rate": 1.3740716632186468e-05,
      "loss": 0.1842,
      "step": 10761
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5390802841348569,
      "learning_rate": 1.3736287901987644e-05,
      "loss": 0.0958,
      "step": 10762
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4561039769372163,
      "learning_rate": 1.3731859615249238e-05,
      "loss": 0.0826,
      "step": 10763
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5234044484142476,
      "learning_rate": 1.3727431772145589e-05,
      "loss": 0.0633,
      "step": 10764
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4061945634618226,
      "learning_rate": 1.3723004372851045e-05,
      "loss": 0.0637,
      "step": 10765
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.42864807823381623,
      "learning_rate": 1.3718577417539889e-05,
      "loss": 0.0518,
      "step": 10766
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7133818788637241,
      "learning_rate": 1.3714150906386424e-05,
      "loss": 0.1283,
      "step": 10767
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8730176422676424,
      "learning_rate": 1.3709724839564928e-05,
      "loss": 0.2314,
      "step": 10768
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3790822794569364,
      "learning_rate": 1.3705299217249665e-05,
      "loss": 0.0496,
      "step": 10769
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3440364473824426,
      "learning_rate": 1.3700874039614859e-05,
      "loss": 0.0696,
      "step": 10770
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.47545198119026144,
      "learning_rate": 1.3696449306834735e-05,
      "loss": 0.1073,
      "step": 10771
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6797296948414107,
      "learning_rate": 1.3692025019083507e-05,
      "loss": 0.149,
      "step": 10772
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.34256102870417016,
      "learning_rate": 1.3687601176535366e-05,
      "loss": 0.04,
      "step": 10773
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.651039367904001,
      "learning_rate": 1.3683177779364464e-05,
      "loss": 0.132,
      "step": 10774
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4930484263477486,
      "learning_rate": 1.3678754827744966e-05,
      "loss": 0.7662,
      "step": 10775
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5426824183255434,
      "learning_rate": 1.3674332321851002e-05,
      "loss": 0.0556,
      "step": 10776
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8673668040945229,
      "learning_rate": 1.36699102618567e-05,
      "loss": 0.2147,
      "step": 10777
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5740613314543208,
      "learning_rate": 1.3665488647936142e-05,
      "loss": 0.1344,
      "step": 10778
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9456770058893526,
      "learning_rate": 1.3661067480263422e-05,
      "loss": 0.3117,
      "step": 10779
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6219520971255115,
      "learning_rate": 1.3656646759012598e-05,
      "loss": 0.1213,
      "step": 10780
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5560913337140896,
      "learning_rate": 1.3652226484357733e-05,
      "loss": 0.1134,
      "step": 10781
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5538176535336976,
      "learning_rate": 1.3647806656472833e-05,
      "loss": 0.1706,
      "step": 10782
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5982685934139987,
      "learning_rate": 1.3643387275531922e-05,
      "loss": 0.0778,
      "step": 10783
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5484026706070919,
      "learning_rate": 1.3638968341708992e-05,
      "loss": 0.0836,
      "step": 10784
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2309250094601851,
      "learning_rate": 1.3634549855178028e-05,
      "loss": 0.3394,
      "step": 10785
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.497595254394961,
      "learning_rate": 1.3630131816112973e-05,
      "loss": 0.0856,
      "step": 10786
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7807364002362487,
      "learning_rate": 1.3625714224687777e-05,
      "loss": 0.1911,
      "step": 10787
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.48588792938460496,
      "learning_rate": 1.3621297081076365e-05,
      "loss": 0.0921,
      "step": 10788
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.061780453077677,
      "learning_rate": 1.3616880385452651e-05,
      "loss": 0.0977,
      "step": 10789
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3919126538963189,
      "learning_rate": 1.3612464137990505e-05,
      "loss": 0.051,
      "step": 10790
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8348082518415918,
      "learning_rate": 1.3608048338863805e-05,
      "loss": 0.1788,
      "step": 10791
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.43168048389635866,
      "learning_rate": 1.3603632988246418e-05,
      "loss": 0.0551,
      "step": 10792
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3665509056523782,
      "learning_rate": 1.3599218086312154e-05,
      "loss": 0.0571,
      "step": 10793
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6280109107267405,
      "learning_rate": 1.3594803633234849e-05,
      "loss": 0.1758,
      "step": 10794
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1604957457433651,
      "learning_rate": 1.3590389629188299e-05,
      "loss": 0.3465,
      "step": 10795
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.48976095568759864,
      "learning_rate": 1.3585976074346294e-05,
      "loss": 0.062,
      "step": 10796
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8382971784327903,
      "learning_rate": 1.3581562968882582e-05,
      "loss": 0.0994,
      "step": 10797
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.598050944154367,
      "learning_rate": 1.3577150312970921e-05,
      "loss": 0.092,
      "step": 10798
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4479107608505358,
      "learning_rate": 1.3572738106785036e-05,
      "loss": 0.1059,
      "step": 10799
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6293040196268913,
      "learning_rate": 1.356832635049865e-05,
      "loss": 0.1209,
      "step": 10800
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.946568516166713,
      "learning_rate": 1.356391504428544e-05,
      "loss": 0.2314,
      "step": 10801
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6370613194046152,
      "learning_rate": 1.3559504188319103e-05,
      "loss": 0.0765,
      "step": 10802
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7398217991199748,
      "learning_rate": 1.3555093782773263e-05,
      "loss": 0.1769,
      "step": 10803
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3291757624813388,
      "learning_rate": 1.3550683827821604e-05,
      "loss": 0.0425,
      "step": 10804
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4264874273636156,
      "learning_rate": 1.3546274323637718e-05,
      "loss": 0.061,
      "step": 10805
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5750947825664491,
      "learning_rate": 1.3541865270395227e-05,
      "loss": 0.1051,
      "step": 10806
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5248331531134586,
      "learning_rate": 1.3537456668267696e-05,
      "loss": 0.075,
      "step": 10807
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3571022653562446,
      "learning_rate": 1.3533048517428731e-05,
      "loss": 0.0629,
      "step": 10808
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6400592932411187,
      "learning_rate": 1.3528640818051853e-05,
      "loss": 0.1219,
      "step": 10809
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3924956283164862,
      "learning_rate": 1.3524233570310618e-05,
      "loss": 0.0696,
      "step": 10810
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.647291705638923,
      "learning_rate": 1.3519826774378508e-05,
      "loss": 0.1866,
      "step": 10811
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5425670229800348,
      "learning_rate": 1.3515420430429071e-05,
      "loss": 0.1127,
      "step": 10812
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.167868489068675,
      "learning_rate": 1.3511014538635747e-05,
      "loss": 0.2309,
      "step": 10813
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.609180142268343,
      "learning_rate": 1.350660909917203e-05,
      "loss": 0.15,
      "step": 10814
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6347554917768341,
      "learning_rate": 1.3502204112211325e-05,
      "loss": 0.1622,
      "step": 10815
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.455126990977602,
      "learning_rate": 1.3497799577927106e-05,
      "loss": 0.0565,
      "step": 10816
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.391290073278159,
      "learning_rate": 1.349339549649275e-05,
      "loss": 0.0601,
      "step": 10817
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.099780974878243,
      "learning_rate": 1.3488991868081675e-05,
      "loss": 0.3846,
      "step": 10818
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5854215535062894,
      "learning_rate": 1.3484588692867211e-05,
      "loss": 0.3487,
      "step": 10819
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7740676814486903,
      "learning_rate": 1.348018597102277e-05,
      "loss": 0.3489,
      "step": 10820
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3882234772445073,
      "learning_rate": 1.3475783702721651e-05,
      "loss": 0.0678,
      "step": 10821
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.45667022678833735,
      "learning_rate": 1.3471381888137197e-05,
      "loss": 0.0534,
      "step": 10822
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4434833731570497,
      "learning_rate": 1.346698052744269e-05,
      "loss": 0.0404,
      "step": 10823
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8128010048305524,
      "learning_rate": 1.3462579620811424e-05,
      "loss": 0.2133,
      "step": 10824
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.40599421348792547,
      "learning_rate": 1.3458179168416667e-05,
      "loss": 0.0801,
      "step": 10825
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7303854013992124,
      "learning_rate": 1.345377917043168e-05,
      "loss": 0.131,
      "step": 10826
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4945386219062396,
      "learning_rate": 1.3449379627029671e-05,
      "loss": 0.1141,
      "step": 10827
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7294437409901865,
      "learning_rate": 1.3444980538383865e-05,
      "loss": 0.158,
      "step": 10828
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5128428573750423,
      "learning_rate": 1.3440581904667465e-05,
      "loss": 0.0707,
      "step": 10829
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5419336465002716,
      "learning_rate": 1.343618372605363e-05,
      "loss": 0.1189,
      "step": 10830
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7747536803120533,
      "learning_rate": 1.3431786002715529e-05,
      "loss": 0.2106,
      "step": 10831
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.668978339811639,
      "learning_rate": 1.3427388734826301e-05,
      "loss": 0.249,
      "step": 10832
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6650995739289133,
      "learning_rate": 1.3422991922559086e-05,
      "loss": 0.1262,
      "step": 10833
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2754568426394294,
      "learning_rate": 1.3418595566086967e-05,
      "loss": 0.0509,
      "step": 10834
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4387069388711009,
      "learning_rate": 1.3414199665583038e-05,
      "loss": 0.0599,
      "step": 10835
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5886192281459189,
      "learning_rate": 1.3409804221220373e-05,
      "loss": 0.2138,
      "step": 10836
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5806438162692045,
      "learning_rate": 1.3405409233172031e-05,
      "loss": 0.1831,
      "step": 10837
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0106713097536233,
      "learning_rate": 1.3401014701611028e-05,
      "loss": 0.2666,
      "step": 10838
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4358674518844906,
      "learning_rate": 1.3396620626710389e-05,
      "loss": 0.0606,
      "step": 10839
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7662911224318535,
      "learning_rate": 1.339222700864311e-05,
      "loss": 0.1546,
      "step": 10840
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6890001803934154,
      "learning_rate": 1.3387833847582185e-05,
      "loss": 0.1522,
      "step": 10841
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2518606976988154,
      "learning_rate": 1.3383441143700554e-05,
      "loss": 0.3977,
      "step": 10842
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6090945785507133,
      "learning_rate": 1.3379048897171167e-05,
      "loss": 0.0875,
      "step": 10843
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3719165425995705,
      "learning_rate": 1.3374657108166954e-05,
      "loss": 0.052,
      "step": 10844
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5384903452916139,
      "learning_rate": 1.3370265776860835e-05,
      "loss": 0.1152,
      "step": 10845
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7026365046280154,
      "learning_rate": 1.3365874903425674e-05,
      "loss": 0.1001,
      "step": 10846
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4752419863308638,
      "learning_rate": 1.3361484488034359e-05,
      "loss": 0.0794,
      "step": 10847
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5603093422725522,
      "learning_rate": 1.335709453085974e-05,
      "loss": 0.0991,
      "step": 10848
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.971597000639922,
      "learning_rate": 1.3352705032074663e-05,
      "loss": 1.0393,
      "step": 10849
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3573885207876086,
      "learning_rate": 1.3348315991851925e-05,
      "loss": 0.4534,
      "step": 10850
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.42325345613305543,
      "learning_rate": 1.3343927410364337e-05,
      "loss": 0.0559,
      "step": 10851
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.42605453646717395,
      "learning_rate": 1.3339539287784684e-05,
      "loss": 0.0707,
      "step": 10852
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6702066952517799,
      "learning_rate": 1.333515162428573e-05,
      "loss": 0.0807,
      "step": 10853
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.49311918442346775,
      "learning_rate": 1.3330764420040212e-05,
      "loss": 0.0457,
      "step": 10854
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6769742621465542,
      "learning_rate": 1.3326377675220858e-05,
      "loss": 0.1448,
      "step": 10855
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7375377714200573,
      "learning_rate": 1.3321991390000382e-05,
      "loss": 0.11,
      "step": 10856
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7603682722909961,
      "learning_rate": 1.331760556455149e-05,
      "loss": 0.1908,
      "step": 10857
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6378118751071042,
      "learning_rate": 1.3313220199046821e-05,
      "loss": 0.08,
      "step": 10858
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.154923957902395,
      "learning_rate": 1.3308835293659053e-05,
      "loss": 0.3856,
      "step": 10859
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.6324411145761144,
      "learning_rate": 1.3304450848560826e-05,
      "loss": 0.5777,
      "step": 10860
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5330196478457085,
      "learning_rate": 1.3300066863924743e-05,
      "loss": 0.0926,
      "step": 10861
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1130283888370107,
      "learning_rate": 1.329568333992341e-05,
      "loss": 0.3899,
      "step": 10862
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1965937187666102,
      "learning_rate": 1.3291300276729415e-05,
      "loss": 0.4662,
      "step": 10863
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5046157654853881,
      "learning_rate": 1.3286917674515329e-05,
      "loss": 0.0883,
      "step": 10864
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6529491598369982,
      "learning_rate": 1.3282535533453677e-05,
      "loss": 0.2572,
      "step": 10865
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4181459960157024,
      "learning_rate": 1.327815385371701e-05,
      "loss": 0.088,
      "step": 10866
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4114885420322035,
      "learning_rate": 1.3273772635477805e-05,
      "loss": 0.0528,
      "step": 10867
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.470057048956102,
      "learning_rate": 1.3269391878908597e-05,
      "loss": 0.041,
      "step": 10868
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3720699030574036,
      "learning_rate": 1.3265011584181825e-05,
      "loss": 0.0985,
      "step": 10869
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6984677138369847,
      "learning_rate": 1.326063175146996e-05,
      "loss": 0.1257,
      "step": 10870
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5634434296272721,
      "learning_rate": 1.3256252380945435e-05,
      "loss": 0.1506,
      "step": 10871
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3444417806092374,
      "learning_rate": 1.3251873472780683e-05,
      "loss": 0.0461,
      "step": 10872
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8689149038632084,
      "learning_rate": 1.3247495027148076e-05,
      "loss": 0.1991,
      "step": 10873
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6497884999305703,
      "learning_rate": 1.324311704422002e-05,
      "loss": 0.1418,
      "step": 10874
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4826194172954893,
      "learning_rate": 1.3238739524168864e-05,
      "loss": 0.0908,
      "step": 10875
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4121276552454661,
      "learning_rate": 1.3234362467166977e-05,
      "loss": 0.0919,
      "step": 10876
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5026551719943836,
      "learning_rate": 1.322998587338666e-05,
      "loss": 0.064,
      "step": 10877
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5421789394661218,
      "learning_rate": 1.3225609743000236e-05,
      "loss": 0.1399,
      "step": 10878
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5061884562885903,
      "learning_rate": 1.3221234076179995e-05,
      "loss": 0.061,
      "step": 10879
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2071850322529114,
      "learning_rate": 1.3216858873098217e-05,
      "loss": 0.2643,
      "step": 10880
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.48547806867522364,
      "learning_rate": 1.3212484133927145e-05,
      "loss": 0.0847,
      "step": 10881
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4883408854662354,
      "learning_rate": 1.3208109858839012e-05,
      "loss": 0.0615,
      "step": 10882
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4712956723290183,
      "learning_rate": 1.3203736048006055e-05,
      "loss": 0.0655,
      "step": 10883
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5356990435405011,
      "learning_rate": 1.3199362701600465e-05,
      "loss": 0.129,
      "step": 10884
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5443762861408566,
      "learning_rate": 1.3194989819794417e-05,
      "loss": 0.0753,
      "step": 10885
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.378453377595164,
      "learning_rate": 1.319061740276008e-05,
      "loss": 0.0451,
      "step": 10886
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4394984330420115,
      "learning_rate": 1.31862454506696e-05,
      "loss": 0.0707,
      "step": 10887
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9231708576878142,
      "learning_rate": 1.3181873963695113e-05,
      "loss": 0.2692,
      "step": 10888
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9275255584934206,
      "learning_rate": 1.3177502942008708e-05,
      "loss": 0.499,
      "step": 10889
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4510613662272846,
      "learning_rate": 1.3173132385782482e-05,
      "loss": 0.066,
      "step": 10890
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7562950955705386,
      "learning_rate": 1.316876229518852e-05,
      "loss": 0.3479,
      "step": 10891
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.054775849606699,
      "learning_rate": 1.316439267039886e-05,
      "loss": 0.2493,
      "step": 10892
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5225976518517714,
      "learning_rate": 1.3160023511585539e-05,
      "loss": 0.1443,
      "step": 10893
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3312055527365067,
      "learning_rate": 1.3155654818920579e-05,
      "loss": 0.0649,
      "step": 10894
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.577730621535413,
      "learning_rate": 1.315128659257599e-05,
      "loss": 0.1591,
      "step": 10895
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5629766368910524,
      "learning_rate": 1.3146918832723725e-05,
      "loss": 0.0731,
      "step": 10896
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.42825322784679165,
      "learning_rate": 1.3142551539535772e-05,
      "loss": 0.041,
      "step": 10897
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9867893341451804,
      "learning_rate": 1.3138184713184042e-05,
      "loss": 0.2296,
      "step": 10898
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.45413654155531213,
      "learning_rate": 1.3133818353840504e-05,
      "loss": 0.0709,
      "step": 10899
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7176675356855107,
      "learning_rate": 1.3129452461677028e-05,
      "loss": 0.1956,
      "step": 10900
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1733793701637558,
      "learning_rate": 1.312508703686553e-05,
      "loss": 0.3761,
      "step": 10901
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6554434349598713,
      "learning_rate": 1.312072207957784e-05,
      "loss": 0.1651,
      "step": 10902
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5757816289418289,
      "learning_rate": 1.3116357589985862e-05,
      "loss": 0.0688,
      "step": 10903
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5402162276528155,
      "learning_rate": 1.3111993568261389e-05,
      "loss": 0.075,
      "step": 10904
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6105603197649467,
      "learning_rate": 1.3107630014576258e-05,
      "loss": 0.2118,
      "step": 10905
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6176333824701488,
      "learning_rate": 1.3103266929102236e-05,
      "loss": 0.1119,
      "step": 10906
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.842074676164939,
      "learning_rate": 1.3098904312011146e-05,
      "loss": 0.6701,
      "step": 10907
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5607225234376341,
      "learning_rate": 1.3094542163474707e-05,
      "loss": 0.0628,
      "step": 10908
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5483761520777004,
      "learning_rate": 1.3090180483664687e-05,
      "loss": 0.0947,
      "step": 10909
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.48331600391751206,
      "learning_rate": 1.3085819272752777e-05,
      "loss": 0.0898,
      "step": 10910
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7166374323227023,
      "learning_rate": 1.3081458530910722e-05,
      "loss": 0.1285,
      "step": 10911
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5554696049088174,
      "learning_rate": 1.3077098258310175e-05,
      "loss": 0.1413,
      "step": 10912
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5859515181658221,
      "learning_rate": 1.3072738455122824e-05,
      "loss": 0.2266,
      "step": 10913
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.36912119766114543,
      "learning_rate": 1.3068379121520291e-05,
      "loss": 0.0615,
      "step": 10914
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5311022610761977,
      "learning_rate": 1.306402025767424e-05,
      "loss": 0.1041,
      "step": 10915
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1171824911869281,
      "learning_rate": 1.3059661863756255e-05,
      "loss": 0.3749,
      "step": 10916
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4805306062975478,
      "learning_rate": 1.3055303939937952e-05,
      "loss": 0.0694,
      "step": 10917
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6843612267953163,
      "learning_rate": 1.3050946486390875e-05,
      "loss": 0.1362,
      "step": 10918
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0806810497161774,
      "learning_rate": 1.3046589503286616e-05,
      "loss": 0.1009,
      "step": 10919
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3448171607219457,
      "learning_rate": 1.3042232990796683e-05,
      "loss": 0.0748,
      "step": 10920
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8081482923793503,
      "learning_rate": 1.303787694909262e-05,
      "loss": 0.2039,
      "step": 10921
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6914385144641179,
      "learning_rate": 1.3033521378345903e-05,
      "loss": 0.1836,
      "step": 10922
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5739042964118094,
      "learning_rate": 1.3029166278728025e-05,
      "loss": 0.1071,
      "step": 10923
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6531894662123012,
      "learning_rate": 1.3024811650410452e-05,
      "loss": 0.0619,
      "step": 10924
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8823234453331242,
      "learning_rate": 1.3020457493564634e-05,
      "loss": 0.4168,
      "step": 10925
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6699905864794734,
      "learning_rate": 1.3016103808361979e-05,
      "loss": 0.0927,
      "step": 10926
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.48260573599283346,
      "learning_rate": 1.3011750594973909e-05,
      "loss": 0.0812,
      "step": 10927
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.83767892761874,
      "learning_rate": 1.3007397853571818e-05,
      "loss": 0.1957,
      "step": 10928
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6685303992477306,
      "learning_rate": 1.3003045584327061e-05,
      "loss": 0.2186,
      "step": 10929
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6174297056239219,
      "learning_rate": 1.2998693787410993e-05,
      "loss": 0.1154,
      "step": 10930
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5489786239791976,
      "learning_rate": 1.2994342462994955e-05,
      "loss": 0.0941,
      "step": 10931
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4598931275462463,
      "learning_rate": 1.2989991611250269e-05,
      "loss": 0.0717,
      "step": 10932
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6260969646161906,
      "learning_rate": 1.2985641232348205e-05,
      "loss": 0.1948,
      "step": 10933
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.42462390502534436,
      "learning_rate": 1.298129132646006e-05,
      "loss": 0.0516,
      "step": 10934
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.48125283385618484,
      "learning_rate": 1.2976941893757089e-05,
      "loss": 0.0745,
      "step": 10935
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.32561531690402,
      "learning_rate": 1.2972592934410543e-05,
      "loss": 0.3459,
      "step": 10936
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5131559679674518,
      "learning_rate": 1.2968244448591621e-05,
      "loss": 0.1746,
      "step": 10937
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.43323146512508526,
      "learning_rate": 1.296389643647154e-05,
      "loss": 0.0656,
      "step": 10938
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1758384377775641,
      "learning_rate": 1.2959548898221479e-05,
      "loss": 0.4278,
      "step": 10939
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7629968181317424,
      "learning_rate": 1.295520183401262e-05,
      "loss": 0.1613,
      "step": 10940
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5688175604897783,
      "learning_rate": 1.2950855244016086e-05,
      "loss": 0.0895,
      "step": 10941
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8315310264458409,
      "learning_rate": 1.2946509128403017e-05,
      "loss": 0.1934,
      "step": 10942
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.39172972451130095,
      "learning_rate": 1.2942163487344521e-05,
      "loss": 0.0312,
      "step": 10943
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3841548477656007,
      "learning_rate": 1.2937818321011702e-05,
      "loss": 0.0487,
      "step": 10944
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.49701424125613297,
      "learning_rate": 1.2933473629575607e-05,
      "loss": 0.0948,
      "step": 10945
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0883021405842361,
      "learning_rate": 1.2929129413207306e-05,
      "loss": 0.2654,
      "step": 10946
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.49667070761339466,
      "learning_rate": 1.292478567207783e-05,
      "loss": 0.0894,
      "step": 10947
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5427216776078689,
      "learning_rate": 1.2920442406358204e-05,
      "loss": 0.1555,
      "step": 10948
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5614140343135692,
      "learning_rate": 1.291609961621941e-05,
      "loss": 0.1242,
      "step": 10949
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9426940470350946,
      "learning_rate": 1.2911757301832433e-05,
      "loss": 0.1885,
      "step": 10950
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8067816302764116,
      "learning_rate": 1.2907415463368236e-05,
      "loss": 0.101,
      "step": 10951
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.42645198494861714,
      "learning_rate": 1.290307410099777e-05,
      "loss": 0.0854,
      "step": 10952
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.37221530498132965,
      "learning_rate": 1.2898733214891931e-05,
      "loss": 0.0577,
      "step": 10953
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8047840766100594,
      "learning_rate": 1.2894392805221642e-05,
      "loss": 0.1403,
      "step": 10954
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.45584170895700277,
      "learning_rate": 1.2890052872157782e-05,
      "loss": 0.0598,
      "step": 10955
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4771989018313927,
      "learning_rate": 1.288571341587123e-05,
      "loss": 0.0784,
      "step": 10956
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3915650607833682,
      "learning_rate": 1.2881374436532814e-05,
      "loss": 0.0653,
      "step": 10957
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.47542068897537587,
      "learning_rate": 1.287703593431337e-05,
      "loss": 0.1015,
      "step": 10958
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6378729874780897,
      "learning_rate": 1.2872697909383718e-05,
      "loss": 0.1426,
      "step": 10959
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7699277209294023,
      "learning_rate": 1.2868360361914633e-05,
      "loss": 0.1815,
      "step": 10960
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8857287667754246,
      "learning_rate": 1.2864023292076898e-05,
      "loss": 0.127,
      "step": 10961
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5454060147476554,
      "learning_rate": 1.285968670004126e-05,
      "loss": 0.1075,
      "step": 10962
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5582242695886133,
      "learning_rate": 1.2855350585978466e-05,
      "loss": 0.0853,
      "step": 10963
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5223128794099261,
      "learning_rate": 1.2851014950059214e-05,
      "loss": 0.0889,
      "step": 10964
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8662729388677753,
      "learning_rate": 1.2846679792454223e-05,
      "loss": 0.1136,
      "step": 10965
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.43529067943261635,
      "learning_rate": 1.2842345113334137e-05,
      "loss": 0.0767,
      "step": 10966
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8173486559388795,
      "learning_rate": 1.2838010912869656e-05,
      "loss": 0.1936,
      "step": 10967
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9761125790859384,
      "learning_rate": 1.2833677191231392e-05,
      "loss": 0.1743,
      "step": 10968
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6710017647759811,
      "learning_rate": 1.2829343948589984e-05,
      "loss": 0.1475,
      "step": 10969
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5622008639088174,
      "learning_rate": 1.2825011185116012e-05,
      "loss": 0.1056,
      "step": 10970
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.540942608350608,
      "learning_rate": 1.2820678900980093e-05,
      "loss": 0.1973,
      "step": 10971
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.39767040240575974,
      "learning_rate": 1.2816347096352762e-05,
      "loss": 0.0669,
      "step": 10972
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5440340680768172,
      "learning_rate": 1.281201577140459e-05,
      "loss": 0.085,
      "step": 10973
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7925404460527085,
      "learning_rate": 1.2807684926306074e-05,
      "loss": 0.1772,
      "step": 10974
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.602248924624445,
      "learning_rate": 1.2803354561227754e-05,
      "loss": 0.1609,
      "step": 10975
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5082676684421938,
      "learning_rate": 1.2799024676340102e-05,
      "loss": 0.1151,
      "step": 10976
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9612550782056907,
      "learning_rate": 1.27946952718136e-05,
      "loss": 0.2644,
      "step": 10977
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0396278255115818,
      "learning_rate": 1.2790366347818671e-05,
      "loss": 0.176,
      "step": 10978
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4902097453056354,
      "learning_rate": 1.2786037904525794e-05,
      "loss": 0.1183,
      "step": 10979
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.48498909907236865,
      "learning_rate": 1.2781709942105341e-05,
      "loss": 0.0762,
      "step": 10980
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4410714438828596,
      "learning_rate": 1.2777382460727739e-05,
      "loss": 0.0809,
      "step": 10981
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8022454912649498,
      "learning_rate": 1.277305546056333e-05,
      "loss": 0.1668,
      "step": 10982
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4248869730700256,
      "learning_rate": 1.276872894178251e-05,
      "loss": 0.0789,
      "step": 10983
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8472118598995905,
      "learning_rate": 1.2764402904555583e-05,
      "loss": 0.1114,
      "step": 10984
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.662325870954618,
      "learning_rate": 1.2760077349052896e-05,
      "loss": 0.1466,
      "step": 10985
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5728913112930811,
      "learning_rate": 1.2755752275444716e-05,
      "loss": 0.1239,
      "step": 10986
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5093095160616401,
      "learning_rate": 1.2751427683901363e-05,
      "loss": 0.0615,
      "step": 10987
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4768911682619014,
      "learning_rate": 1.2747103574593072e-05,
      "loss": 0.0625,
      "step": 10988
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.42698572650096756,
      "learning_rate": 1.2742779947690103e-05,
      "loss": 0.0447,
      "step": 10989
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5947610286316541,
      "learning_rate": 1.273845680336266e-05,
      "loss": 0.1814,
      "step": 10990
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5029417513124697,
      "learning_rate": 1.2734134141780963e-05,
      "loss": 0.0669,
      "step": 10991
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4647249905837897,
      "learning_rate": 1.2729811963115196e-05,
      "loss": 0.067,
      "step": 10992
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6716606031089867,
      "learning_rate": 1.2725490267535534e-05,
      "loss": 0.1606,
      "step": 10993
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6663024586194146,
      "learning_rate": 1.2721169055212106e-05,
      "loss": 0.1738,
      "step": 10994
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6890552489123273,
      "learning_rate": 1.2716848326315057e-05,
      "loss": 0.1374,
      "step": 10995
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.41709462014841403,
      "learning_rate": 1.2712528081014496e-05,
      "loss": 0.0654,
      "step": 10996
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4772610846497863,
      "learning_rate": 1.2708208319480507e-05,
      "loss": 0.1274,
      "step": 10997
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6109211724387482,
      "learning_rate": 1.270388904188316e-05,
      "loss": 0.0969,
      "step": 10998
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6276693837498535,
      "learning_rate": 1.2699570248392518e-05,
      "loss": 0.1245,
      "step": 10999
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6314808775231536,
      "learning_rate": 1.269525193917862e-05,
      "loss": 0.0778,
      "step": 11000
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9077623288016121,
      "learning_rate": 1.2690934114411451e-05,
      "loss": 0.1616,
      "step": 11001
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.47228753926864564,
      "learning_rate": 1.2686616774261045e-05,
      "loss": 0.1035,
      "step": 11002
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7905809831415084,
      "learning_rate": 1.2682299918897356e-05,
      "loss": 0.1428,
      "step": 11003
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6600023661522462,
      "learning_rate": 1.2677983548490352e-05,
      "loss": 0.1214,
      "step": 11004
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8525426129251309,
      "learning_rate": 1.267366766320995e-05,
      "loss": 0.144,
      "step": 11005
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6472667000786728,
      "learning_rate": 1.2669352263226106e-05,
      "loss": 0.1511,
      "step": 11006
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3470109692651878,
      "learning_rate": 1.2665037348708688e-05,
      "loss": 0.0445,
      "step": 11007
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5153929967603456,
      "learning_rate": 1.26607229198276e-05,
      "loss": 0.0853,
      "step": 11008
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6296308438759384,
      "learning_rate": 1.2656408976752677e-05,
      "loss": 0.1293,
      "step": 11009
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3426178019895989,
      "learning_rate": 1.2652095519653798e-05,
      "loss": 0.0463,
      "step": 11010
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.47149527820285886,
      "learning_rate": 1.2647782548700757e-05,
      "loss": 0.0746,
      "step": 11011
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.40226003992518394,
      "learning_rate": 1.2643470064063378e-05,
      "loss": 0.0446,
      "step": 11012
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5320525667659081,
      "learning_rate": 1.2639158065911421e-05,
      "loss": 0.0983,
      "step": 11013
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4499432896327546,
      "learning_rate": 1.2634846554414686e-05,
      "loss": 0.0877,
      "step": 11014
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4573990444687162,
      "learning_rate": 1.2630535529742899e-05,
      "loss": 0.0524,
      "step": 11015
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.557706671951182,
      "learning_rate": 1.26262249920658e-05,
      "loss": 0.1644,
      "step": 11016
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.836099590645033,
      "learning_rate": 1.2621914941553081e-05,
      "loss": 0.2533,
      "step": 11017
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6392169895812021,
      "learning_rate": 1.2617605378374442e-05,
      "loss": 0.102,
      "step": 11018
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.48186470711584506,
      "learning_rate": 1.2613296302699554e-05,
      "loss": 0.0739,
      "step": 11019
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6049854148554392,
      "learning_rate": 1.2608987714698078e-05,
      "loss": 0.21,
      "step": 11020
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4546052049404639,
      "learning_rate": 1.2604679614539624e-05,
      "loss": 0.071,
      "step": 11021
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4493280933766033,
      "learning_rate": 1.2600372002393817e-05,
      "loss": 0.0718,
      "step": 11022
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3614050044013723,
      "learning_rate": 1.2596064878430252e-05,
      "loss": 0.0597,
      "step": 11023
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.762495154233143,
      "learning_rate": 1.2591758242818514e-05,
      "loss": 0.4286,
      "step": 11024
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5517685689767242,
      "learning_rate": 1.2587452095728136e-05,
      "loss": 0.0741,
      "step": 11025
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6456563480833405,
      "learning_rate": 1.2583146437328663e-05,
      "loss": 0.1164,
      "step": 11026
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.269355778762455,
      "learning_rate": 1.2578841267789627e-05,
      "loss": 0.2559,
      "step": 11027
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5019619408849959,
      "learning_rate": 1.2574536587280497e-05,
      "loss": 0.0732,
      "step": 11028
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6067915056168303,
      "learning_rate": 1.2570232395970772e-05,
      "loss": 0.1034,
      "step": 11029
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4685060593714233,
      "learning_rate": 1.2565928694029899e-05,
      "loss": 0.0518,
      "step": 11030
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6640911619576335,
      "learning_rate": 1.256162548162734e-05,
      "loss": 0.1489,
      "step": 11031
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4532586949491522,
      "learning_rate": 1.2557322758932489e-05,
      "loss": 0.101,
      "step": 11032
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6022567143083702,
      "learning_rate": 1.2553020526114755e-05,
      "loss": 0.242,
      "step": 11033
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.579155848416812,
      "learning_rate": 1.2548718783343527e-05,
      "loss": 0.0865,
      "step": 11034
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3714423053232346,
      "learning_rate": 1.2544417530788175e-05,
      "loss": 0.05,
      "step": 11035
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.652995042368128,
      "learning_rate": 1.2540116768618018e-05,
      "loss": 0.079,
      "step": 11036
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7619562377545693,
      "learning_rate": 1.2535816497002396e-05,
      "loss": 0.1215,
      "step": 11037
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7082500496688608,
      "learning_rate": 1.253151671611061e-05,
      "loss": 0.1856,
      "step": 11038
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.557126925461875,
      "learning_rate": 1.252721742611196e-05,
      "loss": 0.0833,
      "step": 11039
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6340139605928067,
      "learning_rate": 1.2522918627175684e-05,
      "loss": 0.2095,
      "step": 11040
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4117156776138639,
      "learning_rate": 1.2518620319471045e-05,
      "loss": 0.1429,
      "step": 11041
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4489853505943247,
      "learning_rate": 1.251432250316727e-05,
      "loss": 0.0578,
      "step": 11042
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5141748308043873,
      "learning_rate": 1.2510025178433576e-05,
      "loss": 0.0612,
      "step": 11043
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.48522934033944676,
      "learning_rate": 1.2505728345439133e-05,
      "loss": 0.1105,
      "step": 11044
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5301655557986319,
      "learning_rate": 1.2501432004353117e-05,
      "loss": 0.0831,
      "step": 11045
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.629361341684693,
      "learning_rate": 1.2497136155344683e-05,
      "loss": 0.1179,
      "step": 11046
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5141588017778564,
      "learning_rate": 1.2492840798582967e-05,
      "loss": 0.1019,
      "step": 11047
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7156012763712564,
      "learning_rate": 1.2488545934237062e-05,
      "loss": 0.2243,
      "step": 11048
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5228306545366126,
      "learning_rate": 1.248425156247607e-05,
      "loss": 0.0958,
      "step": 11049
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.37881113357835466,
      "learning_rate": 1.2479957683469066e-05,
      "loss": 0.05,
      "step": 11050
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.605789114144452,
      "learning_rate": 1.2475664297385109e-05,
      "loss": 0.1118,
      "step": 11051
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6550324974410928,
      "learning_rate": 1.2471371404393217e-05,
      "loss": 0.35,
      "step": 11052
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4825470132436383,
      "learning_rate": 1.246707900466241e-05,
      "loss": 0.0999,
      "step": 11053
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5877983033260972,
      "learning_rate": 1.2462787098361684e-05,
      "loss": 0.1101,
      "step": 11054
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.49939697375097913,
      "learning_rate": 1.2458495685660027e-05,
      "loss": 0.0831,
      "step": 11055
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.45919191345972415,
      "learning_rate": 1.2454204766726371e-05,
      "loss": 0.0485,
      "step": 11056
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.731887590897863,
      "learning_rate": 1.2449914341729665e-05,
      "loss": 0.3198,
      "step": 11057
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6836313309774749,
      "learning_rate": 1.2445624410838836e-05,
      "loss": 0.1673,
      "step": 11058
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4655436372770464,
      "learning_rate": 1.244133497422276e-05,
      "loss": 0.1293,
      "step": 11059
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3975445636697859,
      "learning_rate": 1.2437046032050328e-05,
      "loss": 0.049,
      "step": 11060
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5926165462556852,
      "learning_rate": 1.2432757584490397e-05,
      "loss": 0.1717,
      "step": 11061
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3546177271609638,
      "learning_rate": 1.2428469631711817e-05,
      "loss": 0.0359,
      "step": 11062
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6776968017190409,
      "learning_rate": 1.2424182173883387e-05,
      "loss": 0.0959,
      "step": 11063
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5608440888055836,
      "learning_rate": 1.2419895211173927e-05,
      "loss": 0.0865,
      "step": 11064
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3965548038276129,
      "learning_rate": 1.241560874375219e-05,
      "loss": 0.0713,
      "step": 11065
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6419527371664318,
      "learning_rate": 1.2411322771786973e-05,
      "loss": 0.1049,
      "step": 11066
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5752210810359467,
      "learning_rate": 1.2407037295446994e-05,
      "loss": 0.0903,
      "step": 11067
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8086939282302504,
      "learning_rate": 1.240275231490099e-05,
      "loss": 0.13,
      "step": 11068
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6645267059521176,
      "learning_rate": 1.2398467830317634e-05,
      "loss": 0.378,
      "step": 11069
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7849074255579863,
      "learning_rate": 1.2394183841865652e-05,
      "loss": 0.1265,
      "step": 11070
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.46562154766906616,
      "learning_rate": 1.2389900349713676e-05,
      "loss": 0.0964,
      "step": 11071
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8370974848192525,
      "learning_rate": 1.2385617354030372e-05,
      "loss": 0.145,
      "step": 11072
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.47266234550419284,
      "learning_rate": 1.2381334854984331e-05,
      "loss": 0.0595,
      "step": 11073
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5251332632291557,
      "learning_rate": 1.23770528527442e-05,
      "loss": 0.1,
      "step": 11074
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5726587716816948,
      "learning_rate": 1.2372771347478538e-05,
      "loss": 0.0977,
      "step": 11075
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.577266021348795,
      "learning_rate": 1.2368490339355927e-05,
      "loss": 0.0911,
      "step": 11076
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6579655053891765,
      "learning_rate": 1.2364209828544882e-05,
      "loss": 0.0834,
      "step": 11077
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6945522464164596,
      "learning_rate": 1.2359929815213973e-05,
      "loss": 0.1654,
      "step": 11078
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6104149015241999,
      "learning_rate": 1.2355650299531674e-05,
      "loss": 0.0956,
      "step": 11079
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5058188884318179,
      "learning_rate": 1.2351371281666496e-05,
      "loss": 0.0643,
      "step": 11080
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5954494910710305,
      "learning_rate": 1.2347092761786878e-05,
      "loss": 0.092,
      "step": 11081
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8581128151952807,
      "learning_rate": 1.23428147400613e-05,
      "loss": 0.1887,
      "step": 11082
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3456042344285545,
      "learning_rate": 1.2338537216658172e-05,
      "loss": 0.0422,
      "step": 11083
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7501024954526222,
      "learning_rate": 1.2334260191745917e-05,
      "loss": 0.2164,
      "step": 11084
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6580160119504799,
      "learning_rate": 1.2329983665492906e-05,
      "loss": 0.239,
      "step": 11085
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5295800458341567,
      "learning_rate": 1.2325707638067516e-05,
      "loss": 0.1312,
      "step": 11086
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3100670096682545,
      "learning_rate": 1.2321432109638101e-05,
      "loss": 0.2901,
      "step": 11087
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4121334317039299,
      "learning_rate": 1.2317157080373002e-05,
      "loss": 0.0659,
      "step": 11088
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5493434651434232,
      "learning_rate": 1.2312882550440504e-05,
      "loss": 0.122,
      "step": 11089
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6383614548458107,
      "learning_rate": 1.2308608520008916e-05,
      "loss": 0.1616,
      "step": 11090
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4089356142664181,
      "learning_rate": 1.2304334989246504e-05,
      "loss": 0.0937,
      "step": 11091
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5782787055006066,
      "learning_rate": 1.2300061958321532e-05,
      "loss": 0.1313,
      "step": 11092
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4613893421932271,
      "learning_rate": 1.2295789427402215e-05,
      "loss": 0.0361,
      "step": 11093
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2377736839152933,
      "learning_rate": 1.2291517396656772e-05,
      "loss": 0.2808,
      "step": 11094
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6208510472994145,
      "learning_rate": 1.2287245866253405e-05,
      "loss": 0.1344,
      "step": 11095
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9920551528504646,
      "learning_rate": 1.2282974836360272e-05,
      "loss": 0.2034,
      "step": 11096
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6170678232445074,
      "learning_rate": 1.2278704307145533e-05,
      "loss": 0.1281,
      "step": 11097
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9883261353865773,
      "learning_rate": 1.2274434278777322e-05,
      "loss": 0.1991,
      "step": 11098
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5123018940031157,
      "learning_rate": 1.2270164751423763e-05,
      "loss": 0.0658,
      "step": 11099
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.42456306352544165,
      "learning_rate": 1.2265895725252932e-05,
      "loss": 0.0787,
      "step": 11100
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.45946741597002305,
      "learning_rate": 1.226162720043291e-05,
      "loss": 0.0661,
      "step": 11101
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.660283983828305,
      "learning_rate": 1.2257359177131759e-05,
      "loss": 0.0841,
      "step": 11102
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5880156585739038,
      "learning_rate": 1.2253091655517515e-05,
      "loss": 0.0607,
      "step": 11103
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5682872977404241,
      "learning_rate": 1.224882463575818e-05,
      "loss": 0.0702,
      "step": 11104
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.262924162990557,
      "learning_rate": 1.2244558118021757e-05,
      "loss": 0.3073,
      "step": 11105
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6443755936165347,
      "learning_rate": 1.2240292102476223e-05,
      "loss": 0.1361,
      "step": 11106
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5627620445846562,
      "learning_rate": 1.2236026589289543e-05,
      "loss": 0.106,
      "step": 11107
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5853939795954372,
      "learning_rate": 1.2231761578629633e-05,
      "loss": 0.1131,
      "step": 11108
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6500964713599228,
      "learning_rate": 1.2227497070664421e-05,
      "loss": 0.1673,
      "step": 11109
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.42079804497401735,
      "learning_rate": 1.2223233065561804e-05,
      "loss": 0.1183,
      "step": 11110
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4536801827940526,
      "learning_rate": 1.2218969563489666e-05,
      "loss": 0.124,
      "step": 11111
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3571294194137876,
      "learning_rate": 1.2214706564615847e-05,
      "loss": 0.0413,
      "step": 11112
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6222106664835264,
      "learning_rate": 1.2210444069108196e-05,
      "loss": 0.1466,
      "step": 11113
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5007211027196923,
      "learning_rate": 1.2206182077134523e-05,
      "loss": 0.1025,
      "step": 11114
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5150503730133424,
      "learning_rate": 1.2201920588862645e-05,
      "loss": 0.1194,
      "step": 11115
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4691913880882382,
      "learning_rate": 1.2197659604460312e-05,
      "loss": 0.0942,
      "step": 11116
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.43632080963809383,
      "learning_rate": 1.2193399124095297e-05,
      "loss": 0.0877,
      "step": 11117
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6382397919684369,
      "learning_rate": 1.218913914793534e-05,
      "loss": 0.1257,
      "step": 11118
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3483548408760173,
      "learning_rate": 1.2184879676148162e-05,
      "loss": 0.0512,
      "step": 11119
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8060850592123743,
      "learning_rate": 1.2180620708901446e-05,
      "loss": 0.2292,
      "step": 11120
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6948896260864542,
      "learning_rate": 1.217636224636288e-05,
      "loss": 0.1657,
      "step": 11121
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5423468904543254,
      "learning_rate": 1.2172104288700134e-05,
      "loss": 0.111,
      "step": 11122
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5114037316217237,
      "learning_rate": 1.2167846836080823e-05,
      "loss": 0.0836,
      "step": 11123
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5531756312644521,
      "learning_rate": 1.216358988867258e-05,
      "loss": 0.0947,
      "step": 11124
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8377225041352288,
      "learning_rate": 1.2159333446643003e-05,
      "loss": 0.1982,
      "step": 11125
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.027875501845911,
      "learning_rate": 1.2155077510159682e-05,
      "loss": 0.3724,
      "step": 11126
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.46965563998848253,
      "learning_rate": 1.2150822079390153e-05,
      "loss": 0.0716,
      "step": 11127
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.652914646750864,
      "learning_rate": 1.2146567154501967e-05,
      "loss": 0.1075,
      "step": 11128
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4337030428129551,
      "learning_rate": 1.2142312735662645e-05,
      "loss": 0.1278,
      "step": 11129
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5006462103590522,
      "learning_rate": 1.2138058823039697e-05,
      "loss": 0.093,
      "step": 11130
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7217203644848017,
      "learning_rate": 1.2133805416800579e-05,
      "loss": 0.1785,
      "step": 11131
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.49495108769709634,
      "learning_rate": 1.2129552517112771e-05,
      "loss": 0.0767,
      "step": 11132
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5903519337851454,
      "learning_rate": 1.2125300124143687e-05,
      "loss": 0.0818,
      "step": 11133
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8333579184064615,
      "learning_rate": 1.2121048238060784e-05,
      "loss": 0.2094,
      "step": 11134
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9569385621281944,
      "learning_rate": 1.2116796859031431e-05,
      "loss": 0.1714,
      "step": 11135
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.31992896957346995,
      "learning_rate": 1.211254598722302e-05,
      "loss": 0.0485,
      "step": 11136
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9032064525604203,
      "learning_rate": 1.2108295622802906e-05,
      "loss": 0.2171,
      "step": 11137
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6957451552551529,
      "learning_rate": 1.2104045765938446e-05,
      "loss": 0.1704,
      "step": 11138
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.36015549799065394,
      "learning_rate": 1.2099796416796935e-05,
      "loss": 0.0497,
      "step": 11139
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.44431927066904137,
      "learning_rate": 1.2095547575545686e-05,
      "loss": 0.106,
      "step": 11140
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4402600004678621,
      "learning_rate": 1.2091299242351975e-05,
      "loss": 0.0615,
      "step": 11141
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7296742263302004,
      "learning_rate": 1.2087051417383075e-05,
      "loss": 0.1664,
      "step": 11142
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2120342015530015,
      "learning_rate": 1.2082804100806206e-05,
      "loss": 0.4069,
      "step": 11143
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4599098655484877,
      "learning_rate": 1.2078557292788596e-05,
      "loss": 0.1431,
      "step": 11144
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5077255210142226,
      "learning_rate": 1.2074310993497446e-05,
      "loss": 0.0845,
      "step": 11145
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.43470140346693875,
      "learning_rate": 1.2070065203099945e-05,
      "loss": 0.0789,
      "step": 11146
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2477018724084328,
      "learning_rate": 1.2065819921763236e-05,
      "loss": 0.1964,
      "step": 11147
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.44136014851595434,
      "learning_rate": 1.2061575149654464e-05,
      "loss": 0.0592,
      "step": 11148
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8778691874250786,
      "learning_rate": 1.2057330886940751e-05,
      "loss": 0.2908,
      "step": 11149
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4535762363284654,
      "learning_rate": 1.2053087133789207e-05,
      "loss": 0.0825,
      "step": 11150
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6162099368938381,
      "learning_rate": 1.2048843890366892e-05,
      "loss": 0.0838,
      "step": 11151
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.45083140389944637,
      "learning_rate": 1.2044601156840874e-05,
      "loss": 0.1097,
      "step": 11152
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7125369478043694,
      "learning_rate": 1.2040358933378204e-05,
      "loss": 0.182,
      "step": 11153
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6441374603571842,
      "learning_rate": 1.2036117220145879e-05,
      "loss": 0.1839,
      "step": 11154
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.39539983507238063,
      "learning_rate": 1.203187601731091e-05,
      "loss": 0.0379,
      "step": 11155
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.47992525542817077,
      "learning_rate": 1.202763532504028e-05,
      "loss": 0.0866,
      "step": 11156
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6548186950630611,
      "learning_rate": 1.2023395143500948e-05,
      "loss": 0.0794,
      "step": 11157
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.503730703585689,
      "learning_rate": 1.2019155472859845e-05,
      "loss": 0.5298,
      "step": 11158
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5684531526093574,
      "learning_rate": 1.2014916313283892e-05,
      "loss": 0.1077,
      "step": 11159
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6894670556780119,
      "learning_rate": 1.201067766493999e-05,
      "loss": 0.1769,
      "step": 11160
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4010865283001093,
      "learning_rate": 1.2006439527995026e-05,
      "loss": 0.0717,
      "step": 11161
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.44712988123234304,
      "learning_rate": 1.200220190261584e-05,
      "loss": 0.0776,
      "step": 11162
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.36582154326638233,
      "learning_rate": 1.199796478896929e-05,
      "loss": 0.0437,
      "step": 11163
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.746310334324701,
      "learning_rate": 1.1993728187222167e-05,
      "loss": 0.1693,
      "step": 11164
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3652494828096459,
      "learning_rate": 1.1989492097541305e-05,
      "loss": 0.0482,
      "step": 11165
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6957463567810571,
      "learning_rate": 1.1985256520093454e-05,
      "loss": 0.1018,
      "step": 11166
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.39260014723851266,
      "learning_rate": 1.1981021455045391e-05,
      "loss": 0.0749,
      "step": 11167
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4037453344117229,
      "learning_rate": 1.1976786902563824e-05,
      "loss": 0.0765,
      "step": 11168
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.44760902413711734,
      "learning_rate": 1.1972552862815509e-05,
      "loss": 0.0678,
      "step": 11169
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5391012147724525,
      "learning_rate": 1.1968319335967115e-05,
      "loss": 0.0714,
      "step": 11170
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3903415479812354,
      "learning_rate": 1.1964086322185336e-05,
      "loss": 0.074,
      "step": 11171
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5629767934120423,
      "learning_rate": 1.1959853821636804e-05,
      "loss": 0.1669,
      "step": 11172
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5595650558641972,
      "learning_rate": 1.1955621834488193e-05,
      "loss": 0.1498,
      "step": 11173
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.546272994057365,
      "learning_rate": 1.1951390360906087e-05,
      "loss": 0.0869,
      "step": 11174
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7276964139651932,
      "learning_rate": 1.1947159401057103e-05,
      "loss": 0.1444,
      "step": 11175
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8373320475907415,
      "learning_rate": 1.194292895510779e-05,
      "loss": 0.191,
      "step": 11176
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5472026588154416,
      "learning_rate": 1.1938699023224743e-05,
      "loss": 0.0943,
      "step": 11177
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6333298925157153,
      "learning_rate": 1.1934469605574463e-05,
      "loss": 0.0733,
      "step": 11178
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9337800593938628,
      "learning_rate": 1.193024070232349e-05,
      "loss": 0.2662,
      "step": 11179
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5544523983511155,
      "learning_rate": 1.1926012313638288e-05,
      "loss": 0.0837,
      "step": 11180
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4419695539756849,
      "learning_rate": 1.1921784439685368e-05,
      "loss": 0.0659,
      "step": 11181
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5217020981663104,
      "learning_rate": 1.1917557080631162e-05,
      "loss": 0.0956,
      "step": 11182
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8197666112777432,
      "learning_rate": 1.1913330236642117e-05,
      "loss": 0.1159,
      "step": 11183
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4492306555775715,
      "learning_rate": 1.1909103907884628e-05,
      "loss": 0.0659,
      "step": 11184
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.405760556956992,
      "learning_rate": 1.19048780945251e-05,
      "loss": 0.0606,
      "step": 11185
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2602051671980523,
      "learning_rate": 1.190065279672991e-05,
      "loss": 0.038,
      "step": 11186
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.45250410383929424,
      "learning_rate": 1.1896428014665417e-05,
      "loss": 0.0701,
      "step": 11187
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.482838655320112,
      "learning_rate": 1.1892203748497932e-05,
      "loss": 0.1164,
      "step": 11188
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.46591882360515074,
      "learning_rate": 1.188797999839378e-05,
      "loss": 0.1219,
      "step": 11189
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5523167981120007,
      "learning_rate": 1.188375676451926e-05,
      "loss": 0.1021,
      "step": 11190
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7488710557271147,
      "learning_rate": 1.187953404704063e-05,
      "loss": 0.1854,
      "step": 11191
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6448088605792296,
      "learning_rate": 1.1875311846124145e-05,
      "loss": 0.1063,
      "step": 11192
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.622612944174411,
      "learning_rate": 1.1871090161936043e-05,
      "loss": 0.1352,
      "step": 11193
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4515033669834179,
      "learning_rate": 1.1866868994642535e-05,
      "loss": 0.1035,
      "step": 11194
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5070102015702668,
      "learning_rate": 1.1862648344409802e-05,
      "loss": 0.0602,
      "step": 11195
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.44281964494325143,
      "learning_rate": 1.1858428211404016e-05,
      "loss": 0.0936,
      "step": 11196
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8386387423832027,
      "learning_rate": 1.185420859579133e-05,
      "loss": 0.2673,
      "step": 11197
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.45754711072951393,
      "learning_rate": 1.1849989497737881e-05,
      "loss": 0.0821,
      "step": 11198
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5643313879930507,
      "learning_rate": 1.1845770917409763e-05,
      "loss": 0.0776,
      "step": 11199
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7041252940326062,
      "learning_rate": 1.184155285497307e-05,
      "loss": 0.1305,
      "step": 11200
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6355269513815395,
      "learning_rate": 1.1837335310593871e-05,
      "loss": 0.1024,
      "step": 11201
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9536405897348305,
      "learning_rate": 1.1833118284438224e-05,
      "loss": 0.1792,
      "step": 11202
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.41387620685644627,
      "learning_rate": 1.1828901776672138e-05,
      "loss": 0.0582,
      "step": 11203
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5263617228408723,
      "learning_rate": 1.1824685787461626e-05,
      "loss": 0.1145,
      "step": 11204
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6172970700447796,
      "learning_rate": 1.182047031697268e-05,
      "loss": 0.1238,
      "step": 11205
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5181776361000796,
      "learning_rate": 1.1816255365371271e-05,
      "loss": 0.0997,
      "step": 11206
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4323911792721841,
      "learning_rate": 1.1812040932823327e-05,
      "loss": 0.1016,
      "step": 11207
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.38480446827914166,
      "learning_rate": 1.1807827019494783e-05,
      "loss": 0.0577,
      "step": 11208
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7772270700694927,
      "learning_rate": 1.1803613625551544e-05,
      "loss": 0.2711,
      "step": 11209
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5111590080657841,
      "learning_rate": 1.1799400751159503e-05,
      "loss": 0.0975,
      "step": 11210
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5554113138637227,
      "learning_rate": 1.1795188396484505e-05,
      "loss": 0.1497,
      "step": 11211
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8215899977602179,
      "learning_rate": 1.1790976561692405e-05,
      "loss": 0.2195,
      "step": 11212
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.46647155362099785,
      "learning_rate": 1.1786765246949025e-05,
      "loss": 0.1036,
      "step": 11213
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4423936545237122,
      "learning_rate": 1.1782554452420174e-05,
      "loss": 0.0653,
      "step": 11214
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.823488507990864,
      "learning_rate": 1.1778344178271617e-05,
      "loss": 0.1961,
      "step": 11215
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5083521764937201,
      "learning_rate": 1.1774134424669129e-05,
      "loss": 0.1266,
      "step": 11216
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6278546563061828,
      "learning_rate": 1.1769925191778445e-05,
      "loss": 0.135,
      "step": 11217
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5992447102353526,
      "learning_rate": 1.1765716479765296e-05,
      "loss": 0.1036,
      "step": 11218
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5644069177226636,
      "learning_rate": 1.1761508288795364e-05,
      "loss": 0.065,
      "step": 11219
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6018225835979419,
      "learning_rate": 1.1757300619034343e-05,
      "loss": 0.1391,
      "step": 11220
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5265799447013925,
      "learning_rate": 1.175309347064789e-05,
      "loss": 0.0685,
      "step": 11221
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0196029531980102,
      "learning_rate": 1.1748886843801637e-05,
      "loss": 0.338,
      "step": 11222
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4411854345998911,
      "learning_rate": 1.1744680738661204e-05,
      "loss": 0.0737,
      "step": 11223
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7816026543989165,
      "learning_rate": 1.174047515539219e-05,
      "loss": 0.2029,
      "step": 11224
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6975444169041507,
      "learning_rate": 1.1736270094160181e-05,
      "loss": 0.2129,
      "step": 11225
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.696802314973913,
      "learning_rate": 1.1732065555130715e-05,
      "loss": 0.0962,
      "step": 11226
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1458228321506703,
      "learning_rate": 1.1727861538469348e-05,
      "loss": 0.2889,
      "step": 11227
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4764967460937661,
      "learning_rate": 1.1723658044341563e-05,
      "loss": 0.0881,
      "step": 11228
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6379956157273153,
      "learning_rate": 1.1719455072912893e-05,
      "loss": 0.2541,
      "step": 11229
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.678662428308563,
      "learning_rate": 1.1715252624348788e-05,
      "loss": 0.1854,
      "step": 11230
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4571316724344946,
      "learning_rate": 1.1711050698814716e-05,
      "loss": 0.0915,
      "step": 11231
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4444954320870227,
      "learning_rate": 1.1706849296476085e-05,
      "loss": 0.0837,
      "step": 11232
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4508590591625066,
      "learning_rate": 1.170264841749834e-05,
      "loss": 0.0867,
      "step": 11233
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.46221803685298823,
      "learning_rate": 1.1698448062046848e-05,
      "loss": 0.0779,
      "step": 11234
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8588703826923999,
      "learning_rate": 1.1694248230287e-05,
      "loss": 0.2064,
      "step": 11235
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4569607396146993,
      "learning_rate": 1.1690048922384117e-05,
      "loss": 0.1363,
      "step": 11236
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6412530560597562,
      "learning_rate": 1.1685850138503562e-05,
      "loss": 0.1056,
      "step": 11237
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3694608168539993,
      "learning_rate": 1.1681651878810621e-05,
      "loss": 0.0517,
      "step": 11238
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5143701071403257,
      "learning_rate": 1.16774541434706e-05,
      "loss": 0.0854,
      "step": 11239
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6847444334246722,
      "learning_rate": 1.167325693264874e-05,
      "loss": 0.1098,
      "step": 11240
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6498956073148662,
      "learning_rate": 1.1669060246510322e-05,
      "loss": 0.1684,
      "step": 11241
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.48259537721856327,
      "learning_rate": 1.1664864085220548e-05,
      "loss": 0.0897,
      "step": 11242
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.46302609021872987,
      "learning_rate": 1.1660668448944642e-05,
      "loss": 0.0965,
      "step": 11243
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.45684043806516533,
      "learning_rate": 1.165647333784776e-05,
      "loss": 0.0693,
      "step": 11244
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5138154156670408,
      "learning_rate": 1.1652278752095103e-05,
      "loss": 0.0734,
      "step": 11245
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7572870442419496,
      "learning_rate": 1.1648084691851787e-05,
      "loss": 0.2051,
      "step": 11246
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.41491292101718885,
      "learning_rate": 1.1643891157282957e-05,
      "loss": 0.0434,
      "step": 11247
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.47006659570751064,
      "learning_rate": 1.1639698148553684e-05,
      "loss": 0.1191,
      "step": 11248
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7581948902228935,
      "learning_rate": 1.163550566582909e-05,
      "loss": 0.1747,
      "step": 11249
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4046509735728278,
      "learning_rate": 1.1631313709274208e-05,
      "loss": 0.0698,
      "step": 11250
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8694707630384952,
      "learning_rate": 1.1627122279054096e-05,
      "loss": 0.2028,
      "step": 11251
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6600258721672783,
      "learning_rate": 1.1622931375333751e-05,
      "loss": 0.1098,
      "step": 11252
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8003828683011855,
      "learning_rate": 1.1618740998278185e-05,
      "loss": 0.2247,
      "step": 11253
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7486366878166832,
      "learning_rate": 1.1614551148052377e-05,
      "loss": 0.1868,
      "step": 11254
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5171443035905072,
      "learning_rate": 1.1610361824821295e-05,
      "loss": 0.1392,
      "step": 11255
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3802873589688095,
      "learning_rate": 1.160617302874985e-05,
      "loss": 0.045,
      "step": 11256
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3226810300659052,
      "learning_rate": 1.1601984760002976e-05,
      "loss": 0.0557,
      "step": 11257
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5921918018629626,
      "learning_rate": 1.159779701874557e-05,
      "loss": 0.0781,
      "step": 11258
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3404619795862052,
      "learning_rate": 1.1593609805142494e-05,
      "loss": 0.0483,
      "step": 11259
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9120400752188356,
      "learning_rate": 1.1589423119358605e-05,
      "loss": 0.2047,
      "step": 11260
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1587436733202883,
      "learning_rate": 1.1585236961558743e-05,
      "loss": 0.5587,
      "step": 11261
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4760183718292499,
      "learning_rate": 1.1581051331907722e-05,
      "loss": 0.1193,
      "step": 11262
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.475824550534685,
      "learning_rate": 1.1576866230570318e-05,
      "loss": 0.5751,
      "step": 11263
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5006461673281224,
      "learning_rate": 1.1572681657711312e-05,
      "loss": 0.0587,
      "step": 11264
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3709339750760876,
      "learning_rate": 1.1568497613495452e-05,
      "loss": 0.0514,
      "step": 11265
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5583236063971578,
      "learning_rate": 1.1564314098087476e-05,
      "loss": 0.0863,
      "step": 11266
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5279390386748655,
      "learning_rate": 1.1560131111652073e-05,
      "loss": 0.056,
      "step": 11267
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7926881595742024,
      "learning_rate": 1.1555948654353938e-05,
      "loss": 0.2758,
      "step": 11268
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6878248951340837,
      "learning_rate": 1.1551766726357743e-05,
      "loss": 0.1514,
      "step": 11269
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8022133633016973,
      "learning_rate": 1.1547585327828137e-05,
      "loss": 0.2454,
      "step": 11270
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6025238463906859,
      "learning_rate": 1.1543404458929721e-05,
      "loss": 0.1155,
      "step": 11271
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7909490499585048,
      "learning_rate": 1.1539224119827132e-05,
      "loss": 0.5992,
      "step": 11272
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1552913802265805,
      "learning_rate": 1.1535044310684928e-05,
      "loss": 0.1851,
      "step": 11273
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8574366249402321,
      "learning_rate": 1.153086503166769e-05,
      "loss": 0.1773,
      "step": 11274
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5278327162821004,
      "learning_rate": 1.152668628293993e-05,
      "loss": 0.2984,
      "step": 11275
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.39571283403436547,
      "learning_rate": 1.1522508064666204e-05,
      "loss": 0.0545,
      "step": 11276
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.39478935362026313,
      "learning_rate": 1.1518330377010985e-05,
      "loss": 0.0885,
      "step": 11277
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.49181001070176594,
      "learning_rate": 1.1514153220138771e-05,
      "loss": 0.098,
      "step": 11278
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5191138440297403,
      "learning_rate": 1.1509976594213991e-05,
      "loss": 0.0605,
      "step": 11279
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9049888762633803,
      "learning_rate": 1.1505800499401121e-05,
      "loss": 0.2386,
      "step": 11280
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4559174192144541,
      "learning_rate": 1.1501624935864544e-05,
      "loss": 0.0968,
      "step": 11281
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.513747968538071,
      "learning_rate": 1.1497449903768678e-05,
      "loss": 0.0764,
      "step": 11282
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.535700647406993,
      "learning_rate": 1.1493275403277879e-05,
      "loss": 0.117,
      "step": 11283
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8193031584820992,
      "learning_rate": 1.1489101434556504e-05,
      "loss": 0.2369,
      "step": 11284
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1373377535334432,
      "learning_rate": 1.148492799776889e-05,
      "loss": 0.266,
      "step": 11285
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6150397471207388,
      "learning_rate": 1.1480755093079353e-05,
      "loss": 0.0792,
      "step": 11286
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.43111034374723195,
      "learning_rate": 1.147658272065217e-05,
      "loss": 0.0559,
      "step": 11287
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6754501541655228,
      "learning_rate": 1.1472410880651618e-05,
      "loss": 0.155,
      "step": 11288
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5763438791489721,
      "learning_rate": 1.1468239573241954e-05,
      "loss": 0.1265,
      "step": 11289
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5001541222205091,
      "learning_rate": 1.1464068798587386e-05,
      "loss": 0.0994,
      "step": 11290
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5439759164067468,
      "learning_rate": 1.145989855685213e-05,
      "loss": 0.0897,
      "step": 11291
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4633168284671296,
      "learning_rate": 1.1455728848200373e-05,
      "loss": 0.0988,
      "step": 11292
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4256314379003784,
      "learning_rate": 1.1451559672796283e-05,
      "loss": 0.0793,
      "step": 11293
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4433245594591983,
      "learning_rate": 1.1447391030803995e-05,
      "loss": 0.0844,
      "step": 11294
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6374151678948653,
      "learning_rate": 1.144322292238763e-05,
      "loss": 0.1677,
      "step": 11295
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7897652911250976,
      "learning_rate": 1.1439055347711298e-05,
      "loss": 0.211,
      "step": 11296
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.796129418412253,
      "learning_rate": 1.1434888306939082e-05,
      "loss": 0.179,
      "step": 11297
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4869794107733427,
      "learning_rate": 1.1430721800235029e-05,
      "loss": 0.111,
      "step": 11298
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6211755115646705,
      "learning_rate": 1.142655582776318e-05,
      "loss": 0.1387,
      "step": 11299
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7050346838650892,
      "learning_rate": 1.1422390389687557e-05,
      "loss": 0.0747,
      "step": 11300
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1753870736989322,
      "learning_rate": 1.1418225486172165e-05,
      "loss": 0.0866,
      "step": 11301
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7134879339508421,
      "learning_rate": 1.1414061117380962e-05,
      "loss": 0.2123,
      "step": 11302
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6108019922291847,
      "learning_rate": 1.1409897283477905e-05,
      "loss": 0.1399,
      "step": 11303
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3771378494991655,
      "learning_rate": 1.1405733984626934e-05,
      "loss": 0.0649,
      "step": 11304
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4975266298351954,
      "learning_rate": 1.1401571220991968e-05,
      "loss": 0.1076,
      "step": 11305
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8925367904724681,
      "learning_rate": 1.139740899273688e-05,
      "loss": 0.2919,
      "step": 11306
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3664164349389776,
      "learning_rate": 1.139324730002555e-05,
      "loss": 0.0636,
      "step": 11307
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5780134909575869,
      "learning_rate": 1.1389086143021824e-05,
      "loss": 0.0715,
      "step": 11308
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8733426976105506,
      "learning_rate": 1.1384925521889539e-05,
      "loss": 0.336,
      "step": 11309
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.39131372741450715,
      "learning_rate": 1.1380765436792484e-05,
      "loss": 0.0633,
      "step": 11310
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6295604730385104,
      "learning_rate": 1.1376605887894456e-05,
      "loss": 0.1692,
      "step": 11311
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.235070767709379,
      "learning_rate": 1.1372446875359219e-05,
      "loss": 0.301,
      "step": 11312
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0447449205459496,
      "learning_rate": 1.1368288399350522e-05,
      "loss": 0.0935,
      "step": 11313
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9386214788475169,
      "learning_rate": 1.1364130460032071e-05,
      "loss": 0.4899,
      "step": 11314
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5433824622284787,
      "learning_rate": 1.1359973057567577e-05,
      "loss": 0.1354,
      "step": 11315
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5259793993977725,
      "learning_rate": 1.135581619212072e-05,
      "loss": 0.0707,
      "step": 11316
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7857375670074235,
      "learning_rate": 1.1351659863855165e-05,
      "loss": 0.3381,
      "step": 11317
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5026799229651898,
      "learning_rate": 1.1347504072934536e-05,
      "loss": 0.144,
      "step": 11318
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0825749139421619,
      "learning_rate": 1.1343348819522454e-05,
      "loss": 0.329,
      "step": 11319
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.483313725197331,
      "learning_rate": 1.1339194103782527e-05,
      "loss": 0.1345,
      "step": 11320
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.44048578295082713,
      "learning_rate": 1.1335039925878307e-05,
      "loss": 0.0816,
      "step": 11321
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9120685781918604,
      "learning_rate": 1.133088628597336e-05,
      "loss": 0.2187,
      "step": 11322
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5491439346162549,
      "learning_rate": 1.1326733184231218e-05,
      "loss": 0.1433,
      "step": 11323
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.960656936558948,
      "learning_rate": 1.1322580620815398e-05,
      "loss": 0.2681,
      "step": 11324
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5395849143664708,
      "learning_rate": 1.1318428595889372e-05,
      "loss": 0.126,
      "step": 11325
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5947779263733779,
      "learning_rate": 1.1314277109616626e-05,
      "loss": 0.1262,
      "step": 11326
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8834573935894976,
      "learning_rate": 1.1310126162160578e-05,
      "loss": 0.2181,
      "step": 11327
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5740239805713293,
      "learning_rate": 1.1305975753684695e-05,
      "loss": 0.0891,
      "step": 11328
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8044815982863776,
      "learning_rate": 1.1301825884352352e-05,
      "loss": 0.2613,
      "step": 11329
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6344907569608982,
      "learning_rate": 1.1297676554326948e-05,
      "loss": 0.1083,
      "step": 11330
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3955921438277493,
      "learning_rate": 1.1293527763771821e-05,
      "loss": 0.0843,
      "step": 11331
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.47167479843022,
      "learning_rate": 1.1289379512850348e-05,
      "loss": 0.0625,
      "step": 11332
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6508003615041764,
      "learning_rate": 1.128523180172582e-05,
      "loss": 0.1181,
      "step": 11333
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7285112519068828,
      "learning_rate": 1.1281084630561554e-05,
      "loss": 0.1145,
      "step": 11334
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6524152640196698,
      "learning_rate": 1.12769379995208e-05,
      "loss": 0.2091,
      "step": 11335
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4735024593402887,
      "learning_rate": 1.127279190876685e-05,
      "loss": 0.0525,
      "step": 11336
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5433017082284091,
      "learning_rate": 1.1268646358462912e-05,
      "loss": 0.0556,
      "step": 11337
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7527541611327787,
      "learning_rate": 1.1264501348772216e-05,
      "loss": 0.2232,
      "step": 11338
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4679091523239447,
      "learning_rate": 1.1260356879857928e-05,
      "loss": 0.1111,
      "step": 11339
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4500411230252385,
      "learning_rate": 1.1256212951883255e-05,
      "loss": 0.0959,
      "step": 11340
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3703199334595378,
      "learning_rate": 1.1252069565011317e-05,
      "loss": 0.0553,
      "step": 11341
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4640816529396409,
      "learning_rate": 1.1247926719405264e-05,
      "loss": 0.0546,
      "step": 11342
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5067979600245682,
      "learning_rate": 1.1243784415228173e-05,
      "loss": 0.1022,
      "step": 11343
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7859491466535586,
      "learning_rate": 1.1239642652643168e-05,
      "loss": 0.145,
      "step": 11344
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.533886809325034,
      "learning_rate": 1.123550143181328e-05,
      "loss": 0.1332,
      "step": 11345
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6737984066263454,
      "learning_rate": 1.1231360752901579e-05,
      "loss": 0.1782,
      "step": 11346
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6251126093171475,
      "learning_rate": 1.1227220616071055e-05,
      "loss": 0.1238,
      "step": 11347
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7104071583758531,
      "learning_rate": 1.122308102148474e-05,
      "loss": 0.2239,
      "step": 11348
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.36865314005879635,
      "learning_rate": 1.1218941969305593e-05,
      "loss": 0.0505,
      "step": 11349
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7968899525852156,
      "learning_rate": 1.1214803459696584e-05,
      "loss": 0.1656,
      "step": 11350
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1759098243820476,
      "learning_rate": 1.1210665492820635e-05,
      "loss": 0.2091,
      "step": 11351
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5482099303152368,
      "learning_rate": 1.1206528068840667e-05,
      "loss": 0.1223,
      "step": 11352
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5442942120388953,
      "learning_rate": 1.1202391187919575e-05,
      "loss": 0.0715,
      "step": 11353
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6823059373068504,
      "learning_rate": 1.1198254850220239e-05,
      "loss": 0.1262,
      "step": 11354
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5641118268318488,
      "learning_rate": 1.1194119055905491e-05,
      "loss": 0.0652,
      "step": 11355
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.49666980929398213,
      "learning_rate": 1.118998380513817e-05,
      "loss": 0.0966,
      "step": 11356
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7560224323653294,
      "learning_rate": 1.1185849098081097e-05,
      "loss": 0.2434,
      "step": 11357
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5977801152019389,
      "learning_rate": 1.1181714934897034e-05,
      "loss": 0.1659,
      "step": 11358
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.824361149675473,
      "learning_rate": 1.1177581315748756e-05,
      "loss": 0.2461,
      "step": 11359
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6850899173460084,
      "learning_rate": 1.1173448240799005e-05,
      "loss": 0.1086,
      "step": 11360
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5719606628438301,
      "learning_rate": 1.116931571021052e-05,
      "loss": 0.1083,
      "step": 11361
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6346991201569105,
      "learning_rate": 1.1165183724145978e-05,
      "loss": 0.1214,
      "step": 11362
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0162211055049157,
      "learning_rate": 1.1161052282768064e-05,
      "loss": 0.2951,
      "step": 11363
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5266175049042582,
      "learning_rate": 1.1156921386239444e-05,
      "loss": 0.1777,
      "step": 11364
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5034079977883429,
      "learning_rate": 1.1152791034722757e-05,
      "loss": 0.0969,
      "step": 11365
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7331322341126485,
      "learning_rate": 1.1148661228380603e-05,
      "loss": 0.1332,
      "step": 11366
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.024552050146593,
      "learning_rate": 1.1144531967375581e-05,
      "loss": 0.3019,
      "step": 11367
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9325323288439483,
      "learning_rate": 1.1140403251870266e-05,
      "loss": 0.233,
      "step": 11368
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6725557563584559,
      "learning_rate": 1.1136275082027216e-05,
      "loss": 0.1083,
      "step": 11369
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8393493540998018,
      "learning_rate": 1.1132147458008946e-05,
      "loss": 0.1734,
      "step": 11370
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3975820922480561,
      "learning_rate": 1.112802037997797e-05,
      "loss": 0.109,
      "step": 11371
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7604446674309332,
      "learning_rate": 1.112389384809677e-05,
      "loss": 0.1535,
      "step": 11372
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.986131778325963,
      "learning_rate": 1.1119767862527822e-05,
      "loss": 0.3192,
      "step": 11373
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.44586282775198316,
      "learning_rate": 1.1115642423433557e-05,
      "loss": 0.0735,
      "step": 11374
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4896104092293448,
      "learning_rate": 1.1111517530976394e-05,
      "loss": 0.0902,
      "step": 11375
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5890873107179058,
      "learning_rate": 1.1107393185318743e-05,
      "loss": 0.1241,
      "step": 11376
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6146870586077525,
      "learning_rate": 1.1103269386622989e-05,
      "loss": 0.0924,
      "step": 11377
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0462035118893593,
      "learning_rate": 1.1099146135051466e-05,
      "loss": 0.2388,
      "step": 11378
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6443538096931279,
      "learning_rate": 1.1095023430766524e-05,
      "loss": 0.1468,
      "step": 11379
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9652759126502116,
      "learning_rate": 1.1090901273930474e-05,
      "loss": 0.2061,
      "step": 11380
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3203052089616915,
      "learning_rate": 1.108677966470562e-05,
      "loss": 0.0376,
      "step": 11381
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.49166893514649757,
      "learning_rate": 1.1082658603254212e-05,
      "loss": 0.111,
      "step": 11382
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5336085760594519,
      "learning_rate": 1.1078538089738507e-05,
      "loss": 0.1287,
      "step": 11383
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.44629326036913325,
      "learning_rate": 1.1074418124320738e-05,
      "loss": 0.094,
      "step": 11384
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5896407594168469,
      "learning_rate": 1.1070298707163116e-05,
      "loss": 0.1084,
      "step": 11385
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6338685506973654,
      "learning_rate": 1.1066179838427806e-05,
      "loss": 0.1201,
      "step": 11386
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8670929711957116,
      "learning_rate": 1.106206151827698e-05,
      "loss": 0.1583,
      "step": 11387
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7262098557235103,
      "learning_rate": 1.1057943746872796e-05,
      "loss": 0.1694,
      "step": 11388
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.47518913109943023,
      "learning_rate": 1.1053826524377346e-05,
      "loss": 0.1299,
      "step": 11389
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.45353295443926117,
      "learning_rate": 1.1049709850952738e-05,
      "loss": 0.0616,
      "step": 11390
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.37320693706749164,
      "learning_rate": 1.1045593726761056e-05,
      "loss": 0.0605,
      "step": 11391
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6298601288864524,
      "learning_rate": 1.1041478151964354e-05,
      "loss": 0.1117,
      "step": 11392
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3826694930197804,
      "learning_rate": 1.1037363126724656e-05,
      "loss": 0.0679,
      "step": 11393
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.49269512499101964,
      "learning_rate": 1.1033248651203986e-05,
      "loss": 0.0936,
      "step": 11394
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4327639707991523,
      "learning_rate": 1.1029134725564308e-05,
      "loss": 0.0682,
      "step": 11395
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.642738033380266,
      "learning_rate": 1.1025021349967626e-05,
      "loss": 0.2177,
      "step": 11396
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4509909668902511,
      "learning_rate": 1.102090852457586e-05,
      "loss": 0.0817,
      "step": 11397
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4810746854958185,
      "learning_rate": 1.1016796249550953e-05,
      "loss": 0.077,
      "step": 11398
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5103886644956174,
      "learning_rate": 1.101268452505478e-05,
      "loss": 0.0591,
      "step": 11399
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6880757129924968,
      "learning_rate": 1.1008573351249264e-05,
      "loss": 0.1554,
      "step": 11400
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6511080496698438,
      "learning_rate": 1.100446272829623e-05,
      "loss": 0.116,
      "step": 11401
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7724568755204677,
      "learning_rate": 1.100035265635753e-05,
      "loss": 0.1021,
      "step": 11402
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6204102690122879,
      "learning_rate": 1.0996243135594978e-05,
      "loss": 0.0575,
      "step": 11403
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6528378486469275,
      "learning_rate": 1.0992134166170379e-05,
      "loss": 0.2583,
      "step": 11404
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5017693025200987,
      "learning_rate": 1.0988025748245487e-05,
      "loss": 0.1048,
      "step": 11405
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7148137610742223,
      "learning_rate": 1.0983917881982063e-05,
      "loss": 0.1927,
      "step": 11406
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8293462092525029,
      "learning_rate": 1.097981056754184e-05,
      "loss": 0.1279,
      "step": 11407
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5353865804228561,
      "learning_rate": 1.097570380508653e-05,
      "loss": 0.0728,
      "step": 11408
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6382008636672608,
      "learning_rate": 1.0971597594777803e-05,
      "loss": 0.126,
      "step": 11409
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5844390476175512,
      "learning_rate": 1.0967491936777333e-05,
      "loss": 0.0938,
      "step": 11410
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.38350211638314585,
      "learning_rate": 1.0963386831246761e-05,
      "loss": 0.0581,
      "step": 11411
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6938328537473221,
      "learning_rate": 1.0959282278347718e-05,
      "loss": 0.1928,
      "step": 11412
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.723810011653978,
      "learning_rate": 1.0955178278241788e-05,
      "loss": 0.1809,
      "step": 11413
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5510346231069521,
      "learning_rate": 1.0951074831090552e-05,
      "loss": 0.1703,
      "step": 11414
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3630800445701196,
      "learning_rate": 1.094697193705557e-05,
      "loss": 0.0691,
      "step": 11415
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7208241716086238,
      "learning_rate": 1.0942869596298383e-05,
      "loss": 0.205,
      "step": 11416
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5864380671647538,
      "learning_rate": 1.0938767808980486e-05,
      "loss": 0.0685,
      "step": 11417
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.550979229494491,
      "learning_rate": 1.0934666575263374e-05,
      "loss": 0.0801,
      "step": 11418
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6081478767278383,
      "learning_rate": 1.093056589530853e-05,
      "loss": 0.1356,
      "step": 11419
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6044859461051778,
      "learning_rate": 1.0926465769277383e-05,
      "loss": 0.1227,
      "step": 11420
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5379610172786987,
      "learning_rate": 1.0922366197331366e-05,
      "loss": 0.1195,
      "step": 11421
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6011610759376812,
      "learning_rate": 1.0918267179631878e-05,
      "loss": 0.0765,
      "step": 11422
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4848309404715398,
      "learning_rate": 1.0914168716340312e-05,
      "loss": 0.0881,
      "step": 11423
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.012232553392388,
      "learning_rate": 1.0910070807618011e-05,
      "loss": 0.4605,
      "step": 11424
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5245162108709955,
      "learning_rate": 1.0905973453626328e-05,
      "loss": 0.1452,
      "step": 11425
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6305562849025855,
      "learning_rate": 1.0901876654526555e-05,
      "loss": 0.1006,
      "step": 11426
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0569534464909738,
      "learning_rate": 1.0897780410480018e-05,
      "loss": 0.2767,
      "step": 11427
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5344545047082091,
      "learning_rate": 1.0893684721647964e-05,
      "loss": 0.0674,
      "step": 11428
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.48482921829738795,
      "learning_rate": 1.0889589588191662e-05,
      "loss": 0.0561,
      "step": 11429
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4441417433233801,
      "learning_rate": 1.0885495010272311e-05,
      "loss": 0.095,
      "step": 11430
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.39119919424785715,
      "learning_rate": 1.0881400988051155e-05,
      "loss": 0.0534,
      "step": 11431
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5209494533445089,
      "learning_rate": 1.0877307521689353e-05,
      "loss": 0.0666,
      "step": 11432
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1838986360148465,
      "learning_rate": 1.0873214611348084e-05,
      "loss": 0.4239,
      "step": 11433
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.659211945055778,
      "learning_rate": 1.0869122257188464e-05,
      "loss": 0.1603,
      "step": 11434
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6563111704064811,
      "learning_rate": 1.0865030459371641e-05,
      "loss": 0.1471,
      "step": 11435
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9506508519088762,
      "learning_rate": 1.0860939218058695e-05,
      "loss": 0.0929,
      "step": 11436
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8212013834961112,
      "learning_rate": 1.085684853341071e-05,
      "loss": 0.3857,
      "step": 11437
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3606398696326855,
      "learning_rate": 1.0852758405588717e-05,
      "loss": 0.042,
      "step": 11438
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3480421610215697,
      "learning_rate": 1.0848668834753786e-05,
      "loss": 0.075,
      "step": 11439
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.49046672850155815,
      "learning_rate": 1.0844579821066892e-05,
      "loss": 0.0764,
      "step": 11440
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6050917841764368,
      "learning_rate": 1.0840491364689045e-05,
      "loss": 0.0909,
      "step": 11441
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6403493270128308,
      "learning_rate": 1.0836403465781187e-05,
      "loss": 0.0885,
      "step": 11442
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5469826666687886,
      "learning_rate": 1.0832316124504289e-05,
      "loss": 0.0836,
      "step": 11443
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7758120160052617,
      "learning_rate": 1.082822934101925e-05,
      "loss": 0.1305,
      "step": 11444
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2144208377981238,
      "learning_rate": 1.0824143115486987e-05,
      "loss": 0.3365,
      "step": 11445
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4235408265432598,
      "learning_rate": 1.082005744806836e-05,
      "loss": 0.0506,
      "step": 11446
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4766612833887277,
      "learning_rate": 1.0815972338924235e-05,
      "loss": 0.0734,
      "step": 11447
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.35380323313306533,
      "learning_rate": 1.0811887788215444e-05,
      "loss": 0.0635,
      "step": 11448
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4478770758488957,
      "learning_rate": 1.0807803796102808e-05,
      "loss": 0.0686,
      "step": 11449
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.41789250971209035,
      "learning_rate": 1.0803720362747099e-05,
      "loss": 0.0493,
      "step": 11450
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.36392986396584615,
      "learning_rate": 1.0799637488309094e-05,
      "loss": 0.0497,
      "step": 11451
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7362716067701066,
      "learning_rate": 1.0795555172949538e-05,
      "loss": 0.2177,
      "step": 11452
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.47248123111183066,
      "learning_rate": 1.0791473416829161e-05,
      "loss": 0.0634,
      "step": 11453
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.49460989718544357,
      "learning_rate": 1.0787392220108653e-05,
      "loss": 0.0867,
      "step": 11454
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.40630147608946,
      "learning_rate": 1.0783311582948699e-05,
      "loss": 0.0511,
      "step": 11455
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3318205242496329,
      "learning_rate": 1.0779231505509962e-05,
      "loss": 0.0314,
      "step": 11456
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5209030416110433,
      "learning_rate": 1.0775151987953067e-05,
      "loss": 0.0756,
      "step": 11457
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.663580771951896,
      "learning_rate": 1.0771073030438633e-05,
      "loss": 0.2299,
      "step": 11458
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3787431649734712,
      "learning_rate": 1.0766994633127247e-05,
      "loss": 0.0522,
      "step": 11459
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3750385582637286,
      "learning_rate": 1.0762916796179495e-05,
      "loss": 0.0449,
      "step": 11460
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5191738722909417,
      "learning_rate": 1.0758839519755902e-05,
      "loss": 0.0912,
      "step": 11461
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5600128904592587,
      "learning_rate": 1.0754762804017008e-05,
      "loss": 0.1188,
      "step": 11462
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4679736055884731,
      "learning_rate": 1.0750686649123306e-05,
      "loss": 0.0795,
      "step": 11463
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.39124438183758414,
      "learning_rate": 1.0746611055235293e-05,
      "loss": 0.0586,
      "step": 11464
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4730039132582052,
      "learning_rate": 1.074253602251341e-05,
      "loss": 0.0766,
      "step": 11465
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5791664395887133,
      "learning_rate": 1.0738461551118102e-05,
      "loss": 0.1266,
      "step": 11466
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3763515805278245,
      "learning_rate": 1.0734387641209782e-05,
      "loss": 0.0582,
      "step": 11467
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.34247669494609745,
      "learning_rate": 1.0730314292948853e-05,
      "loss": 0.0517,
      "step": 11468
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3376078452573369,
      "learning_rate": 1.0726241506495668e-05,
      "loss": 0.0621,
      "step": 11469
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4996634062906813,
      "learning_rate": 1.0722169282010586e-05,
      "loss": 0.0629,
      "step": 11470
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6571007724318422,
      "learning_rate": 1.0718097619653932e-05,
      "loss": 0.0991,
      "step": 11471
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.38943908952625983,
      "learning_rate": 1.0714026519586018e-05,
      "loss": 0.0683,
      "step": 11472
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5230632358561306,
      "learning_rate": 1.070995598196711e-05,
      "loss": 0.0473,
      "step": 11473
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4100881450931714,
      "learning_rate": 1.0705886006957478e-05,
      "loss": 0.0711,
      "step": 11474
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6308494064008976,
      "learning_rate": 1.0701816594717353e-05,
      "loss": 0.143,
      "step": 11475
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.38144690196410896,
      "learning_rate": 1.0697747745406969e-05,
      "loss": 0.0366,
      "step": 11476
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5023796062213156,
      "learning_rate": 1.0693679459186498e-05,
      "loss": 0.1253,
      "step": 11477
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4700843995193994,
      "learning_rate": 1.0689611736216118e-05,
      "loss": 0.092,
      "step": 11478
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.42937282554069817,
      "learning_rate": 1.0685544576655978e-05,
      "loss": 0.0666,
      "step": 11479
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4393536372724232,
      "learning_rate": 1.0681477980666218e-05,
      "loss": 0.0889,
      "step": 11480
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4144397240582186,
      "learning_rate": 1.0677411948406923e-05,
      "loss": 0.0496,
      "step": 11481
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6676445468454054,
      "learning_rate": 1.0673346480038185e-05,
      "loss": 0.0525,
      "step": 11482
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.48961016388808937,
      "learning_rate": 1.0669281575720072e-05,
      "loss": 0.0812,
      "step": 11483
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5748083624064773,
      "learning_rate": 1.0665217235612605e-05,
      "loss": 0.0811,
      "step": 11484
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3851698751796496,
      "learning_rate": 1.0661153459875805e-05,
      "loss": 0.037,
      "step": 11485
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.31103408138641614,
      "learning_rate": 1.0657090248669674e-05,
      "loss": 0.0566,
      "step": 11486
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6091060867159925,
      "learning_rate": 1.0653027602154186e-05,
      "loss": 0.1036,
      "step": 11487
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6201684962847465,
      "learning_rate": 1.0648965520489276e-05,
      "loss": 0.2171,
      "step": 11488
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6434866182524639,
      "learning_rate": 1.0644904003834877e-05,
      "loss": 0.1435,
      "step": 11489
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5867530841073444,
      "learning_rate": 1.0640843052350896e-05,
      "loss": 0.1275,
      "step": 11490
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5148524872937527,
      "learning_rate": 1.0636782666197222e-05,
      "loss": 0.1315,
      "step": 11491
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6061010615223421,
      "learning_rate": 1.0632722845533702e-05,
      "loss": 0.1733,
      "step": 11492
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.629682434379588,
      "learning_rate": 1.0628663590520187e-05,
      "loss": 0.159,
      "step": 11493
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6099649183280189,
      "learning_rate": 1.062460490131647e-05,
      "loss": 0.1166,
      "step": 11494
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.554879847669445,
      "learning_rate": 1.062054677808238e-05,
      "loss": 0.1521,
      "step": 11495
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7249152454111469,
      "learning_rate": 1.0616489220977654e-05,
      "loss": 0.3542,
      "step": 11496
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.771031814431309,
      "learning_rate": 1.061243223016207e-05,
      "loss": 0.2103,
      "step": 11497
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6243339244023817,
      "learning_rate": 1.0608375805795321e-05,
      "loss": 0.1129,
      "step": 11498
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6491596896519679,
      "learning_rate": 1.0604319948037147e-05,
      "loss": 0.133,
      "step": 11499
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.44996249323705184,
      "learning_rate": 1.0600264657047207e-05,
      "loss": 0.1372,
      "step": 11500
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.30313173062615517,
      "learning_rate": 1.0596209932985173e-05,
      "loss": 0.0444,
      "step": 11501
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.46428758761906463,
      "learning_rate": 1.0592155776010663e-05,
      "loss": 0.0814,
      "step": 11502
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5750121065258292,
      "learning_rate": 1.058810218628332e-05,
      "loss": 0.0788,
      "step": 11503
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5628224170668923,
      "learning_rate": 1.0584049163962713e-05,
      "loss": 0.1063,
      "step": 11504
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.47641441751255786,
      "learning_rate": 1.0579996709208436e-05,
      "loss": 0.0949,
      "step": 11505
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7605783610902929,
      "learning_rate": 1.0575944822180001e-05,
      "loss": 0.2284,
      "step": 11506
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.482284643053446,
      "learning_rate": 1.0571893503036976e-05,
      "loss": 0.0863,
      "step": 11507
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4086604245679872,
      "learning_rate": 1.0567842751938834e-05,
      "loss": 0.0851,
      "step": 11508
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0302393863269743,
      "learning_rate": 1.0563792569045076e-05,
      "loss": 0.3673,
      "step": 11509
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7452489861069086,
      "learning_rate": 1.0559742954515131e-05,
      "loss": 0.1968,
      "step": 11510
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4171505985896898,
      "learning_rate": 1.0555693908508475e-05,
      "loss": 0.0874,
      "step": 11511
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5356771293201718,
      "learning_rate": 1.055164543118449e-05,
      "loss": 0.1203,
      "step": 11512
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6792175237754599,
      "learning_rate": 1.0547597522702587e-05,
      "loss": 0.0939,
      "step": 11513
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6135007275419176,
      "learning_rate": 1.054355018322212e-05,
      "loss": 0.0891,
      "step": 11514
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.015899758538014,
      "learning_rate": 1.0539503412902444e-05,
      "loss": 0.2519,
      "step": 11515
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3511907727995985,
      "learning_rate": 1.0535457211902883e-05,
      "loss": 0.0412,
      "step": 11516
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6250560276556413,
      "learning_rate": 1.0531411580382744e-05,
      "loss": 0.1336,
      "step": 11517
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5887381342572171,
      "learning_rate": 1.0527366518501294e-05,
      "loss": 0.0899,
      "step": 11518
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.444164436193768,
      "learning_rate": 1.0523322026417795e-05,
      "loss": 0.0914,
      "step": 11519
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5035686205700426,
      "learning_rate": 1.0519278104291483e-05,
      "loss": 0.1118,
      "step": 11520
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8773021956652872,
      "learning_rate": 1.0515234752281577e-05,
      "loss": 0.1289,
      "step": 11521
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.43011547046596527,
      "learning_rate": 1.0511191970547255e-05,
      "loss": 0.0964,
      "step": 11522
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.45028604096629987,
      "learning_rate": 1.0507149759247683e-05,
      "loss": 0.0945,
      "step": 11523
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5216396439895256,
      "learning_rate": 1.0503108118542027e-05,
      "loss": 0.0976,
      "step": 11524
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9924039101646552,
      "learning_rate": 1.0499067048589382e-05,
      "loss": 0.3986,
      "step": 11525
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.41088339493125453,
      "learning_rate": 1.0495026549548859e-05,
      "loss": 0.0965,
      "step": 11526
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7188259252395062,
      "learning_rate": 1.0490986621579538e-05,
      "loss": 0.2341,
      "step": 11527
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7084024778686222,
      "learning_rate": 1.0486947264840477e-05,
      "loss": 0.2696,
      "step": 11528
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4224867729243862,
      "learning_rate": 1.0482908479490699e-05,
      "loss": 0.0712,
      "step": 11529
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0069917815396998,
      "learning_rate": 1.0478870265689216e-05,
      "loss": 0.2167,
      "step": 11530
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5306083106489224,
      "learning_rate": 1.0474832623595018e-05,
      "loss": 0.0846,
      "step": 11531
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5212990219326832,
      "learning_rate": 1.0470795553367078e-05,
      "loss": 0.0769,
      "step": 11532
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6838055969144345,
      "learning_rate": 1.046675905516432e-05,
      "loss": 0.1459,
      "step": 11533
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6184457136251611,
      "learning_rate": 1.0462723129145677e-05,
      "loss": 0.1189,
      "step": 11534
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6442632909053436,
      "learning_rate": 1.0458687775470039e-05,
      "loss": 0.1989,
      "step": 11535
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6243740524703814,
      "learning_rate": 1.0454652994296297e-05,
      "loss": 0.5355,
      "step": 11536
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5036244426586635,
      "learning_rate": 1.045061878578327e-05,
      "loss": 0.0825,
      "step": 11537
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.41685508526237447,
      "learning_rate": 1.0446585150089827e-05,
      "loss": 0.0589,
      "step": 11538
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7157060932351884,
      "learning_rate": 1.044255208737475e-05,
      "loss": 0.1295,
      "step": 11539
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4942742306832014,
      "learning_rate": 1.0438519597796837e-05,
      "loss": 0.0825,
      "step": 11540
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3885933509356474,
      "learning_rate": 1.0434487681514826e-05,
      "loss": 0.1126,
      "step": 11541
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.43063677492404495,
      "learning_rate": 1.0430456338687492e-05,
      "loss": 0.0513,
      "step": 11542
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5371094371047287,
      "learning_rate": 1.0426425569473524e-05,
      "loss": 0.1745,
      "step": 11543
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8043979566843377,
      "learning_rate": 1.0422395374031635e-05,
      "loss": 0.1693,
      "step": 11544
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6630319413134819,
      "learning_rate": 1.0418365752520478e-05,
      "loss": 0.14,
      "step": 11545
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.605456224235328,
      "learning_rate": 1.0414336705098712e-05,
      "loss": 0.1237,
      "step": 11546
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7500207349536345,
      "learning_rate": 1.0410308231924964e-05,
      "loss": 0.1511,
      "step": 11547
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5259031234585682,
      "learning_rate": 1.0406280333157844e-05,
      "loss": 0.1035,
      "step": 11548
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4578868983669972,
      "learning_rate": 1.040225300895592e-05,
      "loss": 0.0946,
      "step": 11549
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6177731693323893,
      "learning_rate": 1.0398226259477757e-05,
      "loss": 0.0717,
      "step": 11550
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.547881208430924,
      "learning_rate": 1.0394200084881898e-05,
      "loss": 0.0879,
      "step": 11551
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5110236858652464,
      "learning_rate": 1.039017448532684e-05,
      "loss": 0.0756,
      "step": 11552
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5300243111133199,
      "learning_rate": 1.038614946097108e-05,
      "loss": 0.1178,
      "step": 11553
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5286571870713238,
      "learning_rate": 1.0382125011973095e-05,
      "loss": 0.0802,
      "step": 11554
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6049630474472735,
      "learning_rate": 1.0378101138491331e-05,
      "loss": 0.1443,
      "step": 11555
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.50877601640961,
      "learning_rate": 1.0374077840684196e-05,
      "loss": 0.1061,
      "step": 11556
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4663278622432872,
      "learning_rate": 1.0370055118710098e-05,
      "loss": 0.0559,
      "step": 11557
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3625702681089236,
      "learning_rate": 1.0366032972727415e-05,
      "loss": 0.0612,
      "step": 11558
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9781070890325655,
      "learning_rate": 1.0362011402894511e-05,
      "loss": 0.2669,
      "step": 11559
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5246514330274981,
      "learning_rate": 1.03579904093697e-05,
      "loss": 0.0741,
      "step": 11560
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2054943967477265,
      "learning_rate": 1.03539699923113e-05,
      "loss": 0.3224,
      "step": 11561
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8926690608637182,
      "learning_rate": 1.0349950151877596e-05,
      "loss": 0.2065,
      "step": 11562
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5600430034553401,
      "learning_rate": 1.0345930888226862e-05,
      "loss": 0.1191,
      "step": 11563
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3992789403113357,
      "learning_rate": 1.0341912201517325e-05,
      "loss": 0.032,
      "step": 11564
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7892460703255761,
      "learning_rate": 1.033789409190721e-05,
      "loss": 0.2375,
      "step": 11565
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7147338800473731,
      "learning_rate": 1.0333876559554712e-05,
      "loss": 0.1729,
      "step": 11566
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.40013639686260594,
      "learning_rate": 1.0329859604618012e-05,
      "loss": 0.0388,
      "step": 11567
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5094819704537147,
      "learning_rate": 1.0325843227255245e-05,
      "loss": 0.0923,
      "step": 11568
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5632172536454629,
      "learning_rate": 1.0321827427624547e-05,
      "loss": 0.1017,
      "step": 11569
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5956361021396184,
      "learning_rate": 1.0317812205884022e-05,
      "loss": 0.0818,
      "step": 11570
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8523423181888087,
      "learning_rate": 1.0313797562191762e-05,
      "loss": 0.3145,
      "step": 11571
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9797602464306225,
      "learning_rate": 1.030978349670581e-05,
      "loss": 0.4427,
      "step": 11572
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.104906660538714,
      "learning_rate": 1.0305770009584208e-05,
      "loss": 0.3412,
      "step": 11573
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6554419658127947,
      "learning_rate": 1.0301757100984973e-05,
      "loss": 0.1834,
      "step": 11574
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.31938362343224846,
      "learning_rate": 1.0297744771066103e-05,
      "loss": 0.055,
      "step": 11575
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3399594357744411,
      "learning_rate": 1.0293733019985549e-05,
      "loss": 0.0432,
      "step": 11576
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4092238152263718,
      "learning_rate": 1.0289721847901266e-05,
      "loss": 0.0669,
      "step": 11577
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.150704925156684,
      "learning_rate": 1.0285711254971175e-05,
      "loss": 0.5135,
      "step": 11578
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5884656205133243,
      "learning_rate": 1.0281701241353186e-05,
      "loss": 0.1048,
      "step": 11579
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.40428382209040314,
      "learning_rate": 1.0277691807205162e-05,
      "loss": 0.0608,
      "step": 11580
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5855561550194139,
      "learning_rate": 1.027368295268496e-05,
      "loss": 0.2136,
      "step": 11581
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7125707947832218,
      "learning_rate": 1.0269674677950422e-05,
      "loss": 0.1879,
      "step": 11582
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2676952916510194,
      "learning_rate": 1.0265666983159342e-05,
      "loss": 0.1732,
      "step": 11583
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5065849517353288,
      "learning_rate": 1.0261659868469511e-05,
      "loss": 0.0798,
      "step": 11584
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8877652814553073,
      "learning_rate": 1.0257653334038695e-05,
      "loss": 0.2894,
      "step": 11585
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4300602615650257,
      "learning_rate": 1.0253647380024642e-05,
      "loss": 0.0557,
      "step": 11586
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.703487405219923,
      "learning_rate": 1.024964200658505e-05,
      "loss": 0.1264,
      "step": 11587
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6157232160351187,
      "learning_rate": 1.0245637213877632e-05,
      "loss": 0.1127,
      "step": 11588
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.34178834140906034,
      "learning_rate": 1.0241633002060034e-05,
      "loss": 0.0371,
      "step": 11589
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9089030045492044,
      "learning_rate": 1.023762937128994e-05,
      "loss": 0.1777,
      "step": 11590
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6405860444921682,
      "learning_rate": 1.0233626321724946e-05,
      "loss": 0.1511,
      "step": 11591
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5095530842420755,
      "learning_rate": 1.0229623853522677e-05,
      "loss": 0.0968,
      "step": 11592
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0082417667148726,
      "learning_rate": 1.0225621966840684e-05,
      "loss": 0.2994,
      "step": 11593
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4780937087070255,
      "learning_rate": 1.0221620661836561e-05,
      "loss": 0.0934,
      "step": 11594
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9010623895536332,
      "learning_rate": 1.0217619938667813e-05,
      "loss": 0.2435,
      "step": 11595
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6083310880075548,
      "learning_rate": 1.0213619797491974e-05,
      "loss": 0.1767,
      "step": 11596
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5639218182941188,
      "learning_rate": 1.0209620238466503e-05,
      "loss": 0.1562,
      "step": 11597
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.061129146331965,
      "learning_rate": 1.0205621261748898e-05,
      "loss": 0.2089,
      "step": 11598
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6121810096637869,
      "learning_rate": 1.020162286749658e-05,
      "loss": 0.1315,
      "step": 11599
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5076774941512747,
      "learning_rate": 1.0197625055866983e-05,
      "loss": 0.1129,
      "step": 11600
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6782614770084159,
      "learning_rate": 1.019362782701748e-05,
      "loss": 0.1482,
      "step": 11601
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2957268653634142,
      "learning_rate": 1.0189631181105478e-05,
      "loss": 0.0499,
      "step": 11602
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4380287739998252,
      "learning_rate": 1.01856351182883e-05,
      "loss": 0.0637,
      "step": 11603
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6006765852553694,
      "learning_rate": 1.0181639638723294e-05,
      "loss": 0.132,
      "step": 11604
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1829582345034806,
      "learning_rate": 1.0177644742567738e-05,
      "loss": 0.2248,
      "step": 11605
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4129542732728346,
      "learning_rate": 1.0173650429978951e-05,
      "loss": 0.0657,
      "step": 11606
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3872579685590807,
      "learning_rate": 1.0169656701114161e-05,
      "loss": 0.4374,
      "step": 11607
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7016326314264513,
      "learning_rate": 1.0165663556130625e-05,
      "loss": 0.1963,
      "step": 11608
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3490732516407152,
      "learning_rate": 1.0161670995185526e-05,
      "loss": 0.0742,
      "step": 11609
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6722389749431057,
      "learning_rate": 1.0157679018436094e-05,
      "loss": 0.126,
      "step": 11610
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5930079517234118,
      "learning_rate": 1.0153687626039468e-05,
      "loss": 0.0994,
      "step": 11611
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6094270073348865,
      "learning_rate": 1.014969681815281e-05,
      "loss": 0.1316,
      "step": 11612
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6628037780045015,
      "learning_rate": 1.0145706594933218e-05,
      "loss": 0.1541,
      "step": 11613
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5983040998910656,
      "learning_rate": 1.0141716956537806e-05,
      "loss": 0.0789,
      "step": 11614
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5951034810593494,
      "learning_rate": 1.0137727903123645e-05,
      "loss": 0.1062,
      "step": 11615
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6441027860548079,
      "learning_rate": 1.01337394348478e-05,
      "loss": 0.1671,
      "step": 11616
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7322300360592452,
      "learning_rate": 1.0129751551867275e-05,
      "loss": 0.1416,
      "step": 11617
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4559967859122199,
      "learning_rate": 1.012576425433909e-05,
      "loss": 0.0704,
      "step": 11618
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5970433490368562,
      "learning_rate": 1.0121777542420235e-05,
      "loss": 0.12,
      "step": 11619
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.368012942207112,
      "learning_rate": 1.0117791416267652e-05,
      "loss": 0.0647,
      "step": 11620
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5276126360377453,
      "learning_rate": 1.0113805876038284e-05,
      "loss": 0.1351,
      "step": 11621
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4920516507004958,
      "learning_rate": 1.0109820921889051e-05,
      "loss": 0.1085,
      "step": 11622
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.37852061712392815,
      "learning_rate": 1.0105836553976846e-05,
      "loss": 0.0399,
      "step": 11623
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4801850452536368,
      "learning_rate": 1.0101852772458526e-05,
      "loss": 0.0786,
      "step": 11624
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3550863347726831,
      "learning_rate": 1.0097869577490935e-05,
      "loss": 0.053,
      "step": 11625
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5438762578471259,
      "learning_rate": 1.0093886969230901e-05,
      "loss": 0.1745,
      "step": 11626
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6572961987274537,
      "learning_rate": 1.0089904947835227e-05,
      "loss": 0.1715,
      "step": 11627
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.35296053225772583,
      "learning_rate": 1.0085923513460672e-05,
      "loss": 0.0345,
      "step": 11628
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.44938032592813476,
      "learning_rate": 1.0081942666263996e-05,
      "loss": 0.0735,
      "step": 11629
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.539063441870109,
      "learning_rate": 1.007796240640193e-05,
      "loss": 0.0617,
      "step": 11630
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.45746979250431613,
      "learning_rate": 1.0073982734031188e-05,
      "loss": 0.0606,
      "step": 11631
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0526437013105372,
      "learning_rate": 1.0070003649308432e-05,
      "loss": 0.3481,
      "step": 11632
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5167694361047895,
      "learning_rate": 1.0066025152390334e-05,
      "loss": 0.0741,
      "step": 11633
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5723158550211661,
      "learning_rate": 1.0062047243433528e-05,
      "loss": 0.1106,
      "step": 11634
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.484800716614948,
      "learning_rate": 1.0058069922594637e-05,
      "loss": 0.0915,
      "step": 11635
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7185329632505344,
      "learning_rate": 1.0054093190030234e-05,
      "loss": 0.3999,
      "step": 11636
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8467505118844333,
      "learning_rate": 1.0050117045896889e-05,
      "loss": 0.2796,
      "step": 11637
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.45206324881042126,
      "learning_rate": 1.0046141490351155e-05,
      "loss": 0.1306,
      "step": 11638
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7816775379811959,
      "learning_rate": 1.0042166523549554e-05,
      "loss": 0.1567,
      "step": 11639
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6796505148788579,
      "learning_rate": 1.0038192145648567e-05,
      "loss": 0.1546,
      "step": 11640
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5721697446840229,
      "learning_rate": 1.0034218356804681e-05,
      "loss": 0.1366,
      "step": 11641
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4852776294507488,
      "learning_rate": 1.0030245157174342e-05,
      "loss": 0.0714,
      "step": 11642
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4445698136253254,
      "learning_rate": 1.002627254691399e-05,
      "loss": 0.1491,
      "step": 11643
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5413361029379472,
      "learning_rate": 1.0022300526180007e-05,
      "loss": 0.0736,
      "step": 11644
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7422008298332461,
      "learning_rate": 1.001832909512879e-05,
      "loss": 0.1488,
      "step": 11645
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3195213883460736,
      "learning_rate": 1.001435825391669e-05,
      "loss": 0.0394,
      "step": 11646
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5390840602159092,
      "learning_rate": 1.0010388002700059e-05,
      "loss": 0.0957,
      "step": 11647
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3267706321470001,
      "learning_rate": 1.0006418341635181e-05,
      "loss": 0.0479,
      "step": 11648
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.44971318424522627,
      "learning_rate": 1.0002449270878362e-05,
      "loss": 0.071,
      "step": 11649
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5421516335304192,
      "learning_rate": 9.998480790585872e-06,
      "loss": 0.1228,
      "step": 11650
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8641322925558632,
      "learning_rate": 9.994512900913933e-06,
      "loss": 0.1573,
      "step": 11651
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3873124629613077,
      "learning_rate": 9.990545602018776e-06,
      "loss": 0.0743,
      "step": 11652
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4705635923301483,
      "learning_rate": 9.986578894056598e-06,
      "loss": 0.0923,
      "step": 11653
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.45510286779574205,
      "learning_rate": 9.982612777183576e-06,
      "loss": 0.0714,
      "step": 11654
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6984427459278891,
      "learning_rate": 9.978647251555843e-06,
      "loss": 0.1396,
      "step": 11655
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3702514111236769,
      "learning_rate": 9.974682317329539e-06,
      "loss": 0.0582,
      "step": 11656
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.49272421908559505,
      "learning_rate": 9.970717974660745e-06,
      "loss": 0.0608,
      "step": 11657
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3401392962254281,
      "learning_rate": 9.966754223705574e-06,
      "loss": 0.0338,
      "step": 11658
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.47874589588901517,
      "learning_rate": 9.962791064620056e-06,
      "loss": 0.0559,
      "step": 11659
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9104698936603386,
      "learning_rate": 9.958828497560235e-06,
      "loss": 0.1491,
      "step": 11660
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.45215646155133105,
      "learning_rate": 9.954866522682103e-06,
      "loss": 0.0513,
      "step": 11661
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5642347732914352,
      "learning_rate": 9.950905140141669e-06,
      "loss": 0.1734,
      "step": 11662
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3295422901874612,
      "learning_rate": 9.946944350094884e-06,
      "loss": 0.3943,
      "step": 11663
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3729874924002069,
      "learning_rate": 9.942984152697693e-06,
      "loss": 0.0767,
      "step": 11664
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8217897252175799,
      "learning_rate": 9.939024548105987e-06,
      "loss": 0.2221,
      "step": 11665
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9627957803491781,
      "learning_rate": 9.935065536475698e-06,
      "loss": 0.1667,
      "step": 11666
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7913967385560655,
      "learning_rate": 9.931107117962668e-06,
      "loss": 0.1685,
      "step": 11667
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3213856182827189,
      "learning_rate": 9.927149292722751e-06,
      "loss": 0.0554,
      "step": 11668
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4207762023246603,
      "learning_rate": 9.923192060911767e-06,
      "loss": 0.0516,
      "step": 11669
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5169595546941325,
      "learning_rate": 9.919235422685525e-06,
      "loss": 0.0997,
      "step": 11670
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9134034124662035,
      "learning_rate": 9.915279378199785e-06,
      "loss": 0.3246,
      "step": 11671
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6247378140975903,
      "learning_rate": 9.911323927610305e-06,
      "loss": 0.0973,
      "step": 11672
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.46531254651307014,
      "learning_rate": 9.907369071072816e-06,
      "loss": 0.0622,
      "step": 11673
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.43000748474713546,
      "learning_rate": 9.903414808743034e-06,
      "loss": 0.0933,
      "step": 11674
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.42509325201230963,
      "learning_rate": 9.899461140776619e-06,
      "loss": 0.0604,
      "step": 11675
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6265430597363361,
      "learning_rate": 9.895508067329244e-06,
      "loss": 0.2126,
      "step": 11676
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4703272100966987,
      "learning_rate": 9.89155558855654e-06,
      "loss": 0.1027,
      "step": 11677
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5268854389794679,
      "learning_rate": 9.887603704614128e-06,
      "loss": 0.0832,
      "step": 11678
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.46724149652253527,
      "learning_rate": 9.883652415657582e-06,
      "loss": 0.092,
      "step": 11679
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7183913605709698,
      "learning_rate": 9.879701721842471e-06,
      "loss": 0.1431,
      "step": 11680
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3895478649469288,
      "learning_rate": 9.87575162332435e-06,
      "loss": 0.0556,
      "step": 11681
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7343556897554365,
      "learning_rate": 9.871802120258719e-06,
      "loss": 0.1239,
      "step": 11682
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7819382836072384,
      "learning_rate": 9.867853212801079e-06,
      "loss": 0.1387,
      "step": 11683
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.362667321713214,
      "learning_rate": 9.863904901106905e-06,
      "loss": 0.0314,
      "step": 11684
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6587338146375771,
      "learning_rate": 9.859957185331647e-06,
      "loss": 0.0886,
      "step": 11685
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4901267529453208,
      "learning_rate": 9.856010065630719e-06,
      "loss": 0.0874,
      "step": 11686
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0619744939764049,
      "learning_rate": 9.852063542159537e-06,
      "loss": 0.2681,
      "step": 11687
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.427237090690265,
      "learning_rate": 9.848117615073454e-06,
      "loss": 0.071,
      "step": 11688
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7814174886921571,
      "learning_rate": 9.844172284527854e-06,
      "loss": 0.2468,
      "step": 11689
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3385577558287541,
      "learning_rate": 9.840227550678044e-06,
      "loss": 0.0278,
      "step": 11690
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6465359539712976,
      "learning_rate": 9.836283413679351e-06,
      "loss": 0.1356,
      "step": 11691
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8346005840524127,
      "learning_rate": 9.83233987368703e-06,
      "loss": 0.2139,
      "step": 11692
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6280396584938045,
      "learning_rate": 9.828396930856373e-06,
      "loss": 0.1811,
      "step": 11693
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0172125551733924,
      "learning_rate": 9.824454585342596e-06,
      "loss": 0.1958,
      "step": 11694
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8412363851221698,
      "learning_rate": 9.820512837300925e-06,
      "loss": 0.2383,
      "step": 11695
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.492217412761934,
      "learning_rate": 9.816571686886524e-06,
      "loss": 0.0732,
      "step": 11696
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.43634511903347595,
      "learning_rate": 9.812631134254596e-06,
      "loss": 0.1091,
      "step": 11697
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5096894807478525,
      "learning_rate": 9.808691179560256e-06,
      "loss": 0.0983,
      "step": 11698
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6182025092262331,
      "learning_rate": 9.80475182295864e-06,
      "loss": 0.1415,
      "step": 11699
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.38648396815937247,
      "learning_rate": 9.800813064604819e-06,
      "loss": 0.0593,
      "step": 11700
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4894674524827658,
      "learning_rate": 9.796874904653896e-06,
      "loss": 0.542,
      "step": 11701
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6157906596901871,
      "learning_rate": 9.792937343260897e-06,
      "loss": 0.1274,
      "step": 11702
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4382465972726382,
      "learning_rate": 9.789000380580859e-06,
      "loss": 0.0991,
      "step": 11703
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8483208034951004,
      "learning_rate": 9.78506401676876e-06,
      "loss": 0.0982,
      "step": 11704
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6325156421847334,
      "learning_rate": 9.781128251979615e-06,
      "loss": 0.0904,
      "step": 11705
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.36554244095025984,
      "learning_rate": 9.777193086368349e-06,
      "loss": 0.0521,
      "step": 11706
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5496137971561093,
      "learning_rate": 9.773258520089909e-06,
      "loss": 0.0969,
      "step": 11707
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8587502403246161,
      "learning_rate": 9.769324553299176e-06,
      "loss": 0.2103,
      "step": 11708
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5091391120017069,
      "learning_rate": 9.765391186151065e-06,
      "loss": 0.0739,
      "step": 11709
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0855458699004672,
      "learning_rate": 9.761458418800418e-06,
      "loss": 0.3568,
      "step": 11710
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6336409698826733,
      "learning_rate": 9.75752625140208e-06,
      "loss": 0.0949,
      "step": 11711
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.46393658889323974,
      "learning_rate": 9.753594684110849e-06,
      "loss": 0.1073,
      "step": 11712
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9227639930281027,
      "learning_rate": 9.74966371708152e-06,
      "loss": 0.2252,
      "step": 11713
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.49038830086292856,
      "learning_rate": 9.745733350468863e-06,
      "loss": 0.0952,
      "step": 11714
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5066173707020714,
      "learning_rate": 9.741803584427623e-06,
      "loss": 0.0961,
      "step": 11715
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6575459069882711,
      "learning_rate": 9.737874419112503e-06,
      "loss": 0.1083,
      "step": 11716
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.45385886547898735,
      "learning_rate": 9.733945854678206e-06,
      "loss": 0.0607,
      "step": 11717
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4783489066257792,
      "learning_rate": 9.73001789127941e-06,
      "loss": 0.0653,
      "step": 11718
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4046828055121752,
      "learning_rate": 9.726090529070744e-06,
      "loss": 0.4663,
      "step": 11719
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5775831193608411,
      "learning_rate": 9.72216376820684e-06,
      "loss": 0.1844,
      "step": 11720
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4183750647686846,
      "learning_rate": 9.718237608842298e-06,
      "loss": 0.0958,
      "step": 11721
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7579531381722523,
      "learning_rate": 9.714312051131702e-06,
      "loss": 0.1731,
      "step": 11722
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4670937598786538,
      "learning_rate": 9.710387095229587e-06,
      "loss": 0.1495,
      "step": 11723
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.598353326492084,
      "learning_rate": 9.70646274129049e-06,
      "loss": 0.1498,
      "step": 11724
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.322285407899601,
      "learning_rate": 9.702538989468915e-06,
      "loss": 0.0448,
      "step": 11725
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.44577625099387025,
      "learning_rate": 9.698615839919353e-06,
      "loss": 0.0825,
      "step": 11726
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5594271030474393,
      "learning_rate": 9.694693292796242e-06,
      "loss": 0.0657,
      "step": 11727
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5358085649634048,
      "learning_rate": 9.690771348254025e-06,
      "loss": 0.0647,
      "step": 11728
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5476693935735133,
      "learning_rate": 9.68685000644711e-06,
      "loss": 0.1601,
      "step": 11729
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4874773553101076,
      "learning_rate": 9.682929267529894e-06,
      "loss": 0.0802,
      "step": 11730
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4166921577247302,
      "learning_rate": 9.679009131656724e-06,
      "loss": 0.1073,
      "step": 11731
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7865432100462404,
      "learning_rate": 9.675089598981942e-06,
      "loss": 0.2287,
      "step": 11732
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4815315840260524,
      "learning_rate": 9.671170669659868e-06,
      "loss": 0.0676,
      "step": 11733
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2233753860971794,
      "learning_rate": 9.667252343844795e-06,
      "loss": 0.2874,
      "step": 11734
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.34524446259458347,
      "learning_rate": 9.663334621690981e-06,
      "loss": 0.0645,
      "step": 11735
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.42757689889032596,
      "learning_rate": 9.659417503352674e-06,
      "loss": 0.0568,
      "step": 11736
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.46935171612456345,
      "learning_rate": 9.655500988984093e-06,
      "loss": 0.1049,
      "step": 11737
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5062628204779245,
      "learning_rate": 9.651585078739444e-06,
      "loss": 0.3251,
      "step": 11738
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6262778455082729,
      "learning_rate": 9.647669772772882e-06,
      "loss": 0.2048,
      "step": 11739
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.39958782698574974,
      "learning_rate": 9.643755071238564e-06,
      "loss": 0.0992,
      "step": 11740
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1404065812217326,
      "learning_rate": 9.639840974290611e-06,
      "loss": 0.5875,
      "step": 11741
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8351649254794973,
      "learning_rate": 9.63592748208314e-06,
      "loss": 0.1854,
      "step": 11742
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7182280159217909,
      "learning_rate": 9.632014594770205e-06,
      "loss": 0.2739,
      "step": 11743
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6932954774849795,
      "learning_rate": 9.628102312505868e-06,
      "loss": 0.267,
      "step": 11744
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.42113924885363313,
      "learning_rate": 9.62419063544416e-06,
      "loss": 0.0595,
      "step": 11745
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6468622850466452,
      "learning_rate": 9.620279563739096e-06,
      "loss": 0.1264,
      "step": 11746
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4258798145969715,
      "learning_rate": 9.61636909754464e-06,
      "loss": 0.0462,
      "step": 11747
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4776064863971488,
      "learning_rate": 9.612459237014756e-06,
      "loss": 0.0659,
      "step": 11748
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4495266120347507,
      "learning_rate": 9.60854998230339e-06,
      "loss": 0.0891,
      "step": 11749
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6481250327797451,
      "learning_rate": 9.604641333564429e-06,
      "loss": 0.1498,
      "step": 11750
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4101441613654699,
      "learning_rate": 9.600733290951775e-06,
      "loss": 0.0504,
      "step": 11751
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.493862001334297,
      "learning_rate": 9.596825854619287e-06,
      "loss": 0.1032,
      "step": 11752
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2284390309596374,
      "learning_rate": 9.592919024720814e-06,
      "loss": 0.2824,
      "step": 11753
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8095448264138172,
      "learning_rate": 9.589012801410152e-06,
      "loss": 0.2422,
      "step": 11754
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.760426164117027,
      "learning_rate": 9.585107184841109e-06,
      "loss": 0.129,
      "step": 11755
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8220040524913297,
      "learning_rate": 9.581202175167426e-06,
      "loss": 0.3222,
      "step": 11756
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6344948588475947,
      "learning_rate": 9.577297772542882e-06,
      "loss": 0.1389,
      "step": 11757
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9406500211886104,
      "learning_rate": 9.573393977121167e-06,
      "loss": 0.2578,
      "step": 11758
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1803224792993245,
      "learning_rate": 9.569490789055998e-06,
      "loss": 0.3201,
      "step": 11759
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9190262519739983,
      "learning_rate": 9.565588208501017e-06,
      "loss": 0.3052,
      "step": 11760
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0428397055305039,
      "learning_rate": 9.561686235609907e-06,
      "loss": 0.2778,
      "step": 11761
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2906722097693594,
      "learning_rate": 9.557784870536265e-06,
      "loss": 0.0504,
      "step": 11762
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4114778307481003,
      "learning_rate": 9.553884113433708e-06,
      "loss": 0.0611,
      "step": 11763
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.545743138084067,
      "learning_rate": 9.549983964455787e-06,
      "loss": 0.1126,
      "step": 11764
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5477955050012311,
      "learning_rate": 9.546084423756089e-06,
      "loss": 0.1007,
      "step": 11765
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.51006721598474,
      "learning_rate": 9.542185491488109e-06,
      "loss": 0.1852,
      "step": 11766
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3921659175928271,
      "learning_rate": 9.538287167805376e-06,
      "loss": 0.0508,
      "step": 11767
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.34551037235037363,
      "learning_rate": 9.534389452861342e-06,
      "loss": 0.0367,
      "step": 11768
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5507650526817279,
      "learning_rate": 9.530492346809495e-06,
      "loss": 0.1429,
      "step": 11769
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8379393829247513,
      "learning_rate": 9.526595849803244e-06,
      "loss": 0.2071,
      "step": 11770
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5328962925662588,
      "learning_rate": 9.52269996199601e-06,
      "loss": 0.1541,
      "step": 11771
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4593918865668792,
      "learning_rate": 9.518804683541155e-06,
      "loss": 0.0515,
      "step": 11772
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4226526115079271,
      "learning_rate": 9.51491001459207e-06,
      "loss": 0.0462,
      "step": 11773
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4726164601716372,
      "learning_rate": 9.51101595530207e-06,
      "loss": 0.0777,
      "step": 11774
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7491058921752924,
      "learning_rate": 9.507122505824479e-06,
      "loss": 0.1042,
      "step": 11775
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7898253429380888,
      "learning_rate": 9.50322966631256e-06,
      "loss": 0.1555,
      "step": 11776
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6402580568342475,
      "learning_rate": 9.499337436919615e-06,
      "loss": 0.0946,
      "step": 11777
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6212595826768149,
      "learning_rate": 9.49544581779885e-06,
      "loss": 0.1819,
      "step": 11778
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5671790320457786,
      "learning_rate": 9.49155480910351e-06,
      "loss": 0.109,
      "step": 11779
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.313739416007363,
      "learning_rate": 9.48766441098676e-06,
      "loss": 0.5128,
      "step": 11780
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6661152306520607,
      "learning_rate": 9.48377462360178e-06,
      "loss": 0.2538,
      "step": 11781
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.42679986142715676,
      "learning_rate": 9.479885447101714e-06,
      "loss": 0.0657,
      "step": 11782
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5117220060550542,
      "learning_rate": 9.475996881639687e-06,
      "loss": 0.0707,
      "step": 11783
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5678755852689993,
      "learning_rate": 9.472108927368783e-06,
      "loss": 0.1112,
      "step": 11784
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6722871837705586,
      "learning_rate": 9.468221584442077e-06,
      "loss": 0.1929,
      "step": 11785
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3751213681859242,
      "learning_rate": 9.464334853012627e-06,
      "loss": 0.0759,
      "step": 11786
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.949346574591901,
      "learning_rate": 9.46044873323344e-06,
      "loss": 1.2109,
      "step": 11787
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5711590945047377,
      "learning_rate": 9.456563225257522e-06,
      "loss": 0.0938,
      "step": 11788
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.624483874983002,
      "learning_rate": 9.452678329237853e-06,
      "loss": 0.1616,
      "step": 11789
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7193521441508843,
      "learning_rate": 9.448794045327383e-06,
      "loss": 0.1048,
      "step": 11790
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3849708976120447,
      "learning_rate": 9.444910373679033e-06,
      "loss": 0.0644,
      "step": 11791
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5557944222079488,
      "learning_rate": 9.441027314445708e-06,
      "loss": 0.1281,
      "step": 11792
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8044978538986898,
      "learning_rate": 9.437144867780288e-06,
      "loss": 0.2084,
      "step": 11793
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5151132437963092,
      "learning_rate": 9.433263033835638e-06,
      "loss": 0.0913,
      "step": 11794
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5844433631291434,
      "learning_rate": 9.429381812764567e-06,
      "loss": 0.0899,
      "step": 11795
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7689571607234459,
      "learning_rate": 9.425501204719897e-06,
      "loss": 0.1131,
      "step": 11796
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5707122897066637,
      "learning_rate": 9.421621209854404e-06,
      "loss": 0.0738,
      "step": 11797
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4485334630855983,
      "learning_rate": 9.417741828320858e-06,
      "loss": 0.061,
      "step": 11798
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6214921045118067,
      "learning_rate": 9.413863060271975e-06,
      "loss": 0.1436,
      "step": 11799
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5563852450588822,
      "learning_rate": 9.409984905860472e-06,
      "loss": 0.1414,
      "step": 11800
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.45505976363110934,
      "learning_rate": 9.406107365239039e-06,
      "loss": 0.0575,
      "step": 11801
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4637094331007487,
      "learning_rate": 9.402230438560341e-06,
      "loss": 0.0635,
      "step": 11802
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4942871302413537,
      "learning_rate": 9.398354125976993e-06,
      "loss": 0.0837,
      "step": 11803
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.663697866774125,
      "learning_rate": 9.39447842764164e-06,
      "loss": 0.1815,
      "step": 11804
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.40562825694219706,
      "learning_rate": 9.39060334370685e-06,
      "loss": 0.0472,
      "step": 11805
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6089220640796901,
      "learning_rate": 9.386728874325199e-06,
      "loss": 0.1606,
      "step": 11806
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.38348780698055224,
      "learning_rate": 9.382855019649214e-06,
      "loss": 0.0484,
      "step": 11807
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.44153167750576305,
      "learning_rate": 9.378981779831419e-06,
      "loss": 0.0546,
      "step": 11808
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5670138558509069,
      "learning_rate": 9.375109155024308e-06,
      "loss": 0.1479,
      "step": 11809
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8366139317971109,
      "learning_rate": 9.371237145380355e-06,
      "loss": 0.218,
      "step": 11810
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7462892049952143,
      "learning_rate": 9.367365751051987e-06,
      "loss": 0.1356,
      "step": 11811
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5124416879392572,
      "learning_rate": 9.363494972191633e-06,
      "loss": 0.0752,
      "step": 11812
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4016786330425167,
      "learning_rate": 9.35962480895169e-06,
      "loss": 0.0955,
      "step": 11813
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5752065282670699,
      "learning_rate": 9.355755261484536e-06,
      "loss": 0.0543,
      "step": 11814
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5085347622043321,
      "learning_rate": 9.3518863299425e-06,
      "loss": 0.0721,
      "step": 11815
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3997717948191542,
      "learning_rate": 9.348018014477913e-06,
      "loss": 0.0653,
      "step": 11816
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5475450010728623,
      "learning_rate": 9.344150315243084e-06,
      "loss": 0.0906,
      "step": 11817
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3534290731355298,
      "learning_rate": 9.340283232390265e-06,
      "loss": 0.043,
      "step": 11818
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5551202262441215,
      "learning_rate": 9.33641676607172e-06,
      "loss": 0.1085,
      "step": 11819
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6377201262969722,
      "learning_rate": 9.332550916439675e-06,
      "loss": 0.0819,
      "step": 11820
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.564151719359871,
      "learning_rate": 9.328685683646334e-06,
      "loss": 0.1298,
      "step": 11821
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3585476660148081,
      "learning_rate": 9.324821067843865e-06,
      "loss": 0.2367,
      "step": 11822
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3895754682692845,
      "learning_rate": 9.320957069184421e-06,
      "loss": 0.0801,
      "step": 11823
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3944094223802579,
      "learning_rate": 9.317093687820133e-06,
      "loss": 0.0729,
      "step": 11824
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5269346561486346,
      "learning_rate": 9.313230923903118e-06,
      "loss": 0.1477,
      "step": 11825
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.39324363313530203,
      "learning_rate": 9.309368777585434e-06,
      "loss": 0.0667,
      "step": 11826
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.32242577065081657,
      "learning_rate": 9.305507249019147e-06,
      "loss": 0.0229,
      "step": 11827
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6229391418172157,
      "learning_rate": 9.301646338356284e-06,
      "loss": 0.1035,
      "step": 11828
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.41194137165915795,
      "learning_rate": 9.297786045748865e-06,
      "loss": 0.0905,
      "step": 11829
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5425713043814825,
      "learning_rate": 9.293926371348857e-06,
      "loss": 0.0713,
      "step": 11830
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.44079748827877885,
      "learning_rate": 9.29006731530822e-06,
      "loss": 0.0722,
      "step": 11831
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.534281369005663,
      "learning_rate": 9.286208877778896e-06,
      "loss": 0.0803,
      "step": 11832
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6686503293129815,
      "learning_rate": 9.282351058912797e-06,
      "loss": 0.2186,
      "step": 11833
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6314613027821674,
      "learning_rate": 9.278493858861791e-06,
      "loss": 0.1443,
      "step": 11834
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7142026734261894,
      "learning_rate": 9.27463727777775e-06,
      "loss": 0.1054,
      "step": 11835
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9040784475809203,
      "learning_rate": 9.27078131581251e-06,
      "loss": 0.1838,
      "step": 11836
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.49655933728296103,
      "learning_rate": 9.26692597311789e-06,
      "loss": 0.0504,
      "step": 11837
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6500437202807814,
      "learning_rate": 9.263071249845665e-06,
      "loss": 0.2934,
      "step": 11838
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4495049865220841,
      "learning_rate": 9.259217146147598e-06,
      "loss": 0.0987,
      "step": 11839
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8563462921673679,
      "learning_rate": 9.255363662175435e-06,
      "loss": 0.2609,
      "step": 11840
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7286284840542293,
      "learning_rate": 9.251510798080897e-06,
      "loss": 0.1595,
      "step": 11841
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.44408100731452493,
      "learning_rate": 9.247658554015656e-06,
      "loss": 0.082,
      "step": 11842
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5523836705391275,
      "learning_rate": 9.24380693013139e-06,
      "loss": 0.1243,
      "step": 11843
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5089569356504736,
      "learning_rate": 9.239955926579744e-06,
      "loss": 0.1271,
      "step": 11844
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6200284330532141,
      "learning_rate": 9.23610554351232e-06,
      "loss": 0.0928,
      "step": 11845
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7711747461940957,
      "learning_rate": 9.232255781080716e-06,
      "loss": 0.1956,
      "step": 11846
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5065299317593858,
      "learning_rate": 9.228406639436505e-06,
      "loss": 0.1147,
      "step": 11847
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7525954632323149,
      "learning_rate": 9.224558118731232e-06,
      "loss": 0.2051,
      "step": 11848
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6092935115553559,
      "learning_rate": 9.220710219116407e-06,
      "loss": 0.1713,
      "step": 11849
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.513086884985844,
      "learning_rate": 9.216862940743529e-06,
      "loss": 0.0907,
      "step": 11850
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5373943236075339,
      "learning_rate": 9.213016283764067e-06,
      "loss": 0.0549,
      "step": 11851
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4181707855960261,
      "learning_rate": 9.20917024832948e-06,
      "loss": 0.0625,
      "step": 11852
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6085112655208715,
      "learning_rate": 9.205324834591164e-06,
      "loss": 0.0864,
      "step": 11853
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5286653419068382,
      "learning_rate": 9.201480042700539e-06,
      "loss": 0.1071,
      "step": 11854
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5375972943839159,
      "learning_rate": 9.197635872808952e-06,
      "loss": 0.1183,
      "step": 11855
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7951592929036091,
      "learning_rate": 9.193792325067782e-06,
      "loss": 0.2452,
      "step": 11856
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2880901648483064,
      "learning_rate": 9.189949399628326e-06,
      "loss": 0.0694,
      "step": 11857
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5174795318015437,
      "learning_rate": 9.186107096641901e-06,
      "loss": 0.1161,
      "step": 11858
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4329930954690959,
      "learning_rate": 9.182265416259759e-06,
      "loss": 0.0694,
      "step": 11859
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5179224424318537,
      "learning_rate": 9.178424358633178e-06,
      "loss": 0.1204,
      "step": 11860
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6159450894741051,
      "learning_rate": 9.174583923913363e-06,
      "loss": 0.1467,
      "step": 11861
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.27551800435191165,
      "learning_rate": 9.170744112251526e-06,
      "loss": 0.0392,
      "step": 11862
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.47808419477075,
      "learning_rate": 9.166904923798821e-06,
      "loss": 0.0417,
      "step": 11863
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4474089666028748,
      "learning_rate": 9.163066358706438e-06,
      "loss": 0.0421,
      "step": 11864
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5595977244863519,
      "learning_rate": 9.159228417125468e-06,
      "loss": 0.0862,
      "step": 11865
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.620078393120581,
      "learning_rate": 9.155391099207039e-06,
      "loss": 0.1157,
      "step": 11866
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3464377708511498,
      "learning_rate": 9.151554405102202e-06,
      "loss": 0.0506,
      "step": 11867
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4908694244380551,
      "learning_rate": 9.147718334962042e-06,
      "loss": 0.0655,
      "step": 11868
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5867364501723269,
      "learning_rate": 9.143882888937563e-06,
      "loss": 0.1282,
      "step": 11869
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4879148949409474,
      "learning_rate": 9.140048067179788e-06,
      "loss": 0.0825,
      "step": 11870
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4248948865053241,
      "learning_rate": 9.136213869839668e-06,
      "loss": 0.0737,
      "step": 11871
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.40113752232198296,
      "learning_rate": 9.132380297068196e-06,
      "loss": 0.0769,
      "step": 11872
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8645037083468254,
      "learning_rate": 9.128547349016276e-06,
      "loss": 0.2898,
      "step": 11873
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3876400532165102,
      "learning_rate": 9.124715025834829e-06,
      "loss": 0.0322,
      "step": 11874
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6429863020071604,
      "learning_rate": 9.12088332767472e-06,
      "loss": 0.1046,
      "step": 11875
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5996788758983767,
      "learning_rate": 9.117052254686813e-06,
      "loss": 0.0981,
      "step": 11876
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6040119282642428,
      "learning_rate": 9.113221807021944e-06,
      "loss": 0.1382,
      "step": 11877
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7568206768199676,
      "learning_rate": 9.109391984830925e-06,
      "loss": 0.2049,
      "step": 11878
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4387654836803199,
      "learning_rate": 9.105562788264523e-06,
      "loss": 0.0554,
      "step": 11879
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5252617596368939,
      "learning_rate": 9.10173421747351e-06,
      "loss": 0.059,
      "step": 11880
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.7064024902137644,
      "learning_rate": 9.09790627260861e-06,
      "loss": 0.5914,
      "step": 11881
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5675626232392226,
      "learning_rate": 9.094078953820545e-06,
      "loss": 0.1426,
      "step": 11882
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.579516171918946,
      "learning_rate": 9.090252261259984e-06,
      "loss": 0.1053,
      "step": 11883
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4982280488474961,
      "learning_rate": 9.086426195077594e-06,
      "loss": 0.0692,
      "step": 11884
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6637241053213847,
      "learning_rate": 9.082600755424017e-06,
      "loss": 0.1651,
      "step": 11885
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3396150248471185,
      "learning_rate": 9.078775942449849e-06,
      "loss": 0.0517,
      "step": 11886
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7499362696132442,
      "learning_rate": 9.074951756305681e-06,
      "loss": 0.1354,
      "step": 11887
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.46736661407181457,
      "learning_rate": 9.071128197142075e-06,
      "loss": 0.0918,
      "step": 11888
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.42981882088945156,
      "learning_rate": 9.067305265109577e-06,
      "loss": 0.0564,
      "step": 11889
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3588816818588593,
      "learning_rate": 9.063482960358683e-06,
      "loss": 0.0745,
      "step": 11890
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5676438064498235,
      "learning_rate": 9.059661283039884e-06,
      "loss": 0.06,
      "step": 11891
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3996733445941161,
      "learning_rate": 9.055840233303641e-06,
      "loss": 0.0402,
      "step": 11892
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6379464815664929,
      "learning_rate": 9.052019811300408e-06,
      "loss": 0.1702,
      "step": 11893
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.37507148875423224,
      "learning_rate": 9.048200017180575e-06,
      "loss": 0.0685,
      "step": 11894
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6696656168405934,
      "learning_rate": 9.04438085109454e-06,
      "loss": 0.1352,
      "step": 11895
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5424662360033476,
      "learning_rate": 9.040562313192664e-06,
      "loss": 0.1063,
      "step": 11896
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4603377230465067,
      "learning_rate": 9.036744403625297e-06,
      "loss": 0.0858,
      "step": 11897
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.717640535492584,
      "learning_rate": 9.032927122542733e-06,
      "loss": 0.1569,
      "step": 11898
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5355768276107679,
      "learning_rate": 9.029110470095275e-06,
      "loss": 0.1083,
      "step": 11899
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7433395574732166,
      "learning_rate": 9.025294446433182e-06,
      "loss": 0.2147,
      "step": 11900
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7272600168188146,
      "learning_rate": 9.021479051706702e-06,
      "loss": 0.1437,
      "step": 11901
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.701021275446587,
      "learning_rate": 9.017664286066038e-06,
      "loss": 0.2261,
      "step": 11902
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8243421844185351,
      "learning_rate": 9.013850149661382e-06,
      "loss": 0.2233,
      "step": 11903
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6451982742814905,
      "learning_rate": 9.010036642642905e-06,
      "loss": 0.2116,
      "step": 11904
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.583941151029198,
      "learning_rate": 9.006223765160753e-06,
      "loss": 0.1051,
      "step": 11905
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.43354662720181947,
      "learning_rate": 9.002411517365026e-06,
      "loss": 0.1239,
      "step": 11906
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2988059356125048,
      "learning_rate": 8.998599899405819e-06,
      "loss": 0.2206,
      "step": 11907
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.622440161984133,
      "learning_rate": 8.994788911433206e-06,
      "loss": 0.1069,
      "step": 11908
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4669330492811693,
      "learning_rate": 8.99097855359723e-06,
      "loss": 0.0666,
      "step": 11909
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4976789006348455,
      "learning_rate": 8.987168826047892e-06,
      "loss": 0.1237,
      "step": 11910
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.30259958594097935,
      "learning_rate": 8.983359728935195e-06,
      "loss": 0.0315,
      "step": 11911
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6091091591157384,
      "learning_rate": 8.979551262409114e-06,
      "loss": 0.1099,
      "step": 11912
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5178153426411771,
      "learning_rate": 8.97574342661957e-06,
      "loss": 0.106,
      "step": 11913
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4998460012901819,
      "learning_rate": 8.971936221716493e-06,
      "loss": 0.0997,
      "step": 11914
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6935298948634296,
      "learning_rate": 8.968129647849774e-06,
      "loss": 0.2194,
      "step": 11915
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8868115987870678,
      "learning_rate": 8.964323705169289e-06,
      "loss": 0.2873,
      "step": 11916
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.923884865316952,
      "learning_rate": 8.960518393824863e-06,
      "loss": 0.1702,
      "step": 11917
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6378439554459948,
      "learning_rate": 8.956713713966325e-06,
      "loss": 0.0987,
      "step": 11918
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7935753829680561,
      "learning_rate": 8.952909665743465e-06,
      "loss": 0.2592,
      "step": 11919
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.464120954020579,
      "learning_rate": 8.949106249306057e-06,
      "loss": 0.312,
      "step": 11920
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5182526241328477,
      "learning_rate": 8.945303464803833e-06,
      "loss": 0.0688,
      "step": 11921
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4327734072424529,
      "learning_rate": 8.941501312386527e-06,
      "loss": 0.1542,
      "step": 11922
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7600908882831208,
      "learning_rate": 8.937699792203805e-06,
      "loss": 0.2143,
      "step": 11923
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0620885594809586,
      "learning_rate": 8.933898904405371e-06,
      "loss": 0.4231,
      "step": 11924
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5799410629662692,
      "learning_rate": 8.930098649140844e-06,
      "loss": 0.1433,
      "step": 11925
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7672404501580674,
      "learning_rate": 8.926299026559857e-06,
      "loss": 0.1566,
      "step": 11926
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7612886179240177,
      "learning_rate": 8.922500036811982e-06,
      "loss": 0.0949,
      "step": 11927
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6626641504312267,
      "learning_rate": 8.918701680046818e-06,
      "loss": 0.1043,
      "step": 11928
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6481299267781135,
      "learning_rate": 8.914903956413884e-06,
      "loss": 0.371,
      "step": 11929
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2765595613360161,
      "learning_rate": 8.911106866062721e-06,
      "loss": 0.0355,
      "step": 11930
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5295869186322462,
      "learning_rate": 8.907310409142791e-06,
      "loss": 0.1139,
      "step": 11931
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7099772976920334,
      "learning_rate": 8.903514585803602e-06,
      "loss": 0.238,
      "step": 11932
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.333849686913549,
      "learning_rate": 8.899719396194573e-06,
      "loss": 0.0433,
      "step": 11933
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5779533461910191,
      "learning_rate": 8.895924840465125e-06,
      "loss": 0.1013,
      "step": 11934
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5083183455429837,
      "learning_rate": 8.892130918764662e-06,
      "loss": 0.1251,
      "step": 11935
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8285609827917687,
      "learning_rate": 8.888337631242555e-06,
      "loss": 0.1702,
      "step": 11936
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3066244953796611,
      "learning_rate": 8.884544978048134e-06,
      "loss": 0.0239,
      "step": 11937
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.784498250095546,
      "learning_rate": 8.880752959330729e-06,
      "loss": 0.1715,
      "step": 11938
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5064560827325058,
      "learning_rate": 8.876961575239629e-06,
      "loss": 0.3434,
      "step": 11939
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6048352615324915,
      "learning_rate": 8.873170825924118e-06,
      "loss": 0.1426,
      "step": 11940
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5722208922315906,
      "learning_rate": 8.869380711533421e-06,
      "loss": 0.0933,
      "step": 11941
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.48546566125000334,
      "learning_rate": 8.865591232216763e-06,
      "loss": 0.0869,
      "step": 11942
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4758359640647123,
      "learning_rate": 8.86180238812335e-06,
      "loss": 0.0753,
      "step": 11943
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.38582177133799705,
      "learning_rate": 8.858014179402336e-06,
      "loss": 0.0543,
      "step": 11944
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7174922228056618,
      "learning_rate": 8.854226606202873e-06,
      "loss": 0.1544,
      "step": 11945
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9999253946174522,
      "learning_rate": 8.850439668674076e-06,
      "loss": 0.3339,
      "step": 11946
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7706530232868049,
      "learning_rate": 8.846653366965055e-06,
      "loss": 0.2634,
      "step": 11947
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.38578613149858565,
      "learning_rate": 8.842867701224861e-06,
      "loss": 0.0478,
      "step": 11948
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9290060951318184,
      "learning_rate": 8.839082671602541e-06,
      "loss": 0.0441,
      "step": 11949
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5205226067797797,
      "learning_rate": 8.835298278247119e-06,
      "loss": 0.0867,
      "step": 11950
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5739670609941496,
      "learning_rate": 8.8315145213076e-06,
      "loss": 0.1643,
      "step": 11951
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7060678656459443,
      "learning_rate": 8.827731400932929e-06,
      "loss": 0.1633,
      "step": 11952
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6271171366710502,
      "learning_rate": 8.823948917272074e-06,
      "loss": 0.0801,
      "step": 11953
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.424393949581971,
      "learning_rate": 8.820167070473926e-06,
      "loss": 0.0606,
      "step": 11954
    },
    {
      "epoch": 0.73,
      "grad_norm": 10.392233761193062,
      "learning_rate": 8.816385860687413e-06,
      "loss": 0.7454,
      "step": 11955
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.390701097610435,
      "learning_rate": 8.812605288061378e-06,
      "loss": 0.0655,
      "step": 11956
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8648472671485614,
      "learning_rate": 8.808825352744685e-06,
      "loss": 0.1331,
      "step": 11957
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7980594247383389,
      "learning_rate": 8.805046054886121e-06,
      "loss": 0.2187,
      "step": 11958
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.394945314569532,
      "learning_rate": 8.801267394634518e-06,
      "loss": 0.0467,
      "step": 11959
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6841634177696938,
      "learning_rate": 8.79748937213862e-06,
      "loss": 0.157,
      "step": 11960
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.44554079486935333,
      "learning_rate": 8.793711987547184e-06,
      "loss": 0.047,
      "step": 11961
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.38944936924254075,
      "learning_rate": 8.789935241008906e-06,
      "loss": 0.0723,
      "step": 11962
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4186157261286029,
      "learning_rate": 8.786159132672514e-06,
      "loss": 0.0889,
      "step": 11963
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6494099153707411,
      "learning_rate": 8.782383662686647e-06,
      "loss": 0.0723,
      "step": 11964
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7757429546800663,
      "learning_rate": 8.778608831199963e-06,
      "loss": 0.2575,
      "step": 11965
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5866854617813928,
      "learning_rate": 8.774834638361062e-06,
      "loss": 0.075,
      "step": 11966
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.507256278512812,
      "learning_rate": 8.771061084318565e-06,
      "loss": 0.063,
      "step": 11967
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5744514476476255,
      "learning_rate": 8.767288169221017e-06,
      "loss": 0.1066,
      "step": 11968
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5017641567495126,
      "learning_rate": 8.763515893216973e-06,
      "loss": 0.1314,
      "step": 11969
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.935587576087493,
      "learning_rate": 8.759744256454928e-06,
      "loss": 0.2359,
      "step": 11970
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.353533491043018,
      "learning_rate": 8.755973259083408e-06,
      "loss": 0.0692,
      "step": 11971
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9638236171367658,
      "learning_rate": 8.752202901250855e-06,
      "loss": 0.2072,
      "step": 11972
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5199884197274507,
      "learning_rate": 8.748433183105726e-06,
      "loss": 0.1029,
      "step": 11973
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.44971920705125146,
      "learning_rate": 8.744664104796422e-06,
      "loss": 0.0794,
      "step": 11974
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5377205285953504,
      "learning_rate": 8.740895666471341e-06,
      "loss": 0.1565,
      "step": 11975
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.45155383307721336,
      "learning_rate": 8.73712786827885e-06,
      "loss": 0.0918,
      "step": 11976
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5202008381745067,
      "learning_rate": 8.7333607103673e-06,
      "loss": 0.0837,
      "step": 11977
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.49611917114867005,
      "learning_rate": 8.729594192884985e-06,
      "loss": 0.0894,
      "step": 11978
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4896456155233409,
      "learning_rate": 8.725828315980212e-06,
      "loss": 0.1169,
      "step": 11979
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.422066477233257,
      "learning_rate": 8.722063079801245e-06,
      "loss": 0.0789,
      "step": 11980
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4047196753055244,
      "learning_rate": 8.718298484496316e-06,
      "loss": 0.0565,
      "step": 11981
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7235734684010013,
      "learning_rate": 8.714534530213642e-06,
      "loss": 0.1135,
      "step": 11982
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.909507134329517,
      "learning_rate": 8.710771217101416e-06,
      "loss": 0.2712,
      "step": 11983
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4960100540345575,
      "learning_rate": 8.707008545307807e-06,
      "loss": 0.0747,
      "step": 11984
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2846188810551067,
      "learning_rate": 8.703246514980942e-06,
      "loss": 0.1357,
      "step": 11985
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7714043760939037,
      "learning_rate": 8.699485126268942e-06,
      "loss": 0.3748,
      "step": 11986
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5705191920042971,
      "learning_rate": 8.695724379319895e-06,
      "loss": 0.1023,
      "step": 11987
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4450848054007717,
      "learning_rate": 8.691964274281871e-06,
      "loss": 0.078,
      "step": 11988
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6858850467320305,
      "learning_rate": 8.688204811302894e-06,
      "loss": 0.1594,
      "step": 11989
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.438515662624206,
      "learning_rate": 8.684445990530982e-06,
      "loss": 0.0727,
      "step": 11990
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4285266036254393,
      "learning_rate": 8.680687812114125e-06,
      "loss": 0.1015,
      "step": 11991
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7535024454093007,
      "learning_rate": 8.676930276200294e-06,
      "loss": 0.1362,
      "step": 11992
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.475118379850952,
      "learning_rate": 8.673173382937408e-06,
      "loss": 0.0555,
      "step": 11993
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3878963906752485,
      "learning_rate": 8.669417132473385e-06,
      "loss": 0.0649,
      "step": 11994
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1626637100027193,
      "learning_rate": 8.665661524956115e-06,
      "loss": 0.2883,
      "step": 11995
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9163192650802207,
      "learning_rate": 8.661906560533464e-06,
      "loss": 0.1985,
      "step": 11996
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.37951703678762155,
      "learning_rate": 8.658152239353254e-06,
      "loss": 0.0542,
      "step": 11997
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5684468664991009,
      "learning_rate": 8.6543985615633e-06,
      "loss": 0.1025,
      "step": 11998
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.33816160795481126,
      "learning_rate": 8.65064552731139e-06,
      "loss": 0.0486,
      "step": 11999
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.48600485899824947,
      "learning_rate": 8.646893136745293e-06,
      "loss": 0.1347,
      "step": 12000
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.682512739271692,
      "learning_rate": 8.643141390012723e-06,
      "loss": 0.1993,
      "step": 12001
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4599157895067417,
      "learning_rate": 8.639390287261398e-06,
      "loss": 0.0762,
      "step": 12002
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8807090624234792,
      "learning_rate": 8.635639828639006e-06,
      "loss": 0.2355,
      "step": 12003
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5230657022958225,
      "learning_rate": 8.631890014293207e-06,
      "loss": 0.0826,
      "step": 12004
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.48166698048612244,
      "learning_rate": 8.628140844371621e-06,
      "loss": 0.0797,
      "step": 12005
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3274558448914734,
      "learning_rate": 8.624392319021866e-06,
      "loss": 0.04,
      "step": 12006
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5534505633867833,
      "learning_rate": 8.620644438391518e-06,
      "loss": 0.1329,
      "step": 12007
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3227723790524693,
      "learning_rate": 8.616897202628147e-06,
      "loss": 0.0383,
      "step": 12008
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4685214879144185,
      "learning_rate": 8.613150611879266e-06,
      "loss": 0.0867,
      "step": 12009
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6190400942110943,
      "learning_rate": 8.60940466629239e-06,
      "loss": 0.1404,
      "step": 12010
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8527056004137089,
      "learning_rate": 8.605659366015006e-06,
      "loss": 0.1998,
      "step": 12011
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.46665954994709047,
      "learning_rate": 8.601914711194553e-06,
      "loss": 0.0716,
      "step": 12012
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.40254012426080144,
      "learning_rate": 8.598170701978472e-06,
      "loss": 0.058,
      "step": 12013
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9814302983983368,
      "learning_rate": 8.594427338514165e-06,
      "loss": 0.181,
      "step": 12014
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.42999484540838523,
      "learning_rate": 8.590684620949019e-06,
      "loss": 0.0889,
      "step": 12015
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4346911445008277,
      "learning_rate": 8.586942549430371e-06,
      "loss": 0.0815,
      "step": 12016
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5084141636492452,
      "learning_rate": 8.583201124105567e-06,
      "loss": 0.0763,
      "step": 12017
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7028089168462982,
      "learning_rate": 8.579460345121884e-06,
      "loss": 0.2512,
      "step": 12018
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5383062533504565,
      "learning_rate": 8.575720212626632e-06,
      "loss": 0.239,
      "step": 12019
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.36109605968792086,
      "learning_rate": 8.571980726767034e-06,
      "loss": 0.0371,
      "step": 12020
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6525817887038621,
      "learning_rate": 8.56824188769034e-06,
      "loss": 0.1636,
      "step": 12021
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4800389244061795,
      "learning_rate": 8.564503695543721e-06,
      "loss": 0.082,
      "step": 12022
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5961098406018802,
      "learning_rate": 8.56076615047439e-06,
      "loss": 0.1824,
      "step": 12023
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5136329454033077,
      "learning_rate": 8.557029252629468e-06,
      "loss": 0.1227,
      "step": 12024
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8274761860817474,
      "learning_rate": 8.553293002156093e-06,
      "loss": 0.1156,
      "step": 12025
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.44435139148174624,
      "learning_rate": 8.549557399201347e-06,
      "loss": 0.0706,
      "step": 12026
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.593290245264075,
      "learning_rate": 8.54582244391233e-06,
      "loss": 0.2132,
      "step": 12027
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8502667399640953,
      "learning_rate": 8.542088136436069e-06,
      "loss": 0.155,
      "step": 12028
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.517672658223753,
      "learning_rate": 8.538354476919599e-06,
      "loss": 0.0704,
      "step": 12029
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7035306281433136,
      "learning_rate": 8.534621465509898e-06,
      "loss": 0.1278,
      "step": 12030
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5829980930632623,
      "learning_rate": 8.530889102353964e-06,
      "loss": 0.0852,
      "step": 12031
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4357840357267032,
      "learning_rate": 8.527157387598722e-06,
      "loss": 0.0666,
      "step": 12032
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0688561309286528,
      "learning_rate": 8.523426321391107e-06,
      "loss": 0.2629,
      "step": 12033
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.29315359026598103,
      "learning_rate": 8.51969590387799e-06,
      "loss": 0.0392,
      "step": 12034
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5807396429987587,
      "learning_rate": 8.515966135206272e-06,
      "loss": 0.1071,
      "step": 12035
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.47037357880595504,
      "learning_rate": 8.512237015522772e-06,
      "loss": 0.1593,
      "step": 12036
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8032062066117893,
      "learning_rate": 8.508508544974327e-06,
      "loss": 0.2026,
      "step": 12037
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6812691069792839,
      "learning_rate": 8.504780723707704e-06,
      "loss": 0.2103,
      "step": 12038
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7942569707174468,
      "learning_rate": 8.5010535518697e-06,
      "loss": 0.2358,
      "step": 12039
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6829781563181319,
      "learning_rate": 8.497327029607035e-06,
      "loss": 0.6339,
      "step": 12040
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1843417901992848,
      "learning_rate": 8.493601157066438e-06,
      "loss": 0.3325,
      "step": 12041
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.35473644100032925,
      "learning_rate": 8.489875934394587e-06,
      "loss": 0.0554,
      "step": 12042
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4655670360395474,
      "learning_rate": 8.486151361738154e-06,
      "loss": 0.0665,
      "step": 12043
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0064015425624524,
      "learning_rate": 8.482427439243772e-06,
      "loss": 0.1211,
      "step": 12044
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3719080742208712,
      "learning_rate": 8.478704167058074e-06,
      "loss": 0.0364,
      "step": 12045
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6667796112208388,
      "learning_rate": 8.47498154532762e-06,
      "loss": 0.1049,
      "step": 12046
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5553849241188062,
      "learning_rate": 8.471259574198989e-06,
      "loss": 0.1017,
      "step": 12047
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4160751293234559,
      "learning_rate": 8.467538253818722e-06,
      "loss": 0.0346,
      "step": 12048
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4650507046213835,
      "learning_rate": 8.463817584333314e-06,
      "loss": 0.0672,
      "step": 12049
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4608076963175244,
      "learning_rate": 8.460097565889258e-06,
      "loss": 0.0659,
      "step": 12050
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5639618131929317,
      "learning_rate": 8.456378198633013e-06,
      "loss": 0.1011,
      "step": 12051
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8841511671752067,
      "learning_rate": 8.452659482711026e-06,
      "loss": 0.2539,
      "step": 12052
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6757636278638686,
      "learning_rate": 8.448941418269684e-06,
      "loss": 0.2407,
      "step": 12053
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.43519657282598273,
      "learning_rate": 8.445224005455382e-06,
      "loss": 0.0752,
      "step": 12054
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5962748115623063,
      "learning_rate": 8.441507244414474e-06,
      "loss": 0.1661,
      "step": 12055
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7537083526696087,
      "learning_rate": 8.4377911352933e-06,
      "loss": 0.0472,
      "step": 12056
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8078778326803123,
      "learning_rate": 8.434075678238152e-06,
      "loss": 0.264,
      "step": 12057
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4935248723886729,
      "learning_rate": 8.430360873395315e-06,
      "loss": 0.0971,
      "step": 12058
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5178780263542705,
      "learning_rate": 8.42664672091105e-06,
      "loss": 0.0716,
      "step": 12059
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6062063290718919,
      "learning_rate": 8.422933220931584e-06,
      "loss": 0.0936,
      "step": 12060
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.317692816088103,
      "learning_rate": 8.419220373603113e-06,
      "loss": 0.1632,
      "step": 12061
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5862994511914009,
      "learning_rate": 8.415508179071818e-06,
      "loss": 0.1065,
      "step": 12062
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3302224328891553,
      "learning_rate": 8.41179663748385e-06,
      "loss": 0.0377,
      "step": 12063
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4679537395973384,
      "learning_rate": 8.408085748985347e-06,
      "loss": 0.0618,
      "step": 12064
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7484563401128945,
      "learning_rate": 8.40437551372239e-06,
      "loss": 0.2598,
      "step": 12065
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6088444654934837,
      "learning_rate": 8.400665931841062e-06,
      "loss": 0.1046,
      "step": 12066
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.29490524247513167,
      "learning_rate": 8.396957003487413e-06,
      "loss": 0.0359,
      "step": 12067
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7776450619041942,
      "learning_rate": 8.393248728807476e-06,
      "loss": 0.2216,
      "step": 12068
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.40004260647228984,
      "learning_rate": 8.38954110794722e-06,
      "loss": 0.0811,
      "step": 12069
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8596736828804215,
      "learning_rate": 8.385834141052651e-06,
      "loss": 0.1146,
      "step": 12070
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.37296690795348175,
      "learning_rate": 8.38212782826969e-06,
      "loss": 0.0888,
      "step": 12071
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.36682129010514547,
      "learning_rate": 8.378422169744277e-06,
      "loss": 0.055,
      "step": 12072
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.42430719675849576,
      "learning_rate": 8.374717165622285e-06,
      "loss": 0.063,
      "step": 12073
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.41689498388594964,
      "learning_rate": 8.371012816049593e-06,
      "loss": 0.0553,
      "step": 12074
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4968742235502345,
      "learning_rate": 8.367309121172046e-06,
      "loss": 0.0689,
      "step": 12075
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4623681272631199,
      "learning_rate": 8.363606081135464e-06,
      "loss": 0.0642,
      "step": 12076
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.685890029320148,
      "learning_rate": 8.359903696085627e-06,
      "loss": 0.1646,
      "step": 12077
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5251595344789296,
      "learning_rate": 8.356201966168308e-06,
      "loss": 0.0648,
      "step": 12078
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.47143404530792093,
      "learning_rate": 8.352500891529252e-06,
      "loss": 0.0667,
      "step": 12079
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.34960428075683747,
      "learning_rate": 8.34880047231416e-06,
      "loss": 0.0761,
      "step": 12080
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6561451767629046,
      "learning_rate": 8.345100708668727e-06,
      "loss": 0.0918,
      "step": 12081
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4142313211555335,
      "learning_rate": 8.341401600738616e-06,
      "loss": 0.0913,
      "step": 12082
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5513698399183474,
      "learning_rate": 8.337703148669468e-06,
      "loss": 0.1112,
      "step": 12083
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5627399873246366,
      "learning_rate": 8.334005352606882e-06,
      "loss": 0.1741,
      "step": 12084
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6152130184938652,
      "learning_rate": 8.33030821269645e-06,
      "loss": 0.1608,
      "step": 12085
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.722728225627061,
      "learning_rate": 8.32661172908373e-06,
      "loss": 0.1466,
      "step": 12086
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5997265242376856,
      "learning_rate": 8.322915901914264e-06,
      "loss": 0.1811,
      "step": 12087
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.36883785991646634,
      "learning_rate": 8.319220731333544e-06,
      "loss": 0.0506,
      "step": 12088
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6026555507244331,
      "learning_rate": 8.315526217487054e-06,
      "loss": 0.1288,
      "step": 12089
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3081604685246553,
      "learning_rate": 8.311832360520258e-06,
      "loss": 0.0565,
      "step": 12090
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6643032189827603,
      "learning_rate": 8.30813916057859e-06,
      "loss": 0.1009,
      "step": 12091
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5544757850931478,
      "learning_rate": 8.304446617807438e-06,
      "loss": 0.1125,
      "step": 12092
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6553250685829833,
      "learning_rate": 8.300754732352187e-06,
      "loss": 0.2024,
      "step": 12093
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6109818865799097,
      "learning_rate": 8.297063504358189e-06,
      "loss": 0.105,
      "step": 12094
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5666647925887506,
      "learning_rate": 8.29337293397078e-06,
      "loss": 0.1315,
      "step": 12095
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.497820560953492,
      "learning_rate": 8.289683021335245e-06,
      "loss": 0.1,
      "step": 12096
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4309349147398519,
      "learning_rate": 8.285993766596865e-06,
      "loss": 0.0637,
      "step": 12097
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.45491540341914466,
      "learning_rate": 8.282305169900886e-06,
      "loss": 0.0556,
      "step": 12098
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5806453494088429,
      "learning_rate": 8.278617231392546e-06,
      "loss": 0.1204,
      "step": 12099
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3514996948922053,
      "learning_rate": 8.27492995121702e-06,
      "loss": 0.0792,
      "step": 12100
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3940589145871534,
      "learning_rate": 8.271243329519487e-06,
      "loss": 0.0856,
      "step": 12101
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.552176780533701,
      "learning_rate": 8.267557366445094e-06,
      "loss": 0.2121,
      "step": 12102
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6458879895576333,
      "learning_rate": 8.263872062138966e-06,
      "loss": 0.0896,
      "step": 12103
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4060520697500902,
      "learning_rate": 8.260187416746184e-06,
      "loss": 0.0568,
      "step": 12104
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7638770290196737,
      "learning_rate": 8.25650343041182e-06,
      "loss": 0.334,
      "step": 12105
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6492400011161981,
      "learning_rate": 8.252820103280912e-06,
      "loss": 0.1222,
      "step": 12106
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4804961812303213,
      "learning_rate": 8.249137435498491e-06,
      "loss": 0.0947,
      "step": 12107
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9002719276153947,
      "learning_rate": 8.245455427209523e-06,
      "loss": 0.2731,
      "step": 12108
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3771052457822753,
      "learning_rate": 8.241774078558984e-06,
      "loss": 0.0677,
      "step": 12109
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7200121948668408,
      "learning_rate": 8.238093389691818e-06,
      "loss": 0.1557,
      "step": 12110
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5599468689668352,
      "learning_rate": 8.234413360752919e-06,
      "loss": 0.083,
      "step": 12111
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5566999725413246,
      "learning_rate": 8.230733991887183e-06,
      "loss": 0.0957,
      "step": 12112
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.36040004757201927,
      "learning_rate": 8.227055283239466e-06,
      "loss": 0.0741,
      "step": 12113
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5642527740264653,
      "learning_rate": 8.223377234954612e-06,
      "loss": 0.0812,
      "step": 12114
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7272704512194548,
      "learning_rate": 8.219699847177411e-06,
      "loss": 0.1703,
      "step": 12115
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6008739530027916,
      "learning_rate": 8.216023120052663e-06,
      "loss": 0.0894,
      "step": 12116
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6377031026750922,
      "learning_rate": 8.212347053725095e-06,
      "loss": 0.0857,
      "step": 12117
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6610789097532218,
      "learning_rate": 8.208671648339473e-06,
      "loss": 0.1401,
      "step": 12118
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6161835493332686,
      "learning_rate": 8.204996904040471e-06,
      "loss": 0.1132,
      "step": 12119
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6520356380905794,
      "learning_rate": 8.20132282097279e-06,
      "loss": 0.2133,
      "step": 12120
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6113748792487983,
      "learning_rate": 8.197649399281048e-06,
      "loss": 0.1262,
      "step": 12121
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4473753170450537,
      "learning_rate": 8.19397663910991e-06,
      "loss": 0.0697,
      "step": 12122
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6040745449843637,
      "learning_rate": 8.190304540603946e-06,
      "loss": 0.1341,
      "step": 12123
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5630468961178491,
      "learning_rate": 8.186633103907752e-06,
      "loss": 0.081,
      "step": 12124
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.734926182259864,
      "learning_rate": 8.182962329165842e-06,
      "loss": 0.2205,
      "step": 12125
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6280550023620106,
      "learning_rate": 8.179292216522775e-06,
      "loss": 0.1867,
      "step": 12126
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.42686053876028784,
      "learning_rate": 8.175622766123017e-06,
      "loss": 0.0953,
      "step": 12127
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3829235418893526,
      "learning_rate": 8.17195397811106e-06,
      "loss": 0.0852,
      "step": 12128
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8750461867930419,
      "learning_rate": 8.168285852631318e-06,
      "loss": 0.136,
      "step": 12129
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.38762348844890004,
      "learning_rate": 8.164618389828241e-06,
      "loss": 0.0435,
      "step": 12130
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7094198505705238,
      "learning_rate": 8.160951589846197e-06,
      "loss": 0.0896,
      "step": 12131
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9538465099159563,
      "learning_rate": 8.157285452829563e-06,
      "loss": 0.1959,
      "step": 12132
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.41678594377146566,
      "learning_rate": 8.153619978922655e-06,
      "loss": 0.0751,
      "step": 12133
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5979866846039242,
      "learning_rate": 8.14995516826982e-06,
      "loss": 0.1352,
      "step": 12134
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.37905998834640964,
      "learning_rate": 8.146291021015318e-06,
      "loss": 0.0471,
      "step": 12135
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7218416859173366,
      "learning_rate": 8.142627537303423e-06,
      "loss": 0.1194,
      "step": 12136
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3221084787721947,
      "learning_rate": 8.13896471727835e-06,
      "loss": 0.0525,
      "step": 12137
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5917859528313112,
      "learning_rate": 8.135302561084332e-06,
      "loss": 0.1254,
      "step": 12138
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7048607688348343,
      "learning_rate": 8.131641068865534e-06,
      "loss": 0.1188,
      "step": 12139
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3730486869181199,
      "learning_rate": 8.127980240766123e-06,
      "loss": 0.0825,
      "step": 12140
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7342344995796442,
      "learning_rate": 8.124320076930215e-06,
      "loss": 0.1871,
      "step": 12141
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.38642392692584177,
      "learning_rate": 8.120660577501917e-06,
      "loss": 0.0686,
      "step": 12142
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0992857516930177,
      "learning_rate": 8.11700174262531e-06,
      "loss": 0.2944,
      "step": 12143
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5502172718350039,
      "learning_rate": 8.113343572444454e-06,
      "loss": 0.0969,
      "step": 12144
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5877126096308574,
      "learning_rate": 8.109686067103352e-06,
      "loss": 0.0828,
      "step": 12145
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6370493782765202,
      "learning_rate": 8.10602922674602e-06,
      "loss": 0.1171,
      "step": 12146
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5656713172841441,
      "learning_rate": 8.102373051516429e-06,
      "loss": 0.1358,
      "step": 12147
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4475560555689149,
      "learning_rate": 8.098717541558515e-06,
      "loss": 0.1163,
      "step": 12148
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6347792664820863,
      "learning_rate": 8.095062697016202e-06,
      "loss": 0.1085,
      "step": 12149
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4645954578889293,
      "learning_rate": 8.091408518033386e-06,
      "loss": 0.1153,
      "step": 12150
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.48876116536933545,
      "learning_rate": 8.087755004753945e-06,
      "loss": 0.0819,
      "step": 12151
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7100058139079065,
      "learning_rate": 8.0841021573217e-06,
      "loss": 0.1203,
      "step": 12152
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.581373190058586,
      "learning_rate": 8.080449975880477e-06,
      "loss": 0.0759,
      "step": 12153
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7640042551423669,
      "learning_rate": 8.076798460574065e-06,
      "loss": 0.0939,
      "step": 12154
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.47811179976087664,
      "learning_rate": 8.07314761154623e-06,
      "loss": 0.086,
      "step": 12155
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5142043579263642,
      "learning_rate": 8.069497428940701e-06,
      "loss": 0.1096,
      "step": 12156
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5451916535591126,
      "learning_rate": 8.065847912901192e-06,
      "loss": 0.0536,
      "step": 12157
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7664140682546373,
      "learning_rate": 8.062199063571386e-06,
      "loss": 0.1551,
      "step": 12158
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5768942986429255,
      "learning_rate": 8.058550881094948e-06,
      "loss": 0.0811,
      "step": 12159
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6813157083297476,
      "learning_rate": 8.054903365615496e-06,
      "loss": 0.1101,
      "step": 12160
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7245734859197223,
      "learning_rate": 8.051256517276643e-06,
      "loss": 0.199,
      "step": 12161
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6126239243964949,
      "learning_rate": 8.04761033622197e-06,
      "loss": 0.1282,
      "step": 12162
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7522268200301782,
      "learning_rate": 8.043964822595031e-06,
      "loss": 0.139,
      "step": 12163
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.44857079236100184,
      "learning_rate": 8.040319976539344e-06,
      "loss": 0.0583,
      "step": 12164
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.43254936224980634,
      "learning_rate": 8.036675798198412e-06,
      "loss": 0.092,
      "step": 12165
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4482128362602178,
      "learning_rate": 8.033032287715713e-06,
      "loss": 0.0757,
      "step": 12166
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.555199223200044,
      "learning_rate": 8.0293894452347e-06,
      "loss": 0.0715,
      "step": 12167
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.440979186212276,
      "learning_rate": 8.025747270898779e-06,
      "loss": 0.0378,
      "step": 12168
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5831163900423906,
      "learning_rate": 8.022105764851354e-06,
      "loss": 0.0953,
      "step": 12169
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8021957629905868,
      "learning_rate": 8.018464927235791e-06,
      "loss": 0.1795,
      "step": 12170
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6040276555360546,
      "learning_rate": 8.014824758195441e-06,
      "loss": 0.1562,
      "step": 12171
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5537467461438541,
      "learning_rate": 8.011185257873607e-06,
      "loss": 0.0746,
      "step": 12172
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7124230365769133,
      "learning_rate": 8.007546426413584e-06,
      "loss": 0.5428,
      "step": 12173
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1324025144230434,
      "learning_rate": 8.003908263958638e-06,
      "loss": 0.3975,
      "step": 12174
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8585221271194275,
      "learning_rate": 8.00027077065201e-06,
      "loss": 0.2186,
      "step": 12175
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4764627526747612,
      "learning_rate": 7.996633946636898e-06,
      "loss": 0.0631,
      "step": 12176
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.44924027416181594,
      "learning_rate": 7.992997792056495e-06,
      "loss": 0.0864,
      "step": 12177
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.615090392891995,
      "learning_rate": 7.989362307053963e-06,
      "loss": 0.1057,
      "step": 12178
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6901213569259018,
      "learning_rate": 7.98572749177242e-06,
      "loss": 0.0562,
      "step": 12179
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5429369102521406,
      "learning_rate": 7.982093346354983e-06,
      "loss": 0.0519,
      "step": 12180
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5194650466266715,
      "learning_rate": 7.978459870944724e-06,
      "loss": 0.1666,
      "step": 12181
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4119275411683412,
      "learning_rate": 7.974827065684706e-06,
      "loss": 0.2196,
      "step": 12182
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.46298475948445916,
      "learning_rate": 7.97119493071794e-06,
      "loss": 0.072,
      "step": 12183
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5876010101268279,
      "learning_rate": 7.967563466187444e-06,
      "loss": 0.1155,
      "step": 12184
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8726808749802152,
      "learning_rate": 7.963932672236163e-06,
      "loss": 0.2376,
      "step": 12185
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4682356236287488,
      "learning_rate": 7.960302549007081e-06,
      "loss": 0.064,
      "step": 12186
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.8085422674386127,
      "learning_rate": 7.956673096643089e-06,
      "loss": 0.6594,
      "step": 12187
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0634809406151429,
      "learning_rate": 7.9530443152871e-06,
      "loss": 0.316,
      "step": 12188
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.41206650848947673,
      "learning_rate": 7.949416205081958e-06,
      "loss": 0.1153,
      "step": 12189
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8748812636766528,
      "learning_rate": 7.945788766170537e-06,
      "loss": 0.2952,
      "step": 12190
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5366136019288289,
      "learning_rate": 7.942161998695624e-06,
      "loss": 0.1102,
      "step": 12191
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8333097452123736,
      "learning_rate": 7.938535902800025e-06,
      "loss": 0.1594,
      "step": 12192
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.42474573597856646,
      "learning_rate": 7.934910478626484e-06,
      "loss": 0.068,
      "step": 12193
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7648477003836689,
      "learning_rate": 7.931285726317758e-06,
      "loss": 0.1692,
      "step": 12194
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1177737884147931,
      "learning_rate": 7.92766164601654e-06,
      "loss": 0.2597,
      "step": 12195
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6294161782524486,
      "learning_rate": 7.924038237865528e-06,
      "loss": 0.2061,
      "step": 12196
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9724156810644933,
      "learning_rate": 7.92041550200735e-06,
      "loss": 0.2765,
      "step": 12197
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5583247981055174,
      "learning_rate": 7.91679343858467e-06,
      "loss": 0.0926,
      "step": 12198
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5953789920304172,
      "learning_rate": 7.913172047740072e-06,
      "loss": 0.0838,
      "step": 12199
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6255410600116551,
      "learning_rate": 7.909551329616133e-06,
      "loss": 0.1841,
      "step": 12200
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6422778388733758,
      "learning_rate": 7.905931284355406e-06,
      "loss": 0.1867,
      "step": 12201
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.34694117788698897,
      "learning_rate": 7.902311912100425e-06,
      "loss": 0.0562,
      "step": 12202
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5427521827667697,
      "learning_rate": 7.898693212993665e-06,
      "loss": 0.0839,
      "step": 12203
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9166413565274014,
      "learning_rate": 7.895075187177613e-06,
      "loss": 0.2907,
      "step": 12204
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4084864794120196,
      "learning_rate": 7.89145783479471e-06,
      "loss": 0.0424,
      "step": 12205
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.32465713246344713,
      "learning_rate": 7.88784115598738e-06,
      "loss": 0.0468,
      "step": 12206
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0308610154469702,
      "learning_rate": 7.884225150898e-06,
      "loss": 0.5576,
      "step": 12207
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4465822960381651,
      "learning_rate": 7.880609819668941e-06,
      "loss": 0.367,
      "step": 12208
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5486286694343351,
      "learning_rate": 7.876995162442551e-06,
      "loss": 0.0896,
      "step": 12209
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5740599035624652,
      "learning_rate": 7.873381179361129e-06,
      "loss": 0.1042,
      "step": 12210
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.36128453140381134,
      "learning_rate": 7.869767870566958e-06,
      "loss": 0.0331,
      "step": 12211
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.35453725223262494,
      "learning_rate": 7.866155236202305e-06,
      "loss": 0.0734,
      "step": 12212
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6070156218143211,
      "learning_rate": 7.86254327640941e-06,
      "loss": 0.082,
      "step": 12213
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.48293493958425926,
      "learning_rate": 7.858931991330456e-06,
      "loss": 0.0849,
      "step": 12214
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.48965969855853375,
      "learning_rate": 7.855321381107645e-06,
      "loss": 0.0942,
      "step": 12215
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.104422087593933,
      "learning_rate": 7.851711445883103e-06,
      "loss": 0.2615,
      "step": 12216
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.638560660835354,
      "learning_rate": 7.848102185798987e-06,
      "loss": 0.2013,
      "step": 12217
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5298637715128917,
      "learning_rate": 7.844493600997369e-06,
      "loss": 0.0785,
      "step": 12218
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.39187343211560877,
      "learning_rate": 7.840885691620345e-06,
      "loss": 0.0499,
      "step": 12219
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7047378891354981,
      "learning_rate": 7.837278457809929e-06,
      "loss": 0.1361,
      "step": 12220
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5870755273108007,
      "learning_rate": 7.833671899708178e-06,
      "loss": 0.0915,
      "step": 12221
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.517648777075366,
      "learning_rate": 7.830066017457061e-06,
      "loss": 0.0791,
      "step": 12222
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6102790341195374,
      "learning_rate": 7.826460811198555e-06,
      "loss": 0.0987,
      "step": 12223
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3910922368328338,
      "learning_rate": 7.82285628107458e-06,
      "loss": 0.0564,
      "step": 12224
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4125743291120674,
      "learning_rate": 7.819252427227078e-06,
      "loss": 0.0649,
      "step": 12225
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5758252252214278,
      "learning_rate": 7.815649249797915e-06,
      "loss": 0.1844,
      "step": 12226
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7845213092292905,
      "learning_rate": 7.812046748928964e-06,
      "loss": 0.1347,
      "step": 12227
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2628633701338356,
      "learning_rate": 7.808444924762035e-06,
      "loss": 0.3859,
      "step": 12228
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8596362066226862,
      "learning_rate": 7.804843777438966e-06,
      "loss": 0.154,
      "step": 12229
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6526711995449186,
      "learning_rate": 7.801243307101514e-06,
      "loss": 0.0928,
      "step": 12230
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.34919088718719243,
      "learning_rate": 7.797643513891446e-06,
      "loss": 0.0419,
      "step": 12231
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0402033886127207,
      "learning_rate": 7.794044397950465e-06,
      "loss": 0.1802,
      "step": 12232
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.311640320235313,
      "learning_rate": 7.7904459594203e-06,
      "loss": 0.0363,
      "step": 12233
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5057234082523594,
      "learning_rate": 7.786848198442609e-06,
      "loss": 0.082,
      "step": 12234
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3662270550849045,
      "learning_rate": 7.783251115159045e-06,
      "loss": 0.0733,
      "step": 12235
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7854280538092122,
      "learning_rate": 7.779654709711216e-06,
      "loss": 0.2535,
      "step": 12236
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6139506227496185,
      "learning_rate": 7.776058982240722e-06,
      "loss": 0.1349,
      "step": 12237
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.54406521643671,
      "learning_rate": 7.772463932889131e-06,
      "loss": 0.1307,
      "step": 12238
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5091045178911283,
      "learning_rate": 7.76886956179799e-06,
      "loss": 0.0907,
      "step": 12239
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5419246934723126,
      "learning_rate": 7.765275869108793e-06,
      "loss": 0.0781,
      "step": 12240
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5323926829551425,
      "learning_rate": 7.76168285496304e-06,
      "loss": 0.0539,
      "step": 12241
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9162394986082606,
      "learning_rate": 7.758090519502183e-06,
      "loss": 0.2603,
      "step": 12242
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8745414808074031,
      "learning_rate": 7.75449886286767e-06,
      "loss": 0.1205,
      "step": 12243
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4306551760625088,
      "learning_rate": 7.750907885200889e-06,
      "loss": 0.0511,
      "step": 12244
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8309458057330921,
      "learning_rate": 7.747317586643226e-06,
      "loss": 0.3747,
      "step": 12245
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4166883739911802,
      "learning_rate": 7.743727967336043e-06,
      "loss": 0.1009,
      "step": 12246
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.776904812395018,
      "learning_rate": 7.740139027420646e-06,
      "loss": 0.2327,
      "step": 12247
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.48341579361660425,
      "learning_rate": 7.736550767038347e-06,
      "loss": 0.0879,
      "step": 12248
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.44577242942055306,
      "learning_rate": 7.732963186330417e-06,
      "loss": 0.1053,
      "step": 12249
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3810496527475557,
      "learning_rate": 7.729376285438108e-06,
      "loss": 0.0572,
      "step": 12250
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5038877715873302,
      "learning_rate": 7.725790064502625e-06,
      "loss": 0.0705,
      "step": 12251
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4242997924491859,
      "learning_rate": 7.722204523665166e-06,
      "loss": 0.0851,
      "step": 12252
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5897975385019039,
      "learning_rate": 7.718619663066898e-06,
      "loss": 0.1143,
      "step": 12253
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4743722196256741,
      "learning_rate": 7.715035482848967e-06,
      "loss": 0.0633,
      "step": 12254
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.39248298449553537,
      "learning_rate": 7.71145198315247e-06,
      "loss": 0.0605,
      "step": 12255
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.37787191604924475,
      "learning_rate": 7.707869164118498e-06,
      "loss": 0.0383,
      "step": 12256
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3790544599555946,
      "learning_rate": 7.70428702588811e-06,
      "loss": 0.0356,
      "step": 12257
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4126116579954215,
      "learning_rate": 7.700705568602343e-06,
      "loss": 0.0418,
      "step": 12258
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.41806790908931374,
      "learning_rate": 7.69712479240219e-06,
      "loss": 0.106,
      "step": 12259
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5617806974884904,
      "learning_rate": 7.693544697428633e-06,
      "loss": 0.0819,
      "step": 12260
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4051726194966915,
      "learning_rate": 7.689965283822626e-06,
      "loss": 0.0383,
      "step": 12261
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5215073464636255,
      "learning_rate": 7.6863865517251e-06,
      "loss": 0.1369,
      "step": 12262
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.47906278828466275,
      "learning_rate": 7.682808501276933e-06,
      "loss": 0.1059,
      "step": 12263
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.398044752424695,
      "learning_rate": 7.679231132619007e-06,
      "loss": 0.0423,
      "step": 12264
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.6480660128257876,
      "learning_rate": 7.675654445892166e-06,
      "loss": 0.461,
      "step": 12265
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6306402386099688,
      "learning_rate": 7.672078441237232e-06,
      "loss": 0.1121,
      "step": 12266
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.155297231634451,
      "learning_rate": 7.668503118794981e-06,
      "loss": 0.2331,
      "step": 12267
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5020218711341289,
      "learning_rate": 7.664928478706183e-06,
      "loss": 0.0768,
      "step": 12268
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5578298074629104,
      "learning_rate": 7.661354521111574e-06,
      "loss": 0.1383,
      "step": 12269
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6155897529504155,
      "learning_rate": 7.657781246151872e-06,
      "loss": 0.1296,
      "step": 12270
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5569264733483089,
      "learning_rate": 7.654208653967742e-06,
      "loss": 0.0632,
      "step": 12271
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1221049833008256,
      "learning_rate": 7.650636744699852e-06,
      "loss": 0.2569,
      "step": 12272
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9568218288650213,
      "learning_rate": 7.647065518488833e-06,
      "loss": 0.3966,
      "step": 12273
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.30534273071870716,
      "learning_rate": 7.643494975475271e-06,
      "loss": 0.0346,
      "step": 12274
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7103854372321562,
      "learning_rate": 7.639925115799754e-06,
      "loss": 0.2257,
      "step": 12275
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0504203247639337,
      "learning_rate": 7.636355939602824e-06,
      "loss": 0.3035,
      "step": 12276
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4046345148707151,
      "learning_rate": 7.632787447025016e-06,
      "loss": 0.0602,
      "step": 12277
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4590532680573338,
      "learning_rate": 7.629219638206802e-06,
      "loss": 0.1088,
      "step": 12278
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.39369589849776565,
      "learning_rate": 7.625652513288662e-06,
      "loss": 0.104,
      "step": 12279
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8490707450237345,
      "learning_rate": 7.6220860724110335e-06,
      "loss": 0.1762,
      "step": 12280
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.42052867275270794,
      "learning_rate": 7.618520315714337e-06,
      "loss": 0.0774,
      "step": 12281
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8263604941999945,
      "learning_rate": 7.614955243338947e-06,
      "loss": 0.184,
      "step": 12282
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.38593440952541963,
      "learning_rate": 7.611390855425235e-06,
      "loss": 0.0757,
      "step": 12283
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2286783662112688,
      "learning_rate": 7.607827152113511e-06,
      "loss": 0.5335,
      "step": 12284
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4630884440056323,
      "learning_rate": 7.604264133544115e-06,
      "loss": 0.1,
      "step": 12285
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6482665994649286,
      "learning_rate": 7.600701799857296e-06,
      "loss": 0.1537,
      "step": 12286
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3382721955470666,
      "learning_rate": 7.597140151193324e-06,
      "loss": 0.0785,
      "step": 12287
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6333620255384415,
      "learning_rate": 7.593579187692404e-06,
      "loss": 0.1435,
      "step": 12288
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.33118254101232164,
      "learning_rate": 7.590018909494761e-06,
      "loss": 0.0338,
      "step": 12289
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5606660941401761,
      "learning_rate": 7.586459316740541e-06,
      "loss": 0.056,
      "step": 12290
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.812123418091228,
      "learning_rate": 7.582900409569907e-06,
      "loss": 0.2063,
      "step": 12291
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6393284033757405,
      "learning_rate": 7.579342188122951e-06,
      "loss": 0.3919,
      "step": 12292
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.46603297776427277,
      "learning_rate": 7.575784652539794e-06,
      "loss": 0.0559,
      "step": 12293
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.34769705349597657,
      "learning_rate": 7.572227802960474e-06,
      "loss": 0.0407,
      "step": 12294
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.35189240587234627,
      "learning_rate": 7.568671639525044e-06,
      "loss": 0.0473,
      "step": 12295
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5297284002916256,
      "learning_rate": 7.565116162373487e-06,
      "loss": 0.1088,
      "step": 12296
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6026214675631433,
      "learning_rate": 7.561561371645817e-06,
      "loss": 0.1344,
      "step": 12297
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1792393582778058,
      "learning_rate": 7.558007267481967e-06,
      "loss": 0.2376,
      "step": 12298
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3941931463819837,
      "learning_rate": 7.554453850021878e-06,
      "loss": 0.0759,
      "step": 12299
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6998395269896889,
      "learning_rate": 7.550901119405429e-06,
      "loss": 0.1592,
      "step": 12300
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7231519661190513,
      "learning_rate": 7.547349075772525e-06,
      "loss": 0.1407,
      "step": 12301
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.37608707750088116,
      "learning_rate": 7.543797719262985e-06,
      "loss": 0.0464,
      "step": 12302
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3869960183648447,
      "learning_rate": 7.54024705001665e-06,
      "loss": 0.0738,
      "step": 12303
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5710333313340459,
      "learning_rate": 7.536697068173293e-06,
      "loss": 0.1786,
      "step": 12304
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5686273423095409,
      "learning_rate": 7.5331477738726865e-06,
      "loss": 0.1274,
      "step": 12305
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7478753038861125,
      "learning_rate": 7.52959916725457e-06,
      "loss": 0.1694,
      "step": 12306
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.547571077938834,
      "learning_rate": 7.526051248458666e-06,
      "loss": 0.0813,
      "step": 12307
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1912541901720206,
      "learning_rate": 7.522504017624635e-06,
      "loss": 0.2303,
      "step": 12308
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.44490528692314885,
      "learning_rate": 7.518957474892149e-06,
      "loss": 0.0822,
      "step": 12309
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9321028864146453,
      "learning_rate": 7.515411620400837e-06,
      "loss": 0.3909,
      "step": 12310
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.31119242313663675,
      "learning_rate": 7.511866454290306e-06,
      "loss": 0.0227,
      "step": 12311
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.38792557955338053,
      "learning_rate": 7.508321976700119e-06,
      "loss": 0.0652,
      "step": 12312
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.45044072760259163,
      "learning_rate": 7.504778187769831e-06,
      "loss": 0.1789,
      "step": 12313
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8483137193867583,
      "learning_rate": 7.501235087638972e-06,
      "loss": 0.3629,
      "step": 12314
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6960122309479775,
      "learning_rate": 7.497692676447021e-06,
      "loss": 0.0507,
      "step": 12315
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4540423326155243,
      "learning_rate": 7.4941509543334505e-06,
      "loss": 0.063,
      "step": 12316
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.46299291597508035,
      "learning_rate": 7.490609921437708e-06,
      "loss": 0.0896,
      "step": 12317
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9643695550872937,
      "learning_rate": 7.487069577899206e-06,
      "loss": 0.1525,
      "step": 12318
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.47598498692781627,
      "learning_rate": 7.483529923857318e-06,
      "loss": 0.0927,
      "step": 12319
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4080177534515914,
      "learning_rate": 7.47999095945141e-06,
      "loss": 0.0852,
      "step": 12320
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6613011294509041,
      "learning_rate": 7.476452684820814e-06,
      "loss": 0.1274,
      "step": 12321
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0495664786129344,
      "learning_rate": 7.472915100104843e-06,
      "loss": 0.3008,
      "step": 12322
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.35295021519152103,
      "learning_rate": 7.469378205442759e-06,
      "loss": 0.0336,
      "step": 12323
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3899664873683582,
      "learning_rate": 7.465842000973819e-06,
      "loss": 0.0371,
      "step": 12324
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6065913691178063,
      "learning_rate": 7.462306486837245e-06,
      "loss": 0.1694,
      "step": 12325
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2657141318949933,
      "learning_rate": 7.45877166317224e-06,
      "loss": 0.0298,
      "step": 12326
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9443791319264343,
      "learning_rate": 7.455237530117959e-06,
      "loss": 0.2512,
      "step": 12327
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6179768683076257,
      "learning_rate": 7.451704087813549e-06,
      "loss": 0.1871,
      "step": 12328
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.45264880585316786,
      "learning_rate": 7.448171336398127e-06,
      "loss": 0.0984,
      "step": 12329
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.39156070991875913,
      "learning_rate": 7.4446392760107866e-06,
      "loss": 0.0535,
      "step": 12330
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5650085772064622,
      "learning_rate": 7.441107906790573e-06,
      "loss": 0.1454,
      "step": 12331
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2293313569473776,
      "learning_rate": 7.437577228876521e-06,
      "loss": 0.1842,
      "step": 12332
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6521343616858246,
      "learning_rate": 7.4340472424076434e-06,
      "loss": 0.1467,
      "step": 12333
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.47008145227942766,
      "learning_rate": 7.430517947522922e-06,
      "loss": 0.0954,
      "step": 12334
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5660276012387073,
      "learning_rate": 7.4269893443612906e-06,
      "loss": 0.2097,
      "step": 12335
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.650440342500806,
      "learning_rate": 7.423461433061685e-06,
      "loss": 0.1265,
      "step": 12336
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4123644876656625,
      "learning_rate": 7.419934213763e-06,
      "loss": 0.075,
      "step": 12337
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5006046456574456,
      "learning_rate": 7.41640768660411e-06,
      "loss": 0.0954,
      "step": 12338
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5539531134655938,
      "learning_rate": 7.4128818517238465e-06,
      "loss": 0.1308,
      "step": 12339
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9085681723190945,
      "learning_rate": 7.409356709261025e-06,
      "loss": 0.2288,
      "step": 12340
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3214481006144668,
      "learning_rate": 7.405832259354445e-06,
      "loss": 0.0469,
      "step": 12341
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4838619742510873,
      "learning_rate": 7.402308502142852e-06,
      "loss": 0.1059,
      "step": 12342
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5819609177266947,
      "learning_rate": 7.398785437764985e-06,
      "loss": 0.2078,
      "step": 12343
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8031938931697266,
      "learning_rate": 7.395263066359551e-06,
      "loss": 0.1475,
      "step": 12344
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4434772107012953,
      "learning_rate": 7.391741388065232e-06,
      "loss": 0.0864,
      "step": 12345
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4121117066402112,
      "learning_rate": 7.388220403020665e-06,
      "loss": 0.0434,
      "step": 12346
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5824657106210865,
      "learning_rate": 7.384700111364487e-06,
      "loss": 0.1006,
      "step": 12347
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4582547308273598,
      "learning_rate": 7.381180513235286e-06,
      "loss": 0.0695,
      "step": 12348
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5820086524447202,
      "learning_rate": 7.377661608771646e-06,
      "loss": 0.0978,
      "step": 12349
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6032264632260934,
      "learning_rate": 7.374143398112087e-06,
      "loss": 0.1196,
      "step": 12350
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.42242102916299124,
      "learning_rate": 7.370625881395138e-06,
      "loss": 0.1109,
      "step": 12351
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4021582979118791,
      "learning_rate": 7.367109058759281e-06,
      "loss": 0.0666,
      "step": 12352
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.37769757521304614,
      "learning_rate": 7.363592930342985e-06,
      "loss": 0.0565,
      "step": 12353
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.44437710323937724,
      "learning_rate": 7.360077496284667e-06,
      "loss": 0.0438,
      "step": 12354
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6611887086539784,
      "learning_rate": 7.35656275672274e-06,
      "loss": 0.1288,
      "step": 12355
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.35411650268294725,
      "learning_rate": 7.353048711795582e-06,
      "loss": 0.0452,
      "step": 12356
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6797555237344933,
      "learning_rate": 7.349535361641552e-06,
      "loss": 0.2501,
      "step": 12357
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5238068511762372,
      "learning_rate": 7.3460227063989555e-06,
      "loss": 0.1093,
      "step": 12358
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.36485675457672195,
      "learning_rate": 7.342510746206097e-06,
      "loss": 0.0407,
      "step": 12359
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3383694767903284,
      "learning_rate": 7.338999481201247e-06,
      "loss": 0.0469,
      "step": 12360
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0165664310674922,
      "learning_rate": 7.33548891152265e-06,
      "loss": 0.2973,
      "step": 12361
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4694862549560759,
      "learning_rate": 7.331979037308509e-06,
      "loss": 0.0879,
      "step": 12362
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4991503189919262,
      "learning_rate": 7.328469858697012e-06,
      "loss": 0.1605,
      "step": 12363
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6645191922138054,
      "learning_rate": 7.324961375826325e-06,
      "loss": 0.1243,
      "step": 12364
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9077305567481757,
      "learning_rate": 7.3214535888345815e-06,
      "loss": 0.2081,
      "step": 12365
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4322909338529231,
      "learning_rate": 7.317946497859873e-06,
      "loss": 0.059,
      "step": 12366
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5888441673493381,
      "learning_rate": 7.3144401030402825e-06,
      "loss": 0.0958,
      "step": 12367
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4960033400048221,
      "learning_rate": 7.3109344045138615e-06,
      "loss": 0.1033,
      "step": 12368
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5853643044098599,
      "learning_rate": 7.307429402418636e-06,
      "loss": 0.1535,
      "step": 12369
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.584484017652451,
      "learning_rate": 7.303925096892589e-06,
      "loss": 0.1073,
      "step": 12370
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.789516925842136,
      "learning_rate": 7.3004214880736915e-06,
      "loss": 0.2359,
      "step": 12371
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6620446329292344,
      "learning_rate": 7.296918576099892e-06,
      "loss": 0.1052,
      "step": 12372
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4758830789447317,
      "learning_rate": 7.29341636110909e-06,
      "loss": 0.0521,
      "step": 12373
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5985191859505496,
      "learning_rate": 7.289914843239176e-06,
      "loss": 0.5482,
      "step": 12374
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5413967568001806,
      "learning_rate": 7.286414022628005e-06,
      "loss": 0.1234,
      "step": 12375
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.49148281368678026,
      "learning_rate": 7.282913899413418e-06,
      "loss": 0.0733,
      "step": 12376
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.39771421015684166,
      "learning_rate": 7.279414473733201e-06,
      "loss": 0.0754,
      "step": 12377
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5828636311218024,
      "learning_rate": 7.275915745725142e-06,
      "loss": 0.108,
      "step": 12378
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.37247437063976047,
      "learning_rate": 7.272417715526966e-06,
      "loss": 0.0399,
      "step": 12379
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7401190023349441,
      "learning_rate": 7.268920383276428e-06,
      "loss": 0.22,
      "step": 12380
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.34982694827807276,
      "learning_rate": 7.265423749111191e-06,
      "loss": 0.039,
      "step": 12381
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.32823689417866064,
      "learning_rate": 7.261927813168942e-06,
      "loss": 0.0504,
      "step": 12382
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1963571397626314,
      "learning_rate": 7.25843257558729e-06,
      "loss": 0.38,
      "step": 12383
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7241062709831524,
      "learning_rate": 7.254938036503878e-06,
      "loss": 0.12,
      "step": 12384
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.541976991757166,
      "learning_rate": 7.2514441960562645e-06,
      "loss": 0.0694,
      "step": 12385
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.48919914073597937,
      "learning_rate": 7.247951054382024e-06,
      "loss": 0.0949,
      "step": 12386
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4621440643818317,
      "learning_rate": 7.244458611618654e-06,
      "loss": 0.0809,
      "step": 12387
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3548126522129262,
      "learning_rate": 7.240966867903689e-06,
      "loss": 0.0438,
      "step": 12388
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8452532533529346,
      "learning_rate": 7.237475823374579e-06,
      "loss": 0.1613,
      "step": 12389
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2968161991120042,
      "learning_rate": 7.233985478168784e-06,
      "loss": 0.0518,
      "step": 12390
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3461822005264473,
      "learning_rate": 7.2304958324236975e-06,
      "loss": 0.0407,
      "step": 12391
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5759827046516824,
      "learning_rate": 7.227006886276741e-06,
      "loss": 0.0894,
      "step": 12392
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.227063132150793,
      "learning_rate": 7.223518639865251e-06,
      "loss": 0.5719,
      "step": 12393
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7334757709769838,
      "learning_rate": 7.2200310933265815e-06,
      "loss": 0.3302,
      "step": 12394
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.419211075812201,
      "learning_rate": 7.216544246798015e-06,
      "loss": 0.0507,
      "step": 12395
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5392073704107958,
      "learning_rate": 7.213058100416861e-06,
      "loss": 0.0898,
      "step": 12396
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3811581663660166,
      "learning_rate": 7.209572654320351e-06,
      "loss": 0.082,
      "step": 12397
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.41999753530584744,
      "learning_rate": 7.206087908645723e-06,
      "loss": 0.0726,
      "step": 12398
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5598962616951716,
      "learning_rate": 7.20260386353015e-06,
      "loss": 0.1435,
      "step": 12399
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7838788737102396,
      "learning_rate": 7.199120519110833e-06,
      "loss": 0.2198,
      "step": 12400
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.35160984618085633,
      "learning_rate": 7.195637875524891e-06,
      "loss": 0.0511,
      "step": 12401
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7370625593965757,
      "learning_rate": 7.192155932909453e-06,
      "loss": 0.0984,
      "step": 12402
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7872679105086143,
      "learning_rate": 7.188674691401592e-06,
      "loss": 0.1697,
      "step": 12403
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4606551430950711,
      "learning_rate": 7.185194151138372e-06,
      "loss": 0.0562,
      "step": 12404
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3550765099841158,
      "learning_rate": 7.181714312256827e-06,
      "loss": 0.3887,
      "step": 12405
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.683919879553967,
      "learning_rate": 7.178235174893965e-06,
      "loss": 0.1352,
      "step": 12406
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5742308853674263,
      "learning_rate": 7.174756739186752e-06,
      "loss": 0.1714,
      "step": 12407
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4507883019025923,
      "learning_rate": 7.171279005272138e-06,
      "loss": 0.0925,
      "step": 12408
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7338787291118943,
      "learning_rate": 7.167801973287058e-06,
      "loss": 0.1217,
      "step": 12409
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.43775100873944206,
      "learning_rate": 7.164325643368383e-06,
      "loss": 0.0833,
      "step": 12410
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4795842079492482,
      "learning_rate": 7.160850015652992e-06,
      "loss": 0.0997,
      "step": 12411
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.498846330626011,
      "learning_rate": 7.157375090277721e-06,
      "loss": 0.1316,
      "step": 12412
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4000260603849264,
      "learning_rate": 7.153900867379387e-06,
      "loss": 0.0592,
      "step": 12413
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.41896463949834445,
      "learning_rate": 7.150427347094757e-06,
      "loss": 0.0809,
      "step": 12414
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6320849915624562,
      "learning_rate": 7.146954529560595e-06,
      "loss": 0.1978,
      "step": 12415
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.28161622679889103,
      "learning_rate": 7.1434824149136295e-06,
      "loss": 0.0584,
      "step": 12416
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8016386296795662,
      "learning_rate": 7.140011003290567e-06,
      "loss": 0.2242,
      "step": 12417
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.509618118970592,
      "learning_rate": 7.136540294828062e-06,
      "loss": 0.5868,
      "step": 12418
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.40285811818388856,
      "learning_rate": 7.133070289662766e-06,
      "loss": 0.0684,
      "step": 12419
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7244706520108141,
      "learning_rate": 7.129600987931301e-06,
      "loss": 0.2181,
      "step": 12420
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5471007851452272,
      "learning_rate": 7.126132389770257e-06,
      "loss": 0.0818,
      "step": 12421
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9507409061070471,
      "learning_rate": 7.122664495316184e-06,
      "loss": 0.4041,
      "step": 12422
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8313830707654436,
      "learning_rate": 7.1191973047056216e-06,
      "loss": 0.1514,
      "step": 12423
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.39780286755178146,
      "learning_rate": 7.115730818075078e-06,
      "loss": 0.0779,
      "step": 12424
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3991697581917701,
      "learning_rate": 7.112265035561036e-06,
      "loss": 0.0693,
      "step": 12425
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6416860449029415,
      "learning_rate": 7.10879995729993e-06,
      "loss": 0.1359,
      "step": 12426
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3679352213194725,
      "learning_rate": 7.105335583428193e-06,
      "loss": 0.0546,
      "step": 12427
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6472776395700389,
      "learning_rate": 7.10187191408222e-06,
      "loss": 0.2046,
      "step": 12428
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4229090901311777,
      "learning_rate": 7.098408949398383e-06,
      "loss": 0.0716,
      "step": 12429
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.540736961311966,
      "learning_rate": 7.094946689513007e-06,
      "loss": 0.1651,
      "step": 12430
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8795410871706113,
      "learning_rate": 7.091485134562414e-06,
      "loss": 0.1999,
      "step": 12431
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7069598916160206,
      "learning_rate": 7.088024284682884e-06,
      "loss": 0.1921,
      "step": 12432
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.49508019181703483,
      "learning_rate": 7.084564140010683e-06,
      "loss": 0.066,
      "step": 12433
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5871120086412229,
      "learning_rate": 7.081104700682023e-06,
      "loss": 0.2021,
      "step": 12434
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.754211556202722,
      "learning_rate": 7.077645966833116e-06,
      "loss": 0.1967,
      "step": 12435
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6383135740247956,
      "learning_rate": 7.07418793860013e-06,
      "loss": 0.1411,
      "step": 12436
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4880441625288707,
      "learning_rate": 7.070730616119217e-06,
      "loss": 0.3129,
      "step": 12437
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5343729183071094,
      "learning_rate": 7.0672739995264855e-06,
      "loss": 0.0713,
      "step": 12438
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4187274942061921,
      "learning_rate": 7.063818088958027e-06,
      "loss": 0.0483,
      "step": 12439
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.24872132878778014,
      "learning_rate": 7.060362884549912e-06,
      "loss": 0.03,
      "step": 12440
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5518940331271474,
      "learning_rate": 7.056908386438157e-06,
      "loss": 0.0796,
      "step": 12441
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5327321976272471,
      "learning_rate": 7.053454594758782e-06,
      "loss": 0.0666,
      "step": 12442
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0312627880544962,
      "learning_rate": 7.0500015096477605e-06,
      "loss": 0.3219,
      "step": 12443
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.471921426289761,
      "learning_rate": 7.046549131241051e-06,
      "loss": 0.0577,
      "step": 12444
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4436840134042343,
      "learning_rate": 7.043097459674561e-06,
      "loss": 0.101,
      "step": 12445
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.603208798153433,
      "learning_rate": 7.039646495084201e-06,
      "loss": 0.193,
      "step": 12446
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5838929684750406,
      "learning_rate": 7.036196237605816e-06,
      "loss": 0.1279,
      "step": 12447
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8860897843747773,
      "learning_rate": 7.032746687375275e-06,
      "loss": 0.2673,
      "step": 12448
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4155411410803713,
      "learning_rate": 7.029297844528366e-06,
      "loss": 0.5891,
      "step": 12449
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.42140081551201053,
      "learning_rate": 7.025849709200885e-06,
      "loss": 0.0814,
      "step": 12450
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.570636284809547,
      "learning_rate": 7.02240228152857e-06,
      "loss": 0.1203,
      "step": 12451
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7576557959516546,
      "learning_rate": 7.018955561647176e-06,
      "loss": 0.0864,
      "step": 12452
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6185324440017045,
      "learning_rate": 7.015509549692378e-06,
      "loss": 0.1026,
      "step": 12453
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6322203127524599,
      "learning_rate": 7.012064245799868e-06,
      "loss": 0.1538,
      "step": 12454
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.579004133495926,
      "learning_rate": 7.008619650105263e-06,
      "loss": 0.1503,
      "step": 12455
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6513355677953662,
      "learning_rate": 7.005175762744215e-06,
      "loss": 0.1432,
      "step": 12456
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.719476560685213,
      "learning_rate": 7.001732583852283e-06,
      "loss": 0.1429,
      "step": 12457
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3539240770722271,
      "learning_rate": 6.998290113565048e-06,
      "loss": 0.081,
      "step": 12458
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3707804632361334,
      "learning_rate": 6.994848352018016e-06,
      "loss": 0.0538,
      "step": 12459
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5577657919975022,
      "learning_rate": 6.991407299346722e-06,
      "loss": 0.0788,
      "step": 12460
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.45566191060025796,
      "learning_rate": 6.987966955686623e-06,
      "loss": 0.0587,
      "step": 12461
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.41735654471792305,
      "learning_rate": 6.984527321173182e-06,
      "loss": 0.0548,
      "step": 12462
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5456100939530687,
      "learning_rate": 6.981088395941796e-06,
      "loss": 0.0813,
      "step": 12463
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3752254449671591,
      "learning_rate": 6.9776501801278895e-06,
      "loss": 0.0453,
      "step": 12464
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5401116207259343,
      "learning_rate": 6.974212673866801e-06,
      "loss": 0.1515,
      "step": 12465
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4616015556620637,
      "learning_rate": 6.9707758772938805e-06,
      "loss": 0.1073,
      "step": 12466
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5063291986361387,
      "learning_rate": 6.967339790544436e-06,
      "loss": 0.0726,
      "step": 12467
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.34137343867729975,
      "learning_rate": 6.963904413753753e-06,
      "loss": 0.0502,
      "step": 12468
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6658847380473676,
      "learning_rate": 6.9604697470570715e-06,
      "loss": 0.0916,
      "step": 12469
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6870624081715597,
      "learning_rate": 6.957035790589625e-06,
      "loss": 0.1274,
      "step": 12470
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9917685903313047,
      "learning_rate": 6.953602544486618e-06,
      "loss": 0.1355,
      "step": 12471
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8461705640728817,
      "learning_rate": 6.950170008883206e-06,
      "loss": 0.0767,
      "step": 12472
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5453667983716591,
      "learning_rate": 6.946738183914533e-06,
      "loss": 0.1164,
      "step": 12473
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6127686066506244,
      "learning_rate": 6.9433070697157185e-06,
      "loss": 0.496,
      "step": 12474
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5799530672462646,
      "learning_rate": 6.939876666421855e-06,
      "loss": 0.1724,
      "step": 12475
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.48799361416881903,
      "learning_rate": 6.936446974167979e-06,
      "loss": 0.0928,
      "step": 12476
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.44680079003577733,
      "learning_rate": 6.933017993089142e-06,
      "loss": 0.066,
      "step": 12477
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5677916240591647,
      "learning_rate": 6.929589723320315e-06,
      "loss": 0.0894,
      "step": 12478
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.44897422983524815,
      "learning_rate": 6.926162164996511e-06,
      "loss": 0.1144,
      "step": 12479
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5496225716426291,
      "learning_rate": 6.9227353182526464e-06,
      "loss": 0.1288,
      "step": 12480
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8922381627203543,
      "learning_rate": 6.919309183223655e-06,
      "loss": 0.3105,
      "step": 12481
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5853061048800166,
      "learning_rate": 6.915883760044406e-06,
      "loss": 0.1336,
      "step": 12482
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.42063368597201356,
      "learning_rate": 6.912459048849787e-06,
      "loss": 0.141,
      "step": 12483
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7233871250514369,
      "learning_rate": 6.909035049774609e-06,
      "loss": 0.1103,
      "step": 12484
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4741639298426107,
      "learning_rate": 6.905611762953698e-06,
      "loss": 0.0596,
      "step": 12485
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4751358581650397,
      "learning_rate": 6.902189188521802e-06,
      "loss": 0.0806,
      "step": 12486
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1242274468643636,
      "learning_rate": 6.8987673266137035e-06,
      "loss": 0.4891,
      "step": 12487
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5458659575394476,
      "learning_rate": 6.895346177364101e-06,
      "loss": 0.0907,
      "step": 12488
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5067635675866222,
      "learning_rate": 6.891925740907701e-06,
      "loss": 0.1024,
      "step": 12489
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5859504663761095,
      "learning_rate": 6.888506017379148e-06,
      "loss": 0.0884,
      "step": 12490
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5082500032867112,
      "learning_rate": 6.8850870069131075e-06,
      "loss": 0.0912,
      "step": 12491
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3894451995923017,
      "learning_rate": 6.881668709644165e-06,
      "loss": 0.0624,
      "step": 12492
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7326725441470433,
      "learning_rate": 6.878251125706922e-06,
      "loss": 0.1852,
      "step": 12493
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7134118779128005,
      "learning_rate": 6.874834255235901e-06,
      "loss": 0.1027,
      "step": 12494
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4529554752642466,
      "learning_rate": 6.871418098365662e-06,
      "loss": 0.102,
      "step": 12495
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.37411533856330037,
      "learning_rate": 6.868002655230676e-06,
      "loss": 0.0524,
      "step": 12496
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.40017391457545864,
      "learning_rate": 6.864587925965429e-06,
      "loss": 0.0564,
      "step": 12497
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.39867668147781177,
      "learning_rate": 6.8611739107043385e-06,
      "loss": 0.0708,
      "step": 12498
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3174006701759513,
      "learning_rate": 6.8577606095818426e-06,
      "loss": 0.052,
      "step": 12499
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.34391679870278,
      "learning_rate": 6.85434802273231e-06,
      "loss": 0.0519,
      "step": 12500
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7257187725035164,
      "learning_rate": 6.8509361502901045e-06,
      "loss": 0.1923,
      "step": 12501
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1716989719097217,
      "learning_rate": 6.847524992389542e-06,
      "loss": 0.6359,
      "step": 12502
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.438499013343856,
      "learning_rate": 6.844114549164931e-06,
      "loss": 0.0663,
      "step": 12503
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6798831365010426,
      "learning_rate": 6.840704820750543e-06,
      "loss": 0.1636,
      "step": 12504
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5508198176372247,
      "learning_rate": 6.837295807280627e-06,
      "loss": 0.1015,
      "step": 12505
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4849942409529209,
      "learning_rate": 6.833887508889386e-06,
      "loss": 0.0548,
      "step": 12506
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.821451358829028,
      "learning_rate": 6.830479925711011e-06,
      "loss": 0.2195,
      "step": 12507
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5854580896330129,
      "learning_rate": 6.827073057879671e-06,
      "loss": 0.0844,
      "step": 12508
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5762459303003524,
      "learning_rate": 6.823666905529483e-06,
      "loss": 0.1051,
      "step": 12509
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6516965260682488,
      "learning_rate": 6.8202614687945535e-06,
      "loss": 0.0612,
      "step": 12510
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6150591343381009,
      "learning_rate": 6.816856747808959e-06,
      "loss": 0.173,
      "step": 12511
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4002339338673706,
      "learning_rate": 6.813452742706752e-06,
      "loss": 0.0489,
      "step": 12512
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3844197141596728,
      "learning_rate": 6.810049453621939e-06,
      "loss": 0.0567,
      "step": 12513
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0094562549681665,
      "learning_rate": 6.806646880688517e-06,
      "loss": 0.2713,
      "step": 12514
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5360612461468752,
      "learning_rate": 6.803245024040442e-06,
      "loss": 0.1429,
      "step": 12515
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8028687168530914,
      "learning_rate": 6.79984388381166e-06,
      "loss": 0.1443,
      "step": 12516
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5433754070654482,
      "learning_rate": 6.796443460136062e-06,
      "loss": 0.1331,
      "step": 12517
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4389148061453095,
      "learning_rate": 6.793043753147529e-06,
      "loss": 0.1519,
      "step": 12518
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6705284568875634,
      "learning_rate": 6.7896447629799135e-06,
      "loss": 0.1434,
      "step": 12519
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5077948254337026,
      "learning_rate": 6.7862464897670424e-06,
      "loss": 0.0725,
      "step": 12520
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4912000621759633,
      "learning_rate": 6.7828489336426905e-06,
      "loss": 0.0818,
      "step": 12521
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5238626257983934,
      "learning_rate": 6.779452094740635e-06,
      "loss": 0.0974,
      "step": 12522
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4430216734895852,
      "learning_rate": 6.776055973194606e-06,
      "loss": 0.096,
      "step": 12523
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3842960811608543,
      "learning_rate": 6.772660569138323e-06,
      "loss": 0.0578,
      "step": 12524
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.39348844306373815,
      "learning_rate": 6.76926588270545e-06,
      "loss": 0.0896,
      "step": 12525
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3250527033830728,
      "learning_rate": 6.765871914029642e-06,
      "loss": 0.0456,
      "step": 12526
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4524415046276254,
      "learning_rate": 6.762478663244526e-06,
      "loss": 0.0387,
      "step": 12527
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3863191056539763,
      "learning_rate": 6.759086130483705e-06,
      "loss": 0.0411,
      "step": 12528
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.38374258660727045,
      "learning_rate": 6.755694315880726e-06,
      "loss": 0.0557,
      "step": 12529
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.40454004761401235,
      "learning_rate": 6.752303219569136e-06,
      "loss": 0.0392,
      "step": 12530
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.562600163302221,
      "learning_rate": 6.748912841682448e-06,
      "loss": 0.1039,
      "step": 12531
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4040153122827156,
      "learning_rate": 6.745523182354147e-06,
      "loss": 0.0589,
      "step": 12532
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3946268441181733,
      "learning_rate": 6.7421342417176735e-06,
      "loss": 0.0724,
      "step": 12533
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.46826311310762025,
      "learning_rate": 6.738746019906461e-06,
      "loss": 0.0894,
      "step": 12534
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4310398338270126,
      "learning_rate": 6.735358517053905e-06,
      "loss": 0.0458,
      "step": 12535
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5061778781873298,
      "learning_rate": 6.7319717332933825e-06,
      "loss": 0.1009,
      "step": 12536
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6478729894505142,
      "learning_rate": 6.728585668758217e-06,
      "loss": 0.1509,
      "step": 12537
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5874428239983688,
      "learning_rate": 6.725200323581726e-06,
      "loss": 0.1954,
      "step": 12538
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8854755280374543,
      "learning_rate": 6.721815697897205e-06,
      "loss": 0.0996,
      "step": 12539
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6517143540548268,
      "learning_rate": 6.718431791837892e-06,
      "loss": 0.1189,
      "step": 12540
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4074330733499088,
      "learning_rate": 6.715048605537022e-06,
      "loss": 0.0561,
      "step": 12541
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.46442960311642045,
      "learning_rate": 6.7116661391277924e-06,
      "loss": 0.0428,
      "step": 12542
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4102703659163858,
      "learning_rate": 6.708284392743381e-06,
      "loss": 0.066,
      "step": 12543
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5620095379924528,
      "learning_rate": 6.7049033665169125e-06,
      "loss": 0.1593,
      "step": 12544
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6898164669036598,
      "learning_rate": 6.701523060581519e-06,
      "loss": 0.203,
      "step": 12545
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.42956784792743047,
      "learning_rate": 6.698143475070262e-06,
      "loss": 0.0615,
      "step": 12546
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5019413568273408,
      "learning_rate": 6.694764610116227e-06,
      "loss": 0.0539,
      "step": 12547
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.43021677926340035,
      "learning_rate": 6.691386465852423e-06,
      "loss": 0.0694,
      "step": 12548
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.33710860032134893,
      "learning_rate": 6.688009042411861e-06,
      "loss": 0.0429,
      "step": 12549
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5310903777265885,
      "learning_rate": 6.6846323399274935e-06,
      "loss": 0.1109,
      "step": 12550
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.47567515232827995,
      "learning_rate": 6.681256358532289e-06,
      "loss": 0.0717,
      "step": 12551
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4512053120443572,
      "learning_rate": 6.677881098359145e-06,
      "loss": 0.0685,
      "step": 12552
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5200483288077984,
      "learning_rate": 6.674506559540961e-06,
      "loss": 0.1641,
      "step": 12553
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5252864227661221,
      "learning_rate": 6.67113274221057e-06,
      "loss": 0.1149,
      "step": 12554
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.36439948096867714,
      "learning_rate": 6.667759646500835e-06,
      "loss": 0.0592,
      "step": 12555
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8617893086412017,
      "learning_rate": 6.664387272544534e-06,
      "loss": 0.1847,
      "step": 12556
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8518030841243627,
      "learning_rate": 6.661015620474453e-06,
      "loss": 0.2122,
      "step": 12557
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4638100777521241,
      "learning_rate": 6.657644690423315e-06,
      "loss": 0.0753,
      "step": 12558
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5220700276925618,
      "learning_rate": 6.654274482523867e-06,
      "loss": 0.1018,
      "step": 12559
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.40906937508779834,
      "learning_rate": 6.6509049969087715e-06,
      "loss": 0.0741,
      "step": 12560
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.423384950909682,
      "learning_rate": 6.647536233710705e-06,
      "loss": 0.1183,
      "step": 12561
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5506213378751331,
      "learning_rate": 6.6441681930622755e-06,
      "loss": 0.0914,
      "step": 12562
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6672847987054957,
      "learning_rate": 6.6408008750961144e-06,
      "loss": 0.1073,
      "step": 12563
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4810263779591929,
      "learning_rate": 6.637434279944771e-06,
      "loss": 0.1198,
      "step": 12564
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6759098563188897,
      "learning_rate": 6.634068407740809e-06,
      "loss": 0.1205,
      "step": 12565
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4148513291249997,
      "learning_rate": 6.630703258616722e-06,
      "loss": 0.0711,
      "step": 12566
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.40782825634758857,
      "learning_rate": 6.6273388327050265e-06,
      "loss": 0.0576,
      "step": 12567
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.44878605936129723,
      "learning_rate": 6.62397513013816e-06,
      "loss": 0.0853,
      "step": 12568
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3886980657180686,
      "learning_rate": 6.620612151048572e-06,
      "loss": 0.0885,
      "step": 12569
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3361936500941527,
      "learning_rate": 6.617249895568647e-06,
      "loss": 0.0319,
      "step": 12570
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4484202856753919,
      "learning_rate": 6.613888363830767e-06,
      "loss": 0.0672,
      "step": 12571
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6713422509743348,
      "learning_rate": 6.61052755596728e-06,
      "loss": 0.1324,
      "step": 12572
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5760547519194976,
      "learning_rate": 6.60716747211051e-06,
      "loss": 0.0946,
      "step": 12573
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5501021914997006,
      "learning_rate": 6.603808112392731e-06,
      "loss": 0.0975,
      "step": 12574
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.688558338213776,
      "learning_rate": 6.60044947694621e-06,
      "loss": 0.2488,
      "step": 12575
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3963158126114243,
      "learning_rate": 6.5970915659031875e-06,
      "loss": 0.0657,
      "step": 12576
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5584634596810751,
      "learning_rate": 6.593734379395852e-06,
      "loss": 0.0823,
      "step": 12577
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.43246472464760966,
      "learning_rate": 6.590377917556384e-06,
      "loss": 0.0893,
      "step": 12578
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4299597509874688,
      "learning_rate": 6.587022180516933e-06,
      "loss": 0.0755,
      "step": 12579
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4780336409280404,
      "learning_rate": 6.583667168409621e-06,
      "loss": 0.1553,
      "step": 12580
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0240585871495667,
      "learning_rate": 6.580312881366524e-06,
      "loss": 0.1951,
      "step": 12581
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8342892239324506,
      "learning_rate": 6.5769593195197075e-06,
      "loss": 0.1321,
      "step": 12582
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4679672557075524,
      "learning_rate": 6.573606483001207e-06,
      "loss": 0.0979,
      "step": 12583
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7539850951795973,
      "learning_rate": 6.570254371943035e-06,
      "loss": 0.144,
      "step": 12584
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6058665304407599,
      "learning_rate": 6.5669029864771466e-06,
      "loss": 0.1595,
      "step": 12585
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4101306970963291,
      "learning_rate": 6.5635523267355e-06,
      "loss": 0.0348,
      "step": 12586
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5817758817921949,
      "learning_rate": 6.560202392850009e-06,
      "loss": 0.1395,
      "step": 12587
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.37893069171680577,
      "learning_rate": 6.556853184952574e-06,
      "loss": 0.0425,
      "step": 12588
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.47161300773832765,
      "learning_rate": 6.55350470317504e-06,
      "loss": 0.0726,
      "step": 12589
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.43540805858238313,
      "learning_rate": 6.550156947649244e-06,
      "loss": 0.0353,
      "step": 12590
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.43773692640865525,
      "learning_rate": 6.546809918506994e-06,
      "loss": 0.0542,
      "step": 12591
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4675584739955747,
      "learning_rate": 6.543463615880069e-06,
      "loss": 0.0663,
      "step": 12592
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5143875853515517,
      "learning_rate": 6.5401180399002005e-06,
      "loss": 0.182,
      "step": 12593
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.539470327059103,
      "learning_rate": 6.536773190699113e-06,
      "loss": 0.1281,
      "step": 12594
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6300449814811292,
      "learning_rate": 6.5334290684085e-06,
      "loss": 0.1289,
      "step": 12595
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4285368065211913,
      "learning_rate": 6.530085673160024e-06,
      "loss": 0.0593,
      "step": 12596
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7360083479997053,
      "learning_rate": 6.526743005085309e-06,
      "loss": 0.1702,
      "step": 12597
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.293640868915589,
      "learning_rate": 6.5234010643159566e-06,
      "loss": 0.0321,
      "step": 12598
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.45006396795844356,
      "learning_rate": 6.520059850983548e-06,
      "loss": 0.0722,
      "step": 12599
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4682272741864517,
      "learning_rate": 6.516719365219634e-06,
      "loss": 0.0819,
      "step": 12600
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8867408694707275,
      "learning_rate": 6.513379607155717e-06,
      "loss": 0.1115,
      "step": 12601
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4734717999019171,
      "learning_rate": 6.510040576923296e-06,
      "loss": 0.0891,
      "step": 12602
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5619714602196485,
      "learning_rate": 6.5067022746538284e-06,
      "loss": 0.0842,
      "step": 12603
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5465694504015779,
      "learning_rate": 6.50336470047875e-06,
      "loss": 0.0848,
      "step": 12604
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4699397744962549,
      "learning_rate": 6.500027854529456e-06,
      "loss": 0.0825,
      "step": 12605
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5492407521389029,
      "learning_rate": 6.496691736937321e-06,
      "loss": 0.1024,
      "step": 12606
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7397895869241848,
      "learning_rate": 6.493356347833704e-06,
      "loss": 0.0825,
      "step": 12607
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9108856248245445,
      "learning_rate": 6.490021687349901e-06,
      "loss": 0.2907,
      "step": 12608
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.47626066084958607,
      "learning_rate": 6.48668775561721e-06,
      "loss": 0.0854,
      "step": 12609
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5105243040521442,
      "learning_rate": 6.4833545527668875e-06,
      "loss": 0.0793,
      "step": 12610
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.42798987166097424,
      "learning_rate": 6.480022078930178e-06,
      "loss": 0.055,
      "step": 12611
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5575114404735991,
      "learning_rate": 6.476690334238264e-06,
      "loss": 0.1413,
      "step": 12612
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3379108538348594,
      "learning_rate": 6.473359318822325e-06,
      "loss": 0.0477,
      "step": 12613
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7295639099810143,
      "learning_rate": 6.47002903281351e-06,
      "loss": 0.2213,
      "step": 12614
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.43678116493153935,
      "learning_rate": 6.466699476342935e-06,
      "loss": 0.0677,
      "step": 12615
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8461289451498203,
      "learning_rate": 6.4633706495416805e-06,
      "loss": 0.2152,
      "step": 12616
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4621706599085954,
      "learning_rate": 6.460042552540807e-06,
      "loss": 0.0617,
      "step": 12617
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7210841138371082,
      "learning_rate": 6.456715185471346e-06,
      "loss": 0.1629,
      "step": 12618
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9581406590631635,
      "learning_rate": 6.453388548464304e-06,
      "loss": 0.1651,
      "step": 12619
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5246866389624071,
      "learning_rate": 6.45006264165064e-06,
      "loss": 0.0927,
      "step": 12620
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.478699207943013,
      "learning_rate": 6.446737465161301e-06,
      "loss": 0.0542,
      "step": 12621
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0583651810058463,
      "learning_rate": 6.443413019127209e-06,
      "loss": 0.4371,
      "step": 12622
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4864015606165486,
      "learning_rate": 6.44008930367925e-06,
      "loss": 0.1173,
      "step": 12623
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4281892725332394,
      "learning_rate": 6.43676631894827e-06,
      "loss": 0.0745,
      "step": 12624
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9583254430059834,
      "learning_rate": 6.433444065065106e-06,
      "loss": 0.1939,
      "step": 12625
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.971602287292877,
      "learning_rate": 6.430122542160552e-06,
      "loss": 0.1741,
      "step": 12626
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.46368374211927144,
      "learning_rate": 6.426801750365391e-06,
      "loss": 0.0532,
      "step": 12627
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.94373966613746,
      "learning_rate": 6.423481689810348e-06,
      "loss": 0.2539,
      "step": 12628
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1201082184708349,
      "learning_rate": 6.420162360626144e-06,
      "loss": 0.1639,
      "step": 12629
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.48873446742026017,
      "learning_rate": 6.416843762943467e-06,
      "loss": 0.0552,
      "step": 12630
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8114281050590183,
      "learning_rate": 6.413525896892972e-06,
      "loss": 0.1967,
      "step": 12631
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.46318634203721515,
      "learning_rate": 6.410208762605277e-06,
      "loss": 0.0802,
      "step": 12632
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.927680597066546,
      "learning_rate": 6.406892360210986e-06,
      "loss": 0.2207,
      "step": 12633
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5852991617173156,
      "learning_rate": 6.403576689840676e-06,
      "loss": 0.4832,
      "step": 12634
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5885188497733339,
      "learning_rate": 6.4002617516248725e-06,
      "loss": 0.1036,
      "step": 12635
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7967014987622567,
      "learning_rate": 6.396947545694093e-06,
      "loss": 0.2535,
      "step": 12636
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8691036996852096,
      "learning_rate": 6.39363407217882e-06,
      "loss": 0.3772,
      "step": 12637
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5558122995491886,
      "learning_rate": 6.390321331209517e-06,
      "loss": 0.1279,
      "step": 12638
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6350185531791533,
      "learning_rate": 6.387009322916593e-06,
      "loss": 0.1157,
      "step": 12639
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4817742841247144,
      "learning_rate": 6.38369804743045e-06,
      "loss": 0.0893,
      "step": 12640
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.536464352150431,
      "learning_rate": 6.3803875048814576e-06,
      "loss": 0.0901,
      "step": 12641
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.811301962685202,
      "learning_rate": 6.377077695399961e-06,
      "loss": 0.2354,
      "step": 12642
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.087686399574937,
      "learning_rate": 6.373768619116255e-06,
      "loss": 0.3965,
      "step": 12643
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5222805205694007,
      "learning_rate": 6.370460276160634e-06,
      "loss": 0.0554,
      "step": 12644
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3409762922517424,
      "learning_rate": 6.367152666663326e-06,
      "loss": 0.0511,
      "step": 12645
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.616948527418499,
      "learning_rate": 6.363845790754588e-06,
      "loss": 0.4605,
      "step": 12646
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5697027962208374,
      "learning_rate": 6.3605396485645925e-06,
      "loss": 0.1152,
      "step": 12647
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3549684501620342,
      "learning_rate": 6.357234240223514e-06,
      "loss": 0.0444,
      "step": 12648
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.798861082912148,
      "learning_rate": 6.353929565861471e-06,
      "loss": 0.073,
      "step": 12649
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.38561605421832196,
      "learning_rate": 6.350625625608597e-06,
      "loss": 0.0703,
      "step": 12650
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.542173862667206,
      "learning_rate": 6.347322419594953e-06,
      "loss": 0.1426,
      "step": 12651
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.33043806107684415,
      "learning_rate": 6.3440199479505965e-06,
      "loss": 0.0546,
      "step": 12652
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.616058720176975,
      "learning_rate": 6.340718210805535e-06,
      "loss": 0.1316,
      "step": 12653
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.36439706299331454,
      "learning_rate": 6.337417208289781e-06,
      "loss": 0.0634,
      "step": 12654
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6802023986037845,
      "learning_rate": 6.334116940533283e-06,
      "loss": 0.1243,
      "step": 12655
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5415545479498128,
      "learning_rate": 6.330817407665982e-06,
      "loss": 0.0699,
      "step": 12656
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5380409907561058,
      "learning_rate": 6.327518609817767e-06,
      "loss": 0.079,
      "step": 12657
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4811323249365963,
      "learning_rate": 6.324220547118539e-06,
      "loss": 0.1187,
      "step": 12658
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.39644645864750827,
      "learning_rate": 6.3209232196981275e-06,
      "loss": 0.077,
      "step": 12659
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7222842207387277,
      "learning_rate": 6.317626627686363e-06,
      "loss": 0.1622,
      "step": 12660
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3665025601049138,
      "learning_rate": 6.314330771213012e-06,
      "loss": 0.0525,
      "step": 12661
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5053336096826889,
      "learning_rate": 6.311035650407862e-06,
      "loss": 0.0692,
      "step": 12662
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.41945798903675735,
      "learning_rate": 6.3077412654006295e-06,
      "loss": 0.0698,
      "step": 12663
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6611445482390167,
      "learning_rate": 6.304447616321024e-06,
      "loss": 0.1178,
      "step": 12664
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.45235833413784493,
      "learning_rate": 6.301154703298709e-06,
      "loss": 0.0819,
      "step": 12665
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5033115680403223,
      "learning_rate": 6.297862526463333e-06,
      "loss": 0.0705,
      "step": 12666
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9981842441390085,
      "learning_rate": 6.294571085944512e-06,
      "loss": 0.1954,
      "step": 12667
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.41790831006034473,
      "learning_rate": 6.291280381871842e-06,
      "loss": 0.0549,
      "step": 12668
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6147559320429522,
      "learning_rate": 6.2879904143748645e-06,
      "loss": 0.1059,
      "step": 12669
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4905858631162584,
      "learning_rate": 6.2847011835831135e-06,
      "loss": 0.1194,
      "step": 12670
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6167115288989554,
      "learning_rate": 6.28141268962609e-06,
      "loss": 0.1173,
      "step": 12671
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2932146363685292,
      "learning_rate": 6.27812493263327e-06,
      "loss": 0.0499,
      "step": 12672
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3402102222200783,
      "learning_rate": 6.274837912734083e-06,
      "loss": 0.0467,
      "step": 12673
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6038205911689069,
      "learning_rate": 6.271551630057948e-06,
      "loss": 0.1564,
      "step": 12674
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3319955001601416,
      "learning_rate": 6.268266084734253e-06,
      "loss": 0.0466,
      "step": 12675
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5321214396002455,
      "learning_rate": 6.264981276892343e-06,
      "loss": 0.1389,
      "step": 12676
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.40268676883365473,
      "learning_rate": 6.261697206661543e-06,
      "loss": 0.0846,
      "step": 12677
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9042152455195787,
      "learning_rate": 6.258413874171154e-06,
      "loss": 0.2284,
      "step": 12678
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.43862436511947006,
      "learning_rate": 6.255131279550452e-06,
      "loss": 0.0797,
      "step": 12679
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6491608315333327,
      "learning_rate": 6.251849422928657e-06,
      "loss": 0.112,
      "step": 12680
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5078910926473432,
      "learning_rate": 6.2485683044349844e-06,
      "loss": 0.0577,
      "step": 12681
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4750647593207426,
      "learning_rate": 6.24528792419862e-06,
      "loss": 0.1089,
      "step": 12682
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5122949743975481,
      "learning_rate": 6.242008282348716e-06,
      "loss": 0.0752,
      "step": 12683
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5716290498357101,
      "learning_rate": 6.238729379014383e-06,
      "loss": 0.1692,
      "step": 12684
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6841591496493495,
      "learning_rate": 6.2354512143247195e-06,
      "loss": 0.14,
      "step": 12685
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4931811020273067,
      "learning_rate": 6.23217378840879e-06,
      "loss": 0.0821,
      "step": 12686
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.574417204411676,
      "learning_rate": 6.2288971013956365e-06,
      "loss": 0.1054,
      "step": 12687
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5354023168814701,
      "learning_rate": 6.225621153414249e-06,
      "loss": 0.0963,
      "step": 12688
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.37554536480608797,
      "learning_rate": 6.22234594459361e-06,
      "loss": 0.0589,
      "step": 12689
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.38234371214666113,
      "learning_rate": 6.219071475062671e-06,
      "loss": 0.0585,
      "step": 12690
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5408666532004632,
      "learning_rate": 6.215797744950355e-06,
      "loss": 0.0793,
      "step": 12691
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.577672033577222,
      "learning_rate": 6.212524754385537e-06,
      "loss": 0.0969,
      "step": 12692
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4571774196899621,
      "learning_rate": 6.209252503497085e-06,
      "loss": 0.0639,
      "step": 12693
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.40216068888521167,
      "learning_rate": 6.2059809924138265e-06,
      "loss": 0.0755,
      "step": 12694
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2644692917063511,
      "learning_rate": 6.202710221264574e-06,
      "loss": 0.2811,
      "step": 12695
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.7685451609715708,
      "learning_rate": 6.1994401901780825e-06,
      "loss": 0.1838,
      "step": 12696
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6212256873210058,
      "learning_rate": 6.196170899283104e-06,
      "loss": 0.0941,
      "step": 12697
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4535878003180469,
      "learning_rate": 6.192902348708357e-06,
      "loss": 0.1018,
      "step": 12698
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4068441569013639,
      "learning_rate": 6.189634538582528e-06,
      "loss": 0.0765,
      "step": 12699
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6345407544413865,
      "learning_rate": 6.186367469034263e-06,
      "loss": 0.1101,
      "step": 12700
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4151202424595275,
      "learning_rate": 6.1831011401921905e-06,
      "loss": 0.0796,
      "step": 12701
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4559200556179314,
      "learning_rate": 6.179835552184923e-06,
      "loss": 0.0791,
      "step": 12702
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5148600858191995,
      "learning_rate": 6.1765707051410075e-06,
      "loss": 0.1362,
      "step": 12703
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7270209298759371,
      "learning_rate": 6.173306599188994e-06,
      "loss": 0.1411,
      "step": 12704
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5089687085006448,
      "learning_rate": 6.170043234457393e-06,
      "loss": 0.0548,
      "step": 12705
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6786803240718663,
      "learning_rate": 6.166780611074693e-06,
      "loss": 0.1384,
      "step": 12706
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5745611475983418,
      "learning_rate": 6.1635187291693295e-06,
      "loss": 0.0944,
      "step": 12707
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5775840487614734,
      "learning_rate": 6.1602575888697334e-06,
      "loss": 0.1058,
      "step": 12708
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5680652189692668,
      "learning_rate": 6.156997190304298e-06,
      "loss": 0.1002,
      "step": 12709
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6343000595030817,
      "learning_rate": 6.153737533601392e-06,
      "loss": 0.111,
      "step": 12710
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4896967145364874,
      "learning_rate": 6.150478618889344e-06,
      "loss": 0.1175,
      "step": 12711
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.49742472564307977,
      "learning_rate": 6.147220446296464e-06,
      "loss": 0.0687,
      "step": 12712
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4133619127879583,
      "learning_rate": 6.143963015951016e-06,
      "loss": 0.0626,
      "step": 12713
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.43779116106601096,
      "learning_rate": 6.140706327981269e-06,
      "loss": 0.0714,
      "step": 12714
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6688741143156087,
      "learning_rate": 6.137450382515422e-06,
      "loss": 0.1543,
      "step": 12715
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4226601571756356,
      "learning_rate": 6.134195179681681e-06,
      "loss": 0.057,
      "step": 12716
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7236678975980118,
      "learning_rate": 6.130940719608183e-06,
      "loss": 0.1173,
      "step": 12717
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.37310437214155295,
      "learning_rate": 6.127687002423085e-06,
      "loss": 0.0523,
      "step": 12718
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5118320546552023,
      "learning_rate": 6.124434028254467e-06,
      "loss": 0.0516,
      "step": 12719
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.36821041378431957,
      "learning_rate": 6.121181797230418e-06,
      "loss": 0.0622,
      "step": 12720
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.45660971147764395,
      "learning_rate": 6.117930309478956e-06,
      "loss": 0.0733,
      "step": 12721
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.898752601146354,
      "learning_rate": 6.114679565128126e-06,
      "loss": 0.1298,
      "step": 12722
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3587981657717145,
      "learning_rate": 6.111429564305887e-06,
      "loss": 0.0525,
      "step": 12723
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.47639823151431,
      "learning_rate": 6.10818030714021e-06,
      "loss": 0.0908,
      "step": 12724
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4310414986225311,
      "learning_rate": 6.104931793759e-06,
      "loss": 0.0691,
      "step": 12725
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4588311378856544,
      "learning_rate": 6.10168402429018e-06,
      "loss": 0.0586,
      "step": 12726
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6280489334288454,
      "learning_rate": 6.098436998861598e-06,
      "loss": 0.0943,
      "step": 12727
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5228505590711785,
      "learning_rate": 6.095190717601107e-06,
      "loss": 0.0866,
      "step": 12728
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6122281524267821,
      "learning_rate": 6.091945180636488e-06,
      "loss": 0.1916,
      "step": 12729
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.553235417349556,
      "learning_rate": 6.088700388095553e-06,
      "loss": 0.0521,
      "step": 12730
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7393511800545878,
      "learning_rate": 6.0854563401060325e-06,
      "loss": 0.0701,
      "step": 12731
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3868177401473273,
      "learning_rate": 6.0822130367956505e-06,
      "loss": 0.075,
      "step": 12732
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8446388085360069,
      "learning_rate": 6.078970478292104e-06,
      "loss": 0.0996,
      "step": 12733
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.40973134023302554,
      "learning_rate": 6.075728664723043e-06,
      "loss": 0.0682,
      "step": 12734
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3217060210421904,
      "learning_rate": 6.072487596216109e-06,
      "loss": 0.0478,
      "step": 12735
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3280825978956794,
      "learning_rate": 6.069247272898903e-06,
      "loss": 0.2954,
      "step": 12736
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6935087647126812,
      "learning_rate": 6.066007694899006e-06,
      "loss": 0.0661,
      "step": 12737
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5807809294008219,
      "learning_rate": 6.0627688623439484e-06,
      "loss": 0.1587,
      "step": 12738
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.40992366113279044,
      "learning_rate": 6.059530775361261e-06,
      "loss": 0.0463,
      "step": 12739
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4399027887059186,
      "learning_rate": 6.056293434078403e-06,
      "loss": 0.0865,
      "step": 12740
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5173880237477592,
      "learning_rate": 6.053056838622867e-06,
      "loss": 0.1256,
      "step": 12741
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5921728473080693,
      "learning_rate": 6.049820989122054e-06,
      "loss": 0.0936,
      "step": 12742
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4789526354200289,
      "learning_rate": 6.046585885703376e-06,
      "loss": 0.3865,
      "step": 12743
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7057540454149597,
      "learning_rate": 6.043351528494182e-06,
      "loss": 0.1629,
      "step": 12744
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.700329586089994,
      "learning_rate": 6.0401179176218345e-06,
      "loss": 0.1873,
      "step": 12745
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4737381480244731,
      "learning_rate": 6.0368850532136315e-06,
      "loss": 0.0771,
      "step": 12746
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.41659572791096794,
      "learning_rate": 6.033652935396858e-06,
      "loss": 0.0637,
      "step": 12747
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6567630827169122,
      "learning_rate": 6.030421564298747e-06,
      "loss": 0.1104,
      "step": 12748
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7463376692763092,
      "learning_rate": 6.027190940046548e-06,
      "loss": 0.1103,
      "step": 12749
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8803295975397152,
      "learning_rate": 6.023961062767433e-06,
      "loss": 0.271,
      "step": 12750
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5323565362770101,
      "learning_rate": 6.020731932588578e-06,
      "loss": 0.0959,
      "step": 12751
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.46784079306735327,
      "learning_rate": 6.017503549637094e-06,
      "loss": 0.0719,
      "step": 12752
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6017765395813074,
      "learning_rate": 6.014275914040113e-06,
      "loss": 0.1729,
      "step": 12753
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5087557551161838,
      "learning_rate": 6.011049025924689e-06,
      "loss": 0.0789,
      "step": 12754
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4460158935305031,
      "learning_rate": 6.007822885417882e-06,
      "loss": 0.0556,
      "step": 12755
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0174211159053177,
      "learning_rate": 6.004597492646682e-06,
      "loss": 0.467,
      "step": 12756
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6714073236742369,
      "learning_rate": 6.00137284773811e-06,
      "loss": 0.2036,
      "step": 12757
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5897043423206018,
      "learning_rate": 5.998148950819096e-06,
      "loss": 0.148,
      "step": 12758
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5211194130111396,
      "learning_rate": 5.994925802016585e-06,
      "loss": 0.1003,
      "step": 12759
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6136174189668293,
      "learning_rate": 5.991703401457449e-06,
      "loss": 0.1226,
      "step": 12760
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3685456163292724,
      "learning_rate": 5.9884817492685875e-06,
      "loss": 0.0592,
      "step": 12761
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.44903910143128917,
      "learning_rate": 5.985260845576818e-06,
      "loss": 0.1156,
      "step": 12762
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7465225376317783,
      "learning_rate": 5.982040690508964e-06,
      "loss": 0.1607,
      "step": 12763
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.44696305873468767,
      "learning_rate": 5.97882128419179e-06,
      "loss": 0.0486,
      "step": 12764
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7203943924161309,
      "learning_rate": 5.975602626752053e-06,
      "loss": 0.1636,
      "step": 12765
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6218414625596027,
      "learning_rate": 5.972384718316476e-06,
      "loss": 0.083,
      "step": 12766
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7100828523269492,
      "learning_rate": 5.969167559011757e-06,
      "loss": 0.1436,
      "step": 12767
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5517955380155081,
      "learning_rate": 5.965951148964544e-06,
      "loss": 0.0971,
      "step": 12768
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.37779775408478267,
      "learning_rate": 5.962735488301471e-06,
      "loss": 0.0538,
      "step": 12769
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.38697718912053597,
      "learning_rate": 5.959520577149156e-06,
      "loss": 0.0438,
      "step": 12770
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.681513163824184,
      "learning_rate": 5.956306415634155e-06,
      "loss": 0.1325,
      "step": 12771
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4262510370237697,
      "learning_rate": 5.953093003883017e-06,
      "loss": 0.0649,
      "step": 12772
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3671784885411925,
      "learning_rate": 5.949880342022257e-06,
      "loss": 0.045,
      "step": 12773
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3487795170075267,
      "learning_rate": 5.94666843017837e-06,
      "loss": 0.0465,
      "step": 12774
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4871605354333294,
      "learning_rate": 5.943457268477795e-06,
      "loss": 0.1081,
      "step": 12775
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7359067918265809,
      "learning_rate": 5.940246857046963e-06,
      "loss": 0.1338,
      "step": 12776
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6123375051523735,
      "learning_rate": 5.937037196012271e-06,
      "loss": 0.0524,
      "step": 12777
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8623608188430935,
      "learning_rate": 5.933828285500096e-06,
      "loss": 0.1667,
      "step": 12778
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4806227235777174,
      "learning_rate": 5.9306201256367575e-06,
      "loss": 0.3938,
      "step": 12779
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6473488323606009,
      "learning_rate": 5.927412716548572e-06,
      "loss": 0.1108,
      "step": 12780
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3986617894805424,
      "learning_rate": 5.924206058361814e-06,
      "loss": 0.0402,
      "step": 12781
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.540776754625923,
      "learning_rate": 5.921000151202743e-06,
      "loss": 0.0795,
      "step": 12782
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4000499437719701,
      "learning_rate": 5.9177949951975646e-06,
      "loss": 0.0803,
      "step": 12783
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5357379588713791,
      "learning_rate": 5.914590590472472e-06,
      "loss": 0.0875,
      "step": 12784
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2063575114302563,
      "learning_rate": 5.911386937153627e-06,
      "loss": 0.461,
      "step": 12785
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6028640347344307,
      "learning_rate": 5.908184035367162e-06,
      "loss": 0.1366,
      "step": 12786
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1825948666974546,
      "learning_rate": 5.9049818852391695e-06,
      "loss": 0.3541,
      "step": 12787
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5387593424390065,
      "learning_rate": 5.901780486895725e-06,
      "loss": 0.0606,
      "step": 12788
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8533201102077814,
      "learning_rate": 5.89857984046287e-06,
      "loss": 0.1903,
      "step": 12789
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5798385564422864,
      "learning_rate": 5.895379946066623e-06,
      "loss": 0.2159,
      "step": 12790
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4989026714945183,
      "learning_rate": 5.89218080383295e-06,
      "loss": 0.0797,
      "step": 12791
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.48493522684954044,
      "learning_rate": 5.888982413887817e-06,
      "loss": 0.0418,
      "step": 12792
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5148460869371569,
      "learning_rate": 5.885784776357139e-06,
      "loss": 0.1006,
      "step": 12793
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3352279471146255,
      "learning_rate": 5.8825878913668225e-06,
      "loss": 0.0334,
      "step": 12794
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3995885593490997,
      "learning_rate": 5.879391759042713e-06,
      "loss": 0.0479,
      "step": 12795
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.43202953505093356,
      "learning_rate": 5.8761963795106536e-06,
      "loss": 0.1227,
      "step": 12796
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4833054803940054,
      "learning_rate": 5.873001752896451e-06,
      "loss": 0.1006,
      "step": 12797
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5369963497366682,
      "learning_rate": 5.869807879325881e-06,
      "loss": 0.0932,
      "step": 12798
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5142605017790172,
      "learning_rate": 5.86661475892468e-06,
      "loss": 0.1093,
      "step": 12799
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.47304957481607274,
      "learning_rate": 5.8634223918185685e-06,
      "loss": 0.0779,
      "step": 12800
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6618955565442902,
      "learning_rate": 5.860230778133236e-06,
      "loss": 0.292,
      "step": 12801
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.9135074001206025,
      "learning_rate": 5.857039917994331e-06,
      "loss": 0.34,
      "step": 12802
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3089417519840208,
      "learning_rate": 5.8538498115274814e-06,
      "loss": 0.0391,
      "step": 12803
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7150309855379128,
      "learning_rate": 5.850660458858287e-06,
      "loss": 0.1704,
      "step": 12804
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.29005988477239003,
      "learning_rate": 5.847471860112322e-06,
      "loss": 0.0397,
      "step": 12805
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1496412455568752,
      "learning_rate": 5.844284015415108e-06,
      "loss": 0.2915,
      "step": 12806
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.31242105113916385,
      "learning_rate": 5.84109692489217e-06,
      "loss": 0.0358,
      "step": 12807
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5922300497741798,
      "learning_rate": 5.83791058866896e-06,
      "loss": 0.1238,
      "step": 12808
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9434572974155135,
      "learning_rate": 5.834725006870959e-06,
      "loss": 0.245,
      "step": 12809
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.920092175412986,
      "learning_rate": 5.8315401796235595e-06,
      "loss": 0.427,
      "step": 12810
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4551777661953979,
      "learning_rate": 5.828356107052171e-06,
      "loss": 0.061,
      "step": 12811
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5318587860372884,
      "learning_rate": 5.825172789282126e-06,
      "loss": 0.0508,
      "step": 12812
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6284606842321209,
      "learning_rate": 5.821990226438786e-06,
      "loss": 0.0542,
      "step": 12813
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6324743105927801,
      "learning_rate": 5.818808418647429e-06,
      "loss": 0.1431,
      "step": 12814
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8422837826087398,
      "learning_rate": 5.815627366033339e-06,
      "loss": 0.1683,
      "step": 12815
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5449168737440003,
      "learning_rate": 5.812447068721729e-06,
      "loss": 0.0809,
      "step": 12816
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7105742279982133,
      "learning_rate": 5.80926752683785e-06,
      "loss": 0.158,
      "step": 12817
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.769244210672756,
      "learning_rate": 5.806088740506851e-06,
      "loss": 0.1151,
      "step": 12818
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4367952497085417,
      "learning_rate": 5.802910709853901e-06,
      "loss": 0.0558,
      "step": 12819
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1149713054742745,
      "learning_rate": 5.799733435004101e-06,
      "loss": 0.2082,
      "step": 12820
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4428105751934076,
      "learning_rate": 5.796556916082571e-06,
      "loss": 0.0709,
      "step": 12821
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0672843012268531,
      "learning_rate": 5.79338115321435e-06,
      "loss": 0.3116,
      "step": 12822
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5951703423792055,
      "learning_rate": 5.790206146524485e-06,
      "loss": 0.1498,
      "step": 12823
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7952539089958708,
      "learning_rate": 5.787031896137959e-06,
      "loss": 0.1808,
      "step": 12824
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4829576727452836,
      "learning_rate": 5.783858402179768e-06,
      "loss": 0.0919,
      "step": 12825
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.45081176340266726,
      "learning_rate": 5.780685664774838e-06,
      "loss": 0.063,
      "step": 12826
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3776083056848915,
      "learning_rate": 5.777513684048094e-06,
      "loss": 0.0315,
      "step": 12827
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7304070922658552,
      "learning_rate": 5.774342460124396e-06,
      "loss": 0.1518,
      "step": 12828
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6508413517856357,
      "learning_rate": 5.771171993128632e-06,
      "loss": 0.1276,
      "step": 12829
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2695156563441428,
      "learning_rate": 5.768002283185597e-06,
      "loss": 0.1908,
      "step": 12830
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6391651673510093,
      "learning_rate": 5.764833330420105e-06,
      "loss": 0.0926,
      "step": 12831
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6833544315174136,
      "learning_rate": 5.7616651349569035e-06,
      "loss": 0.16,
      "step": 12832
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9150443830284044,
      "learning_rate": 5.758497696920731e-06,
      "loss": 0.1362,
      "step": 12833
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3961435619974713,
      "learning_rate": 5.755331016436294e-06,
      "loss": 0.0449,
      "step": 12834
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.502401932033448,
      "learning_rate": 5.752165093628276e-06,
      "loss": 0.1096,
      "step": 12835
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.381120978704116,
      "learning_rate": 5.748999928621304e-06,
      "loss": 0.0498,
      "step": 12836
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5681713267476995,
      "learning_rate": 5.745835521540003e-06,
      "loss": 0.104,
      "step": 12837
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8253640892083357,
      "learning_rate": 5.7426718725089615e-06,
      "loss": 0.2053,
      "step": 12838
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6813197004805099,
      "learning_rate": 5.7395089816527246e-06,
      "loss": 0.2333,
      "step": 12839
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6722131770025485,
      "learning_rate": 5.736346849095819e-06,
      "loss": 0.1397,
      "step": 12840
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4618083694642971,
      "learning_rate": 5.733185474962749e-06,
      "loss": 0.0522,
      "step": 12841
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7469111868996892,
      "learning_rate": 5.730024859377977e-06,
      "loss": 0.2273,
      "step": 12842
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8607926337009438,
      "learning_rate": 5.726865002465933e-06,
      "loss": 0.1362,
      "step": 12843
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7748812575520818,
      "learning_rate": 5.723705904351026e-06,
      "loss": 0.1674,
      "step": 12844
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.39112825210456564,
      "learning_rate": 5.7205475651576305e-06,
      "loss": 0.0842,
      "step": 12845
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7179013198814171,
      "learning_rate": 5.717389985010102e-06,
      "loss": 0.0968,
      "step": 12846
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4309490367891424,
      "learning_rate": 5.714233164032745e-06,
      "loss": 0.0997,
      "step": 12847
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.44905231181865507,
      "learning_rate": 5.711077102349846e-06,
      "loss": 0.058,
      "step": 12848
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8219762142093574,
      "learning_rate": 5.707921800085667e-06,
      "loss": 0.1371,
      "step": 12849
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.362440846937889,
      "learning_rate": 5.70476725736444e-06,
      "loss": 0.0874,
      "step": 12850
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6613893909855236,
      "learning_rate": 5.701613474310346e-06,
      "loss": 0.1768,
      "step": 12851
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7044537326268,
      "learning_rate": 5.698460451047563e-06,
      "loss": 0.1704,
      "step": 12852
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8095487630801521,
      "learning_rate": 5.695308187700224e-06,
      "loss": 0.183,
      "step": 12853
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4910616475792211,
      "learning_rate": 5.692156684392444e-06,
      "loss": 0.1541,
      "step": 12854
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.448873897738072,
      "learning_rate": 5.689005941248285e-06,
      "loss": 0.0702,
      "step": 12855
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8134117857571729,
      "learning_rate": 5.6858559583918e-06,
      "loss": 0.1432,
      "step": 12856
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.29143052379036233,
      "learning_rate": 5.68270673594701e-06,
      "loss": 0.04,
      "step": 12857
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6223273141352124,
      "learning_rate": 5.679558274037908e-06,
      "loss": 0.097,
      "step": 12858
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7328899498818063,
      "learning_rate": 5.6764105727884346e-06,
      "loss": 0.1066,
      "step": 12859
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5485719397505906,
      "learning_rate": 5.673263632322526e-06,
      "loss": 0.0542,
      "step": 12860
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5741263058744618,
      "learning_rate": 5.670117452764082e-06,
      "loss": 0.0933,
      "step": 12861
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4085826391204108,
      "learning_rate": 5.6669720342369694e-06,
      "loss": 0.0464,
      "step": 12862
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5216727981581094,
      "learning_rate": 5.663827376865019e-06,
      "loss": 0.0966,
      "step": 12863
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3152830565137713,
      "learning_rate": 5.660683480772042e-06,
      "loss": 0.0468,
      "step": 12864
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6233241748858632,
      "learning_rate": 5.657540346081816e-06,
      "loss": 0.131,
      "step": 12865
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1036061517092544,
      "learning_rate": 5.654397972918096e-06,
      "loss": 0.2199,
      "step": 12866
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7156951915373613,
      "learning_rate": 5.651256361404586e-06,
      "loss": 0.2377,
      "step": 12867
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5730262840366537,
      "learning_rate": 5.648115511664978e-06,
      "loss": 0.0755,
      "step": 12868
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4172304222630738,
      "learning_rate": 5.64497542382294e-06,
      "loss": 0.0801,
      "step": 12869
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4947803023416236,
      "learning_rate": 5.641836098002081e-06,
      "loss": 0.1549,
      "step": 12870
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9304444750619746,
      "learning_rate": 5.6386975343260125e-06,
      "loss": 0.1899,
      "step": 12871
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2499728769783547,
      "learning_rate": 5.635559732918297e-06,
      "loss": 0.029,
      "step": 12872
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5063321109926083,
      "learning_rate": 5.6324226939024786e-06,
      "loss": 0.1321,
      "step": 12873
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6770634662565215,
      "learning_rate": 5.6292864174020525e-06,
      "loss": 0.2254,
      "step": 12874
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3871677850396689,
      "learning_rate": 5.6261509035405e-06,
      "loss": 0.0481,
      "step": 12875
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.40692736883839054,
      "learning_rate": 5.623016152441274e-06,
      "loss": 0.0467,
      "step": 12876
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3014270206474935,
      "learning_rate": 5.619882164227796e-06,
      "loss": 0.0513,
      "step": 12877
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2648599488024854,
      "learning_rate": 5.616748939023439e-06,
      "loss": 0.2827,
      "step": 12878
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4387479232981567,
      "learning_rate": 5.6136164769515695e-06,
      "loss": 0.1011,
      "step": 12879
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5350714794158283,
      "learning_rate": 5.61048477813551e-06,
      "loss": 0.1208,
      "step": 12880
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5311223477685256,
      "learning_rate": 5.6073538426985675e-06,
      "loss": 0.0799,
      "step": 12881
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.47985778285696856,
      "learning_rate": 5.604223670763998e-06,
      "loss": 0.0881,
      "step": 12882
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5089196814116005,
      "learning_rate": 5.601094262455042e-06,
      "loss": 0.0946,
      "step": 12883
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.43767387776533495,
      "learning_rate": 5.597965617894907e-06,
      "loss": 0.0728,
      "step": 12884
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8708314235590856,
      "learning_rate": 5.594837737206779e-06,
      "loss": 0.1182,
      "step": 12885
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6391386166845389,
      "learning_rate": 5.5917106205137875e-06,
      "loss": 0.1738,
      "step": 12886
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7551010939624452,
      "learning_rate": 5.588584267939062e-06,
      "loss": 0.0674,
      "step": 12887
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5564764266457817,
      "learning_rate": 5.585458679605682e-06,
      "loss": 0.1317,
      "step": 12888
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.38093532699321275,
      "learning_rate": 5.582333855636715e-06,
      "loss": 0.0564,
      "step": 12889
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4575658877174505,
      "learning_rate": 5.579209796155174e-06,
      "loss": 0.0488,
      "step": 12890
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7185856245573075,
      "learning_rate": 5.576086501284061e-06,
      "loss": 0.0938,
      "step": 12891
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4651972055509496,
      "learning_rate": 5.572963971146344e-06,
      "loss": 0.0928,
      "step": 12892
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9360189222235119,
      "learning_rate": 5.5698422058649665e-06,
      "loss": 0.1623,
      "step": 12893
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4022527296960155,
      "learning_rate": 5.566721205562816e-06,
      "loss": 0.0598,
      "step": 12894
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8621878919822511,
      "learning_rate": 5.563600970362781e-06,
      "loss": 0.0778,
      "step": 12895
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7806958541049216,
      "learning_rate": 5.560481500387707e-06,
      "loss": 0.1276,
      "step": 12896
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5984814678608378,
      "learning_rate": 5.557362795760412e-06,
      "loss": 0.0756,
      "step": 12897
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5243336417635895,
      "learning_rate": 5.554244856603674e-06,
      "loss": 0.1018,
      "step": 12898
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5648363271565507,
      "learning_rate": 5.55112768304025e-06,
      "loss": 0.0825,
      "step": 12899
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.631235044998675,
      "learning_rate": 5.548011275192874e-06,
      "loss": 0.1491,
      "step": 12900
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.449318387190304,
      "learning_rate": 5.54489563318423e-06,
      "loss": 0.0689,
      "step": 12901
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4868770518718401,
      "learning_rate": 5.541780757136986e-06,
      "loss": 0.0793,
      "step": 12902
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5651239972391267,
      "learning_rate": 5.538666647173779e-06,
      "loss": 0.0605,
      "step": 12903
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5918710949750438,
      "learning_rate": 5.5355533034172236e-06,
      "loss": 0.1805,
      "step": 12904
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7229511594048367,
      "learning_rate": 5.532440725989874e-06,
      "loss": 0.1761,
      "step": 12905
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8120795771269438,
      "learning_rate": 5.5293289150142925e-06,
      "loss": 0.1395,
      "step": 12906
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.39932177783127354,
      "learning_rate": 5.5262178706129735e-06,
      "loss": 0.0749,
      "step": 12907
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.27411090089793244,
      "learning_rate": 5.523107592908424e-06,
      "loss": 0.0516,
      "step": 12908
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.39555784321449033,
      "learning_rate": 5.519998082023084e-06,
      "loss": 0.0548,
      "step": 12909
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6631518725421254,
      "learning_rate": 5.516889338079384e-06,
      "loss": 0.1559,
      "step": 12910
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5692529272059992,
      "learning_rate": 5.513781361199702e-06,
      "loss": 0.0899,
      "step": 12911
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.46697104942052464,
      "learning_rate": 5.510674151506426e-06,
      "loss": 0.0759,
      "step": 12912
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6404090958517576,
      "learning_rate": 5.5075677091218695e-06,
      "loss": 0.2138,
      "step": 12913
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9276648794872054,
      "learning_rate": 5.5044620341683475e-06,
      "loss": 0.1999,
      "step": 12914
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8087787065994738,
      "learning_rate": 5.501357126768117e-06,
      "loss": 0.1035,
      "step": 12915
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3859233342549517,
      "learning_rate": 5.498252987043442e-06,
      "loss": 0.0596,
      "step": 12916
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6579246333721871,
      "learning_rate": 5.4951496151165155e-06,
      "loss": 0.1139,
      "step": 12917
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8978475695913879,
      "learning_rate": 5.492047011109535e-06,
      "loss": 0.3297,
      "step": 12918
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3709342348416054,
      "learning_rate": 5.488945175144633e-06,
      "loss": 0.0756,
      "step": 12919
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7687813924495963,
      "learning_rate": 5.485844107343954e-06,
      "loss": 0.1702,
      "step": 12920
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.41597650899054645,
      "learning_rate": 5.482743807829572e-06,
      "loss": 0.0776,
      "step": 12921
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.40923469905999804,
      "learning_rate": 5.4796442767235585e-06,
      "loss": 0.0459,
      "step": 12922
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4238903527917504,
      "learning_rate": 5.476545514147926e-06,
      "loss": 0.0825,
      "step": 12923
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.170967307386524,
      "learning_rate": 5.473447520224703e-06,
      "loss": 0.5713,
      "step": 12924
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4774428007317167,
      "learning_rate": 5.470350295075841e-06,
      "loss": 0.086,
      "step": 12925
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5918647423561461,
      "learning_rate": 5.467253838823289e-06,
      "loss": 0.1262,
      "step": 12926
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4013513586845784,
      "learning_rate": 5.464158151588938e-06,
      "loss": 0.0427,
      "step": 12927
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3054361761869865,
      "learning_rate": 5.461063233494695e-06,
      "loss": 0.0298,
      "step": 12928
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.577201398091539,
      "learning_rate": 5.45796908466239e-06,
      "loss": 0.4239,
      "step": 12929
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.40033197102716883,
      "learning_rate": 5.454875705213855e-06,
      "loss": 0.0565,
      "step": 12930
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4955253664195214,
      "learning_rate": 5.45178309527086e-06,
      "loss": 0.0782,
      "step": 12931
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3956583061953619,
      "learning_rate": 5.4486912549551794e-06,
      "loss": 0.0433,
      "step": 12932
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5495314781979711,
      "learning_rate": 5.445600184388533e-06,
      "loss": 0.1793,
      "step": 12933
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.48408188998879315,
      "learning_rate": 5.4425098836926285e-06,
      "loss": 0.0863,
      "step": 12934
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5474044868507894,
      "learning_rate": 5.4394203529891194e-06,
      "loss": 0.1066,
      "step": 12935
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4126769589466812,
      "learning_rate": 5.436331592399649e-06,
      "loss": 0.0559,
      "step": 12936
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6128376741676655,
      "learning_rate": 5.433243602045834e-06,
      "loss": 0.1204,
      "step": 12937
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4878596356397278,
      "learning_rate": 5.430156382049231e-06,
      "loss": 0.1182,
      "step": 12938
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6135446635172132,
      "learning_rate": 5.4270699325314e-06,
      "loss": 0.135,
      "step": 12939
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.47407241859557137,
      "learning_rate": 5.423984253613851e-06,
      "loss": 0.0933,
      "step": 12940
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8091884946397502,
      "learning_rate": 5.420899345418079e-06,
      "loss": 0.182,
      "step": 12941
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7801875825338305,
      "learning_rate": 5.417815208065524e-06,
      "loss": 0.0853,
      "step": 12942
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.776703773111974,
      "learning_rate": 5.41473184167762e-06,
      "loss": 0.1726,
      "step": 12943
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5881242842043792,
      "learning_rate": 5.4116492463757575e-06,
      "loss": 0.1027,
      "step": 12944
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5426890296816482,
      "learning_rate": 5.408567422281311e-06,
      "loss": 0.0871,
      "step": 12945
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3376368656322054,
      "learning_rate": 5.4054863695156e-06,
      "loss": 0.0415,
      "step": 12946
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4985712139916689,
      "learning_rate": 5.402406088199932e-06,
      "loss": 0.0919,
      "step": 12947
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3932778857899312,
      "learning_rate": 5.39932657845558e-06,
      "loss": 0.0962,
      "step": 12948
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3514210771415406,
      "learning_rate": 5.396247840403798e-06,
      "loss": 0.0619,
      "step": 12949
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0374588364219761,
      "learning_rate": 5.39316987416578e-06,
      "loss": 0.2069,
      "step": 12950
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4265137253969457,
      "learning_rate": 5.390092679862713e-06,
      "loss": 0.0576,
      "step": 12951
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5098027964443856,
      "learning_rate": 5.387016257615754e-06,
      "loss": 0.0746,
      "step": 12952
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3971544456623426,
      "learning_rate": 5.383940607546029e-06,
      "loss": 0.1262,
      "step": 12953
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6291017284831214,
      "learning_rate": 5.380865729774609e-06,
      "loss": 0.0904,
      "step": 12954
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.48042009066169783,
      "learning_rate": 5.377791624422571e-06,
      "loss": 0.079,
      "step": 12955
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9318088564140997,
      "learning_rate": 5.374718291610936e-06,
      "loss": 0.2252,
      "step": 12956
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6187093578845761,
      "learning_rate": 5.371645731460714e-06,
      "loss": 0.1019,
      "step": 12957
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3450758460440964,
      "learning_rate": 5.368573944092861e-06,
      "loss": 0.0595,
      "step": 12958
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8322453856191866,
      "learning_rate": 5.365502929628322e-06,
      "loss": 0.2843,
      "step": 12959
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3695965398365064,
      "learning_rate": 5.3624326881880046e-06,
      "loss": 0.0655,
      "step": 12960
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3473001013332512,
      "learning_rate": 5.359363219892791e-06,
      "loss": 0.0656,
      "step": 12961
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4600175994781669,
      "learning_rate": 5.356294524863517e-06,
      "loss": 0.0603,
      "step": 12962
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5659383707511744,
      "learning_rate": 5.3532266032210035e-06,
      "loss": 0.0909,
      "step": 12963
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.741843241482211,
      "learning_rate": 5.3501594550860425e-06,
      "loss": 0.229,
      "step": 12964
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5433243825916074,
      "learning_rate": 5.34709308057939e-06,
      "loss": 0.0658,
      "step": 12965
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6718208668076926,
      "learning_rate": 5.3440274798217645e-06,
      "loss": 0.1751,
      "step": 12966
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8053514511471045,
      "learning_rate": 5.340962652933862e-06,
      "loss": 0.3155,
      "step": 12967
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4501224298502191,
      "learning_rate": 5.3378986000363565e-06,
      "loss": 0.0797,
      "step": 12968
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.38714343865957773,
      "learning_rate": 5.334835321249864e-06,
      "loss": 0.0619,
      "step": 12969
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1191079057271938,
      "learning_rate": 5.331772816695005e-06,
      "loss": 0.245,
      "step": 12970
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.46336684532153033,
      "learning_rate": 5.328711086492341e-06,
      "loss": 0.0799,
      "step": 12971
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.29931913895267404,
      "learning_rate": 5.325650130762428e-06,
      "loss": 0.0284,
      "step": 12972
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5214735908023918,
      "learning_rate": 5.322589949625761e-06,
      "loss": 0.0825,
      "step": 12973
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6304214133968642,
      "learning_rate": 5.319530543202839e-06,
      "loss": 0.2036,
      "step": 12974
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.48416640145862544,
      "learning_rate": 5.3164719116140905e-06,
      "loss": 0.1147,
      "step": 12975
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.36898289781092325,
      "learning_rate": 5.313414054979962e-06,
      "loss": 0.0604,
      "step": 12976
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.35967480973149724,
      "learning_rate": 5.310356973420824e-06,
      "loss": 0.0617,
      "step": 12977
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.41214941022046697,
      "learning_rate": 5.307300667057049e-06,
      "loss": 0.0594,
      "step": 12978
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6534946212714816,
      "learning_rate": 5.304245136008948e-06,
      "loss": 0.0805,
      "step": 12979
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4410357635419059,
      "learning_rate": 5.301190380396845e-06,
      "loss": 0.1014,
      "step": 12980
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3498427430239184,
      "learning_rate": 5.298136400340986e-06,
      "loss": 0.2316,
      "step": 12981
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.41989250081260593,
      "learning_rate": 5.295083195961625e-06,
      "loss": 0.054,
      "step": 12982
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.49102111534918713,
      "learning_rate": 5.292030767378947e-06,
      "loss": 0.1042,
      "step": 12983
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5656173574805828,
      "learning_rate": 5.288979114713155e-06,
      "loss": 0.1447,
      "step": 12984
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.42312109186346153,
      "learning_rate": 5.285928238084375e-06,
      "loss": 0.0415,
      "step": 12985
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6441894512779821,
      "learning_rate": 5.282878137612737e-06,
      "loss": 0.0932,
      "step": 12986
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6999778737935798,
      "learning_rate": 5.279828813418303e-06,
      "loss": 0.1756,
      "step": 12987
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.39822690800044935,
      "learning_rate": 5.276780265621154e-06,
      "loss": 0.0617,
      "step": 12988
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.34607267514176354,
      "learning_rate": 5.273732494341294e-06,
      "loss": 0.0506,
      "step": 12989
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.48793737540878673,
      "learning_rate": 5.270685499698733e-06,
      "loss": 0.0888,
      "step": 12990
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7894646196017234,
      "learning_rate": 5.267639281813405e-06,
      "loss": 0.1069,
      "step": 12991
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6359411581460421,
      "learning_rate": 5.264593840805276e-06,
      "loss": 0.1388,
      "step": 12992
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5319179532385976,
      "learning_rate": 5.2615491767942255e-06,
      "loss": 0.1343,
      "step": 12993
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8628800248691619,
      "learning_rate": 5.2585052899001366e-06,
      "loss": 0.1708,
      "step": 12994
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5560478338133237,
      "learning_rate": 5.2554621802428345e-06,
      "loss": 0.1623,
      "step": 12995
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.29925513267272974,
      "learning_rate": 5.252419847942139e-06,
      "loss": 0.0388,
      "step": 12996
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.950893118425588,
      "learning_rate": 5.2493782931178265e-06,
      "loss": 0.1808,
      "step": 12997
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.534942457177182,
      "learning_rate": 5.246337515889644e-06,
      "loss": 0.1471,
      "step": 12998
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3878171383124967,
      "learning_rate": 5.243297516377316e-06,
      "loss": 0.0897,
      "step": 12999
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9562189553028325,
      "learning_rate": 5.24025829470052e-06,
      "loss": 0.487,
      "step": 13000
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.27206272496802325,
      "learning_rate": 5.237219850978914e-06,
      "loss": 0.0416,
      "step": 13001
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4972867279869785,
      "learning_rate": 5.234182185332126e-06,
      "loss": 0.1053,
      "step": 13002
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.49400885951473983,
      "learning_rate": 5.231145297879758e-06,
      "loss": 0.1067,
      "step": 13003
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.33968336370077495,
      "learning_rate": 5.228109188741362e-06,
      "loss": 0.0522,
      "step": 13004
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7472636945192347,
      "learning_rate": 5.225073858036483e-06,
      "loss": 0.0598,
      "step": 13005
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5066451394722387,
      "learning_rate": 5.222039305884607e-06,
      "loss": 0.0747,
      "step": 13006
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6508803958311273,
      "learning_rate": 5.219005532405227e-06,
      "loss": 0.1327,
      "step": 13007
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8738404926691703,
      "learning_rate": 5.215972537717773e-06,
      "loss": 0.1211,
      "step": 13008
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4856261784459591,
      "learning_rate": 5.212940321941665e-06,
      "loss": 0.0704,
      "step": 13009
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.30693416952430913,
      "learning_rate": 5.209908885196263e-06,
      "loss": 0.0428,
      "step": 13010
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.41192389473535956,
      "learning_rate": 5.2068782276009444e-06,
      "loss": 0.0642,
      "step": 13011
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6372954876626636,
      "learning_rate": 5.203848349275009e-06,
      "loss": 0.1849,
      "step": 13012
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7247870339721629,
      "learning_rate": 5.200819250337757e-06,
      "loss": 0.1729,
      "step": 13013
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6242721368371852,
      "learning_rate": 5.1977909309084305e-06,
      "loss": 0.1202,
      "step": 13014
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4882020376551192,
      "learning_rate": 5.194763391106278e-06,
      "loss": 0.1017,
      "step": 13015
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.442404562967513,
      "learning_rate": 5.191736631050476e-06,
      "loss": 0.0775,
      "step": 13016
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7854744094674432,
      "learning_rate": 5.18871065086021e-06,
      "loss": 0.155,
      "step": 13017
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.863939103447759,
      "learning_rate": 5.185685450654587e-06,
      "loss": 0.285,
      "step": 13018
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5166208515358045,
      "learning_rate": 5.182661030552743e-06,
      "loss": 0.0552,
      "step": 13019
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6501026338213309,
      "learning_rate": 5.179637390673731e-06,
      "loss": 0.126,
      "step": 13020
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7075578990415727,
      "learning_rate": 5.1766145311366035e-06,
      "loss": 0.1365,
      "step": 13021
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5489704008299193,
      "learning_rate": 5.173592452060358e-06,
      "loss": 0.0926,
      "step": 13022
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5416062023204025,
      "learning_rate": 5.170571153563999e-06,
      "loss": 0.1511,
      "step": 13023
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5621922175000037,
      "learning_rate": 5.167550635766457e-06,
      "loss": 0.0896,
      "step": 13024
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7458016363378273,
      "learning_rate": 5.164530898786671e-06,
      "loss": 0.2406,
      "step": 13025
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4570104461870889,
      "learning_rate": 5.161511942743508e-06,
      "loss": 0.0427,
      "step": 13026
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3554290478258039,
      "learning_rate": 5.1584937677558395e-06,
      "loss": 0.0506,
      "step": 13027
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6977136204823698,
      "learning_rate": 5.155476373942492e-06,
      "loss": 0.1367,
      "step": 13028
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3414861734987261,
      "learning_rate": 5.152459761422266e-06,
      "loss": 0.0437,
      "step": 13029
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6085467170148692,
      "learning_rate": 5.149443930313919e-06,
      "loss": 0.0989,
      "step": 13030
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5592822769026781,
      "learning_rate": 5.1464288807361895e-06,
      "loss": 0.1132,
      "step": 13031
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5597542772293685,
      "learning_rate": 5.143414612807785e-06,
      "loss": 0.0629,
      "step": 13032
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4111351789949758,
      "learning_rate": 5.140401126647381e-06,
      "loss": 0.0593,
      "step": 13033
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6570710642436219,
      "learning_rate": 5.137388422373615e-06,
      "loss": 0.1534,
      "step": 13034
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.49719041685898197,
      "learning_rate": 5.134376500105101e-06,
      "loss": 0.1317,
      "step": 13035
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3699770107843029,
      "learning_rate": 5.131365359960427e-06,
      "loss": 0.0729,
      "step": 13036
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5573888376811251,
      "learning_rate": 5.12835500205813e-06,
      "loss": 0.0881,
      "step": 13037
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6047055471822255,
      "learning_rate": 5.125345426516742e-06,
      "loss": 0.0902,
      "step": 13038
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.42866796134221147,
      "learning_rate": 5.122336633454744e-06,
      "loss": 0.0758,
      "step": 13039
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7522319528898836,
      "learning_rate": 5.119328622990608e-06,
      "loss": 0.1106,
      "step": 13040
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.49318311972233175,
      "learning_rate": 5.1163213952427455e-06,
      "loss": 0.0564,
      "step": 13041
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4298904891256959,
      "learning_rate": 5.113314950329559e-06,
      "loss": 0.0951,
      "step": 13042
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4883686499942589,
      "learning_rate": 5.110309288369414e-06,
      "loss": 0.1037,
      "step": 13043
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.507031474755777,
      "learning_rate": 5.1073044094806535e-06,
      "loss": 0.1553,
      "step": 13044
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5090644200395272,
      "learning_rate": 5.104300313781571e-06,
      "loss": 0.1552,
      "step": 13045
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4587341566439563,
      "learning_rate": 5.101297001390443e-06,
      "loss": 0.0681,
      "step": 13046
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.517057943786207,
      "learning_rate": 5.098294472425511e-06,
      "loss": 0.0773,
      "step": 13047
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4813552831590435,
      "learning_rate": 5.095292727004996e-06,
      "loss": 0.0959,
      "step": 13048
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6701458339257943,
      "learning_rate": 5.092291765247068e-06,
      "loss": 0.1796,
      "step": 13049
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.39038167556070347,
      "learning_rate": 5.089291587269876e-06,
      "loss": 0.1052,
      "step": 13050
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.41550387349117507,
      "learning_rate": 5.086292193191547e-06,
      "loss": 0.0611,
      "step": 13051
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6772778523700943,
      "learning_rate": 5.083293583130169e-06,
      "loss": 0.1956,
      "step": 13052
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7965653311602826,
      "learning_rate": 5.080295757203793e-06,
      "loss": 0.1716,
      "step": 13053
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.36795430603522616,
      "learning_rate": 5.077298715530449e-06,
      "loss": 0.0404,
      "step": 13054
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.219605258683183,
      "learning_rate": 5.074302458228131e-06,
      "loss": 0.3181,
      "step": 13055
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8236670790956423,
      "learning_rate": 5.071306985414814e-06,
      "loss": 0.1489,
      "step": 13056
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5398282618850269,
      "learning_rate": 5.068312297208414e-06,
      "loss": 0.1053,
      "step": 13057
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4503933352335961,
      "learning_rate": 5.065318393726845e-06,
      "loss": 0.1257,
      "step": 13058
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5883169538339287,
      "learning_rate": 5.062325275087979e-06,
      "loss": 0.0831,
      "step": 13059
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.685284351415933,
      "learning_rate": 5.0593329414096615e-06,
      "loss": 0.1701,
      "step": 13060
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5335632219020325,
      "learning_rate": 5.0563413928096885e-06,
      "loss": 0.0682,
      "step": 13061
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1044684763436492,
      "learning_rate": 5.053350629405848e-06,
      "loss": 0.219,
      "step": 13062
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4926740739332688,
      "learning_rate": 5.050360651315897e-06,
      "loss": 0.0645,
      "step": 13063
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6383534493943444,
      "learning_rate": 5.047371458657538e-06,
      "loss": 0.1223,
      "step": 13064
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5323040028612178,
      "learning_rate": 5.044383051548462e-06,
      "loss": 0.0771,
      "step": 13065
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3776095712640841,
      "learning_rate": 5.04139543010633e-06,
      "loss": 0.0539,
      "step": 13066
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8076939725759016,
      "learning_rate": 5.038408594448768e-06,
      "loss": 0.2019,
      "step": 13067
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6391071733321907,
      "learning_rate": 5.0354225446933606e-06,
      "loss": 0.168,
      "step": 13068
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5564217502551148,
      "learning_rate": 5.032437280957675e-06,
      "loss": 0.1133,
      "step": 13069
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.55330772072951,
      "learning_rate": 5.029452803359244e-06,
      "loss": 0.0826,
      "step": 13070
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.34469583113410424,
      "learning_rate": 5.026469112015575e-06,
      "loss": 0.0492,
      "step": 13071
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.610284515265603,
      "learning_rate": 5.023486207044126e-06,
      "loss": 0.1214,
      "step": 13072
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.39477565330102676,
      "learning_rate": 5.020504088562347e-06,
      "loss": 0.0465,
      "step": 13073
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4492959608536208,
      "learning_rate": 5.017522756687626e-06,
      "loss": 0.1079,
      "step": 13074
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5879611078502957,
      "learning_rate": 5.014542211537371e-06,
      "loss": 0.126,
      "step": 13075
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.48632846769088217,
      "learning_rate": 5.011562453228904e-06,
      "loss": 0.0512,
      "step": 13076
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4886349833783207,
      "learning_rate": 5.008583481879556e-06,
      "loss": 0.0749,
      "step": 13077
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5185366969557568,
      "learning_rate": 5.005605297606592e-06,
      "loss": 0.0884,
      "step": 13078
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3947347291090661,
      "learning_rate": 5.0026279005272875e-06,
      "loss": 0.0508,
      "step": 13079
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8541276672723925,
      "learning_rate": 4.999651290758847e-06,
      "loss": 0.144,
      "step": 13080
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3215124759078518,
      "learning_rate": 4.996675468418477e-06,
      "loss": 0.0637,
      "step": 13081
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9189083032870169,
      "learning_rate": 4.993700433623316e-06,
      "loss": 0.3175,
      "step": 13082
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5507726174614646,
      "learning_rate": 4.990726186490521e-06,
      "loss": 0.0917,
      "step": 13083
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9196112365911285,
      "learning_rate": 4.987752727137171e-06,
      "loss": 0.314,
      "step": 13084
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5663955813915253,
      "learning_rate": 4.9847800556803405e-06,
      "loss": 0.0922,
      "step": 13085
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7311634269687153,
      "learning_rate": 4.981808172237054e-06,
      "loss": 0.167,
      "step": 13086
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6926703309029638,
      "learning_rate": 4.978837076924339e-06,
      "loss": 0.168,
      "step": 13087
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.277012804233329,
      "learning_rate": 4.97586676985915e-06,
      "loss": 0.4033,
      "step": 13088
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3806471433954883,
      "learning_rate": 4.972897251158442e-06,
      "loss": 0.0579,
      "step": 13089
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6261920487250625,
      "learning_rate": 4.969928520939113e-06,
      "loss": 0.1247,
      "step": 13090
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6505628101825982,
      "learning_rate": 4.966960579318061e-06,
      "loss": 0.1522,
      "step": 13091
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9534090582750127,
      "learning_rate": 4.9639934264121255e-06,
      "loss": 0.1996,
      "step": 13092
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1284150176233134,
      "learning_rate": 4.961027062338134e-06,
      "loss": 0.2049,
      "step": 13093
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.42391551413145423,
      "learning_rate": 4.958061487212862e-06,
      "loss": 0.0931,
      "step": 13094
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3800476073607,
      "learning_rate": 4.955096701153072e-06,
      "loss": 0.041,
      "step": 13095
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.44772669215223543,
      "learning_rate": 4.952132704275489e-06,
      "loss": 0.085,
      "step": 13096
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5354033340881844,
      "learning_rate": 4.949169496696818e-06,
      "loss": 0.1415,
      "step": 13097
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4466237887460236,
      "learning_rate": 4.946207078533708e-06,
      "loss": 0.0708,
      "step": 13098
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.36218402195506627,
      "learning_rate": 4.943245449902794e-06,
      "loss": 0.0381,
      "step": 13099
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5307229874433925,
      "learning_rate": 4.9402846109206924e-06,
      "loss": 0.0819,
      "step": 13100
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8577140404112388,
      "learning_rate": 4.937324561703952e-06,
      "loss": 0.1493,
      "step": 13101
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5802880097097955,
      "learning_rate": 4.934365302369124e-06,
      "loss": 0.0779,
      "step": 13102
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3946182400298645,
      "learning_rate": 4.931406833032712e-06,
      "loss": 0.078,
      "step": 13103
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7484136882651027,
      "learning_rate": 4.928449153811202e-06,
      "loss": 0.2567,
      "step": 13104
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6366010313078488,
      "learning_rate": 4.92549226482103e-06,
      "loss": 0.1072,
      "step": 13105
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.303640146099669,
      "learning_rate": 4.922536166178612e-06,
      "loss": 0.2347,
      "step": 13106
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5937589790252706,
      "learning_rate": 4.919580858000333e-06,
      "loss": 0.1258,
      "step": 13107
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3669355753940003,
      "learning_rate": 4.916626340402558e-06,
      "loss": 0.0713,
      "step": 13108
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5674203047014554,
      "learning_rate": 4.913672613501588e-06,
      "loss": 0.0851,
      "step": 13109
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4988615377345183,
      "learning_rate": 4.910719677413722e-06,
      "loss": 0.0841,
      "step": 13110
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.2146530103265682,
      "learning_rate": 4.90776753225522e-06,
      "loss": 0.7624,
      "step": 13111
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7341721553915017,
      "learning_rate": 4.904816178142316e-06,
      "loss": 0.1833,
      "step": 13112
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5225810483709492,
      "learning_rate": 4.901865615191195e-06,
      "loss": 0.0683,
      "step": 13113
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6252792971820706,
      "learning_rate": 4.898915843518029e-06,
      "loss": 0.2141,
      "step": 13114
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3280659320217712,
      "learning_rate": 4.895966863238951e-06,
      "loss": 0.0556,
      "step": 13115
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.50014920361633,
      "learning_rate": 4.893018674470074e-06,
      "loss": 0.0493,
      "step": 13116
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4608238902569597,
      "learning_rate": 4.890071277327454e-06,
      "loss": 0.1329,
      "step": 13117
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.642217113624105,
      "learning_rate": 4.88712467192714e-06,
      "loss": 0.1698,
      "step": 13118
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3925834543242093,
      "learning_rate": 4.884178858385141e-06,
      "loss": 0.0722,
      "step": 13119
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8249135481084859,
      "learning_rate": 4.881233836817445e-06,
      "loss": 0.2653,
      "step": 13120
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8232308849658584,
      "learning_rate": 4.878289607339987e-06,
      "loss": 0.1379,
      "step": 13121
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5481642722289359,
      "learning_rate": 4.875346170068684e-06,
      "loss": 0.146,
      "step": 13122
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3989717318123334,
      "learning_rate": 4.872403525119426e-06,
      "loss": 0.0714,
      "step": 13123
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.21232848119050948,
      "learning_rate": 4.869461672608075e-06,
      "loss": 0.0286,
      "step": 13124
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5632643377995626,
      "learning_rate": 4.8665206126504355e-06,
      "loss": 0.075,
      "step": 13125
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5770169637892286,
      "learning_rate": 4.8635803453623095e-06,
      "loss": 0.1582,
      "step": 13126
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3241760836788994,
      "learning_rate": 4.860640870859456e-06,
      "loss": 0.0991,
      "step": 13127
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.553267558196162,
      "learning_rate": 4.857702189257613e-06,
      "loss": 0.1055,
      "step": 13128
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.3301598956135536,
      "learning_rate": 4.854764300672462e-06,
      "loss": 0.4683,
      "step": 13129
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4260191435765234,
      "learning_rate": 4.851827205219681e-06,
      "loss": 0.0842,
      "step": 13130
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5359682283019513,
      "learning_rate": 4.848890903014908e-06,
      "loss": 0.0968,
      "step": 13131
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.35078836439741334,
      "learning_rate": 4.8459553941737345e-06,
      "loss": 0.0341,
      "step": 13132
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7632980505197721,
      "learning_rate": 4.843020678811741e-06,
      "loss": 0.2182,
      "step": 13133
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4923381956201439,
      "learning_rate": 4.840086757044468e-06,
      "loss": 0.0796,
      "step": 13134
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7686110936495048,
      "learning_rate": 4.837153628987437e-06,
      "loss": 0.2599,
      "step": 13135
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.44750121152659256,
      "learning_rate": 4.834221294756108e-06,
      "loss": 0.0566,
      "step": 13136
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5122888629053572,
      "learning_rate": 4.831289754465942e-06,
      "loss": 0.1066,
      "step": 13137
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3326287953989014,
      "learning_rate": 4.828359008232353e-06,
      "loss": 0.0485,
      "step": 13138
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.709829525891044,
      "learning_rate": 4.82542905617073e-06,
      "loss": 0.2356,
      "step": 13139
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.33896067510080724,
      "learning_rate": 4.8224998983964185e-06,
      "loss": 0.0498,
      "step": 13140
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7024093998036088,
      "learning_rate": 4.819571535024747e-06,
      "loss": 0.0674,
      "step": 13141
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.41406557502510166,
      "learning_rate": 4.816643966171005e-06,
      "loss": 0.1011,
      "step": 13142
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3283589817094123,
      "learning_rate": 4.8137171919504645e-06,
      "loss": 0.1052,
      "step": 13143
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.46915316586719114,
      "learning_rate": 4.810791212478339e-06,
      "loss": 0.118,
      "step": 13144
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4381030819257262,
      "learning_rate": 4.807866027869828e-06,
      "loss": 0.0785,
      "step": 13145
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3118623491953876,
      "learning_rate": 4.804941638240107e-06,
      "loss": 0.05,
      "step": 13146
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.39579355854647147,
      "learning_rate": 4.802018043704312e-06,
      "loss": 0.0766,
      "step": 13147
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6088805600108601,
      "learning_rate": 4.799095244377536e-06,
      "loss": 0.1536,
      "step": 13148
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.28206497178418927,
      "learning_rate": 4.796173240374857e-06,
      "loss": 0.0329,
      "step": 13149
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.46146046630979976,
      "learning_rate": 4.793252031811318e-06,
      "loss": 0.1145,
      "step": 13150
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5305306028420663,
      "learning_rate": 4.7903316188019325e-06,
      "loss": 0.0974,
      "step": 13151
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6012427531864902,
      "learning_rate": 4.787412001461672e-06,
      "loss": 0.199,
      "step": 13152
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5102404668618334,
      "learning_rate": 4.784493179905483e-06,
      "loss": 0.0965,
      "step": 13153
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.39184922978498204,
      "learning_rate": 4.781575154248289e-06,
      "loss": 0.0609,
      "step": 13154
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6945927920547779,
      "learning_rate": 4.778657924604976e-06,
      "loss": 0.1426,
      "step": 13155
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4771778760255442,
      "learning_rate": 4.775741491090385e-06,
      "loss": 0.1176,
      "step": 13156
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5813551822925684,
      "learning_rate": 4.772825853819346e-06,
      "loss": 0.1132,
      "step": 13157
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5464679908181806,
      "learning_rate": 4.76991101290665e-06,
      "loss": 0.1157,
      "step": 13158
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.37698007423754526,
      "learning_rate": 4.7669969684670636e-06,
      "loss": 0.0806,
      "step": 13159
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.525434735888283,
      "learning_rate": 4.764083720615301e-06,
      "loss": 0.1171,
      "step": 13160
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4157839289149121,
      "learning_rate": 4.761171269466061e-06,
      "loss": 0.0764,
      "step": 13161
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8169366896631203,
      "learning_rate": 4.758259615134022e-06,
      "loss": 0.1756,
      "step": 13162
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.34659783865927357,
      "learning_rate": 4.755348757733802e-06,
      "loss": 0.0536,
      "step": 13163
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5177685368734902,
      "learning_rate": 4.75243869738001e-06,
      "loss": 0.0911,
      "step": 13164
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.506730674476157,
      "learning_rate": 4.7495294341872195e-06,
      "loss": 0.0675,
      "step": 13165
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.823543349967443,
      "learning_rate": 4.746620968269972e-06,
      "loss": 0.0932,
      "step": 13166
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5119586889179027,
      "learning_rate": 4.743713299742767e-06,
      "loss": 0.0703,
      "step": 13167
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6332264776291453,
      "learning_rate": 4.74080642872009e-06,
      "loss": 0.2049,
      "step": 13168
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.45228508872074386,
      "learning_rate": 4.737900355316374e-06,
      "loss": 0.1003,
      "step": 13169
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8629332168262381,
      "learning_rate": 4.734995079646054e-06,
      "loss": 0.283,
      "step": 13170
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.43550455160487445,
      "learning_rate": 4.732090601823494e-06,
      "loss": 0.0692,
      "step": 13171
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.29534446582242313,
      "learning_rate": 4.729186921963058e-06,
      "loss": 0.0518,
      "step": 13172
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5783093624438336,
      "learning_rate": 4.726284040179049e-06,
      "loss": 0.1096,
      "step": 13173
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5221420468175891,
      "learning_rate": 4.7233819565857815e-06,
      "loss": 0.106,
      "step": 13174
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.49335371418893453,
      "learning_rate": 4.720480671297489e-06,
      "loss": 0.0901,
      "step": 13175
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4898470094381112,
      "learning_rate": 4.717580184428414e-06,
      "loss": 0.0927,
      "step": 13176
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.429401607398349,
      "learning_rate": 4.71468049609273e-06,
      "loss": 0.0669,
      "step": 13177
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.30446220381596567,
      "learning_rate": 4.711781606404625e-06,
      "loss": 0.0541,
      "step": 13178
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6300238950456641,
      "learning_rate": 4.708883515478213e-06,
      "loss": 0.0968,
      "step": 13179
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8626683333568597,
      "learning_rate": 4.705986223427605e-06,
      "loss": 0.3754,
      "step": 13180
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.727119871845014,
      "learning_rate": 4.70308973036685e-06,
      "loss": 0.1426,
      "step": 13181
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7545739700447959,
      "learning_rate": 4.700194036410011e-06,
      "loss": 0.0878,
      "step": 13182
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8426220828520119,
      "learning_rate": 4.697299141671077e-06,
      "loss": 0.1907,
      "step": 13183
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.36170603512568517,
      "learning_rate": 4.6944050462640305e-06,
      "loss": 0.0831,
      "step": 13184
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5084270083587851,
      "learning_rate": 4.6915117503027986e-06,
      "loss": 0.0903,
      "step": 13185
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.39204903856273465,
      "learning_rate": 4.6886192539013165e-06,
      "loss": 0.0592,
      "step": 13186
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.36736341291762337,
      "learning_rate": 4.685727557173444e-06,
      "loss": 0.0746,
      "step": 13187
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7623266312760626,
      "learning_rate": 4.682836660233042e-06,
      "loss": 0.1172,
      "step": 13188
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5469141919285921,
      "learning_rate": 4.67994656319391e-06,
      "loss": 0.0496,
      "step": 13189
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2609038701615742,
      "learning_rate": 4.677057266169857e-06,
      "loss": 0.5101,
      "step": 13190
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6912401741895808,
      "learning_rate": 4.674168769274617e-06,
      "loss": 0.1773,
      "step": 13191
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5359498379908503,
      "learning_rate": 4.671281072621928e-06,
      "loss": 0.1254,
      "step": 13192
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.36445937709872583,
      "learning_rate": 4.668394176325463e-06,
      "loss": 0.0635,
      "step": 13193
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3300462949244186,
      "learning_rate": 4.665508080498892e-06,
      "loss": 0.0329,
      "step": 13194
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4835255555405874,
      "learning_rate": 4.66262278525584e-06,
      "loss": 0.0763,
      "step": 13195
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5608603197902539,
      "learning_rate": 4.659738290709914e-06,
      "loss": 0.1013,
      "step": 13196
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5331410372721224,
      "learning_rate": 4.656854596974658e-06,
      "loss": 0.0874,
      "step": 13197
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2630141297329907,
      "learning_rate": 4.653971704163618e-06,
      "loss": 0.0262,
      "step": 13198
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6150185856295423,
      "learning_rate": 4.651089612390303e-06,
      "loss": 0.1036,
      "step": 13199
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7933046220943075,
      "learning_rate": 4.648208321768163e-06,
      "loss": 0.2038,
      "step": 13200
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.029493749814837,
      "learning_rate": 4.645327832410648e-06,
      "loss": 0.3201,
      "step": 13201
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.887588157172276,
      "learning_rate": 4.6424481444311634e-06,
      "loss": 0.1544,
      "step": 13202
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7273941972822073,
      "learning_rate": 4.639569257943091e-06,
      "loss": 0.1141,
      "step": 13203
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2982632026468084,
      "learning_rate": 4.636691173059765e-06,
      "loss": 0.2424,
      "step": 13204
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4692391013986207,
      "learning_rate": 4.633813889894498e-06,
      "loss": 0.0945,
      "step": 13205
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6024859209656234,
      "learning_rate": 4.630937408560576e-06,
      "loss": 0.0716,
      "step": 13206
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4427288514770383,
      "learning_rate": 4.628061729171254e-06,
      "loss": 0.095,
      "step": 13207
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5202706374175244,
      "learning_rate": 4.625186851839733e-06,
      "loss": 0.0694,
      "step": 13208
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1947338448960283,
      "learning_rate": 4.622312776679208e-06,
      "loss": 0.474,
      "step": 13209
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5433552490941995,
      "learning_rate": 4.619439503802836e-06,
      "loss": 0.1777,
      "step": 13210
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7707521001785532,
      "learning_rate": 4.61656703332374e-06,
      "loss": 0.1342,
      "step": 13211
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4628123334372581,
      "learning_rate": 4.613695365355003e-06,
      "loss": 0.0618,
      "step": 13212
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.34442233065095323,
      "learning_rate": 4.610824500009689e-06,
      "loss": 0.0448,
      "step": 13213
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.29276249249772923,
      "learning_rate": 4.607954437400827e-06,
      "loss": 0.0336,
      "step": 13214
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5376004153430582,
      "learning_rate": 4.605085177641422e-06,
      "loss": 0.1245,
      "step": 13215
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1695080084224925,
      "learning_rate": 4.602216720844421e-06,
      "loss": 0.361,
      "step": 13216
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3499268473791548,
      "learning_rate": 4.5993490671227685e-06,
      "loss": 0.0398,
      "step": 13217
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5241500871361702,
      "learning_rate": 4.596482216589362e-06,
      "loss": 0.0692,
      "step": 13218
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.414916696983325,
      "learning_rate": 4.593616169357079e-06,
      "loss": 0.0579,
      "step": 13219
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3501497566808854,
      "learning_rate": 4.590750925538748e-06,
      "loss": 0.0704,
      "step": 13220
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0988009921902413,
      "learning_rate": 4.587886485247178e-06,
      "loss": 0.3014,
      "step": 13221
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3076818978529452,
      "learning_rate": 4.5850228485951465e-06,
      "loss": 0.0476,
      "step": 13222
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5834087495750455,
      "learning_rate": 4.5821600156954025e-06,
      "loss": 0.1462,
      "step": 13223
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.22131499057819437,
      "learning_rate": 4.579297986660646e-06,
      "loss": 0.0254,
      "step": 13224
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5636770398335458,
      "learning_rate": 4.576436761603564e-06,
      "loss": 0.0932,
      "step": 13225
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5861957503941693,
      "learning_rate": 4.573576340636801e-06,
      "loss": 0.1111,
      "step": 13226
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.41841408690168963,
      "learning_rate": 4.570716723872983e-06,
      "loss": 0.0822,
      "step": 13227
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4709916028186601,
      "learning_rate": 4.5678579114246865e-06,
      "loss": 0.0875,
      "step": 13228
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5993535690319314,
      "learning_rate": 4.564999903404463e-06,
      "loss": 0.1514,
      "step": 13229
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.465507819081353,
      "learning_rate": 4.5621426999248455e-06,
      "loss": 0.129,
      "step": 13230
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.38963649266202566,
      "learning_rate": 4.559286301098312e-06,
      "loss": 0.0547,
      "step": 13231
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.46837756322517593,
      "learning_rate": 4.556430707037326e-06,
      "loss": 0.0861,
      "step": 13232
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.510049279265492,
      "learning_rate": 4.553575917854314e-06,
      "loss": 0.1095,
      "step": 13233
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8573239287525034,
      "learning_rate": 4.550721933661678e-06,
      "loss": 0.1583,
      "step": 13234
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5938155228134494,
      "learning_rate": 4.5478687545717685e-06,
      "loss": 0.2124,
      "step": 13235
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.37170683176008473,
      "learning_rate": 4.545016380696931e-06,
      "loss": 0.041,
      "step": 13236
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3328735093123951,
      "learning_rate": 4.542164812149444e-06,
      "loss": 0.093,
      "step": 13237
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3291965393517869,
      "learning_rate": 4.539314049041604e-06,
      "loss": 0.0513,
      "step": 13238
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4402823505026195,
      "learning_rate": 4.536464091485626e-06,
      "loss": 0.0661,
      "step": 13239
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0923825356322154,
      "learning_rate": 4.5336149395937305e-06,
      "loss": 0.2219,
      "step": 13240
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4034763914525587,
      "learning_rate": 4.530766593478067e-06,
      "loss": 0.0745,
      "step": 13241
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.348026390992197,
      "learning_rate": 4.527919053250806e-06,
      "loss": 0.0578,
      "step": 13242
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5045651772804522,
      "learning_rate": 4.52507231902404e-06,
      "loss": 0.0813,
      "step": 13243
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.44248830815050677,
      "learning_rate": 4.522226390909853e-06,
      "loss": 0.1074,
      "step": 13244
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9117464661011461,
      "learning_rate": 4.519381269020279e-06,
      "loss": 0.252,
      "step": 13245
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3297501416798733,
      "learning_rate": 4.516536953467354e-06,
      "loss": 0.0524,
      "step": 13246
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.369532136608069,
      "learning_rate": 4.5136934443630426e-06,
      "loss": 0.0393,
      "step": 13247
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4423843428042565,
      "learning_rate": 4.5108507418193115e-06,
      "loss": 0.0731,
      "step": 13248
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.589159976465129,
      "learning_rate": 4.508008845948056e-06,
      "loss": 0.1248,
      "step": 13249
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6189516352706523,
      "learning_rate": 4.505167756861189e-06,
      "loss": 0.1434,
      "step": 13250
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7810244753037673,
      "learning_rate": 4.502327474670551e-06,
      "loss": 0.1986,
      "step": 13251
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6009237409422421,
      "learning_rate": 4.499487999487978e-06,
      "loss": 0.0889,
      "step": 13252
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6822502923428717,
      "learning_rate": 4.49664933142524e-06,
      "loss": 0.1434,
      "step": 13253
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5059431542161874,
      "learning_rate": 4.493811470594125e-06,
      "loss": 0.1033,
      "step": 13254
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3483723757333077,
      "learning_rate": 4.490974417106345e-06,
      "loss": 0.0624,
      "step": 13255
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6953368692239917,
      "learning_rate": 4.488138171073603e-06,
      "loss": 0.1327,
      "step": 13256
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.34813499643623874,
      "learning_rate": 4.4853027326075516e-06,
      "loss": 0.0552,
      "step": 13257
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.43438306394209547,
      "learning_rate": 4.482468101819845e-06,
      "loss": 0.0662,
      "step": 13258
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.36939052317063475,
      "learning_rate": 4.479634278822068e-06,
      "loss": 0.0642,
      "step": 13259
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.47976075986965966,
      "learning_rate": 4.476801263725802e-06,
      "loss": 0.0749,
      "step": 13260
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5301126320161897,
      "learning_rate": 4.473969056642574e-06,
      "loss": 0.0914,
      "step": 13261
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3938120549575066,
      "learning_rate": 4.471137657683891e-06,
      "loss": 0.0751,
      "step": 13262
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3965398856082179,
      "learning_rate": 4.468307066961233e-06,
      "loss": 0.0736,
      "step": 13263
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6228813756931638,
      "learning_rate": 4.465477284586042e-06,
      "loss": 0.1071,
      "step": 13264
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5664398689724276,
      "learning_rate": 4.4626483106697316e-06,
      "loss": 0.0868,
      "step": 13265
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5520019799747234,
      "learning_rate": 4.4598201453236705e-06,
      "loss": 0.0709,
      "step": 13266
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5740331517595295,
      "learning_rate": 4.456992788659215e-06,
      "loss": 0.097,
      "step": 13267
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3474835015657418,
      "learning_rate": 4.4541662407876636e-06,
      "loss": 0.0581,
      "step": 13268
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8715631566222652,
      "learning_rate": 4.451340501820328e-06,
      "loss": 0.193,
      "step": 13269
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5906598229365612,
      "learning_rate": 4.448515571868434e-06,
      "loss": 0.1334,
      "step": 13270
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.442563563303682,
      "learning_rate": 4.445691451043218e-06,
      "loss": 0.0472,
      "step": 13271
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3981680815898019,
      "learning_rate": 4.442868139455847e-06,
      "loss": 0.0354,
      "step": 13272
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6859106973272212,
      "learning_rate": 4.440045637217505e-06,
      "loss": 0.2979,
      "step": 13273
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6915726631591427,
      "learning_rate": 4.437223944439292e-06,
      "loss": 0.1402,
      "step": 13274
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.48144394255206885,
      "learning_rate": 4.434403061232318e-06,
      "loss": 0.0806,
      "step": 13275
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.45377313433019,
      "learning_rate": 4.431582987707619e-06,
      "loss": 0.0898,
      "step": 13276
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2880916573167194,
      "learning_rate": 4.428763723976251e-06,
      "loss": 0.0411,
      "step": 13277
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9339481688592265,
      "learning_rate": 4.4259452701491925e-06,
      "loss": 0.2162,
      "step": 13278
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.43929936156324606,
      "learning_rate": 4.4231276263374225e-06,
      "loss": 0.127,
      "step": 13279
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0707249698132357,
      "learning_rate": 4.4203107926518495e-06,
      "loss": 0.2832,
      "step": 13280
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4427776249708332,
      "learning_rate": 4.4174947692034036e-06,
      "loss": 0.0894,
      "step": 13281
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4801746348241215,
      "learning_rate": 4.4146795561029325e-06,
      "loss": 0.0915,
      "step": 13282
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0716787744116827,
      "learning_rate": 4.411865153461287e-06,
      "loss": 0.3774,
      "step": 13283
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.438434603801516,
      "learning_rate": 4.409051561389252e-06,
      "loss": 0.0892,
      "step": 13284
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.39077587298564226,
      "learning_rate": 4.4062387799976265e-06,
      "loss": 0.0636,
      "step": 13285
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.548178999891066,
      "learning_rate": 4.403426809397135e-06,
      "loss": 0.09,
      "step": 13286
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9267114095646881,
      "learning_rate": 4.400615649698495e-06,
      "loss": 0.3004,
      "step": 13287
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.711695976173739,
      "learning_rate": 4.39780530101237e-06,
      "loss": 0.1982,
      "step": 13288
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.31536399546302746,
      "learning_rate": 4.394995763449425e-06,
      "loss": 0.0397,
      "step": 13289
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.36552564289010636,
      "learning_rate": 4.39218703712026e-06,
      "loss": 0.068,
      "step": 13290
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7867594356818101,
      "learning_rate": 4.389379122135464e-06,
      "loss": 0.2716,
      "step": 13291
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5049868162926974,
      "learning_rate": 4.386572018605578e-06,
      "loss": 0.06,
      "step": 13292
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6413636392942337,
      "learning_rate": 4.383765726641126e-06,
      "loss": 0.2322,
      "step": 13293
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9430405938357177,
      "learning_rate": 4.380960246352592e-06,
      "loss": 0.4009,
      "step": 13294
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6429685810475214,
      "learning_rate": 4.378155577850435e-06,
      "loss": 0.1327,
      "step": 13295
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5944422385239605,
      "learning_rate": 4.375351721245069e-06,
      "loss": 0.1542,
      "step": 13296
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3726464502875111,
      "learning_rate": 4.372548676646884e-06,
      "loss": 0.0419,
      "step": 13297
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4645553938954955,
      "learning_rate": 4.369746444166248e-06,
      "loss": 0.0819,
      "step": 13298
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2512800320970357,
      "learning_rate": 4.366945023913472e-06,
      "loss": 0.0219,
      "step": 13299
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8240747408887713,
      "learning_rate": 4.364144415998858e-06,
      "loss": 0.2666,
      "step": 13300
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.45146260233931995,
      "learning_rate": 4.36134462053267e-06,
      "loss": 0.0783,
      "step": 13301
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4256204754297563,
      "learning_rate": 4.358545637625138e-06,
      "loss": 0.0639,
      "step": 13302
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.171900978194544,
      "learning_rate": 4.355747467386456e-06,
      "loss": 0.2821,
      "step": 13303
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8017319562469549,
      "learning_rate": 4.352950109926787e-06,
      "loss": 0.1474,
      "step": 13304
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5199892859789012,
      "learning_rate": 4.350153565356272e-06,
      "loss": 0.1094,
      "step": 13305
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4414490527332686,
      "learning_rate": 4.347357833785015e-06,
      "loss": 0.0815,
      "step": 13306
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3372070223044189,
      "learning_rate": 4.344562915323078e-06,
      "loss": 0.0361,
      "step": 13307
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5187804005022951,
      "learning_rate": 4.3417688100804984e-06,
      "loss": 0.0728,
      "step": 13308
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4432007573587087,
      "learning_rate": 4.338975518167288e-06,
      "loss": 0.0781,
      "step": 13309
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.762508101166838,
      "learning_rate": 4.336183039693425e-06,
      "loss": 0.1427,
      "step": 13310
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4811008264993409,
      "learning_rate": 4.333391374768838e-06,
      "loss": 0.0905,
      "step": 13311
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5504573445593188,
      "learning_rate": 4.330600523503445e-06,
      "loss": 0.0737,
      "step": 13312
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6387039281310161,
      "learning_rate": 4.32781048600712e-06,
      "loss": 0.1414,
      "step": 13313
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3717439439250092,
      "learning_rate": 4.325021262389719e-06,
      "loss": 0.0397,
      "step": 13314
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0574484590697293,
      "learning_rate": 4.322232852761043e-06,
      "loss": 0.2815,
      "step": 13315
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0415217152117335,
      "learning_rate": 4.319445257230878e-06,
      "loss": 0.2936,
      "step": 13316
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4746168652348339,
      "learning_rate": 4.316658475908975e-06,
      "loss": 0.1107,
      "step": 13317
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7707598406203611,
      "learning_rate": 4.3138725089050565e-06,
      "loss": 0.1899,
      "step": 13318
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.50762603848127,
      "learning_rate": 4.3110873563287966e-06,
      "loss": 0.3689,
      "step": 13319
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.821688048522092,
      "learning_rate": 4.308303018289855e-06,
      "loss": 0.139,
      "step": 13320
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7065920211267741,
      "learning_rate": 4.305519494897853e-06,
      "loss": 0.0874,
      "step": 13321
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.47401993370519985,
      "learning_rate": 4.302736786262384e-06,
      "loss": 0.0619,
      "step": 13322
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.43392508323007495,
      "learning_rate": 4.299954892492997e-06,
      "loss": 0.0973,
      "step": 13323
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8052275812230321,
      "learning_rate": 4.297173813699218e-06,
      "loss": 0.3203,
      "step": 13324
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5314435911842442,
      "learning_rate": 4.294393549990547e-06,
      "loss": 0.158,
      "step": 13325
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.46740767423853424,
      "learning_rate": 4.291614101476446e-06,
      "loss": 0.0816,
      "step": 13326
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4441946028405981,
      "learning_rate": 4.288835468266331e-06,
      "loss": 0.0825,
      "step": 13327
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7402436323346954,
      "learning_rate": 4.286057650469608e-06,
      "loss": 0.1091,
      "step": 13328
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7632527065609749,
      "learning_rate": 4.283280648195648e-06,
      "loss": 0.239,
      "step": 13329
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.41090324542511986,
      "learning_rate": 4.280504461553769e-06,
      "loss": 0.0386,
      "step": 13330
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6443384330431349,
      "learning_rate": 4.277729090653277e-06,
      "loss": 0.2052,
      "step": 13331
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4435847214035243,
      "learning_rate": 4.274954535603443e-06,
      "loss": 0.0614,
      "step": 13332
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.947427181465423,
      "learning_rate": 4.272180796513508e-06,
      "loss": 0.2322,
      "step": 13333
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.37275552905255566,
      "learning_rate": 4.269407873492664e-06,
      "loss": 0.073,
      "step": 13334
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7041156610839022,
      "learning_rate": 4.2666357666500955e-06,
      "loss": 0.2077,
      "step": 13335
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5507416785015035,
      "learning_rate": 4.2638644760949255e-06,
      "loss": 0.1043,
      "step": 13336
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3799301122394704,
      "learning_rate": 4.261094001936281e-06,
      "loss": 0.0513,
      "step": 13337
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4324642340855128,
      "learning_rate": 4.258324344283224e-06,
      "loss": 0.0671,
      "step": 13338
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5028727844591191,
      "learning_rate": 4.25555550324481e-06,
      "loss": 0.0514,
      "step": 13339
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7938288666128158,
      "learning_rate": 4.2527874789300285e-06,
      "loss": 0.2548,
      "step": 13340
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4885136950112647,
      "learning_rate": 4.250020271447885e-06,
      "loss": 0.0807,
      "step": 13341
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9578549656953731,
      "learning_rate": 4.2472538809073124e-06,
      "loss": 0.2406,
      "step": 13342
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.48591904130360875,
      "learning_rate": 4.24448830741723e-06,
      "loss": 0.0767,
      "step": 13343
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5448792497348135,
      "learning_rate": 4.241723551086504e-06,
      "loss": 0.1154,
      "step": 13344
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4267360710303933,
      "learning_rate": 4.238959612024015e-06,
      "loss": 0.0538,
      "step": 13345
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6375447184722879,
      "learning_rate": 4.2361964903385565e-06,
      "loss": 0.1446,
      "step": 13346
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.45480313901576824,
      "learning_rate": 4.23343418613893e-06,
      "loss": 0.047,
      "step": 13347
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5286599830697987,
      "learning_rate": 4.230672699533869e-06,
      "loss": 0.192,
      "step": 13348
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3157937485083029,
      "learning_rate": 4.2279120306321225e-06,
      "loss": 0.053,
      "step": 13349
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6546456716792665,
      "learning_rate": 4.22515217954236e-06,
      "loss": 0.2391,
      "step": 13350
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8546692653764526,
      "learning_rate": 4.222393146373249e-06,
      "loss": 0.1709,
      "step": 13351
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3759050239696401,
      "learning_rate": 4.219634931233402e-06,
      "loss": 0.0551,
      "step": 13352
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4145053920827125,
      "learning_rate": 4.216877534231431e-06,
      "loss": 0.0811,
      "step": 13353
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6397845533055858,
      "learning_rate": 4.214120955475878e-06,
      "loss": 0.1176,
      "step": 13354
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3445404297484043,
      "learning_rate": 4.211365195075287e-06,
      "loss": 0.0414,
      "step": 13355
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.38750625590254534,
      "learning_rate": 4.208610253138143e-06,
      "loss": 0.043,
      "step": 13356
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4970416459366156,
      "learning_rate": 4.205856129772912e-06,
      "loss": 0.0878,
      "step": 13357
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.47782015488202606,
      "learning_rate": 4.203102825088029e-06,
      "loss": 0.053,
      "step": 13358
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.374797064769595,
      "learning_rate": 4.200350339191897e-06,
      "loss": 0.0526,
      "step": 13359
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.595797745628366,
      "learning_rate": 4.197598672192873e-06,
      "loss": 0.0916,
      "step": 13360
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3945344072660657,
      "learning_rate": 4.194847824199299e-06,
      "loss": 0.0443,
      "step": 13361
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.42586620365998057,
      "learning_rate": 4.192097795319474e-06,
      "loss": 0.0799,
      "step": 13362
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4037258521296367,
      "learning_rate": 4.189348585661681e-06,
      "loss": 0.0531,
      "step": 13363
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.313868301436046,
      "learning_rate": 4.186600195334139e-06,
      "loss": 0.0543,
      "step": 13364
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5755450797108862,
      "learning_rate": 4.1838526244450675e-06,
      "loss": 0.1616,
      "step": 13365
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3333776794302117,
      "learning_rate": 4.181105873102639e-06,
      "loss": 0.0481,
      "step": 13366
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4719971556654693,
      "learning_rate": 4.1783599414149885e-06,
      "loss": 0.1095,
      "step": 13367
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5632960495453289,
      "learning_rate": 4.175614829490229e-06,
      "loss": 0.0731,
      "step": 13368
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5508371138844084,
      "learning_rate": 4.172870537436438e-06,
      "loss": 0.1072,
      "step": 13369
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6034195671543957,
      "learning_rate": 4.170127065361667e-06,
      "loss": 0.1191,
      "step": 13370
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4182996733942362,
      "learning_rate": 4.1673844133739144e-06,
      "loss": 0.0694,
      "step": 13371
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5802386218618161,
      "learning_rate": 4.164642581581169e-06,
      "loss": 0.0879,
      "step": 13372
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6969383891940344,
      "learning_rate": 4.161901570091378e-06,
      "loss": 0.1117,
      "step": 13373
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6276444004700118,
      "learning_rate": 4.15916137901246e-06,
      "loss": 0.1817,
      "step": 13374
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8714752313621928,
      "learning_rate": 4.15642200845229e-06,
      "loss": 0.3807,
      "step": 13375
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.883083372904027,
      "learning_rate": 4.153683458518723e-06,
      "loss": 0.3683,
      "step": 13376
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.358403108737029,
      "learning_rate": 4.150945729319578e-06,
      "loss": 0.0501,
      "step": 13377
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8751069857696377,
      "learning_rate": 4.1482088209626485e-06,
      "loss": 0.1757,
      "step": 13378
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9118126542658235,
      "learning_rate": 4.145472733555674e-06,
      "loss": 0.229,
      "step": 13379
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.29134636726745805,
      "learning_rate": 4.142737467206387e-06,
      "loss": 0.0247,
      "step": 13380
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4370247368431361,
      "learning_rate": 4.1400030220224704e-06,
      "loss": 0.0945,
      "step": 13381
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5906790132744407,
      "learning_rate": 4.137269398111593e-06,
      "loss": 0.111,
      "step": 13382
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.804677872506211,
      "learning_rate": 4.134536595581365e-06,
      "loss": 0.2352,
      "step": 13383
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9207659852794428,
      "learning_rate": 4.131804614539384e-06,
      "loss": 0.4251,
      "step": 13384
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.706683019387781,
      "learning_rate": 4.129073455093213e-06,
      "loss": 0.2201,
      "step": 13385
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.915051508682364,
      "learning_rate": 4.126343117350381e-06,
      "loss": 0.188,
      "step": 13386
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6802716137612947,
      "learning_rate": 4.1236136014183745e-06,
      "loss": 0.1393,
      "step": 13387
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.36186485981773076,
      "learning_rate": 4.120884907404663e-06,
      "loss": 0.0359,
      "step": 13388
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.44822686654702837,
      "learning_rate": 4.118157035416676e-06,
      "loss": 0.1009,
      "step": 13389
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.45421239313658307,
      "learning_rate": 4.1154299855618194e-06,
      "loss": 0.0803,
      "step": 13390
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5499864868252197,
      "learning_rate": 4.112703757947445e-06,
      "loss": 0.1131,
      "step": 13391
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.571586962440784,
      "learning_rate": 4.109978352680891e-06,
      "loss": 0.1054,
      "step": 13392
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7937207245650628,
      "learning_rate": 4.107253769869462e-06,
      "loss": 0.1972,
      "step": 13393
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5132984321670954,
      "learning_rate": 4.104530009620431e-06,
      "loss": 0.0622,
      "step": 13394
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.37601057365817103,
      "learning_rate": 4.101807072041022e-06,
      "loss": 0.073,
      "step": 13395
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.360544050287961,
      "learning_rate": 4.099084957238447e-06,
      "loss": 0.0587,
      "step": 13396
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5862404716801722,
      "learning_rate": 4.096363665319883e-06,
      "loss": 0.0943,
      "step": 13397
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5512008083125906,
      "learning_rate": 4.093643196392458e-06,
      "loss": 0.1878,
      "step": 13398
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4701755871794399,
      "learning_rate": 4.0909235505632805e-06,
      "loss": 0.0267,
      "step": 13399
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3854330347409188,
      "learning_rate": 4.088204727939429e-06,
      "loss": 0.084,
      "step": 13400
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.627070215401104,
      "learning_rate": 4.085486728627949e-06,
      "loss": 0.1438,
      "step": 13401
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.383037631346011,
      "learning_rate": 4.082769552735841e-06,
      "loss": 0.1218,
      "step": 13402
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4485040607735635,
      "learning_rate": 4.0800532003700845e-06,
      "loss": 0.0662,
      "step": 13403
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5013148566249558,
      "learning_rate": 4.077337671637629e-06,
      "loss": 0.0743,
      "step": 13404
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8799370845893915,
      "learning_rate": 4.074622966645389e-06,
      "loss": 0.3514,
      "step": 13405
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4473568141202583,
      "learning_rate": 4.071909085500234e-06,
      "loss": 0.0422,
      "step": 13406
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.45962320793804556,
      "learning_rate": 4.069196028309017e-06,
      "loss": 0.0862,
      "step": 13407
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.41489471008440376,
      "learning_rate": 4.066483795178549e-06,
      "loss": 0.0979,
      "step": 13408
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.41945003679934506,
      "learning_rate": 4.063772386215628e-06,
      "loss": 0.0581,
      "step": 13409
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5843688787202536,
      "learning_rate": 4.061061801526983e-06,
      "loss": 0.0778,
      "step": 13410
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.38882956632215804,
      "learning_rate": 4.058352041219343e-06,
      "loss": 0.0291,
      "step": 13411
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7305763381471728,
      "learning_rate": 4.05564310539939e-06,
      "loss": 0.1247,
      "step": 13412
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.40870353442016705,
      "learning_rate": 4.052934994173788e-06,
      "loss": 0.053,
      "step": 13413
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.35334806582153766,
      "learning_rate": 4.050227707649138e-06,
      "loss": 0.0506,
      "step": 13414
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6157047660035544,
      "learning_rate": 4.0475212459320375e-06,
      "loss": 0.0948,
      "step": 13415
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4586939890968127,
      "learning_rate": 4.04481560912904e-06,
      "loss": 0.0739,
      "step": 13416
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.41773740205793114,
      "learning_rate": 4.0421107973466795e-06,
      "loss": 0.0799,
      "step": 13417
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8315930383953098,
      "learning_rate": 4.039406810691432e-06,
      "loss": 0.2238,
      "step": 13418
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.49338826488298315,
      "learning_rate": 4.036703649269758e-06,
      "loss": 0.1339,
      "step": 13419
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6187647560866779,
      "learning_rate": 4.034001313188085e-06,
      "loss": 0.1702,
      "step": 13420
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5457765315017586,
      "learning_rate": 4.03129980255281e-06,
      "loss": 0.1783,
      "step": 13421
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.063485320081383,
      "learning_rate": 4.028599117470286e-06,
      "loss": 0.2512,
      "step": 13422
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5421907810035302,
      "learning_rate": 4.025899258046845e-06,
      "loss": 0.0988,
      "step": 13423
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8524613293805282,
      "learning_rate": 4.023200224388787e-06,
      "loss": 0.2779,
      "step": 13424
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4786924671860292,
      "learning_rate": 4.0205020166023645e-06,
      "loss": 0.0581,
      "step": 13425
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4029323636556343,
      "learning_rate": 4.017804634793812e-06,
      "loss": 0.0843,
      "step": 13426
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.39515700564660056,
      "learning_rate": 4.01510807906933e-06,
      "loss": 0.0483,
      "step": 13427
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3969808100259868,
      "learning_rate": 4.012412349535086e-06,
      "loss": 0.0794,
      "step": 13428
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.502442370754547,
      "learning_rate": 4.009717446297203e-06,
      "loss": 0.0665,
      "step": 13429
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.34021113683016724,
      "learning_rate": 4.007023369461787e-06,
      "loss": 0.0663,
      "step": 13430
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3703650076920115,
      "learning_rate": 4.004330119134908e-06,
      "loss": 0.0633,
      "step": 13431
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6163033189973569,
      "learning_rate": 4.001637695422605e-06,
      "loss": 0.1514,
      "step": 13432
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6392855264434135,
      "learning_rate": 3.99894609843087e-06,
      "loss": 0.1465,
      "step": 13433
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6137339978991548,
      "learning_rate": 3.996255328265683e-06,
      "loss": 0.0621,
      "step": 13434
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3600254142720372,
      "learning_rate": 3.993565385032963e-06,
      "loss": 0.0494,
      "step": 13435
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3413135291637228,
      "learning_rate": 3.990876268838642e-06,
      "loss": 0.0565,
      "step": 13436
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8260676132974886,
      "learning_rate": 3.988187979788574e-06,
      "loss": 0.1081,
      "step": 13437
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8485238398295935,
      "learning_rate": 3.985500517988608e-06,
      "loss": 0.1381,
      "step": 13438
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0408187536635312,
      "learning_rate": 3.982813883544537e-06,
      "loss": 0.3735,
      "step": 13439
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2943373585120483,
      "learning_rate": 3.980128076562156e-06,
      "loss": 0.0374,
      "step": 13440
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4712016197049241,
      "learning_rate": 3.977443097147193e-06,
      "loss": 0.1009,
      "step": 13441
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.332953194629631,
      "learning_rate": 3.9747589454053655e-06,
      "loss": 0.0373,
      "step": 13442
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.42067561446795243,
      "learning_rate": 3.972075621442337e-06,
      "loss": 0.0737,
      "step": 13443
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6314780957654518,
      "learning_rate": 3.969393125363771e-06,
      "loss": 0.1347,
      "step": 13444
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4310716503790906,
      "learning_rate": 3.966711457275263e-06,
      "loss": 0.0934,
      "step": 13445
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.626226326101054,
      "learning_rate": 3.964030617282408e-06,
      "loss": 0.1359,
      "step": 13446
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5636710209176756,
      "learning_rate": 3.9613506054907296e-06,
      "loss": 0.0893,
      "step": 13447
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8345826147048007,
      "learning_rate": 3.958671422005769e-06,
      "loss": 0.1803,
      "step": 13448
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.605233844372563,
      "learning_rate": 3.955993066932986e-06,
      "loss": 0.1512,
      "step": 13449
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4291127958401177,
      "learning_rate": 3.953315540377847e-06,
      "loss": 0.0531,
      "step": 13450
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2572454946153612,
      "learning_rate": 3.950638842445744e-06,
      "loss": 0.0357,
      "step": 13451
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.38108303262541393,
      "learning_rate": 3.947962973242089e-06,
      "loss": 0.0531,
      "step": 13452
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7727300993108637,
      "learning_rate": 3.945287932872213e-06,
      "loss": 0.2306,
      "step": 13453
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.40306033221788073,
      "learning_rate": 3.942613721441446e-06,
      "loss": 0.0617,
      "step": 13454
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.32910250979731326,
      "learning_rate": 3.939940339055062e-06,
      "loss": 0.0567,
      "step": 13455
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.40937400774959404,
      "learning_rate": 3.937267785818322e-06,
      "loss": 0.0861,
      "step": 13456
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4676676827937486,
      "learning_rate": 3.93459606183644e-06,
      "loss": 0.062,
      "step": 13457
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6774841184097771,
      "learning_rate": 3.931925167214618e-06,
      "loss": 0.1694,
      "step": 13458
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.331046093895516,
      "learning_rate": 3.929255102057994e-06,
      "loss": 0.0563,
      "step": 13459
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5528375882781852,
      "learning_rate": 3.9265858664716966e-06,
      "loss": 0.0541,
      "step": 13460
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7230155348087174,
      "learning_rate": 3.923917460560822e-06,
      "loss": 0.132,
      "step": 13461
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4894780428162583,
      "learning_rate": 3.921249884430417e-06,
      "loss": 0.0836,
      "step": 13462
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.40377986874896327,
      "learning_rate": 3.918583138185511e-06,
      "loss": 0.0631,
      "step": 13463
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5693157669771765,
      "learning_rate": 3.915917221931096e-06,
      "loss": 0.071,
      "step": 13464
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.48056548446520486,
      "learning_rate": 3.913252135772136e-06,
      "loss": 0.0694,
      "step": 13465
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6011552624838123,
      "learning_rate": 3.910587879813543e-06,
      "loss": 0.0894,
      "step": 13466
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.48859995518352434,
      "learning_rate": 3.9079244541602225e-06,
      "loss": 0.1213,
      "step": 13467
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.49447575399135446,
      "learning_rate": 3.905261858917031e-06,
      "loss": 0.0919,
      "step": 13468
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5856730397896448,
      "learning_rate": 3.902600094188805e-06,
      "loss": 0.0761,
      "step": 13469
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6077777789557322,
      "learning_rate": 3.899939160080326e-06,
      "loss": 0.1424,
      "step": 13470
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5461653826465577,
      "learning_rate": 3.8972790566963655e-06,
      "loss": 0.081,
      "step": 13471
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5379661914759397,
      "learning_rate": 3.894619784141651e-06,
      "loss": 0.1654,
      "step": 13472
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5921769484017512,
      "learning_rate": 3.891961342520889e-06,
      "loss": 0.1189,
      "step": 13473
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5874082292943406,
      "learning_rate": 3.889303731938729e-06,
      "loss": 0.1188,
      "step": 13474
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5850576083722624,
      "learning_rate": 3.8866469524998115e-06,
      "loss": 0.1608,
      "step": 13475
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3041697791355063,
      "learning_rate": 3.883991004308735e-06,
      "loss": 0.0298,
      "step": 13476
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6102635150793781,
      "learning_rate": 3.881335887470069e-06,
      "loss": 0.1609,
      "step": 13477
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4058513786992002,
      "learning_rate": 3.878681602088341e-06,
      "loss": 0.0549,
      "step": 13478
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5503976895425663,
      "learning_rate": 3.876028148268057e-06,
      "loss": 0.1001,
      "step": 13479
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.45171326006614687,
      "learning_rate": 3.8733755261136816e-06,
      "loss": 0.0607,
      "step": 13480
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6355266377726677,
      "learning_rate": 3.870723735729659e-06,
      "loss": 0.1083,
      "step": 13481
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3653399428452109,
      "learning_rate": 3.86807277722038e-06,
      "loss": 0.0502,
      "step": 13482
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3575120485433015,
      "learning_rate": 3.86542265069022e-06,
      "loss": 0.0452,
      "step": 13483
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6454382542727979,
      "learning_rate": 3.862773356243518e-06,
      "loss": 0.1836,
      "step": 13484
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.686546496947017,
      "learning_rate": 3.860124893984582e-06,
      "loss": 0.1176,
      "step": 13485
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4148049259443823,
      "learning_rate": 3.857477264017673e-06,
      "loss": 0.0448,
      "step": 13486
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2990661178338936,
      "learning_rate": 3.854830466447038e-06,
      "loss": 0.5158,
      "step": 13487
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4599969481212714,
      "learning_rate": 3.852184501376882e-06,
      "loss": 0.0554,
      "step": 13488
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4838776707546453,
      "learning_rate": 3.849539368911384e-06,
      "loss": 0.0722,
      "step": 13489
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.42201179900941765,
      "learning_rate": 3.846895069154671e-06,
      "loss": 0.0619,
      "step": 13490
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5133208584178627,
      "learning_rate": 3.8442516022108606e-06,
      "loss": 0.1023,
      "step": 13491
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.51083228039655,
      "learning_rate": 3.8416089681840324e-06,
      "loss": 0.1341,
      "step": 13492
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.485641562976472,
      "learning_rate": 3.8389671671782174e-06,
      "loss": 0.1632,
      "step": 13493
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6839111153767218,
      "learning_rate": 3.8363261992974296e-06,
      "loss": 0.0746,
      "step": 13494
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.44952349862739743,
      "learning_rate": 3.833686064645647e-06,
      "loss": 0.0503,
      "step": 13495
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.613464438546848,
      "learning_rate": 3.831046763326821e-06,
      "loss": 0.3543,
      "step": 13496
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.41077615083353486,
      "learning_rate": 3.828408295444846e-06,
      "loss": 0.0963,
      "step": 13497
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3676495145461748,
      "learning_rate": 3.825770661103612e-06,
      "loss": 0.0434,
      "step": 13498
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4909579853396535,
      "learning_rate": 3.823133860406961e-06,
      "loss": 0.0571,
      "step": 13499
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4008650177747504,
      "learning_rate": 3.820497893458713e-06,
      "loss": 0.0826,
      "step": 13500
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4638843192483846,
      "learning_rate": 3.817862760362637e-06,
      "loss": 0.0807,
      "step": 13501
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4229007278275628,
      "learning_rate": 3.8152284612224894e-06,
      "loss": 0.0787,
      "step": 13502
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4668263911240819,
      "learning_rate": 3.812594996141966e-06,
      "loss": 0.0903,
      "step": 13503
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8410799234867801,
      "learning_rate": 3.8099623652247785e-06,
      "loss": 0.239,
      "step": 13504
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0137451772414943,
      "learning_rate": 3.807330568574549e-06,
      "loss": 0.3524,
      "step": 13505
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5556118427650635,
      "learning_rate": 3.804699606294912e-06,
      "loss": 0.0799,
      "step": 13506
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5520100505968832,
      "learning_rate": 3.8020694784894284e-06,
      "loss": 0.1045,
      "step": 13507
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5499627078054552,
      "learning_rate": 3.799440185261674e-06,
      "loss": 0.0922,
      "step": 13508
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0255212139373875,
      "learning_rate": 3.7968117267151463e-06,
      "loss": 0.2943,
      "step": 13509
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3030246325567507,
      "learning_rate": 3.794184102953344e-06,
      "loss": 0.0467,
      "step": 13510
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.43650379867884276,
      "learning_rate": 3.7915573140796997e-06,
      "loss": 0.117,
      "step": 13511
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6532207207210677,
      "learning_rate": 3.7889313601976535e-06,
      "loss": 0.1437,
      "step": 13512
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7657276867895513,
      "learning_rate": 3.786306241410578e-06,
      "loss": 0.1335,
      "step": 13513
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.48677906621738737,
      "learning_rate": 3.783681957821833e-06,
      "loss": 0.1469,
      "step": 13514
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5145979329257792,
      "learning_rate": 3.781058509534721e-06,
      "loss": 0.0817,
      "step": 13515
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9783487574289417,
      "learning_rate": 3.7784358966525575e-06,
      "loss": 0.0932,
      "step": 13516
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0761594956168907,
      "learning_rate": 3.7758141192785756e-06,
      "loss": 0.2538,
      "step": 13517
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5117285398536668,
      "learning_rate": 3.7731931775160072e-06,
      "loss": 0.0876,
      "step": 13518
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6249714499014525,
      "learning_rate": 3.7705730714680253e-06,
      "loss": 0.1183,
      "step": 13519
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.80849711410945,
      "learning_rate": 3.7679538012378055e-06,
      "loss": 0.3156,
      "step": 13520
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.47797698466723326,
      "learning_rate": 3.7653353669284573e-06,
      "loss": 0.0871,
      "step": 13521
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4150089151649231,
      "learning_rate": 3.7627177686430815e-06,
      "loss": 0.0577,
      "step": 13522
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4207094077925299,
      "learning_rate": 3.7601010064847174e-06,
      "loss": 0.0764,
      "step": 13523
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.38201267176363024,
      "learning_rate": 3.7574850805564025e-06,
      "loss": 0.0564,
      "step": 13524
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.48438236454136313,
      "learning_rate": 3.754869990961121e-06,
      "loss": 0.0887,
      "step": 13525
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.39349644409284285,
      "learning_rate": 3.7522557378018407e-06,
      "loss": 0.0581,
      "step": 13526
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5195364271189371,
      "learning_rate": 3.749642321181476e-06,
      "loss": 0.0814,
      "step": 13527
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6749736975272322,
      "learning_rate": 3.7470297412029204e-06,
      "loss": 0.1488,
      "step": 13528
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.40243994642091513,
      "learning_rate": 3.744417997969041e-06,
      "loss": 0.0606,
      "step": 13529
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7693301373440263,
      "learning_rate": 3.741807091582647e-06,
      "loss": 0.271,
      "step": 13530
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2533385034691213,
      "learning_rate": 3.7391970221465567e-06,
      "loss": 0.2744,
      "step": 13531
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7382235232499551,
      "learning_rate": 3.73658778976351e-06,
      "loss": 0.1147,
      "step": 13532
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5357037317739848,
      "learning_rate": 3.733979394536244e-06,
      "loss": 0.0649,
      "step": 13533
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1530854124684684,
      "learning_rate": 3.7313718365674404e-06,
      "loss": 0.3405,
      "step": 13534
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6378013548990462,
      "learning_rate": 3.728765115959784e-06,
      "loss": 0.1479,
      "step": 13535
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4400857197773881,
      "learning_rate": 3.726159232815882e-06,
      "loss": 0.083,
      "step": 13536
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6149231092891713,
      "learning_rate": 3.723554187238343e-06,
      "loss": 0.1058,
      "step": 13537
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.34728946545894096,
      "learning_rate": 3.7209499793297105e-06,
      "loss": 0.0584,
      "step": 13538
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.48230731845015284,
      "learning_rate": 3.718346609192541e-06,
      "loss": 0.0504,
      "step": 13539
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4432775006164839,
      "learning_rate": 3.7157440769293117e-06,
      "loss": 0.0894,
      "step": 13540
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.41545648122469436,
      "learning_rate": 3.713142382642498e-06,
      "loss": 0.08,
      "step": 13541
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8765694539178209,
      "learning_rate": 3.7105415264345135e-06,
      "loss": 0.3321,
      "step": 13542
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3473896066141387,
      "learning_rate": 3.707941508407775e-06,
      "loss": 0.0606,
      "step": 13543
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7128258522540536,
      "learning_rate": 3.7053423286646377e-06,
      "loss": 0.5422,
      "step": 13544
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4601932716287908,
      "learning_rate": 3.702743987307436e-06,
      "loss": 0.0521,
      "step": 13545
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.41667981682737226,
      "learning_rate": 3.7001464844384577e-06,
      "loss": 0.0522,
      "step": 13546
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6520727658130556,
      "learning_rate": 3.6975498201599872e-06,
      "loss": 0.2024,
      "step": 13547
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.49949816976570854,
      "learning_rate": 3.6949539945742434e-06,
      "loss": 0.0932,
      "step": 13548
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5032554189653984,
      "learning_rate": 3.692359007783433e-06,
      "loss": 0.067,
      "step": 13549
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.41544307989413626,
      "learning_rate": 3.6897648598897106e-06,
      "loss": 0.0846,
      "step": 13550
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5722469679894255,
      "learning_rate": 3.6871715509952277e-06,
      "loss": 0.1065,
      "step": 13551
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.361275905474782,
      "learning_rate": 3.684579081202069e-06,
      "loss": 0.051,
      "step": 13552
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.47588649087949775,
      "learning_rate": 3.6819874506123147e-06,
      "loss": 0.0823,
      "step": 13553
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3026312928684393,
      "learning_rate": 3.6793966593279857e-06,
      "loss": 0.0511,
      "step": 13554
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4874344884799034,
      "learning_rate": 3.6768067074510916e-06,
      "loss": 0.1287,
      "step": 13555
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6064880543340548,
      "learning_rate": 3.6742175950835982e-06,
      "loss": 0.1229,
      "step": 13556
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5024044665931579,
      "learning_rate": 3.6716293223274494e-06,
      "loss": 0.1467,
      "step": 13557
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.43049654242999075,
      "learning_rate": 3.669041889284533e-06,
      "loss": 0.0961,
      "step": 13558
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4313627737266951,
      "learning_rate": 3.666455296056723e-06,
      "loss": 0.0761,
      "step": 13559
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5393091885897453,
      "learning_rate": 3.663869542745865e-06,
      "loss": 0.1105,
      "step": 13560
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.43605707922541886,
      "learning_rate": 3.6612846294537484e-06,
      "loss": 0.0562,
      "step": 13561
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4163840984290676,
      "learning_rate": 3.6587005562821464e-06,
      "loss": 0.0621,
      "step": 13562
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4740891219058328,
      "learning_rate": 3.6561173233328e-06,
      "loss": 0.0766,
      "step": 13563
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8315845234861082,
      "learning_rate": 3.6535349307074166e-06,
      "loss": 0.2026,
      "step": 13564
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.43088041889183015,
      "learning_rate": 3.6509533785076573e-06,
      "loss": 0.0847,
      "step": 13565
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9784473833394586,
      "learning_rate": 3.648372666835162e-06,
      "loss": 0.2693,
      "step": 13566
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9211807720980655,
      "learning_rate": 3.645792795791536e-06,
      "loss": 0.3565,
      "step": 13567
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5579797638126248,
      "learning_rate": 3.6432137654783593e-06,
      "loss": 0.1304,
      "step": 13568
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5188840225965028,
      "learning_rate": 3.640635575997156e-06,
      "loss": 0.1367,
      "step": 13569
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7616768736592006,
      "learning_rate": 3.6380582274494396e-06,
      "loss": 0.2174,
      "step": 13570
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3902273317577968,
      "learning_rate": 3.6354817199366787e-06,
      "loss": 0.0564,
      "step": 13571
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7528255565434991,
      "learning_rate": 3.6329060535603197e-06,
      "loss": 0.1941,
      "step": 13572
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.49081657564354736,
      "learning_rate": 3.630331228421757e-06,
      "loss": 0.1211,
      "step": 13573
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5947478565107776,
      "learning_rate": 3.6277572446223707e-06,
      "loss": 0.1785,
      "step": 13574
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1525542133226394,
      "learning_rate": 3.6251841022634956e-06,
      "loss": 0.3168,
      "step": 13575
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5972099394057832,
      "learning_rate": 3.622611801446449e-06,
      "loss": 0.0798,
      "step": 13576
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.37171595800906837,
      "learning_rate": 3.6200403422724906e-06,
      "loss": 0.0557,
      "step": 13577
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4211364687516788,
      "learning_rate": 3.6174697248428647e-06,
      "loss": 0.0512,
      "step": 13578
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8234819998948745,
      "learning_rate": 3.614899949258779e-06,
      "loss": 0.1919,
      "step": 13579
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5153936745093401,
      "learning_rate": 3.6123310156214173e-06,
      "loss": 0.0977,
      "step": 13580
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.297864262138355,
      "learning_rate": 3.6097629240319064e-06,
      "loss": 0.3604,
      "step": 13581
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6283041682308645,
      "learning_rate": 3.607195674591357e-06,
      "loss": 0.126,
      "step": 13582
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8059000424998063,
      "learning_rate": 3.604629267400844e-06,
      "loss": 0.217,
      "step": 13583
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9272342822134809,
      "learning_rate": 3.6020637025614203e-06,
      "loss": 0.3132,
      "step": 13584
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4507089020527153,
      "learning_rate": 3.5994989801740734e-06,
      "loss": 0.0936,
      "step": 13585
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.47951425609580345,
      "learning_rate": 3.5969351003397933e-06,
      "loss": 0.1048,
      "step": 13586
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4465372681255665,
      "learning_rate": 3.594372063159515e-06,
      "loss": 0.1129,
      "step": 13587
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7443875150542795,
      "learning_rate": 3.5918098687341555e-06,
      "loss": 0.1456,
      "step": 13588
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.38554088851317647,
      "learning_rate": 3.5892485171645785e-06,
      "loss": 0.0544,
      "step": 13589
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8054440591754405,
      "learning_rate": 3.5866880085516306e-06,
      "loss": 0.1428,
      "step": 13590
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2209435000385245,
      "learning_rate": 3.584128342996129e-06,
      "loss": 0.3664,
      "step": 13591
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3604168137405774,
      "learning_rate": 3.5815695205988376e-06,
      "loss": 0.322,
      "step": 13592
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5359578332549004,
      "learning_rate": 3.579011541460503e-06,
      "loss": 0.1199,
      "step": 13593
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.47696913880023656,
      "learning_rate": 3.5764544056818366e-06,
      "loss": 0.0895,
      "step": 13594
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0727439361906614,
      "learning_rate": 3.573898113363522e-06,
      "loss": 0.2695,
      "step": 13595
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6864328299126197,
      "learning_rate": 3.5713426646061865e-06,
      "loss": 0.1031,
      "step": 13596
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.47867281386454447,
      "learning_rate": 3.568788059510453e-06,
      "loss": 0.0799,
      "step": 13597
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.33188141230341067,
      "learning_rate": 3.566234298176882e-06,
      "loss": 0.0544,
      "step": 13598
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6314430215285004,
      "learning_rate": 3.563681380706038e-06,
      "loss": 0.1863,
      "step": 13599
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6647224691585227,
      "learning_rate": 3.561129307198416e-06,
      "loss": 0.1821,
      "step": 13600
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4392908355852,
      "learning_rate": 3.5585780777545015e-06,
      "loss": 0.1016,
      "step": 13601
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.586311293045874,
      "learning_rate": 3.5560276924747255e-06,
      "loss": 0.086,
      "step": 13602
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6392676684115294,
      "learning_rate": 3.5534781514595188e-06,
      "loss": 0.1772,
      "step": 13603
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4907142123228032,
      "learning_rate": 3.550929454809243e-06,
      "loss": 0.0858,
      "step": 13604
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4075884563112669,
      "learning_rate": 3.548381602624251e-06,
      "loss": 0.0553,
      "step": 13605
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3162047611460234,
      "learning_rate": 3.5458345950048376e-06,
      "loss": 0.0307,
      "step": 13606
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6558499851504489,
      "learning_rate": 3.543288432051303e-06,
      "loss": 0.223,
      "step": 13607
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.34464359140441203,
      "learning_rate": 3.540743113863873e-06,
      "loss": 0.0673,
      "step": 13608
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5718861913474795,
      "learning_rate": 3.5381986405427753e-06,
      "loss": 0.1072,
      "step": 13609
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8862883550139543,
      "learning_rate": 3.535655012188166e-06,
      "loss": 0.2225,
      "step": 13610
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6890451755713205,
      "learning_rate": 3.533112228900212e-06,
      "loss": 0.1734,
      "step": 13611
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3675644586714768,
      "learning_rate": 3.5305702907790087e-06,
      "loss": 0.0432,
      "step": 13612
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.39475377882564544,
      "learning_rate": 3.5280291979246456e-06,
      "loss": 0.0828,
      "step": 13613
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.392880096278819,
      "learning_rate": 3.5254889504371502e-06,
      "loss": 0.082,
      "step": 13614
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8632908944407921,
      "learning_rate": 3.5229495484165542e-06,
      "loss": 0.1825,
      "step": 13615
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4364263879214018,
      "learning_rate": 3.520410991962825e-06,
      "loss": 0.0575,
      "step": 13616
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4634556123605624,
      "learning_rate": 3.5178732811759095e-06,
      "loss": 0.1179,
      "step": 13617
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2273080554973368,
      "learning_rate": 3.515336416155712e-06,
      "loss": 0.2591,
      "step": 13618
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.41811566512936904,
      "learning_rate": 3.512800397002128e-06,
      "loss": 0.054,
      "step": 13619
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2210392732472903,
      "learning_rate": 3.510265223814985e-06,
      "loss": 0.3272,
      "step": 13620
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5330125598696515,
      "learning_rate": 3.507730896694106e-06,
      "loss": 0.1029,
      "step": 13621
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.49119973203063877,
      "learning_rate": 3.5051974157392586e-06,
      "loss": 0.0758,
      "step": 13622
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9766494709663239,
      "learning_rate": 3.502664781050194e-06,
      "loss": 0.2415,
      "step": 13623
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2961360601249347,
      "learning_rate": 3.5001329927266264e-06,
      "loss": 0.034,
      "step": 13624
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8178039643718948,
      "learning_rate": 3.4976020508682344e-06,
      "loss": 0.2579,
      "step": 13625
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.42967020372208803,
      "learning_rate": 3.4950719555746553e-06,
      "loss": 0.0742,
      "step": 13626
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.527792657367472,
      "learning_rate": 3.4925427069455036e-06,
      "loss": 0.0489,
      "step": 13627
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5007780248722525,
      "learning_rate": 3.4900143050803662e-06,
      "loss": 0.1213,
      "step": 13628
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7675094830550586,
      "learning_rate": 3.487486750078775e-06,
      "loss": 0.1863,
      "step": 13629
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7187553998946098,
      "learning_rate": 3.48496004204025e-06,
      "loss": 0.1563,
      "step": 13630
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7823515064757545,
      "learning_rate": 3.4824341810642645e-06,
      "loss": 0.2256,
      "step": 13631
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7858404175937499,
      "learning_rate": 3.479909167250275e-06,
      "loss": 0.355,
      "step": 13632
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7271905898188157,
      "learning_rate": 3.4773850006976775e-06,
      "loss": 0.1825,
      "step": 13633
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4554274948384628,
      "learning_rate": 3.474861681505856e-06,
      "loss": 0.0688,
      "step": 13634
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6049799483182445,
      "learning_rate": 3.4723392097741585e-06,
      "loss": 0.0606,
      "step": 13635
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3956149888109491,
      "learning_rate": 3.4698175856019007e-06,
      "loss": 0.0726,
      "step": 13636
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4081446866796106,
      "learning_rate": 3.467296809088347e-06,
      "loss": 0.072,
      "step": 13637
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.434718972367052,
      "learning_rate": 3.4647768803327512e-06,
      "loss": 0.0841,
      "step": 13638
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2700038168093148,
      "learning_rate": 3.462257799434321e-06,
      "loss": 0.4882,
      "step": 13639
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4039980109859438,
      "learning_rate": 3.4597395664922454e-06,
      "loss": 0.055,
      "step": 13640
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.512014370665115,
      "learning_rate": 3.4572221816056544e-06,
      "loss": 0.0675,
      "step": 13641
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4193911023992336,
      "learning_rate": 3.4547056448736652e-06,
      "loss": 0.0909,
      "step": 13642
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.44269525157159884,
      "learning_rate": 3.452189956395355e-06,
      "loss": 0.0883,
      "step": 13643
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2994946436015636,
      "learning_rate": 3.4496751162697738e-06,
      "loss": 0.0476,
      "step": 13644
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7067889913164611,
      "learning_rate": 3.447161124595924e-06,
      "loss": 0.2391,
      "step": 13645
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.293274860682334,
      "learning_rate": 3.4446479814727845e-06,
      "loss": 0.0205,
      "step": 13646
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.44015840847636245,
      "learning_rate": 3.4421356869993037e-06,
      "loss": 0.1319,
      "step": 13647
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6212255250031714,
      "learning_rate": 3.4396242412743974e-06,
      "loss": 0.1644,
      "step": 13648
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9913536147498982,
      "learning_rate": 3.4371136443969303e-06,
      "loss": 0.3451,
      "step": 13649
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8319398555748335,
      "learning_rate": 3.434603896465752e-06,
      "loss": 0.2077,
      "step": 13650
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5415337060728445,
      "learning_rate": 3.432094997579674e-06,
      "loss": 0.0565,
      "step": 13651
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.431452575040957,
      "learning_rate": 3.429586947837479e-06,
      "loss": 0.1083,
      "step": 13652
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6844528731277515,
      "learning_rate": 3.4270797473378997e-06,
      "loss": 0.2168,
      "step": 13653
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.44235007620946704,
      "learning_rate": 3.4245733961796505e-06,
      "loss": 0.0951,
      "step": 13654
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6041079592155015,
      "learning_rate": 3.422067894461409e-06,
      "loss": 0.2018,
      "step": 13655
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.278819927496684,
      "learning_rate": 3.4195632422818258e-06,
      "loss": 0.1917,
      "step": 13656
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4190087470480264,
      "learning_rate": 3.4170594397394977e-06,
      "loss": 0.0741,
      "step": 13657
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4156456146273335,
      "learning_rate": 3.414556486933007e-06,
      "loss": 0.0751,
      "step": 13658
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5209420981266726,
      "learning_rate": 3.4120543839609027e-06,
      "loss": 0.1221,
      "step": 13659
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4379857542781484,
      "learning_rate": 3.409553130921683e-06,
      "loss": 0.0823,
      "step": 13660
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.47674881841446987,
      "learning_rate": 3.4070527279138313e-06,
      "loss": 0.0907,
      "step": 13661
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4763841315099741,
      "learning_rate": 3.404553175035788e-06,
      "loss": 0.0864,
      "step": 13662
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4072630226380383,
      "learning_rate": 3.4020544723859684e-06,
      "loss": 0.0606,
      "step": 13663
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.38016199509370086,
      "learning_rate": 3.3995566200627383e-06,
      "loss": 0.0571,
      "step": 13664
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.45692613431327744,
      "learning_rate": 3.397059618164444e-06,
      "loss": 0.0668,
      "step": 13665
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.45380674785777925,
      "learning_rate": 3.394563466789394e-06,
      "loss": 0.0623,
      "step": 13666
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4646126126726592,
      "learning_rate": 3.3920681660358723e-06,
      "loss": 0.1498,
      "step": 13667
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3628117525509777,
      "learning_rate": 3.3895737160021037e-06,
      "loss": 0.0458,
      "step": 13668
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5838387611794679,
      "learning_rate": 3.387080116786309e-06,
      "loss": 0.1244,
      "step": 13669
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7021471149772954,
      "learning_rate": 3.384587368486658e-06,
      "loss": 0.1175,
      "step": 13670
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5278296908947282,
      "learning_rate": 3.382095471201299e-06,
      "loss": 0.097,
      "step": 13671
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3965023588206317,
      "learning_rate": 3.3796044250283265e-06,
      "loss": 0.0431,
      "step": 13672
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.41318885285512774,
      "learning_rate": 3.3771142300658255e-06,
      "loss": 0.0897,
      "step": 13673
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4984350700626886,
      "learning_rate": 3.374624886411834e-06,
      "loss": 0.0698,
      "step": 13674
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8571691587030649,
      "learning_rate": 3.372136394164363e-06,
      "loss": 0.1027,
      "step": 13675
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4906101908665395,
      "learning_rate": 3.369648753421378e-06,
      "loss": 0.0656,
      "step": 13676
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7959522581533326,
      "learning_rate": 3.367161964280824e-06,
      "loss": 0.137,
      "step": 13677
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.38404704249069344,
      "learning_rate": 3.3646760268406046e-06,
      "loss": 0.0478,
      "step": 13678
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3847285659500392,
      "learning_rate": 3.3621909411986036e-06,
      "loss": 0.0404,
      "step": 13679
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5393324400529373,
      "learning_rate": 3.359706707452645e-06,
      "loss": 0.1427,
      "step": 13680
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5731033141619636,
      "learning_rate": 3.3572233257005454e-06,
      "loss": 0.0984,
      "step": 13681
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4090911284122046,
      "learning_rate": 3.3547407960400706e-06,
      "loss": 0.0509,
      "step": 13682
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3212758501755159,
      "learning_rate": 3.3522591185689705e-06,
      "loss": 0.0569,
      "step": 13683
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.30105073782992287,
      "learning_rate": 3.3497782933849392e-06,
      "loss": 0.0503,
      "step": 13684
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7120903771972484,
      "learning_rate": 3.3472983205856512e-06,
      "loss": 0.313,
      "step": 13685
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5921271126526898,
      "learning_rate": 3.3448192002687453e-06,
      "loss": 0.0615,
      "step": 13686
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5572643912171199,
      "learning_rate": 3.3423409325318346e-06,
      "loss": 0.061,
      "step": 13687
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5121177700229518,
      "learning_rate": 3.3398635174724747e-06,
      "loss": 0.1083,
      "step": 13688
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6515101271969597,
      "learning_rate": 3.3373869551882126e-06,
      "loss": 0.1975,
      "step": 13689
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5800524593354222,
      "learning_rate": 3.334911245776559e-06,
      "loss": 0.1362,
      "step": 13690
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4741252111772554,
      "learning_rate": 3.3324363893349696e-06,
      "loss": 0.1148,
      "step": 13691
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4889062925647463,
      "learning_rate": 3.3299623859608857e-06,
      "loss": 0.071,
      "step": 13692
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5116898608915555,
      "learning_rate": 3.327489235751716e-06,
      "loss": 0.1205,
      "step": 13693
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5373792819943235,
      "learning_rate": 3.3250169388048314e-06,
      "loss": 0.0916,
      "step": 13694
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.31906681753388283,
      "learning_rate": 3.322545495217558e-06,
      "loss": 0.0232,
      "step": 13695
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.38169072046504876,
      "learning_rate": 3.3200749050872115e-06,
      "loss": 0.0483,
      "step": 13696
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.40243282612630993,
      "learning_rate": 3.31760516851104e-06,
      "loss": 0.0584,
      "step": 13697
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.382915565100351,
      "learning_rate": 3.315136285586304e-06,
      "loss": 0.288,
      "step": 13698
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5990722116502121,
      "learning_rate": 3.31266825641019e-06,
      "loss": 0.108,
      "step": 13699
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4819508428731095,
      "learning_rate": 3.3102010810798723e-06,
      "loss": 0.0713,
      "step": 13700
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.45495419321849545,
      "learning_rate": 3.307734759692474e-06,
      "loss": 0.0424,
      "step": 13701
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5109203718433161,
      "learning_rate": 3.3052692923451146e-06,
      "loss": 0.0951,
      "step": 13702
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8062348489811981,
      "learning_rate": 3.302804679134844e-06,
      "loss": 0.2112,
      "step": 13703
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7104487335011151,
      "learning_rate": 3.300340920158709e-06,
      "loss": 0.2239,
      "step": 13704
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6019154934235157,
      "learning_rate": 3.2978780155136948e-06,
      "loss": 0.138,
      "step": 13705
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7615648613029746,
      "learning_rate": 3.295415965296786e-06,
      "loss": 0.1544,
      "step": 13706
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8483797438824621,
      "learning_rate": 3.2929547696049003e-06,
      "loss": 0.1344,
      "step": 13707
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.49599692936924594,
      "learning_rate": 3.2904944285349485e-06,
      "loss": 0.0652,
      "step": 13708
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.660194320941057,
      "learning_rate": 3.2880349421837765e-06,
      "loss": 0.1222,
      "step": 13709
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6595871608799043,
      "learning_rate": 3.2855763106482417e-06,
      "loss": 0.152,
      "step": 13710
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6795395433869488,
      "learning_rate": 3.283118534025126e-06,
      "loss": 0.2007,
      "step": 13711
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6469751580768844,
      "learning_rate": 3.280661612411201e-06,
      "loss": 0.203,
      "step": 13712
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.659895273065248,
      "learning_rate": 3.278205545903182e-06,
      "loss": 0.1465,
      "step": 13713
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.520262235582933,
      "learning_rate": 3.2757503345977906e-06,
      "loss": 0.0702,
      "step": 13714
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0118465209590197,
      "learning_rate": 3.273295978591673e-06,
      "loss": 0.2494,
      "step": 13715
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5187006186760589,
      "learning_rate": 3.270842477981467e-06,
      "loss": 0.127,
      "step": 13716
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.33198150074062704,
      "learning_rate": 3.268389832863761e-06,
      "loss": 0.0349,
      "step": 13717
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2997299105398675,
      "learning_rate": 3.2659380433351235e-06,
      "loss": 0.0393,
      "step": 13718
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5734762224233668,
      "learning_rate": 3.2634871094920778e-06,
      "loss": 0.1054,
      "step": 13719
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4776724335495495,
      "learning_rate": 3.26103703143113e-06,
      "loss": 0.0904,
      "step": 13720
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7280056024056253,
      "learning_rate": 3.258587809248731e-06,
      "loss": 0.0783,
      "step": 13721
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9245876062414681,
      "learning_rate": 3.256139443041309e-06,
      "loss": 0.2668,
      "step": 13722
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.34268825291805466,
      "learning_rate": 3.253691932905259e-06,
      "loss": 0.0441,
      "step": 13723
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6246398470705794,
      "learning_rate": 3.2512452789369514e-06,
      "loss": 0.0825,
      "step": 13724
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6647073635966723,
      "learning_rate": 3.248799481232695e-06,
      "loss": 0.1976,
      "step": 13725
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7245070552594698,
      "learning_rate": 3.2463545398887933e-06,
      "loss": 0.1389,
      "step": 13726
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9518684996714273,
      "learning_rate": 3.243910455001509e-06,
      "loss": 0.1712,
      "step": 13727
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5831768626971079,
      "learning_rate": 3.241467226667058e-06,
      "loss": 0.1082,
      "step": 13728
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6936254752979752,
      "learning_rate": 3.2390248549816345e-06,
      "loss": 0.1592,
      "step": 13729
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6367978092783844,
      "learning_rate": 3.2365833400413994e-06,
      "loss": 0.1177,
      "step": 13730
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4560609196632217,
      "learning_rate": 3.2341426819424785e-06,
      "loss": 0.0676,
      "step": 13731
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.529670610482187,
      "learning_rate": 3.231702880780957e-06,
      "loss": 0.0525,
      "step": 13732
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6073566598800524,
      "learning_rate": 3.2292639366528902e-06,
      "loss": 0.1812,
      "step": 13733
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.35331138597316947,
      "learning_rate": 3.226825849654308e-06,
      "loss": 0.0575,
      "step": 13734
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.43508371354587705,
      "learning_rate": 3.2243886198811996e-06,
      "loss": 0.0875,
      "step": 13735
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5365994391762444,
      "learning_rate": 3.2219522474295134e-06,
      "loss": 0.1181,
      "step": 13736
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.46457855635839734,
      "learning_rate": 3.2195167323951726e-06,
      "loss": 0.1229,
      "step": 13737
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8296295465878251,
      "learning_rate": 3.21708207487407e-06,
      "loss": 0.3061,
      "step": 13738
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0413492551793142,
      "learning_rate": 3.2146482749620622e-06,
      "loss": 0.2436,
      "step": 13739
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2790401280599158,
      "learning_rate": 3.212215332754959e-06,
      "loss": 0.0345,
      "step": 13740
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.38256424330764716,
      "learning_rate": 3.2097832483485523e-06,
      "loss": 0.0628,
      "step": 13741
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3885897749161298,
      "learning_rate": 3.2073520218385972e-06,
      "loss": 0.0755,
      "step": 13742
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.717314731435376,
      "learning_rate": 3.2049216533208164e-06,
      "loss": 0.3382,
      "step": 13743
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.45675818511379,
      "learning_rate": 3.2024921428908838e-06,
      "loss": 0.055,
      "step": 13744
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5411020206544631,
      "learning_rate": 3.2000634906444586e-06,
      "loss": 0.0873,
      "step": 13745
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4763968199009402,
      "learning_rate": 3.1976356966771567e-06,
      "loss": 0.063,
      "step": 13746
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4379597059701367,
      "learning_rate": 3.195208761084567e-06,
      "loss": 0.0988,
      "step": 13747
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.49224406862333403,
      "learning_rate": 3.192782683962231e-06,
      "loss": 0.0898,
      "step": 13748
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6709980838577959,
      "learning_rate": 3.190357465405669e-06,
      "loss": 0.1495,
      "step": 13749
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5357916475254494,
      "learning_rate": 3.187933105510363e-06,
      "loss": 0.082,
      "step": 13750
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5678267158524732,
      "learning_rate": 3.18550960437177e-06,
      "loss": 0.1161,
      "step": 13751
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5818401125283128,
      "learning_rate": 3.183086962085291e-06,
      "loss": 0.1281,
      "step": 13752
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5888191307095528,
      "learning_rate": 3.1806651787463144e-06,
      "loss": 0.1249,
      "step": 13753
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5201560779810429,
      "learning_rate": 3.1782442544501854e-06,
      "loss": 0.1165,
      "step": 13754
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5822452749709303,
      "learning_rate": 3.1758241892922286e-06,
      "loss": 0.1314,
      "step": 13755
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.537166263578851,
      "learning_rate": 3.173404983367706e-06,
      "loss": 0.041,
      "step": 13756
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7329578066471187,
      "learning_rate": 3.170986636771872e-06,
      "loss": 0.2651,
      "step": 13757
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5019386755598375,
      "learning_rate": 3.1685691495999427e-06,
      "loss": 0.0962,
      "step": 13758
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5116932012336934,
      "learning_rate": 3.166152521947088e-06,
      "loss": 0.0723,
      "step": 13759
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5521553412731702,
      "learning_rate": 3.163736753908458e-06,
      "loss": 0.1341,
      "step": 13760
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2714367532635093,
      "learning_rate": 3.1613218455791597e-06,
      "loss": 0.0263,
      "step": 13761
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5334737833010275,
      "learning_rate": 3.158907797054278e-06,
      "loss": 0.0801,
      "step": 13762
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6224671829502946,
      "learning_rate": 3.1564946084288423e-06,
      "loss": 0.0939,
      "step": 13763
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3477476807691328,
      "learning_rate": 3.1540822797978735e-06,
      "loss": 0.0492,
      "step": 13764
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6389856735685049,
      "learning_rate": 3.1516708112563324e-06,
      "loss": 0.1388,
      "step": 13765
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3108149505814186,
      "learning_rate": 3.1492602028991817e-06,
      "loss": 0.0302,
      "step": 13766
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5506961502207995,
      "learning_rate": 3.1468504548213098e-06,
      "loss": 0.1101,
      "step": 13767
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8193715657607479,
      "learning_rate": 3.144441567117601e-06,
      "loss": 0.1528,
      "step": 13768
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4032732029184211,
      "learning_rate": 3.1420335398828827e-06,
      "loss": 0.0437,
      "step": 13769
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.48542325164369615,
      "learning_rate": 3.139626373211979e-06,
      "loss": 0.1136,
      "step": 13770
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4549610654731842,
      "learning_rate": 3.137220067199645e-06,
      "loss": 0.1039,
      "step": 13771
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5469258371307184,
      "learning_rate": 3.1348146219406322e-06,
      "loss": 0.0918,
      "step": 13772
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7596485590140103,
      "learning_rate": 3.132410037529626e-06,
      "loss": 0.1544,
      "step": 13773
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.28887127988726924,
      "learning_rate": 3.1300063140613205e-06,
      "loss": 0.0291,
      "step": 13774
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8932274281517223,
      "learning_rate": 3.1276034516303342e-06,
      "loss": 0.2449,
      "step": 13775
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7665927955893277,
      "learning_rate": 3.1252014503312802e-06,
      "loss": 0.1697,
      "step": 13776
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5217698089636266,
      "learning_rate": 3.122800310258714e-06,
      "loss": 0.1142,
      "step": 13777
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.35228529711725043,
      "learning_rate": 3.120400031507187e-06,
      "loss": 0.065,
      "step": 13778
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4526431329225386,
      "learning_rate": 3.1180006141711874e-06,
      "loss": 0.1102,
      "step": 13779
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1751271140506365,
      "learning_rate": 3.115602058345188e-06,
      "loss": 0.3432,
      "step": 13780
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5502381589942427,
      "learning_rate": 3.113204364123612e-06,
      "loss": 0.0875,
      "step": 13781
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9867591352988387,
      "learning_rate": 3.110807531600876e-06,
      "loss": 0.3469,
      "step": 13782
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7357053544150385,
      "learning_rate": 3.10841156087133e-06,
      "loss": 0.1479,
      "step": 13783
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.501105555099792,
      "learning_rate": 3.1060164520293145e-06,
      "loss": 0.0463,
      "step": 13784
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4103907791807006,
      "learning_rate": 3.1036222051691154e-06,
      "loss": 0.0495,
      "step": 13785
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0392173085766754,
      "learning_rate": 3.101228820385005e-06,
      "loss": 0.3199,
      "step": 13786
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3724776728375238,
      "learning_rate": 3.0988362977712097e-06,
      "loss": 0.0726,
      "step": 13787
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4621365243763148,
      "learning_rate": 3.09644463742193e-06,
      "loss": 0.0754,
      "step": 13788
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.49015393027756876,
      "learning_rate": 3.0940538394313186e-06,
      "loss": 0.0735,
      "step": 13789
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5724560271486219,
      "learning_rate": 3.0916639038935046e-06,
      "loss": 0.0916,
      "step": 13790
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6631536391942241,
      "learning_rate": 3.0892748309025858e-06,
      "loss": 0.1849,
      "step": 13791
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8242494784122838,
      "learning_rate": 3.086886620552626e-06,
      "loss": 0.2938,
      "step": 13792
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6106066845093132,
      "learning_rate": 3.0844992729376356e-06,
      "loss": 0.1081,
      "step": 13793
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6156220717429003,
      "learning_rate": 3.082112788151617e-06,
      "loss": 0.1928,
      "step": 13794
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4829612656426197,
      "learning_rate": 3.079727166288532e-06,
      "loss": 0.0821,
      "step": 13795
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5153356191370541,
      "learning_rate": 3.077342407442288e-06,
      "loss": 0.0748,
      "step": 13796
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7543367639227596,
      "learning_rate": 3.0749585117067937e-06,
      "loss": 0.1523,
      "step": 13797
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0464491910627483,
      "learning_rate": 3.0725754791758933e-06,
      "loss": 0.1343,
      "step": 13798
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4736691953735482,
      "learning_rate": 3.0701933099434143e-06,
      "loss": 0.0729,
      "step": 13799
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.42185260215059417,
      "learning_rate": 3.0678120041031292e-06,
      "loss": 0.0532,
      "step": 13800
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4280404176006453,
      "learning_rate": 3.065431561748816e-06,
      "loss": 0.0988,
      "step": 13801
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3960950621238158,
      "learning_rate": 3.0630519829741773e-06,
      "loss": 0.0671,
      "step": 13802
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4937211468905643,
      "learning_rate": 3.06067326787291e-06,
      "loss": 0.0777,
      "step": 13803
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5970569166853708,
      "learning_rate": 3.058295416538645e-06,
      "loss": 0.113,
      "step": 13804
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3515233781357017,
      "learning_rate": 3.05591842906503e-06,
      "loss": 0.0746,
      "step": 13805
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9987640945629602,
      "learning_rate": 3.0535423055456257e-06,
      "loss": 0.1745,
      "step": 13806
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5330747264517506,
      "learning_rate": 3.051167046073994e-06,
      "loss": 0.0992,
      "step": 13807
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.823397645307644,
      "learning_rate": 3.048792650743634e-06,
      "loss": 0.2375,
      "step": 13808
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4398670767569663,
      "learning_rate": 3.046419119648053e-06,
      "loss": 0.0903,
      "step": 13809
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4674182053126783,
      "learning_rate": 3.044046452880678e-06,
      "loss": 0.0989,
      "step": 13810
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.62600998858311,
      "learning_rate": 3.041674650534934e-06,
      "loss": 0.216,
      "step": 13811
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3949916689958269,
      "learning_rate": 3.039303712704186e-06,
      "loss": 0.0639,
      "step": 13812
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6883919542324309,
      "learning_rate": 3.036933639481801e-06,
      "loss": 0.1734,
      "step": 13813
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5316565601225632,
      "learning_rate": 3.0345644309610733e-06,
      "loss": 0.0817,
      "step": 13814
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.43107651587978435,
      "learning_rate": 3.0321960872352895e-06,
      "loss": 0.0691,
      "step": 13815
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4240923114811931,
      "learning_rate": 3.0298286083976864e-06,
      "loss": 0.0593,
      "step": 13816
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6151321090257482,
      "learning_rate": 3.027461994541475e-06,
      "loss": 0.1198,
      "step": 13817
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8072927578324216,
      "learning_rate": 3.025096245759834e-06,
      "loss": 0.1414,
      "step": 13818
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.37011165150606623,
      "learning_rate": 3.0227313621459056e-06,
      "loss": 0.098,
      "step": 13819
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4032129708966875,
      "learning_rate": 3.0203673437927904e-06,
      "loss": 0.0687,
      "step": 13820
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6074624860405865,
      "learning_rate": 3.0180041907935635e-06,
      "loss": 0.1215,
      "step": 13821
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.607216902003117,
      "learning_rate": 3.0156419032412676e-06,
      "loss": 0.0627,
      "step": 13822
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7506196102065195,
      "learning_rate": 3.0132804812289113e-06,
      "loss": 0.1583,
      "step": 13823
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7149815834836593,
      "learning_rate": 3.010919924849451e-06,
      "loss": 0.1667,
      "step": 13824
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6118657865558096,
      "learning_rate": 3.0085602341958343e-06,
      "loss": 0.145,
      "step": 13825
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0973989917548588,
      "learning_rate": 3.0062014093609674e-06,
      "loss": 0.2128,
      "step": 13826
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7343243468314423,
      "learning_rate": 3.00384345043771e-06,
      "loss": 0.0844,
      "step": 13827
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7738454228595645,
      "learning_rate": 3.0014863575188957e-06,
      "loss": 0.2629,
      "step": 13828
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8478976772499263,
      "learning_rate": 2.9991301306973335e-06,
      "loss": 0.1906,
      "step": 13829
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5073293139835051,
      "learning_rate": 2.996774770065788e-06,
      "loss": 0.0955,
      "step": 13830
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3181044850125395,
      "learning_rate": 2.994420275716983e-06,
      "loss": 0.0532,
      "step": 13831
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.618565727510842,
      "learning_rate": 2.992066647743624e-06,
      "loss": 0.1274,
      "step": 13832
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4113976424574112,
      "learning_rate": 2.989713886238377e-06,
      "loss": 0.0537,
      "step": 13833
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6697243779924084,
      "learning_rate": 2.987361991293869e-06,
      "loss": 0.1549,
      "step": 13834
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6988697329452892,
      "learning_rate": 2.985010963002691e-06,
      "loss": 0.1908,
      "step": 13835
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.369002571410968,
      "learning_rate": 2.98266080145741e-06,
      "loss": 0.0425,
      "step": 13836
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4108213611084724,
      "learning_rate": 2.980311506750552e-06,
      "loss": 0.0686,
      "step": 13837
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6223021354873302,
      "learning_rate": 2.977963078974616e-06,
      "loss": 0.1707,
      "step": 13838
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.312872626864176,
      "learning_rate": 2.9756155182220523e-06,
      "loss": 0.0491,
      "step": 13839
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4141790367473009,
      "learning_rate": 2.9732688245852874e-06,
      "loss": 0.0621,
      "step": 13840
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.605706366897126,
      "learning_rate": 2.9709229981567168e-06,
      "loss": 0.1178,
      "step": 13841
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6188039588268452,
      "learning_rate": 2.9685780390286995e-06,
      "loss": 0.1108,
      "step": 13842
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7553924479523726,
      "learning_rate": 2.9662339472935508e-06,
      "loss": 0.1826,
      "step": 13843
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4829625209027531,
      "learning_rate": 2.963890723043561e-06,
      "loss": 0.0901,
      "step": 13844
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.37491803605300716,
      "learning_rate": 2.9615483663709863e-06,
      "loss": 0.0546,
      "step": 13845
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5581297415335875,
      "learning_rate": 2.959206877368054e-06,
      "loss": 0.0762,
      "step": 13846
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8845318723161776,
      "learning_rate": 2.956866256126936e-06,
      "loss": 0.195,
      "step": 13847
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.41333480237195225,
      "learning_rate": 2.9545265027397906e-06,
      "loss": 0.0577,
      "step": 13848
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.635144327000666,
      "learning_rate": 2.9521876172987377e-06,
      "loss": 0.138,
      "step": 13849
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.411700389974454,
      "learning_rate": 2.9498495998958624e-06,
      "loss": 0.0549,
      "step": 13850
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5442822466879912,
      "learning_rate": 2.9475124506232077e-06,
      "loss": 0.1353,
      "step": 13851
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.28679940858145475,
      "learning_rate": 2.9451761695727912e-06,
      "loss": 0.0603,
      "step": 13852
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.46735440241407483,
      "learning_rate": 2.9428407568366014e-06,
      "loss": 0.1054,
      "step": 13853
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.225185812151069,
      "learning_rate": 2.9405062125065745e-06,
      "loss": 0.3612,
      "step": 13854
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7244246330222815,
      "learning_rate": 2.938172536674627e-06,
      "loss": 0.154,
      "step": 13855
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3675772900547617,
      "learning_rate": 2.935839729432635e-06,
      "loss": 0.0686,
      "step": 13856
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7108338429076564,
      "learning_rate": 2.933507790872456e-06,
      "loss": 0.1687,
      "step": 13857
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7666309668581566,
      "learning_rate": 2.931176721085879e-06,
      "loss": 0.2864,
      "step": 13858
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4906167118508043,
      "learning_rate": 2.9288465201646937e-06,
      "loss": 0.0568,
      "step": 13859
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.45425597328005896,
      "learning_rate": 2.926517188200639e-06,
      "loss": 0.0995,
      "step": 13860
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4940166321196576,
      "learning_rate": 2.924188725285426e-06,
      "loss": 0.0941,
      "step": 13861
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.095860886806033,
      "learning_rate": 2.92186113151072e-06,
      "loss": 0.3662,
      "step": 13862
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4992954556507843,
      "learning_rate": 2.919534406968169e-06,
      "loss": 0.1059,
      "step": 13863
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4512595180730159,
      "learning_rate": 2.917208551749362e-06,
      "loss": 0.0675,
      "step": 13864
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.49573288036663726,
      "learning_rate": 2.9148835659458917e-06,
      "loss": 0.0787,
      "step": 13865
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4598177301176435,
      "learning_rate": 2.912559449649277e-06,
      "loss": 0.1081,
      "step": 13866
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.311755706196296,
      "learning_rate": 2.9102362029510333e-06,
      "loss": 0.034,
      "step": 13867
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4765375044246443,
      "learning_rate": 2.90791382594261e-06,
      "loss": 0.2985,
      "step": 13868
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6265371699792479,
      "learning_rate": 2.905592318715461e-06,
      "loss": 0.1401,
      "step": 13869
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.41648244264853623,
      "learning_rate": 2.9032716813609723e-06,
      "loss": 0.0625,
      "step": 13870
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3855579602690442,
      "learning_rate": 2.9009519139705204e-06,
      "loss": 0.621,
      "step": 13871
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4537744565912038,
      "learning_rate": 2.8986330166354187e-06,
      "loss": 0.0559,
      "step": 13872
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4455634305505474,
      "learning_rate": 2.8963149894469854e-06,
      "loss": 0.0666,
      "step": 13873
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5702697471156503,
      "learning_rate": 2.8939978324964646e-06,
      "loss": 0.1374,
      "step": 13874
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9520905091849582,
      "learning_rate": 2.8916815458750994e-06,
      "loss": 0.2319,
      "step": 13875
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5171801144194119,
      "learning_rate": 2.8893661296740647e-06,
      "loss": 0.1501,
      "step": 13876
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3781088033217413,
      "learning_rate": 2.887051583984543e-06,
      "loss": 0.0752,
      "step": 13877
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.372867722715497,
      "learning_rate": 2.884737908897642e-06,
      "loss": 0.0645,
      "step": 13878
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5592936343489119,
      "learning_rate": 2.882425104504466e-06,
      "loss": 0.1099,
      "step": 13879
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6197143610872833,
      "learning_rate": 2.880113170896051e-06,
      "loss": 0.1492,
      "step": 13880
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5741614004500379,
      "learning_rate": 2.8778021081634467e-06,
      "loss": 0.1011,
      "step": 13881
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6858560065256482,
      "learning_rate": 2.875491916397621e-06,
      "loss": 0.2004,
      "step": 13882
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5819290019505315,
      "learning_rate": 2.8731825956895415e-06,
      "loss": 0.1383,
      "step": 13883
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7980139247219102,
      "learning_rate": 2.8708741461301148e-06,
      "loss": 0.0996,
      "step": 13884
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7166236330582604,
      "learning_rate": 2.8685665678102297e-06,
      "loss": 0.1227,
      "step": 13885
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3710412312860889,
      "learning_rate": 2.866259860820744e-06,
      "loss": 0.0471,
      "step": 13886
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.43348602016940746,
      "learning_rate": 2.8639540252524743e-06,
      "loss": 0.0725,
      "step": 13887
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.516701010103723,
      "learning_rate": 2.861649061196192e-06,
      "loss": 0.0677,
      "step": 13888
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1991824126002432,
      "learning_rate": 2.8593449687426526e-06,
      "loss": 0.0226,
      "step": 13889
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6317121909349196,
      "learning_rate": 2.857041747982578e-06,
      "loss": 0.1184,
      "step": 13890
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.493741764854011,
      "learning_rate": 2.8547393990066283e-06,
      "loss": 0.095,
      "step": 13891
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.49399327760068323,
      "learning_rate": 2.8524379219054624e-06,
      "loss": 0.1233,
      "step": 13892
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5141742593661259,
      "learning_rate": 2.850137316769688e-06,
      "loss": 0.141,
      "step": 13893
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.36597215011328377,
      "learning_rate": 2.8478375836898886e-06,
      "loss": 0.1088,
      "step": 13894
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.51809062134269,
      "learning_rate": 2.8455387227565937e-06,
      "loss": 0.0842,
      "step": 13895
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4221393380132157,
      "learning_rate": 2.843240734060315e-06,
      "loss": 0.0709,
      "step": 13896
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2536778312824468,
      "learning_rate": 2.8409436176915274e-06,
      "loss": 0.3755,
      "step": 13897
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5757597454751089,
      "learning_rate": 2.838647373740677e-06,
      "loss": 0.1105,
      "step": 13898
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6914745236217821,
      "learning_rate": 2.8363520022981567e-06,
      "loss": 0.205,
      "step": 13899
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1410179727090335,
      "learning_rate": 2.834057503454343e-06,
      "loss": 0.3319,
      "step": 13900
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.37450098504756724,
      "learning_rate": 2.8317638772995704e-06,
      "loss": 0.0834,
      "step": 13901
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3803245966924029,
      "learning_rate": 2.829471123924146e-06,
      "loss": 0.0588,
      "step": 13902
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7820957517845232,
      "learning_rate": 2.8271792434183263e-06,
      "loss": 0.1296,
      "step": 13903
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9541737814063996,
      "learning_rate": 2.824888235872353e-06,
      "loss": 0.3224,
      "step": 13904
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.521067725261364,
      "learning_rate": 2.82259810137642e-06,
      "loss": 0.0844,
      "step": 13905
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6739924400891484,
      "learning_rate": 2.8203088400206996e-06,
      "loss": 0.1288,
      "step": 13906
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7409913948462354,
      "learning_rate": 2.818020451895312e-06,
      "loss": 0.1259,
      "step": 13907
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6356281535308591,
      "learning_rate": 2.815732937090354e-06,
      "loss": 0.1563,
      "step": 13908
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4856728848353088,
      "learning_rate": 2.813446295695893e-06,
      "loss": 0.0577,
      "step": 13909
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.39569457904629907,
      "learning_rate": 2.8111605278019538e-06,
      "loss": 0.0617,
      "step": 13910
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.46250234795248174,
      "learning_rate": 2.808875633498523e-06,
      "loss": 0.1572,
      "step": 13911
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6362323555024127,
      "learning_rate": 2.8065916128755615e-06,
      "loss": 0.1549,
      "step": 13912
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5709008987575032,
      "learning_rate": 2.8043084660229923e-06,
      "loss": 0.123,
      "step": 13913
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.41641214067629084,
      "learning_rate": 2.802026193030713e-06,
      "loss": 0.0768,
      "step": 13914
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.48538274833624107,
      "learning_rate": 2.799744793988565e-06,
      "loss": 0.137,
      "step": 13915
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.37495096585059645,
      "learning_rate": 2.7974642689863774e-06,
      "loss": 0.0766,
      "step": 13916
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.44631896955678513,
      "learning_rate": 2.79518461811393e-06,
      "loss": 0.0912,
      "step": 13917
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6906750334370335,
      "learning_rate": 2.7929058414609853e-06,
      "loss": 0.0847,
      "step": 13918
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5641794503741455,
      "learning_rate": 2.790627939117246e-06,
      "loss": 0.1057,
      "step": 13919
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2825285176380991,
      "learning_rate": 2.788350911172405e-06,
      "loss": 0.0283,
      "step": 13920
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1172719703404261,
      "learning_rate": 2.786074757716109e-06,
      "loss": 0.2308,
      "step": 13921
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.70338343710522,
      "learning_rate": 2.7837994788379646e-06,
      "loss": 0.1949,
      "step": 13922
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3733464920794608,
      "learning_rate": 2.7815250746275583e-06,
      "loss": 0.0634,
      "step": 13923
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.35920494280319126,
      "learning_rate": 2.7792515451744326e-06,
      "loss": 0.0907,
      "step": 13924
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4261084700393297,
      "learning_rate": 2.7769788905681047e-06,
      "loss": 0.1954,
      "step": 13925
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6640181497281835,
      "learning_rate": 2.774707110898039e-06,
      "loss": 0.1111,
      "step": 13926
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.27888065628431835,
      "learning_rate": 2.7724362062536815e-06,
      "loss": 0.0238,
      "step": 13927
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.43339000929570926,
      "learning_rate": 2.7701661767244396e-06,
      "loss": 0.0995,
      "step": 13928
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.48970229429049467,
      "learning_rate": 2.767897022399696e-06,
      "loss": 0.0898,
      "step": 13929
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6298067040556693,
      "learning_rate": 2.7656287433687698e-06,
      "loss": 0.1919,
      "step": 13930
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6052219740349121,
      "learning_rate": 2.7633613397209798e-06,
      "loss": 0.1145,
      "step": 13931
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.472477230050974,
      "learning_rate": 2.7610948115455867e-06,
      "loss": 0.0627,
      "step": 13932
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4504569061469071,
      "learning_rate": 2.758829158931833e-06,
      "loss": 0.0533,
      "step": 13933
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.537754667142861,
      "learning_rate": 2.7565643819689124e-06,
      "loss": 0.1565,
      "step": 13934
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8998176566780975,
      "learning_rate": 2.7543004807459943e-06,
      "loss": 0.194,
      "step": 13935
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5108803080970838,
      "learning_rate": 2.752037455352208e-06,
      "loss": 0.0888,
      "step": 13936
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3855087252100445,
      "learning_rate": 2.7497753058766562e-06,
      "loss": 0.0773,
      "step": 13937
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5040742686945191,
      "learning_rate": 2.747514032408391e-06,
      "loss": 0.0738,
      "step": 13938
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4107056171542718,
      "learning_rate": 2.745253635036449e-06,
      "loss": 0.0759,
      "step": 13939
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4174118098005116,
      "learning_rate": 2.742994113849817e-06,
      "loss": 0.0544,
      "step": 13940
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4417693635034541,
      "learning_rate": 2.7407354689374666e-06,
      "loss": 0.0699,
      "step": 13941
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5627081323291134,
      "learning_rate": 2.7384777003883063e-06,
      "loss": 0.1083,
      "step": 13942
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4305188911677063,
      "learning_rate": 2.7362208082912318e-06,
      "loss": 0.0704,
      "step": 13943
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8012257259935496,
      "learning_rate": 2.7339647927351024e-06,
      "loss": 0.1271,
      "step": 13944
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.41212034181972934,
      "learning_rate": 2.731709653808742e-06,
      "loss": 0.0495,
      "step": 13945
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5012425530375098,
      "learning_rate": 2.729455391600924e-06,
      "loss": 0.0788,
      "step": 13946
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3753709615544312,
      "learning_rate": 2.7272020062004107e-06,
      "loss": 0.0541,
      "step": 13947
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7443011117330223,
      "learning_rate": 2.724949497695914e-06,
      "loss": 0.1694,
      "step": 13948
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.40090255521930246,
      "learning_rate": 2.7226978661761272e-06,
      "loss": 0.1329,
      "step": 13949
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6038765705810952,
      "learning_rate": 2.7204471117296847e-06,
      "loss": 0.1394,
      "step": 13950
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.34779299050764695,
      "learning_rate": 2.7181972344452077e-06,
      "loss": 0.0504,
      "step": 13951
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3164295455775958,
      "learning_rate": 2.715948234411278e-06,
      "loss": 0.2723,
      "step": 13952
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7579321199534717,
      "learning_rate": 2.7137001117164328e-06,
      "loss": 0.2571,
      "step": 13953
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4475015688963611,
      "learning_rate": 2.711452866449185e-06,
      "loss": 0.0867,
      "step": 13954
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.48194489769415966,
      "learning_rate": 2.709206498698011e-06,
      "loss": 0.1004,
      "step": 13955
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3145750271517704,
      "learning_rate": 2.7069610085513593e-06,
      "loss": 0.0473,
      "step": 13956
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.36331470927936904,
      "learning_rate": 2.704716396097623e-06,
      "loss": 0.079,
      "step": 13957
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.58609660657277,
      "learning_rate": 2.702472661425187e-06,
      "loss": 0.1024,
      "step": 13958
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3965247745891101,
      "learning_rate": 2.7002298046223703e-06,
      "loss": 0.0694,
      "step": 13959
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3709057646729451,
      "learning_rate": 2.697987825777501e-06,
      "loss": 0.0702,
      "step": 13960
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5132512395427165,
      "learning_rate": 2.695746724978826e-06,
      "loss": 0.0849,
      "step": 13961
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.33567587199671395,
      "learning_rate": 2.6935065023145938e-06,
      "loss": 0.0694,
      "step": 13962
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.34160591224139686,
      "learning_rate": 2.6912671578729896e-06,
      "loss": 0.0615,
      "step": 13963
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3871528392010324,
      "learning_rate": 2.689028691742193e-06,
      "loss": 0.0423,
      "step": 13964
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6695002706346358,
      "learning_rate": 2.6867911040103196e-06,
      "loss": 0.2599,
      "step": 13965
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4575365891468129,
      "learning_rate": 2.684554394765479e-06,
      "loss": 0.0692,
      "step": 13966
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4692865471378182,
      "learning_rate": 2.682318564095715e-06,
      "loss": 0.0465,
      "step": 13967
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5469692670594873,
      "learning_rate": 2.680083612089074e-06,
      "loss": 0.0501,
      "step": 13968
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.421203745999463,
      "learning_rate": 2.6778495388335324e-06,
      "loss": 0.0723,
      "step": 13969
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.560047107826326,
      "learning_rate": 2.6756163444170564e-06,
      "loss": 0.1942,
      "step": 13970
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4177572205129155,
      "learning_rate": 2.673384028927553e-06,
      "loss": 0.0723,
      "step": 13971
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8294565113930286,
      "learning_rate": 2.671152592452933e-06,
      "loss": 0.1282,
      "step": 13972
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2714149973664445,
      "learning_rate": 2.6689220350810314e-06,
      "loss": 0.0227,
      "step": 13973
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8999734969604328,
      "learning_rate": 2.6666923568996803e-06,
      "loss": 0.2007,
      "step": 13974
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5517718158278979,
      "learning_rate": 2.6644635579966464e-06,
      "loss": 0.1478,
      "step": 13975
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5682945150927882,
      "learning_rate": 2.662235638459698e-06,
      "loss": 0.0953,
      "step": 13976
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6245645641048613,
      "learning_rate": 2.660008598376537e-06,
      "loss": 0.1164,
      "step": 13977
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5943791108210419,
      "learning_rate": 2.6577824378348546e-06,
      "loss": 0.135,
      "step": 13978
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0591690798674218,
      "learning_rate": 2.6555571569222783e-06,
      "loss": 0.2236,
      "step": 13979
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4363724448621422,
      "learning_rate": 2.65333275572644e-06,
      "loss": 0.0768,
      "step": 13980
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.352195994283108,
      "learning_rate": 2.651109234334903e-06,
      "loss": 0.0597,
      "step": 13981
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5107747169967807,
      "learning_rate": 2.6488865928352175e-06,
      "loss": 0.0864,
      "step": 13982
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4308995507507879,
      "learning_rate": 2.646664831314877e-06,
      "loss": 0.0685,
      "step": 13983
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4986991867985132,
      "learning_rate": 2.6444439498613666e-06,
      "loss": 0.0607,
      "step": 13984
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.601828464669113,
      "learning_rate": 2.642223948562117e-06,
      "loss": 0.1905,
      "step": 13985
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5353996611504425,
      "learning_rate": 2.6400048275045385e-06,
      "loss": 0.1145,
      "step": 13986
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5853631944627131,
      "learning_rate": 2.637786586775989e-06,
      "loss": 0.0683,
      "step": 13987
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8649216889228948,
      "learning_rate": 2.6355692264638104e-06,
      "loss": 0.2356,
      "step": 13988
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8722618378946438,
      "learning_rate": 2.6333527466553045e-06,
      "loss": 0.2084,
      "step": 13989
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8542088418681013,
      "learning_rate": 2.631137147437723e-06,
      "loss": 0.3155,
      "step": 13990
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5831194558169641,
      "learning_rate": 2.628922428898306e-06,
      "loss": 0.0997,
      "step": 13991
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3374292193740327,
      "learning_rate": 2.6267085911242438e-06,
      "loss": 0.0485,
      "step": 13992
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.44080608989412406,
      "learning_rate": 2.624495634202703e-06,
      "loss": 0.0674,
      "step": 13993
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7792548507362312,
      "learning_rate": 2.6222835582208034e-06,
      "loss": 0.3956,
      "step": 13994
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5020834594291521,
      "learning_rate": 2.6200723632656354e-06,
      "loss": 0.0749,
      "step": 13995
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.528055419292543,
      "learning_rate": 2.617862049424258e-06,
      "loss": 0.0736,
      "step": 13996
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5845945081624878,
      "learning_rate": 2.615652616783698e-06,
      "loss": 0.1108,
      "step": 13997
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9836131767248358,
      "learning_rate": 2.6134440654309313e-06,
      "loss": 0.2448,
      "step": 13998
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.34407923217268754,
      "learning_rate": 2.611236395452915e-06,
      "loss": 0.0682,
      "step": 13999
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5757019744642536,
      "learning_rate": 2.609029606936567e-06,
      "loss": 0.1578,
      "step": 14000
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6130389374252403,
      "learning_rate": 2.6068236999687773e-06,
      "loss": 0.0817,
      "step": 14001
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3534893636671299,
      "learning_rate": 2.6046186746363803e-06,
      "loss": 0.0735,
      "step": 14002
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5270438497957747,
      "learning_rate": 2.6024145310261955e-06,
      "loss": 0.1041,
      "step": 14003
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.49102306887349795,
      "learning_rate": 2.6002112692250056e-06,
      "loss": 0.0749,
      "step": 14004
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5835573290004904,
      "learning_rate": 2.5980088893195532e-06,
      "loss": 0.1272,
      "step": 14005
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6152732671240028,
      "learning_rate": 2.595807391396543e-06,
      "loss": 0.1216,
      "step": 14006
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.34407559782427716,
      "learning_rate": 2.5936067755426505e-06,
      "loss": 0.0429,
      "step": 14007
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6260641084956633,
      "learning_rate": 2.591407041844518e-06,
      "loss": 0.1665,
      "step": 14008
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4470368408322157,
      "learning_rate": 2.589208190388756e-06,
      "loss": 0.1041,
      "step": 14009
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3050225081089652,
      "learning_rate": 2.5870102212619234e-06,
      "loss": 0.0226,
      "step": 14010
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8064419982870301,
      "learning_rate": 2.5848131345505594e-06,
      "loss": 0.2356,
      "step": 14011
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.515547555813911,
      "learning_rate": 2.5826169303411695e-06,
      "loss": 0.08,
      "step": 14012
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.37445235985383596,
      "learning_rate": 2.580421608720221e-06,
      "loss": 0.0644,
      "step": 14013
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.45682291198273284,
      "learning_rate": 2.578227169774136e-06,
      "loss": 0.0776,
      "step": 14014
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.36339202223127587,
      "learning_rate": 2.5760336135893154e-06,
      "loss": 0.0399,
      "step": 14015
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.689684567305236,
      "learning_rate": 2.573840940252123e-06,
      "loss": 0.1015,
      "step": 14016
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.39773226143834567,
      "learning_rate": 2.57164914984889e-06,
      "loss": 0.0538,
      "step": 14017
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4487051777298011,
      "learning_rate": 2.5694582424658995e-06,
      "loss": 0.0617,
      "step": 14018
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.650690245262958,
      "learning_rate": 2.567268218189414e-06,
      "loss": 0.1339,
      "step": 14019
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4347151285874965,
      "learning_rate": 2.565079077105659e-06,
      "loss": 0.0706,
      "step": 14020
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8694461872896229,
      "learning_rate": 2.5628908193008143e-06,
      "loss": 0.1986,
      "step": 14021
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5218402018081384,
      "learning_rate": 2.56070344486104e-06,
      "loss": 0.1239,
      "step": 14022
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2925022068079317,
      "learning_rate": 2.55851695387245e-06,
      "loss": 0.5592,
      "step": 14023
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8991981135604192,
      "learning_rate": 2.5563313464211367e-06,
      "loss": 0.1793,
      "step": 14024
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4373373488103191,
      "learning_rate": 2.5541466225931397e-06,
      "loss": 0.0395,
      "step": 14025
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.44367451995152474,
      "learning_rate": 2.5519627824744792e-06,
      "loss": 0.062,
      "step": 14026
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3707697847296466,
      "learning_rate": 2.5497798261511247e-06,
      "loss": 0.0506,
      "step": 14027
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7023371328769827,
      "learning_rate": 2.547597753709036e-06,
      "loss": 0.1698,
      "step": 14028
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5087835758669218,
      "learning_rate": 2.54541656523411e-06,
      "loss": 0.0736,
      "step": 14029
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5053326457135672,
      "learning_rate": 2.5432362608122316e-06,
      "loss": 0.1158,
      "step": 14030
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.31246997392502274,
      "learning_rate": 2.541056840529227e-06,
      "loss": 0.0459,
      "step": 14031
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5633283877916017,
      "learning_rate": 2.538878304470918e-06,
      "loss": 0.084,
      "step": 14032
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.789582348412604,
      "learning_rate": 2.5367006527230646e-06,
      "loss": 0.1725,
      "step": 14033
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.34001965192685407,
      "learning_rate": 2.5345238853714093e-06,
      "loss": 0.0488,
      "step": 14034
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4319955399484425,
      "learning_rate": 2.532348002501639e-06,
      "loss": 0.042,
      "step": 14035
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5664731362952791,
      "learning_rate": 2.5301730041994405e-06,
      "loss": 0.1142,
      "step": 14036
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8618129980351292,
      "learning_rate": 2.527998890550429e-06,
      "loss": 0.1821,
      "step": 14037
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5891731970683307,
      "learning_rate": 2.52582566164021e-06,
      "loss": 0.0865,
      "step": 14038
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.975423302529996,
      "learning_rate": 2.5236533175543336e-06,
      "loss": 0.2877,
      "step": 14039
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4089527436803613,
      "learning_rate": 2.5214818583783437e-06,
      "loss": 0.3871,
      "step": 14040
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5155377487481176,
      "learning_rate": 2.5193112841977174e-06,
      "loss": 0.087,
      "step": 14041
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.33865873095289045,
      "learning_rate": 2.5171415950979215e-06,
      "loss": 0.0447,
      "step": 14042
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.41115500563774454,
      "learning_rate": 2.514972791164366e-06,
      "loss": 0.0822,
      "step": 14043
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5164239428163478,
      "learning_rate": 2.512804872482452e-06,
      "loss": 0.0898,
      "step": 14044
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.47925103437456223,
      "learning_rate": 2.510637839137525e-06,
      "loss": 0.0935,
      "step": 14045
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9281445297337029,
      "learning_rate": 2.5084716912149085e-06,
      "loss": 0.2659,
      "step": 14046
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3570793535136901,
      "learning_rate": 2.5063064287998706e-06,
      "loss": 0.0463,
      "step": 14047
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6891112952822032,
      "learning_rate": 2.5041420519776785e-06,
      "loss": 0.1499,
      "step": 14048
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7316175247337302,
      "learning_rate": 2.5019785608335288e-06,
      "loss": 0.1074,
      "step": 14049
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6052623215238537,
      "learning_rate": 2.499815955452617e-06,
      "loss": 0.1485,
      "step": 14050
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5276682142401643,
      "learning_rate": 2.497654235920066e-06,
      "loss": 0.0908,
      "step": 14051
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4961399746100399,
      "learning_rate": 2.4954934023209976e-06,
      "loss": 0.066,
      "step": 14052
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6725703201528399,
      "learning_rate": 2.493333454740482e-06,
      "loss": 0.0983,
      "step": 14053
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5208336873277327,
      "learning_rate": 2.491174393263562e-06,
      "loss": 0.1078,
      "step": 14054
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8757713548009758,
      "learning_rate": 2.489016217975235e-06,
      "loss": 0.2833,
      "step": 14055
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7547126501836081,
      "learning_rate": 2.4868589289604726e-06,
      "loss": 0.2388,
      "step": 14056
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5212773305992372,
      "learning_rate": 2.484702526304211e-06,
      "loss": 0.1309,
      "step": 14057
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1778205570648714,
      "learning_rate": 2.482547010091343e-06,
      "loss": 0.28,
      "step": 14058
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6538765880740792,
      "learning_rate": 2.4803923804067375e-06,
      "loss": 0.1495,
      "step": 14059
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.379619464860034,
      "learning_rate": 2.4782386373352257e-06,
      "loss": 0.0701,
      "step": 14060
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.574077226742708,
      "learning_rate": 2.4760857809616013e-06,
      "loss": 0.1041,
      "step": 14061
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5234830715318635,
      "learning_rate": 2.473933811370613e-06,
      "loss": 0.1678,
      "step": 14062
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7280622268962029,
      "learning_rate": 2.4717827286470042e-06,
      "loss": 0.1265,
      "step": 14063
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.398315485414732,
      "learning_rate": 2.4696325328754487e-06,
      "loss": 0.0582,
      "step": 14064
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.584655515766442,
      "learning_rate": 2.4674832241406128e-06,
      "loss": 0.0795,
      "step": 14065
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.34930554342984826,
      "learning_rate": 2.465334802527103e-06,
      "loss": 0.0493,
      "step": 14066
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.424924148707339,
      "learning_rate": 2.4631872681195184e-06,
      "loss": 0.071,
      "step": 14067
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7166793558681821,
      "learning_rate": 2.4610406210024e-06,
      "loss": 0.2403,
      "step": 14068
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.47577608421754797,
      "learning_rate": 2.458894861260272e-06,
      "loss": 0.0913,
      "step": 14069
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4113036066489487,
      "learning_rate": 2.456749988977594e-06,
      "loss": 0.0755,
      "step": 14070
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1788812473571475,
      "learning_rate": 2.454606004238838e-06,
      "loss": 0.2097,
      "step": 14071
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5517563297353489,
      "learning_rate": 2.4524629071283946e-06,
      "loss": 0.1243,
      "step": 14072
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9072369742234031,
      "learning_rate": 2.4503206977306515e-06,
      "loss": 0.1428,
      "step": 14073
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.39261861001821347,
      "learning_rate": 2.448179376129936e-06,
      "loss": 0.0776,
      "step": 14074
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4382539116915731,
      "learning_rate": 2.4460389424105666e-06,
      "loss": 0.0886,
      "step": 14075
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.38297173357164427,
      "learning_rate": 2.443899396656804e-06,
      "loss": 0.0849,
      "step": 14076
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0293618631258914,
      "learning_rate": 2.4417607389528944e-06,
      "loss": 0.2536,
      "step": 14077
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6942520559277227,
      "learning_rate": 2.4396229693830208e-06,
      "loss": 0.1329,
      "step": 14078
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5351296824127424,
      "learning_rate": 2.437486088031368e-06,
      "loss": 0.0831,
      "step": 14079
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4518479408452415,
      "learning_rate": 2.435350094982053e-06,
      "loss": 0.0604,
      "step": 14080
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4215969676467933,
      "learning_rate": 2.4332149903191826e-06,
      "loss": 0.0717,
      "step": 14081
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3408162449410903,
      "learning_rate": 2.4310807741268036e-06,
      "loss": 0.0737,
      "step": 14082
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6507239704911029,
      "learning_rate": 2.4289474464889524e-06,
      "loss": 0.1095,
      "step": 14083
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7291761803201618,
      "learning_rate": 2.4268150074896135e-06,
      "loss": 0.1568,
      "step": 14084
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4216966227736295,
      "learning_rate": 2.424683457212754e-06,
      "loss": 0.0527,
      "step": 14085
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7990309360732413,
      "learning_rate": 2.4225527957422777e-06,
      "loss": 0.3895,
      "step": 14086
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.41238522352130996,
      "learning_rate": 2.420423023162077e-06,
      "loss": 0.0608,
      "step": 14087
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.398041721799002,
      "learning_rate": 2.4182941395560126e-06,
      "loss": 0.0382,
      "step": 14088
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.47955135934928445,
      "learning_rate": 2.4161661450078865e-06,
      "loss": 0.0595,
      "step": 14089
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.30563100648287517,
      "learning_rate": 2.414039039601482e-06,
      "loss": 0.0248,
      "step": 14090
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.43926208032810726,
      "learning_rate": 2.4119128234205508e-06,
      "loss": 0.0822,
      "step": 14091
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5861746239930512,
      "learning_rate": 2.4097874965488017e-06,
      "loss": 0.1034,
      "step": 14092
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.41870790400544033,
      "learning_rate": 2.4076630590699062e-06,
      "loss": 0.0788,
      "step": 14093
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2167222785545049,
      "learning_rate": 2.405539511067506e-06,
      "loss": 0.4507,
      "step": 14094
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.24634152659206546,
      "learning_rate": 2.4034168526252093e-06,
      "loss": 0.021,
      "step": 14095
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.475200345432172,
      "learning_rate": 2.401295083826588e-06,
      "loss": 0.0532,
      "step": 14096
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6408660419589373,
      "learning_rate": 2.3991742047551725e-06,
      "loss": 0.186,
      "step": 14097
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5722245505676824,
      "learning_rate": 2.397054215494465e-06,
      "loss": 0.1475,
      "step": 14098
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.49041765300207196,
      "learning_rate": 2.3949351161279334e-06,
      "loss": 0.0629,
      "step": 14099
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.40331576297807575,
      "learning_rate": 2.3928169067390095e-06,
      "loss": 0.0572,
      "step": 14100
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.45449526244278676,
      "learning_rate": 2.3906995874110826e-06,
      "loss": 0.1339,
      "step": 14101
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6669649743464542,
      "learning_rate": 2.388583158227517e-06,
      "loss": 0.1514,
      "step": 14102
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6343000407002085,
      "learning_rate": 2.3864676192716345e-06,
      "loss": 0.172,
      "step": 14103
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7248006182612068,
      "learning_rate": 2.384352970626735e-06,
      "loss": 0.1783,
      "step": 14104
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.40030550278041843,
      "learning_rate": 2.3822392123760638e-06,
      "loss": 0.0511,
      "step": 14105
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.45743785072260373,
      "learning_rate": 2.380126344602843e-06,
      "loss": 0.0748,
      "step": 14106
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.49215727203849424,
      "learning_rate": 2.3780143673902578e-06,
      "loss": 0.1033,
      "step": 14107
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.521180642071697,
      "learning_rate": 2.375903280821468e-06,
      "loss": 0.1064,
      "step": 14108
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5268752609504401,
      "learning_rate": 2.3737930849795724e-06,
      "loss": 0.1189,
      "step": 14109
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.546880208071758,
      "learning_rate": 2.3716837799476586e-06,
      "loss": 0.1453,
      "step": 14110
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7100141781951199,
      "learning_rate": 2.3695753658087738e-06,
      "loss": 0.1307,
      "step": 14111
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0095143693642357,
      "learning_rate": 2.367467842645929e-06,
      "loss": 0.2394,
      "step": 14112
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7251520187424358,
      "learning_rate": 2.3653612105420887e-06,
      "loss": 0.1929,
      "step": 14113
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.47784451017043367,
      "learning_rate": 2.3632554695802002e-06,
      "loss": 0.0826,
      "step": 14114
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.45989672324263553,
      "learning_rate": 2.3611506198431693e-06,
      "loss": 0.0649,
      "step": 14115
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8699516318164006,
      "learning_rate": 2.359046661413866e-06,
      "loss": 0.1616,
      "step": 14116
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6582339774477517,
      "learning_rate": 2.356943594375116e-06,
      "loss": 0.1692,
      "step": 14117
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3720101345696851,
      "learning_rate": 2.354841418809728e-06,
      "loss": 0.0774,
      "step": 14118
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6419036434977529,
      "learning_rate": 2.3527401348004657e-06,
      "loss": 0.1825,
      "step": 14119
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5263691210391089,
      "learning_rate": 2.3506397424300496e-06,
      "loss": 0.0805,
      "step": 14120
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9028120712518009,
      "learning_rate": 2.3485402417811776e-06,
      "loss": 0.0808,
      "step": 14121
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5644472633934564,
      "learning_rate": 2.3464416329365134e-06,
      "loss": 0.1392,
      "step": 14122
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9489594923497723,
      "learning_rate": 2.344343915978678e-06,
      "loss": 0.2379,
      "step": 14123
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.46751635027853794,
      "learning_rate": 2.342247090990257e-06,
      "loss": 0.0876,
      "step": 14124
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.36355789308621,
      "learning_rate": 2.3401511580538134e-06,
      "loss": 0.0828,
      "step": 14125
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.37733702832249655,
      "learning_rate": 2.338056117251844e-06,
      "loss": 0.0572,
      "step": 14126
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.48620955873102945,
      "learning_rate": 2.3359619686668583e-06,
      "loss": 0.0743,
      "step": 14127
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3629499652734554,
      "learning_rate": 2.3338687123812904e-06,
      "loss": 0.0584,
      "step": 14128
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.611762257136319,
      "learning_rate": 2.331776348477557e-06,
      "loss": 0.0912,
      "step": 14129
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7336867730095213,
      "learning_rate": 2.329684877038027e-06,
      "loss": 0.1185,
      "step": 14130
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0530771022635768,
      "learning_rate": 2.3275942981450584e-06,
      "loss": 0.3015,
      "step": 14131
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.41012880644260175,
      "learning_rate": 2.3255046118809464e-06,
      "loss": 0.1118,
      "step": 14132
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5609968595713388,
      "learning_rate": 2.323415818327973e-06,
      "loss": 0.0851,
      "step": 14133
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7156605698179657,
      "learning_rate": 2.3213279175683613e-06,
      "loss": 0.1958,
      "step": 14134
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7374403723291332,
      "learning_rate": 2.319240909684334e-06,
      "loss": 0.8022,
      "step": 14135
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3461599120116898,
      "learning_rate": 2.317154794758039e-06,
      "loss": 0.0396,
      "step": 14136
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.45907143830271946,
      "learning_rate": 2.3150695728716226e-06,
      "loss": 0.069,
      "step": 14137
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6436814492423,
      "learning_rate": 2.3129852441071657e-06,
      "loss": 0.1849,
      "step": 14138
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4473157906523256,
      "learning_rate": 2.3109018085467475e-06,
      "loss": 0.0761,
      "step": 14139
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0509380898294365,
      "learning_rate": 2.30881926627238e-06,
      "loss": 0.3349,
      "step": 14140
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6244474142465908,
      "learning_rate": 2.306737617366067e-06,
      "loss": 0.0832,
      "step": 14141
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6743895131945229,
      "learning_rate": 2.304656861909746e-06,
      "loss": 0.1155,
      "step": 14142
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6177518512338513,
      "learning_rate": 2.3025769999853618e-06,
      "loss": 0.0691,
      "step": 14143
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7634134416405629,
      "learning_rate": 2.3004980316747803e-06,
      "loss": 0.2011,
      "step": 14144
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5717377158208319,
      "learning_rate": 2.2984199570598663e-06,
      "loss": 0.1544,
      "step": 14145
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5042319764941515,
      "learning_rate": 2.296342776222421e-06,
      "loss": 0.1285,
      "step": 14146
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6356729917778494,
      "learning_rate": 2.2942664892442344e-06,
      "loss": 0.1813,
      "step": 14147
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6590291134120472,
      "learning_rate": 2.2921910962070468e-06,
      "loss": 0.1534,
      "step": 14148
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5538905166285771,
      "learning_rate": 2.2901165971925766e-06,
      "loss": 0.1414,
      "step": 14149
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6488159743918457,
      "learning_rate": 2.288042992282485e-06,
      "loss": 0.1532,
      "step": 14150
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5351502596352729,
      "learning_rate": 2.2859702815584195e-06,
      "loss": 0.105,
      "step": 14151
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.434830029363311,
      "learning_rate": 2.2838984651019803e-06,
      "loss": 0.0739,
      "step": 14152
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4680978222498159,
      "learning_rate": 2.2818275429947437e-06,
      "loss": 0.0842,
      "step": 14153
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3723800194070346,
      "learning_rate": 2.279757515318237e-06,
      "loss": 0.062,
      "step": 14154
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5878294933041819,
      "learning_rate": 2.2776883821539557e-06,
      "loss": 0.1274,
      "step": 14155
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5258396541620011,
      "learning_rate": 2.2756201435833734e-06,
      "loss": 0.0763,
      "step": 14156
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6306393821521605,
      "learning_rate": 2.2735527996879086e-06,
      "loss": 0.0996,
      "step": 14157
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.511476516418344,
      "learning_rate": 2.271486350548957e-06,
      "loss": 0.1401,
      "step": 14158
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5534868524255275,
      "learning_rate": 2.269420796247876e-06,
      "loss": 0.0766,
      "step": 14159
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4663907604127815,
      "learning_rate": 2.2673561368659944e-06,
      "loss": 0.0851,
      "step": 14160
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8603473119364137,
      "learning_rate": 2.265292372484587e-06,
      "loss": 0.0758,
      "step": 14161
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.910351252058616,
      "learning_rate": 2.2632295031849134e-06,
      "loss": 0.2191,
      "step": 14162
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.796903632777432,
      "learning_rate": 2.261167529048189e-06,
      "loss": 0.2673,
      "step": 14163
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.33017762947181645,
      "learning_rate": 2.2591064501556012e-06,
      "loss": 0.0516,
      "step": 14164
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7114129009292056,
      "learning_rate": 2.2570462665882835e-06,
      "loss": 0.2413,
      "step": 14165
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.559349800707858,
      "learning_rate": 2.2549869784273565e-06,
      "loss": 0.1088,
      "step": 14166
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7674823098396085,
      "learning_rate": 2.2529285857538914e-06,
      "loss": 0.1597,
      "step": 14167
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3718893606247249,
      "learning_rate": 2.2508710886489345e-06,
      "loss": 0.0593,
      "step": 14168
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5292436689630282,
      "learning_rate": 2.248814487193482e-06,
      "loss": 0.1485,
      "step": 14169
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.38523919919527544,
      "learning_rate": 2.246758781468508e-06,
      "loss": 0.0944,
      "step": 14170
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5558662372593833,
      "learning_rate": 2.2447039715549484e-06,
      "loss": 0.1483,
      "step": 14171
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5241586419331882,
      "learning_rate": 2.2426500575337046e-06,
      "loss": 0.0908,
      "step": 14172
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.47433086920888,
      "learning_rate": 2.240597039485634e-06,
      "loss": 0.0533,
      "step": 14173
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5523052360605785,
      "learning_rate": 2.2385449174915697e-06,
      "loss": 0.0777,
      "step": 14174
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4165030842866483,
      "learning_rate": 2.2364936916323022e-06,
      "loss": 0.0898,
      "step": 14175
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6497905924413669,
      "learning_rate": 2.2344433619886003e-06,
      "loss": 0.1574,
      "step": 14176
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4379006726736463,
      "learning_rate": 2.2323939286411718e-06,
      "loss": 0.0554,
      "step": 14177
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3659552836387052,
      "learning_rate": 2.23034539167071e-06,
      "loss": 0.0484,
      "step": 14178
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5201338235421459,
      "learning_rate": 2.2282977511578708e-06,
      "loss": 0.1003,
      "step": 14179
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3971656581987746,
      "learning_rate": 2.226251007183272e-06,
      "loss": 0.0521,
      "step": 14180
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5495326437849959,
      "learning_rate": 2.2242051598274915e-06,
      "loss": 0.0944,
      "step": 14181
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5684061979224662,
      "learning_rate": 2.222160209171073e-06,
      "loss": 0.1467,
      "step": 14182
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4136302367311614,
      "learning_rate": 2.220116155294533e-06,
      "loss": 0.0629,
      "step": 14183
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3860429650248117,
      "learning_rate": 2.218072998278348e-06,
      "loss": 0.0731,
      "step": 14184
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6250563679673893,
      "learning_rate": 2.2160307382029537e-06,
      "loss": 0.0961,
      "step": 14185
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5098629962574235,
      "learning_rate": 2.213989375148759e-06,
      "loss": 0.0866,
      "step": 14186
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7890573973463237,
      "learning_rate": 2.211948909196135e-06,
      "loss": 0.1954,
      "step": 14187
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7769244152830962,
      "learning_rate": 2.209909340425412e-06,
      "loss": 0.26,
      "step": 14188
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5393803594071062,
      "learning_rate": 2.2078706689168894e-06,
      "loss": 0.0689,
      "step": 14189
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6121005479664444,
      "learning_rate": 2.205832894750834e-06,
      "loss": 0.1507,
      "step": 14190
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5043152130680527,
      "learning_rate": 2.2037960180074756e-06,
      "loss": 0.1039,
      "step": 14191
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6955376578432327,
      "learning_rate": 2.2017600387670035e-06,
      "loss": 0.1548,
      "step": 14192
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.775184967066096,
      "learning_rate": 2.1997249571095833e-06,
      "loss": 0.2551,
      "step": 14193
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5699046458090326,
      "learning_rate": 2.197690773115321e-06,
      "loss": 0.095,
      "step": 14194
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3627689264348227,
      "learning_rate": 2.195657486864322e-06,
      "loss": 0.0444,
      "step": 14195
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.726598437411128,
      "learning_rate": 2.19362509843663e-06,
      "loss": 0.1081,
      "step": 14196
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3632381482922459,
      "learning_rate": 2.1915936079122597e-06,
      "loss": 0.063,
      "step": 14197
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4180280425478823,
      "learning_rate": 2.1895630153711986e-06,
      "loss": 0.1074,
      "step": 14198
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3891922600288311,
      "learning_rate": 2.1875333208933923e-06,
      "loss": 0.0586,
      "step": 14199
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8404668001299018,
      "learning_rate": 2.185504524558743e-06,
      "loss": 0.2565,
      "step": 14200
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5075311906642802,
      "learning_rate": 2.1834766264471314e-06,
      "loss": 0.1139,
      "step": 14201
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.43955929823239187,
      "learning_rate": 2.181449626638396e-06,
      "loss": 0.0595,
      "step": 14202
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.551867553434909,
      "learning_rate": 2.1794235252123486e-06,
      "loss": 0.0628,
      "step": 14203
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.243296338677688,
      "learning_rate": 2.177398322248747e-06,
      "loss": 0.2514,
      "step": 14204
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6995348769750428,
      "learning_rate": 2.175374017827328e-06,
      "loss": 0.1147,
      "step": 14205
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4896302891077376,
      "learning_rate": 2.173350612027794e-06,
      "loss": 0.1263,
      "step": 14206
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8539072433177526,
      "learning_rate": 2.1713281049298094e-06,
      "loss": 0.2926,
      "step": 14207
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.36150158420416845,
      "learning_rate": 2.169306496612994e-06,
      "loss": 0.0625,
      "step": 14208
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8790055617759472,
      "learning_rate": 2.167285787156945e-06,
      "loss": 0.3147,
      "step": 14209
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.42086498415417317,
      "learning_rate": 2.1652659766412146e-06,
      "loss": 0.0485,
      "step": 14210
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.36134817585229756,
      "learning_rate": 2.163247065145335e-06,
      "loss": 0.0393,
      "step": 14211
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.418969668420278,
      "learning_rate": 2.161229052748781e-06,
      "loss": 0.0781,
      "step": 14212
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5406961376658341,
      "learning_rate": 2.1592119395310056e-06,
      "loss": 0.0917,
      "step": 14213
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4323483836060825,
      "learning_rate": 2.1571957255714316e-06,
      "loss": 0.101,
      "step": 14214
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2978249027677837,
      "learning_rate": 2.1551804109494257e-06,
      "loss": 0.0314,
      "step": 14215
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.49323444818281703,
      "learning_rate": 2.153165995744341e-06,
      "loss": 0.0734,
      "step": 14216
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9636776040863477,
      "learning_rate": 2.151152480035484e-06,
      "loss": 0.3222,
      "step": 14217
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3845483401189827,
      "learning_rate": 2.149139863902136e-06,
      "loss": 0.0579,
      "step": 14218
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6404433074979871,
      "learning_rate": 2.1471281474235214e-06,
      "loss": 0.0961,
      "step": 14219
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.40999742066887956,
      "learning_rate": 2.1451173306788503e-06,
      "loss": 0.0625,
      "step": 14220
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5822259612842571,
      "learning_rate": 2.143107413747289e-06,
      "loss": 0.1163,
      "step": 14221
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5490535635512082,
      "learning_rate": 2.1410983967079745e-06,
      "loss": 0.1367,
      "step": 14222
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4149941239077846,
      "learning_rate": 2.1390902796399935e-06,
      "loss": 0.0901,
      "step": 14223
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5763005796449764,
      "learning_rate": 2.1370830626224163e-06,
      "loss": 0.149,
      "step": 14224
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.452393800795297,
      "learning_rate": 2.1350767457342576e-06,
      "loss": 0.0556,
      "step": 14225
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.39761804173861637,
      "learning_rate": 2.13307132905452e-06,
      "loss": 0.0715,
      "step": 14226
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.45512867161505033,
      "learning_rate": 2.1310668126621495e-06,
      "loss": 0.0751,
      "step": 14227
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6954670783884193,
      "learning_rate": 2.1290631966360717e-06,
      "loss": 0.1709,
      "step": 14228
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5877146096163999,
      "learning_rate": 2.1270604810551597e-06,
      "loss": 0.2036,
      "step": 14229
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5097596228245211,
      "learning_rate": 2.125058665998278e-06,
      "loss": 0.0399,
      "step": 14230
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.38835407920684256,
      "learning_rate": 2.1230577515442244e-06,
      "loss": 0.0536,
      "step": 14231
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7003919788437345,
      "learning_rate": 2.1210577377717866e-06,
      "loss": 0.1559,
      "step": 14232
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6530048017516792,
      "learning_rate": 2.1190586247596954e-06,
      "loss": 0.1005,
      "step": 14233
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6970517980698745,
      "learning_rate": 2.1170604125866715e-06,
      "loss": 0.1972,
      "step": 14234
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3980048302686163,
      "learning_rate": 2.1150631013313772e-06,
      "loss": 0.1021,
      "step": 14235
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5627999332709481,
      "learning_rate": 2.1130666910724515e-06,
      "loss": 0.1317,
      "step": 14236
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.43939680369314127,
      "learning_rate": 2.1110711818884825e-06,
      "loss": 0.0742,
      "step": 14237
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.766539759592459,
      "learning_rate": 2.109076573858057e-06,
      "loss": 0.2177,
      "step": 14238
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5278373008407532,
      "learning_rate": 2.107082867059687e-06,
      "loss": 0.0901,
      "step": 14239
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.46419288188811675,
      "learning_rate": 2.1050900615718737e-06,
      "loss": 0.1114,
      "step": 14240
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.45976587046734163,
      "learning_rate": 2.1030981574730656e-06,
      "loss": 0.0786,
      "step": 14241
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4469655385192036,
      "learning_rate": 2.1011071548417e-06,
      "loss": 0.0931,
      "step": 14242
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3643451570939029,
      "learning_rate": 2.0991170537561555e-06,
      "loss": 0.0656,
      "step": 14243
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.550305565913776,
      "learning_rate": 2.0971278542947863e-06,
      "loss": 0.112,
      "step": 14244
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6604308096107784,
      "learning_rate": 2.095139556535905e-06,
      "loss": 0.1078,
      "step": 14245
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4715401840623882,
      "learning_rate": 2.093152160557793e-06,
      "loss": 0.0528,
      "step": 14246
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.39018077924300854,
      "learning_rate": 2.0911656664386993e-06,
      "loss": 0.0738,
      "step": 14247
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7685169536029148,
      "learning_rate": 2.089180074256836e-06,
      "loss": 0.1642,
      "step": 14248
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.64132788675172,
      "learning_rate": 2.087195384090365e-06,
      "loss": 0.0911,
      "step": 14249
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5687668516509577,
      "learning_rate": 2.0852115960174363e-06,
      "loss": 0.1304,
      "step": 14250
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.698978933650718,
      "learning_rate": 2.0832287101161528e-06,
      "loss": 0.1658,
      "step": 14251
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4430545799332738,
      "learning_rate": 2.0812467264645725e-06,
      "loss": 0.0774,
      "step": 14252
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5754189942803845,
      "learning_rate": 2.079265645140738e-06,
      "loss": 0.159,
      "step": 14253
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.41516950036729633,
      "learning_rate": 2.0772854662226367e-06,
      "loss": 0.0528,
      "step": 14254
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3825564793966985,
      "learning_rate": 2.0753061897882427e-06,
      "loss": 0.0389,
      "step": 14255
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5178723358940681,
      "learning_rate": 2.0733278159154685e-06,
      "loss": 0.072,
      "step": 14256
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.44118967737053755,
      "learning_rate": 2.0713503446822076e-06,
      "loss": 0.0432,
      "step": 14257
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0057637724056046,
      "learning_rate": 2.069373776166317e-06,
      "loss": 0.2121,
      "step": 14258
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4018226621735587,
      "learning_rate": 2.0673981104456202e-06,
      "loss": 0.0642,
      "step": 14259
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.49189660084131254,
      "learning_rate": 2.065423347597889e-06,
      "loss": 0.1079,
      "step": 14260
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4363267396913635,
      "learning_rate": 2.0634494877008775e-06,
      "loss": 0.1093,
      "step": 14261
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7490056278062791,
      "learning_rate": 2.0614765308322987e-06,
      "loss": 0.2808,
      "step": 14262
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0742664430614266,
      "learning_rate": 2.059504477069832e-06,
      "loss": 0.3561,
      "step": 14263
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.28989283631345186,
      "learning_rate": 2.0575333264911124e-06,
      "loss": 0.0359,
      "step": 14264
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7940622768455843,
      "learning_rate": 2.0555630791737445e-06,
      "loss": 0.1661,
      "step": 14265
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4302310724201021,
      "learning_rate": 2.0535937351953054e-06,
      "loss": 0.0817,
      "step": 14266
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6719029756707324,
      "learning_rate": 2.0516252946333274e-06,
      "loss": 0.119,
      "step": 14267
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.272962612080235,
      "learning_rate": 2.0496577575653043e-06,
      "loss": 0.2658,
      "step": 14268
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.454629467134906,
      "learning_rate": 2.047691124068704e-06,
      "loss": 0.0514,
      "step": 14269
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.606901305602613,
      "learning_rate": 2.0457253942209542e-06,
      "loss": 0.1475,
      "step": 14270
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.36133612312265195,
      "learning_rate": 2.043760568099451e-06,
      "loss": 0.0605,
      "step": 14271
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.32146661337693644,
      "learning_rate": 2.041796645781541e-06,
      "loss": 0.043,
      "step": 14272
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5702292680674976,
      "learning_rate": 2.0398336273445506e-06,
      "loss": 0.092,
      "step": 14273
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6782111623589064,
      "learning_rate": 2.0378715128657634e-06,
      "loss": 0.1548,
      "step": 14274
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5800868465296748,
      "learning_rate": 2.0359103024224392e-06,
      "loss": 0.1288,
      "step": 14275
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4453059819393742,
      "learning_rate": 2.0339499960917775e-06,
      "loss": 0.0436,
      "step": 14276
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4561150272145558,
      "learning_rate": 2.0319905939509637e-06,
      "loss": 0.6265,
      "step": 14277
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7208683580341319,
      "learning_rate": 2.030032096077142e-06,
      "loss": 0.1753,
      "step": 14278
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.576507761491351,
      "learning_rate": 2.0280745025474223e-06,
      "loss": 0.1082,
      "step": 14279
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3984632095562796,
      "learning_rate": 2.0261178134388657e-06,
      "loss": 0.0565,
      "step": 14280
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5005436106546859,
      "learning_rate": 2.024162028828519e-06,
      "loss": 0.064,
      "step": 14281
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6064715661843545,
      "learning_rate": 2.022207148793384e-06,
      "loss": 0.1306,
      "step": 14282
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.567278033760558,
      "learning_rate": 2.0202531734104165e-06,
      "loss": 0.1208,
      "step": 14283
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4824666278388841,
      "learning_rate": 2.0183001027565486e-06,
      "loss": 0.1196,
      "step": 14284
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.33603938552876655,
      "learning_rate": 2.0163479369086774e-06,
      "loss": 0.0353,
      "step": 14285
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.37733429264406776,
      "learning_rate": 2.0143966759436638e-06,
      "loss": 0.0923,
      "step": 14286
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6162865780779486,
      "learning_rate": 2.0124463199383215e-06,
      "loss": 0.2198,
      "step": 14287
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1553205744166262,
      "learning_rate": 2.010496868969444e-06,
      "loss": 0.2051,
      "step": 14288
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.47981632355522175,
      "learning_rate": 2.0085483231137794e-06,
      "loss": 0.0892,
      "step": 14289
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5017802152247047,
      "learning_rate": 2.0066006824480484e-06,
      "loss": 0.1006,
      "step": 14290
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8545695977737067,
      "learning_rate": 2.0046539470489233e-06,
      "loss": 0.2299,
      "step": 14291
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5769825701032222,
      "learning_rate": 2.00270811699306e-06,
      "loss": 0.1148,
      "step": 14292
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.449587991561083,
      "learning_rate": 2.0007631923570467e-06,
      "loss": 0.0834,
      "step": 14293
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5196907156991895,
      "learning_rate": 1.998819173217481e-06,
      "loss": 0.0888,
      "step": 14294
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.31637908193003944,
      "learning_rate": 1.9968760596508817e-06,
      "loss": 0.0693,
      "step": 14295
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5244454111233857,
      "learning_rate": 1.994933851733766e-06,
      "loss": 0.1349,
      "step": 14296
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5711361060349616,
      "learning_rate": 1.99299254954258e-06,
      "loss": 0.0979,
      "step": 14297
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.38953836375559997,
      "learning_rate": 1.9910521531537744e-06,
      "loss": 0.0675,
      "step": 14298
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.39492623634153456,
      "learning_rate": 1.989112662643733e-06,
      "loss": 0.0721,
      "step": 14299
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4808580932112627,
      "learning_rate": 1.9871740780888186e-06,
      "loss": 0.096,
      "step": 14300
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7494227882631436,
      "learning_rate": 1.985236399565349e-06,
      "loss": 0.0519,
      "step": 14301
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.337652533362142,
      "learning_rate": 1.983299627149621e-06,
      "loss": 0.0491,
      "step": 14302
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6213229769575732,
      "learning_rate": 1.9813637609178814e-06,
      "loss": 0.1518,
      "step": 14303
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5941774169772006,
      "learning_rate": 1.979428800946348e-06,
      "loss": 0.2077,
      "step": 14304
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.47936947495328264,
      "learning_rate": 1.9774947473111955e-06,
      "loss": 0.0821,
      "step": 14305
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5320209755102344,
      "learning_rate": 1.9755616000885825e-06,
      "loss": 0.0678,
      "step": 14306
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.37810504539007767,
      "learning_rate": 1.9736293593546065e-06,
      "loss": 0.0468,
      "step": 14307
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.46717249599000904,
      "learning_rate": 1.971698025185348e-06,
      "loss": 0.0656,
      "step": 14308
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6273229220614922,
      "learning_rate": 1.9697675976568324e-06,
      "loss": 0.1651,
      "step": 14309
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3974796444652732,
      "learning_rate": 1.9678380768450827e-06,
      "loss": 0.0588,
      "step": 14310
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4619529507985269,
      "learning_rate": 1.965909462826049e-06,
      "loss": 0.1039,
      "step": 14311
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7766246503814609,
      "learning_rate": 1.9639817556756703e-06,
      "loss": 0.2543,
      "step": 14312
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.47700014000052043,
      "learning_rate": 1.9620549554698386e-06,
      "loss": 0.0561,
      "step": 14313
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7902480035581008,
      "learning_rate": 1.9601290622844102e-06,
      "loss": 0.153,
      "step": 14314
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.43471819281875923,
      "learning_rate": 1.9582040761952162e-06,
      "loss": 0.0691,
      "step": 14315
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7022508329332398,
      "learning_rate": 1.956279997278043e-06,
      "loss": 0.1318,
      "step": 14316
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5253568139458552,
      "learning_rate": 1.9543568256086388e-06,
      "loss": 0.1125,
      "step": 14317
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4624633787837087,
      "learning_rate": 1.9524345612627226e-06,
      "loss": 0.0703,
      "step": 14318
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.486728156000823,
      "learning_rate": 1.950513204315976e-06,
      "loss": 0.0746,
      "step": 14319
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.36036345100739753,
      "learning_rate": 1.948592754844042e-06,
      "loss": 0.0666,
      "step": 14320
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4311521542300351,
      "learning_rate": 1.946673212922534e-06,
      "loss": 0.0907,
      "step": 14321
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5319983279745834,
      "learning_rate": 1.94475457862702e-06,
      "loss": 0.1069,
      "step": 14322
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4167472754738447,
      "learning_rate": 1.9428368520330446e-06,
      "loss": 0.0663,
      "step": 14323
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3735091910254966,
      "learning_rate": 1.9409200332161065e-06,
      "loss": 0.0355,
      "step": 14324
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5162633141694991,
      "learning_rate": 1.93900412225167e-06,
      "loss": 0.0569,
      "step": 14325
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.41722904183796317,
      "learning_rate": 1.9370891192151685e-06,
      "loss": 0.1242,
      "step": 14326
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5141583205635276,
      "learning_rate": 1.935175024182001e-06,
      "loss": 0.0929,
      "step": 14327
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6050536814498333,
      "learning_rate": 1.9332618372275153e-06,
      "loss": 0.0857,
      "step": 14328
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.42067466379114243,
      "learning_rate": 1.9313495584270507e-06,
      "loss": 0.0604,
      "step": 14329
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5351711335451952,
      "learning_rate": 1.929438187855881e-06,
      "loss": 0.0831,
      "step": 14330
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.48409946994028763,
      "learning_rate": 1.9275277255892705e-06,
      "loss": 0.0569,
      "step": 14331
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7965032884433264,
      "learning_rate": 1.92561817170242e-06,
      "loss": 0.1869,
      "step": 14332
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5662557047441626,
      "learning_rate": 1.9237095262705257e-06,
      "loss": 0.0826,
      "step": 14333
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3206613500137502,
      "learning_rate": 1.921801789368724e-06,
      "loss": 0.0372,
      "step": 14334
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.48824153080035443,
      "learning_rate": 1.9198949610721273e-06,
      "loss": 0.0801,
      "step": 14335
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7290049861744197,
      "learning_rate": 1.917989041455798e-06,
      "loss": 0.1473,
      "step": 14336
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4796646165352465,
      "learning_rate": 1.916084030594792e-06,
      "loss": 0.0663,
      "step": 14337
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4959621975649642,
      "learning_rate": 1.914179928564097e-06,
      "loss": 0.0906,
      "step": 14338
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.37930033526494406,
      "learning_rate": 1.9122767354386893e-06,
      "loss": 0.0509,
      "step": 14339
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.119114140847205,
      "learning_rate": 1.9103744512934814e-06,
      "loss": 0.2583,
      "step": 14340
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6187644094761031,
      "learning_rate": 1.90847307620339e-06,
      "loss": 0.0829,
      "step": 14341
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.338285214133985,
      "learning_rate": 1.906572610243257e-06,
      "loss": 0.0485,
      "step": 14342
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.479874722664911,
      "learning_rate": 1.9046730534879153e-06,
      "loss": 0.0889,
      "step": 14343
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5085442344521741,
      "learning_rate": 1.9027744060121421e-06,
      "loss": 0.0935,
      "step": 14344
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4606937325014242,
      "learning_rate": 1.900876667890697e-06,
      "loss": 0.091,
      "step": 14345
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.156931847906228,
      "learning_rate": 1.8989798391982893e-06,
      "loss": 0.2933,
      "step": 14346
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5196408239120788,
      "learning_rate": 1.897083920009607e-06,
      "loss": 0.1438,
      "step": 14347
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7408442768159714,
      "learning_rate": 1.8951889103992815e-06,
      "loss": 0.1746,
      "step": 14348
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.46344587832017026,
      "learning_rate": 1.8932948104419284e-06,
      "loss": 0.0783,
      "step": 14349
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9493944595468439,
      "learning_rate": 1.8914016202121243e-06,
      "loss": 0.2238,
      "step": 14350
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0497616689121254,
      "learning_rate": 1.8895093397843927e-06,
      "loss": 0.1802,
      "step": 14351
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.550419265188196,
      "learning_rate": 1.8876179692332434e-06,
      "loss": 0.0928,
      "step": 14352
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.31099422113804287,
      "learning_rate": 1.8857275086331366e-06,
      "loss": 0.038,
      "step": 14353
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.428101901998419,
      "learning_rate": 1.8838379580585098e-06,
      "loss": 0.0422,
      "step": 14354
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5397262611827847,
      "learning_rate": 1.881949317583745e-06,
      "loss": 0.087,
      "step": 14355
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5233331720174867,
      "learning_rate": 1.8800615872832022e-06,
      "loss": 0.1166,
      "step": 14356
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4345186867706191,
      "learning_rate": 1.8781747672312056e-06,
      "loss": 0.1024,
      "step": 14357
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.47689897720268354,
      "learning_rate": 1.8762888575020426e-06,
      "loss": 0.0913,
      "step": 14358
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6155224277163912,
      "learning_rate": 1.8744038581699542e-06,
      "loss": 0.0944,
      "step": 14359
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5803558324306528,
      "learning_rate": 1.8725197693091612e-06,
      "loss": 0.1583,
      "step": 14360
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6463624673228202,
      "learning_rate": 1.8706365909938378e-06,
      "loss": 0.0689,
      "step": 14361
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5553350891705662,
      "learning_rate": 1.868754323298136e-06,
      "loss": 0.1067,
      "step": 14362
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7087426290129527,
      "learning_rate": 1.8668729662961459e-06,
      "loss": 0.1763,
      "step": 14363
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3456273925747931,
      "learning_rate": 1.8649925200619479e-06,
      "loss": 0.0645,
      "step": 14364
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4086019380215545,
      "learning_rate": 1.863112984669574e-06,
      "loss": 0.0592,
      "step": 14365
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7272035751212212,
      "learning_rate": 1.8612343601930288e-06,
      "loss": 0.107,
      "step": 14366
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8797640809218197,
      "learning_rate": 1.8593566467062645e-06,
      "loss": 0.2763,
      "step": 14367
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6595103717368903,
      "learning_rate": 1.8574798442832137e-06,
      "loss": 0.1111,
      "step": 14368
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4117588068736977,
      "learning_rate": 1.855603952997767e-06,
      "loss": 0.0562,
      "step": 14369
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.44431643017191574,
      "learning_rate": 1.8537289729237822e-06,
      "loss": 0.0631,
      "step": 14370
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8795045550296763,
      "learning_rate": 1.8518549041350752e-06,
      "loss": 0.3205,
      "step": 14371
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5649919964208489,
      "learning_rate": 1.8499817467054287e-06,
      "loss": 0.0917,
      "step": 14372
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.47550282508178415,
      "learning_rate": 1.8481095007085919e-06,
      "loss": 0.0708,
      "step": 14373
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9248684056991127,
      "learning_rate": 1.846238166218281e-06,
      "loss": 0.2942,
      "step": 14374
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5461635095966777,
      "learning_rate": 1.8443677433081619e-06,
      "loss": 0.0984,
      "step": 14375
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5193422112761421,
      "learning_rate": 1.8424982320518813e-06,
      "loss": 0.0509,
      "step": 14376
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2951115646487065,
      "learning_rate": 1.8406296325230387e-06,
      "loss": 0.0326,
      "step": 14377
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7131686804633781,
      "learning_rate": 1.8387619447952116e-06,
      "loss": 0.1491,
      "step": 14378
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6413293130271678,
      "learning_rate": 1.836895168941924e-06,
      "loss": 0.2177,
      "step": 14379
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.489751454407185,
      "learning_rate": 1.83502930503667e-06,
      "loss": 0.1142,
      "step": 14380
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7129982666511454,
      "learning_rate": 1.8331643531529191e-06,
      "loss": 0.1823,
      "step": 14381
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6058944830884141,
      "learning_rate": 1.831300313364087e-06,
      "loss": 0.1557,
      "step": 14382
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9485457718847633,
      "learning_rate": 1.8294371857435682e-06,
      "loss": 0.3094,
      "step": 14383
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3400744639076714,
      "learning_rate": 1.8275749703647094e-06,
      "loss": 0.0505,
      "step": 14384
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3794101684381711,
      "learning_rate": 1.825713667300838e-06,
      "loss": 0.0832,
      "step": 14385
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5718050657740221,
      "learning_rate": 1.8238532766252231e-06,
      "loss": 0.1242,
      "step": 14386
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9337473335303696,
      "learning_rate": 1.8219937984111174e-06,
      "loss": 0.2066,
      "step": 14387
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7091493899261311,
      "learning_rate": 1.8201352327317207e-06,
      "loss": 0.1016,
      "step": 14388
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3892283422985091,
      "learning_rate": 1.8182775796602187e-06,
      "loss": 0.0567,
      "step": 14389
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0147370348502276,
      "learning_rate": 1.8164208392697391e-06,
      "loss": 0.1917,
      "step": 14390
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.891667733146889,
      "learning_rate": 1.81456501163339e-06,
      "loss": 0.1065,
      "step": 14391
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4482445041554358,
      "learning_rate": 1.8127100968242245e-06,
      "loss": 0.0683,
      "step": 14392
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.39132517471524725,
      "learning_rate": 1.8108560949152893e-06,
      "loss": 0.0512,
      "step": 14393
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0590835306018143,
      "learning_rate": 1.809003005979562e-06,
      "loss": 0.1789,
      "step": 14394
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6492266873584044,
      "learning_rate": 1.8071508300900125e-06,
      "loss": 0.1526,
      "step": 14395
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4291078981465724,
      "learning_rate": 1.8052995673195488e-06,
      "loss": 0.0574,
      "step": 14396
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.48065384332041283,
      "learning_rate": 1.8034492177410739e-06,
      "loss": 0.0815,
      "step": 14397
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6355081016430428,
      "learning_rate": 1.801599781427421e-06,
      "loss": 0.1247,
      "step": 14398
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5408217876076227,
      "learning_rate": 1.7997512584514153e-06,
      "loss": 0.1025,
      "step": 14399
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.594919677964783,
      "learning_rate": 1.7979036488858208e-06,
      "loss": 0.1344,
      "step": 14400
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6461656759482822,
      "learning_rate": 1.796056952803396e-06,
      "loss": 0.1375,
      "step": 14401
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.49277500964849946,
      "learning_rate": 1.7942111702768355e-06,
      "loss": 0.1108,
      "step": 14402
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.48037764117613957,
      "learning_rate": 1.7923663013788173e-06,
      "loss": 0.084,
      "step": 14403
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7650878197092795,
      "learning_rate": 1.790522346181961e-06,
      "loss": 0.182,
      "step": 14404
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5379159639588528,
      "learning_rate": 1.7886793047588808e-06,
      "loss": 0.0994,
      "step": 14405
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9329326596020895,
      "learning_rate": 1.7868371771821268e-06,
      "loss": 0.1699,
      "step": 14406
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6932861995458761,
      "learning_rate": 1.7849959635242358e-06,
      "loss": 0.1622,
      "step": 14407
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.70957058989747,
      "learning_rate": 1.783155663857683e-06,
      "loss": 0.2206,
      "step": 14408
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9676240213490299,
      "learning_rate": 1.7813162782549353e-06,
      "loss": 0.3016,
      "step": 14409
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.39910034533703653,
      "learning_rate": 1.7794778067884043e-06,
      "loss": 0.0621,
      "step": 14410
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0445065701794571,
      "learning_rate": 1.7776402495304795e-06,
      "loss": 0.2597,
      "step": 14411
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4301545791938257,
      "learning_rate": 1.7758036065534945e-06,
      "loss": 0.0465,
      "step": 14412
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5167023364383111,
      "learning_rate": 1.7739678779297664e-06,
      "loss": 0.0883,
      "step": 14413
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.26962582014476255,
      "learning_rate": 1.772133063731568e-06,
      "loss": 0.037,
      "step": 14414
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5319735947658365,
      "learning_rate": 1.7702991640311418e-06,
      "loss": 0.1066,
      "step": 14415
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4799541684021444,
      "learning_rate": 1.7684661789006823e-06,
      "loss": 0.0879,
      "step": 14416
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3801997092496498,
      "learning_rate": 1.76663410841236e-06,
      "loss": 0.0953,
      "step": 14417
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.58621232521432,
      "learning_rate": 1.764802952638306e-06,
      "loss": 0.1818,
      "step": 14418
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8340577407396361,
      "learning_rate": 1.7629727116506068e-06,
      "loss": 0.1613,
      "step": 14419
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.542122510838923,
      "learning_rate": 1.761143385521327e-06,
      "loss": 0.1484,
      "step": 14420
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5415137474926893,
      "learning_rate": 1.7593149743224846e-06,
      "loss": 0.0859,
      "step": 14421
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4166193333030983,
      "learning_rate": 1.757487478126074e-06,
      "loss": 0.0467,
      "step": 14422
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3559047471076985,
      "learning_rate": 1.7556608970040356e-06,
      "loss": 0.1026,
      "step": 14423
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4963427523860023,
      "learning_rate": 1.7538352310282835e-06,
      "loss": 0.0726,
      "step": 14424
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.45894744792601994,
      "learning_rate": 1.7520104802706993e-06,
      "loss": 0.0725,
      "step": 14425
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6380340536198074,
      "learning_rate": 1.7501866448031285e-06,
      "loss": 0.134,
      "step": 14426
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6082131085722348,
      "learning_rate": 1.748363724697366e-06,
      "loss": 0.0798,
      "step": 14427
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4925935547937041,
      "learning_rate": 1.746541720025191e-06,
      "loss": 0.0812,
      "step": 14428
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.33595203923692596,
      "learning_rate": 1.744720630858332e-06,
      "loss": 0.0436,
      "step": 14429
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6503185294094359,
      "learning_rate": 1.74290045726849e-06,
      "loss": 0.1538,
      "step": 14430
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6372450824253743,
      "learning_rate": 1.741081199327324e-06,
      "loss": 0.1442,
      "step": 14431
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8110107023022143,
      "learning_rate": 1.739262857106458e-06,
      "loss": 0.156,
      "step": 14432
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4311431439220912,
      "learning_rate": 1.737445430677484e-06,
      "loss": 0.0624,
      "step": 14433
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.71475084723509,
      "learning_rate": 1.735628920111962e-06,
      "loss": 0.1608,
      "step": 14434
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.38214769961876804,
      "learning_rate": 1.7338133254813955e-06,
      "loss": 0.0542,
      "step": 14435
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6327542662507347,
      "learning_rate": 1.731998646857272e-06,
      "loss": 0.1973,
      "step": 14436
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4365652674108535,
      "learning_rate": 1.7301848843110396e-06,
      "loss": 0.0687,
      "step": 14437
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6580758962982738,
      "learning_rate": 1.7283720379141084e-06,
      "loss": 0.1014,
      "step": 14438
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4946594155400197,
      "learning_rate": 1.726560107737843e-06,
      "loss": 0.0834,
      "step": 14439
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5699774377326104,
      "learning_rate": 1.7247490938535865e-06,
      "loss": 0.066,
      "step": 14440
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5195372088802263,
      "learning_rate": 1.7229389963326403e-06,
      "loss": 0.0863,
      "step": 14441
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3155724870362447,
      "learning_rate": 1.7211298152462695e-06,
      "loss": 0.0362,
      "step": 14442
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2814828491402004,
      "learning_rate": 1.7193215506656978e-06,
      "loss": 0.4959,
      "step": 14443
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4749606699478346,
      "learning_rate": 1.7175142026621239e-06,
      "loss": 0.105,
      "step": 14444
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.48596420931620044,
      "learning_rate": 1.7157077713066988e-06,
      "loss": 0.0759,
      "step": 14445
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3040166348870463,
      "learning_rate": 1.713902256670552e-06,
      "loss": 0.0427,
      "step": 14446
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.40690294464045595,
      "learning_rate": 1.7120976588247573e-06,
      "loss": 0.0613,
      "step": 14447
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8492431911487294,
      "learning_rate": 1.710293977840366e-06,
      "loss": 0.1379,
      "step": 14448
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6255689171912194,
      "learning_rate": 1.708491213788399e-06,
      "loss": 0.1607,
      "step": 14449
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6833654845988018,
      "learning_rate": 1.7066893667398192e-06,
      "loss": 0.1556,
      "step": 14450
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7922496685045142,
      "learning_rate": 1.7048884367655727e-06,
      "loss": 0.1331,
      "step": 14451
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5541391303383402,
      "learning_rate": 1.7030884239365636e-06,
      "loss": 0.1333,
      "step": 14452
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4284510085742532,
      "learning_rate": 1.701289328323663e-06,
      "loss": 0.064,
      "step": 14453
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6100262109281173,
      "learning_rate": 1.6994911499976978e-06,
      "loss": 0.1549,
      "step": 14454
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.33613434372213546,
      "learning_rate": 1.6976938890294668e-06,
      "loss": 0.0325,
      "step": 14455
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.475796756395447,
      "learning_rate": 1.6958975454897191e-06,
      "loss": 0.0915,
      "step": 14456
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5142910304025339,
      "learning_rate": 1.694102119449195e-06,
      "loss": 0.0816,
      "step": 14457
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7747104381550542,
      "learning_rate": 1.6923076109785718e-06,
      "loss": 0.1767,
      "step": 14458
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3762700961405306,
      "learning_rate": 1.6905140201485037e-06,
      "loss": 0.0446,
      "step": 14459
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4311668295439081,
      "learning_rate": 1.6887213470295954e-06,
      "loss": 0.0777,
      "step": 14460
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6228808856607,
      "learning_rate": 1.6869295916924433e-06,
      "loss": 0.1353,
      "step": 14461
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6627679108049611,
      "learning_rate": 1.6851387542075743e-06,
      "loss": 0.0871,
      "step": 14462
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5026595677276948,
      "learning_rate": 1.683348834645504e-06,
      "loss": 0.0907,
      "step": 14463
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.465777612733776,
      "learning_rate": 1.6815598330767014e-06,
      "loss": 0.0664,
      "step": 14464
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5794713027418923,
      "learning_rate": 1.6797717495716042e-06,
      "loss": 0.1126,
      "step": 14465
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3924097345153506,
      "learning_rate": 1.677984584200601e-06,
      "loss": 0.0536,
      "step": 14466
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3443628792405501,
      "learning_rate": 1.6761983370340572e-06,
      "loss": 0.0601,
      "step": 14467
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.29593793528416895,
      "learning_rate": 1.6744130081423032e-06,
      "loss": 0.0461,
      "step": 14468
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4884469970655478,
      "learning_rate": 1.6726285975956297e-06,
      "loss": 0.0537,
      "step": 14469
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.41058122661708313,
      "learning_rate": 1.6708451054642803e-06,
      "loss": 0.08,
      "step": 14470
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.48142683757141913,
      "learning_rate": 1.6690625318184799e-06,
      "loss": 0.0459,
      "step": 14471
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4358903720669125,
      "learning_rate": 1.667280876728408e-06,
      "loss": 0.0591,
      "step": 14472
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4434429228310648,
      "learning_rate": 1.6655001402642146e-06,
      "loss": 0.0752,
      "step": 14473
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3876739939078643,
      "learning_rate": 1.6637203224960014e-06,
      "loss": 0.0636,
      "step": 14474
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.40203032491831986,
      "learning_rate": 1.6619414234938403e-06,
      "loss": 0.0852,
      "step": 14475
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6816120365442431,
      "learning_rate": 1.6601634433277724e-06,
      "loss": 0.1122,
      "step": 14476
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6098483459108709,
      "learning_rate": 1.658386382067803e-06,
      "loss": 0.1012,
      "step": 14477
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4468540450947406,
      "learning_rate": 1.6566102397838844e-06,
      "loss": 0.0747,
      "step": 14478
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7844171439787628,
      "learning_rate": 1.6548350165459498e-06,
      "loss": 0.2163,
      "step": 14479
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4541715169441803,
      "learning_rate": 1.6530607124238956e-06,
      "loss": 0.072,
      "step": 14480
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3771065526489663,
      "learning_rate": 1.651287327487569e-06,
      "loss": 0.0605,
      "step": 14481
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.4697900714935264,
      "learning_rate": 1.6495148618067945e-06,
      "loss": 0.4756,
      "step": 14482
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.45846231555180716,
      "learning_rate": 1.6477433154513555e-06,
      "loss": 0.0469,
      "step": 14483
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.703316096139915,
      "learning_rate": 1.6459726884909988e-06,
      "loss": 0.0896,
      "step": 14484
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4270576158872268,
      "learning_rate": 1.6442029809954351e-06,
      "loss": 0.0515,
      "step": 14485
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.47972298945352976,
      "learning_rate": 1.6424341930343395e-06,
      "loss": 0.0894,
      "step": 14486
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.42051254022713985,
      "learning_rate": 1.6406663246773424e-06,
      "loss": 0.0973,
      "step": 14487
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6712486393001372,
      "learning_rate": 1.6388993759940601e-06,
      "loss": 0.1057,
      "step": 14488
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4671362648743784,
      "learning_rate": 1.6371333470540484e-06,
      "loss": 0.0568,
      "step": 14489
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7602736643817597,
      "learning_rate": 1.6353682379268458e-06,
      "loss": 0.1437,
      "step": 14490
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5327553249266487,
      "learning_rate": 1.6336040486819303e-06,
      "loss": 0.1094,
      "step": 14491
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0565405610387077,
      "learning_rate": 1.6318407793887797e-06,
      "loss": 0.3686,
      "step": 14492
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.38665613412864236,
      "learning_rate": 1.6300784301167992e-06,
      "loss": 0.0681,
      "step": 14493
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2623026290579843,
      "learning_rate": 1.6283170009353838e-06,
      "loss": 0.0345,
      "step": 14494
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4741845523002425,
      "learning_rate": 1.6265564919138692e-06,
      "loss": 0.0507,
      "step": 14495
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.34519291578465633,
      "learning_rate": 1.6247969031215865e-06,
      "loss": 0.0405,
      "step": 14496
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7713133036240847,
      "learning_rate": 1.6230382346277966e-06,
      "loss": 0.1986,
      "step": 14497
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3119863484190106,
      "learning_rate": 1.6212804865017501e-06,
      "loss": 0.05,
      "step": 14498
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9718656656058865,
      "learning_rate": 1.6195236588126383e-06,
      "loss": 0.3156,
      "step": 14499
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.44845711261107646,
      "learning_rate": 1.6177677516296424e-06,
      "loss": 0.0877,
      "step": 14500
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4748819263434951,
      "learning_rate": 1.616012765021885e-06,
      "loss": 0.0721,
      "step": 14501
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5191859175555934,
      "learning_rate": 1.6142586990584685e-06,
      "loss": 0.1271,
      "step": 14502
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4500103553486106,
      "learning_rate": 1.6125055538084383e-06,
      "loss": 0.0871,
      "step": 14503
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7867049894600013,
      "learning_rate": 1.6107533293408305e-06,
      "loss": 0.1829,
      "step": 14504
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6041627900978401,
      "learning_rate": 1.6090020257246263e-06,
      "loss": 0.2051,
      "step": 14505
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4483499999103673,
      "learning_rate": 1.6072516430287787e-06,
      "loss": 0.5827,
      "step": 14506
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4047731143903295,
      "learning_rate": 1.6055021813221966e-06,
      "loss": 0.1036,
      "step": 14507
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8985727442455284,
      "learning_rate": 1.603753640673758e-06,
      "loss": 0.2298,
      "step": 14508
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3739556526346419,
      "learning_rate": 1.602006021152308e-06,
      "loss": 0.0409,
      "step": 14509
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.1318691291715335,
      "learning_rate": 1.60025932282665e-06,
      "loss": 0.4565,
      "step": 14510
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5741175920798286,
      "learning_rate": 1.5985135457655508e-06,
      "loss": 0.134,
      "step": 14511
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.31067668065409476,
      "learning_rate": 1.5967686900377449e-06,
      "loss": 0.042,
      "step": 14512
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2021552334749515,
      "learning_rate": 1.5950247557119296e-06,
      "loss": 0.3479,
      "step": 14513
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4154038796066838,
      "learning_rate": 1.5932817428567642e-06,
      "loss": 0.1047,
      "step": 14514
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.36221318203239283,
      "learning_rate": 1.5915396515408687e-06,
      "loss": 0.0393,
      "step": 14515
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4231024730135867,
      "learning_rate": 1.5897984818328355e-06,
      "loss": 0.0576,
      "step": 14516
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4111846676242567,
      "learning_rate": 1.5880582338012151e-06,
      "loss": 0.0649,
      "step": 14517
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0552567238360662,
      "learning_rate": 1.5863189075145169e-06,
      "loss": 0.1891,
      "step": 14518
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.34698519470851413,
      "learning_rate": 1.5845805030412248e-06,
      "loss": 0.0415,
      "step": 14519
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4769394274315441,
      "learning_rate": 1.5828430204497784e-06,
      "loss": 0.1091,
      "step": 14520
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.593038023072269,
      "learning_rate": 1.5811064598085872e-06,
      "loss": 0.1504,
      "step": 14521
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4030540760357377,
      "learning_rate": 1.5793708211860158e-06,
      "loss": 0.0661,
      "step": 14522
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.47499447621415053,
      "learning_rate": 1.5776361046504013e-06,
      "loss": 0.0768,
      "step": 14523
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.44081376483485923,
      "learning_rate": 1.5759023102700387e-06,
      "loss": 0.0819,
      "step": 14524
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5142718463648288,
      "learning_rate": 1.5741694381131932e-06,
      "loss": 0.0853,
      "step": 14525
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.528660719975442,
      "learning_rate": 1.5724374882480824e-06,
      "loss": 0.1256,
      "step": 14526
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4329711929008,
      "learning_rate": 1.570706460742899e-06,
      "loss": 0.0433,
      "step": 14527
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5532758670668255,
      "learning_rate": 1.568976355665791e-06,
      "loss": 0.092,
      "step": 14528
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4075465172323082,
      "learning_rate": 1.567247173084882e-06,
      "loss": 0.087,
      "step": 14529
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7330493525358351,
      "learning_rate": 1.5655189130682397e-06,
      "loss": 0.2269,
      "step": 14530
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6586630095331643,
      "learning_rate": 1.5637915756839122e-06,
      "loss": 0.102,
      "step": 14531
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7601726287585985,
      "learning_rate": 1.5620651609999093e-06,
      "loss": 0.1133,
      "step": 14532
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7133371336341436,
      "learning_rate": 1.5603396690842016e-06,
      "loss": 0.165,
      "step": 14533
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3856702615461795,
      "learning_rate": 1.5586151000047122e-06,
      "loss": 0.0506,
      "step": 14534
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4690902796006068,
      "learning_rate": 1.556891453829351e-06,
      "loss": 0.0859,
      "step": 14535
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3782101429670828,
      "learning_rate": 1.5551687306259722e-06,
      "loss": 0.0668,
      "step": 14536
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5625592431621985,
      "learning_rate": 1.5534469304624072e-06,
      "loss": 0.1304,
      "step": 14537
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.34249298203090656,
      "learning_rate": 1.5517260534064355e-06,
      "loss": 0.0742,
      "step": 14538
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5332089589084583,
      "learning_rate": 1.5500060995258137e-06,
      "loss": 0.1342,
      "step": 14539
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3817388903045932,
      "learning_rate": 1.5482870688882571e-06,
      "loss": 0.0852,
      "step": 14540
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.36784685980144083,
      "learning_rate": 1.546568961561451e-06,
      "loss": 0.0603,
      "step": 14541
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.634904226177101,
      "learning_rate": 1.5448517776130294e-06,
      "loss": 0.1345,
      "step": 14542
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5044435818039826,
      "learning_rate": 1.5431355171106027e-06,
      "loss": 0.0621,
      "step": 14543
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6070940743207428,
      "learning_rate": 1.5414201801217444e-06,
      "loss": 0.0825,
      "step": 14544
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6844806857981829,
      "learning_rate": 1.5397057667139863e-06,
      "loss": 0.2472,
      "step": 14545
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8229812404888672,
      "learning_rate": 1.5379922769548222e-06,
      "loss": 0.3193,
      "step": 14546
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.38912185580247993,
      "learning_rate": 1.5362797109117171e-06,
      "loss": 0.0722,
      "step": 14547
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7691735101103941,
      "learning_rate": 1.5345680686521007e-06,
      "loss": 0.1178,
      "step": 14548
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5338697083698797,
      "learning_rate": 1.5328573502433497e-06,
      "loss": 0.0699,
      "step": 14549
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.505562935114367,
      "learning_rate": 1.5311475557528264e-06,
      "loss": 0.0861,
      "step": 14550
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6612607241589231,
      "learning_rate": 1.5294386852478416e-06,
      "loss": 0.088,
      "step": 14551
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3421574183960395,
      "learning_rate": 1.5277307387956796e-06,
      "loss": 0.0333,
      "step": 14552
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4331667439944046,
      "learning_rate": 1.5260237164635787e-06,
      "loss": 0.0773,
      "step": 14553
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3395359146475816,
      "learning_rate": 1.524317618318749e-06,
      "loss": 0.0436,
      "step": 14554
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1392870470114975,
      "learning_rate": 1.5226124444283535e-06,
      "loss": 0.3046,
      "step": 14555
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7016988700558571,
      "learning_rate": 1.5209081948595383e-06,
      "loss": 0.1852,
      "step": 14556
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4166254742221122,
      "learning_rate": 1.5192048696793886e-06,
      "loss": 0.0656,
      "step": 14557
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7117774108922974,
      "learning_rate": 1.5175024689549788e-06,
      "loss": 0.1762,
      "step": 14558
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7876449152821757,
      "learning_rate": 1.5158009927533161e-06,
      "loss": 0.1148,
      "step": 14559
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7137158883509038,
      "learning_rate": 1.5141004411414056e-06,
      "loss": 0.1227,
      "step": 14560
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.46911898002962277,
      "learning_rate": 1.5124008141861907e-06,
      "loss": 0.1388,
      "step": 14561
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.36014057342359396,
      "learning_rate": 1.5107021119545927e-06,
      "loss": 0.0494,
      "step": 14562
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8528115003927628,
      "learning_rate": 1.509004334513478e-06,
      "loss": 0.1382,
      "step": 14563
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5740552069313043,
      "learning_rate": 1.5073074819297066e-06,
      "loss": 0.117,
      "step": 14564
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4482052208876034,
      "learning_rate": 1.5056115542700727e-06,
      "loss": 0.0582,
      "step": 14565
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3451956138023701,
      "learning_rate": 1.5039165516013503e-06,
      "loss": 0.0465,
      "step": 14566
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8045838415344709,
      "learning_rate": 1.5022224739902697e-06,
      "loss": 0.2365,
      "step": 14567
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.263510351142576,
      "learning_rate": 1.5005293215035355e-06,
      "loss": 0.3643,
      "step": 14568
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.178702086173783,
      "learning_rate": 1.4988370942077973e-06,
      "loss": 0.2233,
      "step": 14569
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3414380716262222,
      "learning_rate": 1.4971457921696934e-06,
      "loss": 0.0513,
      "step": 14570
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4103788355836008,
      "learning_rate": 1.4954554154557932e-06,
      "loss": 0.0818,
      "step": 14571
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6424011144115434,
      "learning_rate": 1.4937659641326678e-06,
      "loss": 0.1028,
      "step": 14572
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7371714273490331,
      "learning_rate": 1.4920774382668173e-06,
      "loss": 0.1814,
      "step": 14573
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.022965321740835,
      "learning_rate": 1.49038983792473e-06,
      "loss": 0.1851,
      "step": 14574
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6805650789586738,
      "learning_rate": 1.4887031631728415e-06,
      "loss": 0.1642,
      "step": 14575
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4183777602577164,
      "learning_rate": 1.4870174140775573e-06,
      "loss": 0.0974,
      "step": 14576
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.32868825705733257,
      "learning_rate": 1.485332590705249e-06,
      "loss": 0.0469,
      "step": 14577
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6669705066523909,
      "learning_rate": 1.4836486931222527e-06,
      "loss": 0.1476,
      "step": 14578
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4502433402914839,
      "learning_rate": 1.4819657213948567e-06,
      "loss": 0.0478,
      "step": 14579
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0745713422373184,
      "learning_rate": 1.480283675589325e-06,
      "loss": 0.4155,
      "step": 14580
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.44904624985669117,
      "learning_rate": 1.4786025557718825e-06,
      "loss": 0.154,
      "step": 14581
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5002830354043035,
      "learning_rate": 1.4769223620087175e-06,
      "loss": 0.1063,
      "step": 14582
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4919633663488219,
      "learning_rate": 1.4752430943659745e-06,
      "loss": 0.1277,
      "step": 14583
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3482236101000843,
      "learning_rate": 1.4735647529097674e-06,
      "loss": 0.0541,
      "step": 14584
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4336283073102247,
      "learning_rate": 1.4718873377061849e-06,
      "loss": 0.0992,
      "step": 14585
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3503532373823421,
      "learning_rate": 1.4702108488212547e-06,
      "loss": 0.052,
      "step": 14586
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5101511754275238,
      "learning_rate": 1.468535286320985e-06,
      "loss": 0.0811,
      "step": 14587
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5150747698045213,
      "learning_rate": 1.4668606502713456e-06,
      "loss": 0.0627,
      "step": 14588
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4109180276776602,
      "learning_rate": 1.4651869407382722e-06,
      "loss": 0.0554,
      "step": 14589
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4549572085569526,
      "learning_rate": 1.4635141577876543e-06,
      "loss": 0.0598,
      "step": 14590
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6304630689363615,
      "learning_rate": 1.46184230148535e-06,
      "loss": 0.1015,
      "step": 14591
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5075250195800574,
      "learning_rate": 1.4601713718971816e-06,
      "loss": 0.0786,
      "step": 14592
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6381382226677249,
      "learning_rate": 1.4585013690889438e-06,
      "loss": 0.192,
      "step": 14593
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4693220689859648,
      "learning_rate": 1.45683229312637e-06,
      "loss": 0.0837,
      "step": 14594
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5273545787532347,
      "learning_rate": 1.4551641440751884e-06,
      "loss": 0.1844,
      "step": 14595
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.40255947726386754,
      "learning_rate": 1.4534969220010658e-06,
      "loss": 0.0797,
      "step": 14596
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6727613563892246,
      "learning_rate": 1.4518306269696468e-06,
      "loss": 0.1421,
      "step": 14597
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.355173418118868,
      "learning_rate": 1.4501652590465264e-06,
      "loss": 0.429,
      "step": 14598
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3800618106229972,
      "learning_rate": 1.4485008182972825e-06,
      "loss": 0.0641,
      "step": 14599
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4379557702797035,
      "learning_rate": 1.4468373047874407e-06,
      "loss": 0.0626,
      "step": 14600
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.47188236577844933,
      "learning_rate": 1.4451747185824954e-06,
      "loss": 0.0826,
      "step": 14601
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5334280946748812,
      "learning_rate": 1.4435130597478946e-06,
      "loss": 0.1145,
      "step": 14602
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.49022010236517183,
      "learning_rate": 1.4418523283490748e-06,
      "loss": 0.0691,
      "step": 14603
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3754726374520472,
      "learning_rate": 1.4401925244514086e-06,
      "loss": 0.0482,
      "step": 14604
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.46657370833052675,
      "learning_rate": 1.4385336481202494e-06,
      "loss": 0.1194,
      "step": 14605
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4970638691435218,
      "learning_rate": 1.436875699420906e-06,
      "loss": 0.1147,
      "step": 14606
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5634189588566524,
      "learning_rate": 1.4352186784186511e-06,
      "loss": 0.1107,
      "step": 14607
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6670177002741766,
      "learning_rate": 1.4335625851787272e-06,
      "loss": 0.1741,
      "step": 14608
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6957637322140692,
      "learning_rate": 1.4319074197663346e-06,
      "loss": 0.192,
      "step": 14609
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5059713132433652,
      "learning_rate": 1.4302531822466353e-06,
      "loss": 0.0601,
      "step": 14610
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4706187502147743,
      "learning_rate": 1.4285998726847604e-06,
      "loss": 0.0625,
      "step": 14611
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7084659333787648,
      "learning_rate": 1.4269474911458053e-06,
      "loss": 0.1393,
      "step": 14612
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3760253434340931,
      "learning_rate": 1.4252960376948176e-06,
      "loss": 0.0603,
      "step": 14613
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4922256359490161,
      "learning_rate": 1.4236455123968207e-06,
      "loss": 0.0735,
      "step": 14614
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5885372047837828,
      "learning_rate": 1.4219959153167956e-06,
      "loss": 0.0984,
      "step": 14615
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2840053838805707,
      "learning_rate": 1.420347246519693e-06,
      "loss": 0.7292,
      "step": 14616
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8322158543033504,
      "learning_rate": 1.4186995060704172e-06,
      "loss": 0.1641,
      "step": 14617
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3142178396921726,
      "learning_rate": 1.417052694033838e-06,
      "loss": 0.0337,
      "step": 14618
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.41947421654991224,
      "learning_rate": 1.4154068104747981e-06,
      "loss": 0.0829,
      "step": 14619
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4915696632709934,
      "learning_rate": 1.4137618554580984e-06,
      "loss": 0.0828,
      "step": 14620
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3305235097179054,
      "learning_rate": 1.4121178290484926e-06,
      "loss": 0.0588,
      "step": 14621
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5766947374145345,
      "learning_rate": 1.4104747313107124e-06,
      "loss": 0.1183,
      "step": 14622
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8150885329486922,
      "learning_rate": 1.4088325623094507e-06,
      "loss": 0.2416,
      "step": 14623
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5191223435310672,
      "learning_rate": 1.4071913221093608e-06,
      "loss": 0.1223,
      "step": 14624
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3265234531928341,
      "learning_rate": 1.4055510107750524e-06,
      "loss": 0.0581,
      "step": 14625
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8321903814691481,
      "learning_rate": 1.4039116283711102e-06,
      "loss": 0.1905,
      "step": 14626
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7198711861028533,
      "learning_rate": 1.4022731749620793e-06,
      "loss": 0.1607,
      "step": 14627
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.47511199631131906,
      "learning_rate": 1.4006356506124668e-06,
      "loss": 0.1085,
      "step": 14628
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.46044431631228155,
      "learning_rate": 1.3989990553867406e-06,
      "loss": 0.048,
      "step": 14629
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.41700851173159775,
      "learning_rate": 1.3973633893493348e-06,
      "loss": 0.0496,
      "step": 14630
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4314889808302687,
      "learning_rate": 1.3957286525646485e-06,
      "loss": 0.0949,
      "step": 14631
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5197868885557926,
      "learning_rate": 1.3940948450970437e-06,
      "loss": 0.104,
      "step": 14632
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8935430757492522,
      "learning_rate": 1.3924619670108384e-06,
      "loss": 0.2602,
      "step": 14633
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.907599588288361,
      "learning_rate": 1.390830018370326e-06,
      "loss": 0.1948,
      "step": 14634
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8879142608659983,
      "learning_rate": 1.3891989992397576e-06,
      "loss": 0.1887,
      "step": 14635
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4625738129171642,
      "learning_rate": 1.3875689096833456e-06,
      "loss": 0.46,
      "step": 14636
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.27626427315608426,
      "learning_rate": 1.3859397497652666e-06,
      "loss": 0.0573,
      "step": 14637
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.46936833729314353,
      "learning_rate": 1.3843115195496636e-06,
      "loss": 0.0584,
      "step": 14638
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.43069699762274993,
      "learning_rate": 1.3826842191006412e-06,
      "loss": 0.0535,
      "step": 14639
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6214897475320545,
      "learning_rate": 1.38105784848227e-06,
      "loss": 0.1574,
      "step": 14640
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5220961359046337,
      "learning_rate": 1.3794324077585741e-06,
      "loss": 0.1118,
      "step": 14641
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4557579054941593,
      "learning_rate": 1.377807896993552e-06,
      "loss": 0.1012,
      "step": 14642
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.742782010285327,
      "learning_rate": 1.3761843162511668e-06,
      "loss": 0.1717,
      "step": 14643
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5199563850845865,
      "learning_rate": 1.3745616655953336e-06,
      "loss": 0.1003,
      "step": 14644
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3723851692011611,
      "learning_rate": 1.3729399450899377e-06,
      "loss": 0.0515,
      "step": 14645
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0018570678663787,
      "learning_rate": 1.371319154798828e-06,
      "loss": 0.2871,
      "step": 14646
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6455767016008271,
      "learning_rate": 1.3696992947858227e-06,
      "loss": 0.1283,
      "step": 14647
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8287533461009292,
      "learning_rate": 1.368080365114688e-06,
      "loss": 0.1751,
      "step": 14648
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6534588554403304,
      "learning_rate": 1.3664623658491638e-06,
      "loss": 0.1445,
      "step": 14649
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.43354037004829454,
      "learning_rate": 1.3648452970529552e-06,
      "loss": 0.0709,
      "step": 14650
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9628991343079296,
      "learning_rate": 1.3632291587897306e-06,
      "loss": 0.2221,
      "step": 14651
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2876840858546081,
      "learning_rate": 1.3616139511231085e-06,
      "loss": 0.0296,
      "step": 14652
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4991218271555447,
      "learning_rate": 1.3599996741166908e-06,
      "loss": 0.0874,
      "step": 14653
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4342083720142374,
      "learning_rate": 1.3583863278340215e-06,
      "loss": 0.0958,
      "step": 14654
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9155122887921011,
      "learning_rate": 1.3567739123386324e-06,
      "loss": 0.1439,
      "step": 14655
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6673765258136007,
      "learning_rate": 1.3551624276939929e-06,
      "loss": 0.2528,
      "step": 14656
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4707583066105099,
      "learning_rate": 1.35355187396356e-06,
      "loss": 0.0517,
      "step": 14657
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.39066841230103777,
      "learning_rate": 1.3519422512107306e-06,
      "loss": 0.0726,
      "step": 14658
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.596742328582996,
      "learning_rate": 1.350333559498887e-06,
      "loss": 0.1116,
      "step": 14659
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.48465788871921955,
      "learning_rate": 1.3487257988913593e-06,
      "loss": 0.0989,
      "step": 14660
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3724112757518343,
      "learning_rate": 1.3471189694514496e-06,
      "loss": 0.0597,
      "step": 14661
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6906348036166989,
      "learning_rate": 1.3455130712424102e-06,
      "loss": 0.231,
      "step": 14662
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5324532909978421,
      "learning_rate": 1.3439081043274821e-06,
      "loss": 0.1261,
      "step": 14663
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4941635654873222,
      "learning_rate": 1.3423040687698424e-06,
      "loss": 0.074,
      "step": 14664
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.640894055125635,
      "learning_rate": 1.3407009646326462e-06,
      "loss": 0.1337,
      "step": 14665
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8235673120508352,
      "learning_rate": 1.3390987919790072e-06,
      "loss": 0.2237,
      "step": 14666
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6094404903907668,
      "learning_rate": 1.3374975508720078e-06,
      "loss": 0.0989,
      "step": 14667
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6388543822744426,
      "learning_rate": 1.3358972413746895e-06,
      "loss": 0.1247,
      "step": 14668
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4719097561206721,
      "learning_rate": 1.3342978635500575e-06,
      "loss": 0.0757,
      "step": 14669
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3833742198588169,
      "learning_rate": 1.3326994174610695e-06,
      "loss": 0.0594,
      "step": 14670
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.29947541689729557,
      "learning_rate": 1.3311019031706785e-06,
      "loss": 0.0227,
      "step": 14671
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5342800589853655,
      "learning_rate": 1.329505320741764e-06,
      "loss": 0.1391,
      "step": 14672
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.940478821007155,
      "learning_rate": 1.3279096702371901e-06,
      "loss": 0.3189,
      "step": 14673
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6364419323636131,
      "learning_rate": 1.326314951719776e-06,
      "loss": 0.1455,
      "step": 14674
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.35644840567501124,
      "learning_rate": 1.32472116525231e-06,
      "loss": 0.0364,
      "step": 14675
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3149653155210156,
      "learning_rate": 1.3231283108975367e-06,
      "loss": 0.0315,
      "step": 14676
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.47767386679135804,
      "learning_rate": 1.3215363887181753e-06,
      "loss": 0.0987,
      "step": 14677
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.33126137838234637,
      "learning_rate": 1.3199453987768923e-06,
      "loss": 0.0566,
      "step": 14678
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.53196593277024,
      "learning_rate": 1.3183553411363292e-06,
      "loss": 0.0988,
      "step": 14679
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6364445634521809,
      "learning_rate": 1.3167662158590915e-06,
      "loss": 0.0886,
      "step": 14680
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7224359598694914,
      "learning_rate": 1.3151780230077403e-06,
      "loss": 0.1372,
      "step": 14681
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5956777590316984,
      "learning_rate": 1.3135907626448008e-06,
      "loss": 0.1017,
      "step": 14682
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5860271720983212,
      "learning_rate": 1.3120044348327726e-06,
      "loss": 0.1057,
      "step": 14683
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3330568455983914,
      "learning_rate": 1.310419039634106e-06,
      "loss": 0.0449,
      "step": 14684
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5173524238577686,
      "learning_rate": 1.3088345771112176e-06,
      "loss": 0.1259,
      "step": 14685
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.48191652354368947,
      "learning_rate": 1.307251047326491e-06,
      "loss": 0.0869,
      "step": 14686
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5518759069486329,
      "learning_rate": 1.3056684503422705e-06,
      "loss": 0.0961,
      "step": 14687
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.43662006585388047,
      "learning_rate": 1.3040867862208677e-06,
      "loss": 0.0609,
      "step": 14688
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.40649784100234204,
      "learning_rate": 1.3025060550245467e-06,
      "loss": 0.048,
      "step": 14689
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7457509125159482,
      "learning_rate": 1.300926256815546e-06,
      "loss": 0.1128,
      "step": 14690
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6150768202739414,
      "learning_rate": 1.299347391656064e-06,
      "loss": 0.103,
      "step": 14691
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4055456484964573,
      "learning_rate": 1.297769459608264e-06,
      "loss": 0.0477,
      "step": 14692
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.397724156112647,
      "learning_rate": 1.2961924607342606e-06,
      "loss": 0.0713,
      "step": 14693
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6583743516473811,
      "learning_rate": 1.2946163950961515e-06,
      "loss": 0.1332,
      "step": 14694
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.45668012067917274,
      "learning_rate": 1.293041262755984e-06,
      "loss": 0.0587,
      "step": 14695
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.41627624268866803,
      "learning_rate": 1.2914670637757726e-06,
      "loss": 0.0688,
      "step": 14696
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.37545155897107585,
      "learning_rate": 1.2898937982174925e-06,
      "loss": 0.0377,
      "step": 14697
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3635856386601026,
      "learning_rate": 1.2883214661430858e-06,
      "loss": 0.0554,
      "step": 14698
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7110002712168535,
      "learning_rate": 1.286750067614459e-06,
      "loss": 0.1506,
      "step": 14699
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5595381641999759,
      "learning_rate": 1.285179602693476e-06,
      "loss": 0.0949,
      "step": 14700
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5954105065717556,
      "learning_rate": 1.283610071441968e-06,
      "loss": 0.1431,
      "step": 14701
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3758332097818349,
      "learning_rate": 1.2820414739217274e-06,
      "loss": 0.2934,
      "step": 14702
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4881901389403072,
      "learning_rate": 1.2804738101945102e-06,
      "loss": 0.0818,
      "step": 14703
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4605697961890583,
      "learning_rate": 1.2789070803220421e-06,
      "loss": 0.0692,
      "step": 14704
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5872700067968569,
      "learning_rate": 1.2773412843660015e-06,
      "loss": 0.1086,
      "step": 14705
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5951784360701357,
      "learning_rate": 1.2757764223880336e-06,
      "loss": 0.1423,
      "step": 14706
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4478016173016114,
      "learning_rate": 1.2742124944497502e-06,
      "loss": 0.0732,
      "step": 14707
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3425352014398272,
      "learning_rate": 1.2726495006127298e-06,
      "loss": 0.0415,
      "step": 14708
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.557871070877947,
      "learning_rate": 1.271087440938501e-06,
      "loss": 0.1042,
      "step": 14709
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.49134092637427057,
      "learning_rate": 1.2695263154885618e-06,
      "loss": 0.0607,
      "step": 14710
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7736351513622703,
      "learning_rate": 1.2679661243243852e-06,
      "loss": 0.1757,
      "step": 14711
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0066344486250682,
      "learning_rate": 1.266406867507386e-06,
      "loss": 0.3542,
      "step": 14712
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5086756249760903,
      "learning_rate": 1.2648485450989567e-06,
      "loss": 0.0327,
      "step": 14713
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5690730388045205,
      "learning_rate": 1.2632911571604512e-06,
      "loss": 0.0782,
      "step": 14714
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5229006594744816,
      "learning_rate": 1.2617347037531897e-06,
      "loss": 0.0998,
      "step": 14715
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5507657574222886,
      "learning_rate": 1.2601791849384398e-06,
      "loss": 0.1137,
      "step": 14716
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.32234757148662657,
      "learning_rate": 1.2586246007774528e-06,
      "loss": 0.043,
      "step": 14717
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8365009156213228,
      "learning_rate": 1.2570709513314239e-06,
      "loss": 0.178,
      "step": 14718
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5005495148120513,
      "learning_rate": 1.2555182366615348e-06,
      "loss": 0.0812,
      "step": 14719
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5016553623663245,
      "learning_rate": 1.2539664568289033e-06,
      "loss": 0.113,
      "step": 14720
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.40934361472946157,
      "learning_rate": 1.252415611894639e-06,
      "loss": 0.084,
      "step": 14721
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.43162203098578644,
      "learning_rate": 1.2508657019197818e-06,
      "loss": 0.093,
      "step": 14722
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5961739576569884,
      "learning_rate": 1.249316726965369e-06,
      "loss": 0.1296,
      "step": 14723
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7541718875028617,
      "learning_rate": 1.2477686870923738e-06,
      "loss": 0.1863,
      "step": 14724
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4406754990623127,
      "learning_rate": 1.2462215823617536e-06,
      "loss": 0.0723,
      "step": 14725
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5478427154258128,
      "learning_rate": 1.2446754128344036e-06,
      "loss": 0.1073,
      "step": 14726
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4547982932189166,
      "learning_rate": 1.2431301785712168e-06,
      "loss": 0.1046,
      "step": 14727
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.48324095637698183,
      "learning_rate": 1.2415858796330172e-06,
      "loss": 0.0998,
      "step": 14728
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.44510970827738705,
      "learning_rate": 1.2400425160806084e-06,
      "loss": 0.0722,
      "step": 14729
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5101659921516347,
      "learning_rate": 1.2385000879747533e-06,
      "loss": 0.1495,
      "step": 14730
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4970153213718314,
      "learning_rate": 1.2369585953761837e-06,
      "loss": 0.1093,
      "step": 14731
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5547921326829254,
      "learning_rate": 1.2354180383455787e-06,
      "loss": 0.1346,
      "step": 14732
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6034456687705341,
      "learning_rate": 1.2338784169436013e-06,
      "loss": 0.0982,
      "step": 14733
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2047822706796918,
      "learning_rate": 1.2323397312308583e-06,
      "loss": 0.2802,
      "step": 14734
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.37136168486054555,
      "learning_rate": 1.23080198126794e-06,
      "loss": 0.0501,
      "step": 14735
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2507101842430668,
      "learning_rate": 1.2292651671153788e-06,
      "loss": 0.0367,
      "step": 14736
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.36791436660134674,
      "learning_rate": 1.2277292888336844e-06,
      "loss": 0.0461,
      "step": 14737
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.525716945667002,
      "learning_rate": 1.2261943464833225e-06,
      "loss": 0.1214,
      "step": 14738
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.49039119415423577,
      "learning_rate": 1.2246603401247337e-06,
      "loss": 0.0627,
      "step": 14739
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6110863202774138,
      "learning_rate": 1.2231272698183e-06,
      "loss": 0.0943,
      "step": 14740
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3588595358919095,
      "learning_rate": 1.221595135624387e-06,
      "loss": 0.0488,
      "step": 14741
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3358598984414884,
      "learning_rate": 1.220063937603319e-06,
      "loss": 0.0345,
      "step": 14742
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3752640118653619,
      "learning_rate": 1.2185336758153727e-06,
      "loss": 0.0599,
      "step": 14743
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6801029585108576,
      "learning_rate": 1.2170043503207996e-06,
      "loss": 0.0901,
      "step": 14744
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5285008029110851,
      "learning_rate": 1.21547596117981e-06,
      "loss": 0.0937,
      "step": 14745
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.39864632557014146,
      "learning_rate": 1.2139485084525808e-06,
      "loss": 0.0407,
      "step": 14746
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.37480753777612924,
      "learning_rate": 1.2124219921992418e-06,
      "loss": 0.0595,
      "step": 14747
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.33847326841934444,
      "learning_rate": 1.2108964124799005e-06,
      "loss": 0.0655,
      "step": 14748
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.314525962268837,
      "learning_rate": 1.2093717693546086e-06,
      "loss": 0.0443,
      "step": 14749
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8156252093449866,
      "learning_rate": 1.207848062883407e-06,
      "loss": 0.2694,
      "step": 14750
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.38521942119904423,
      "learning_rate": 1.2063252931262786e-06,
      "loss": 0.0621,
      "step": 14751
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9736655750120085,
      "learning_rate": 1.2048034601431752e-06,
      "loss": 0.2991,
      "step": 14752
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6575117855010338,
      "learning_rate": 1.203282563994007e-06,
      "loss": 0.2148,
      "step": 14753
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.39424380556652694,
      "learning_rate": 1.2017626047386682e-06,
      "loss": 0.0606,
      "step": 14754
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.502247573656232,
      "learning_rate": 1.2002435824369857e-06,
      "loss": 0.1035,
      "step": 14755
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.26570136440921505,
      "learning_rate": 1.1987254971487728e-06,
      "loss": 0.0745,
      "step": 14756
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.42861181308622043,
      "learning_rate": 1.197208348933787e-06,
      "loss": 0.0925,
      "step": 14757
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.33457142722937944,
      "learning_rate": 1.1956921378517782e-06,
      "loss": 0.0272,
      "step": 14758
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5216975554682729,
      "learning_rate": 1.1941768639624234e-06,
      "loss": 0.1031,
      "step": 14759
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3502207082634232,
      "learning_rate": 1.1926625273253917e-06,
      "loss": 0.2345,
      "step": 14760
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5227773490638953,
      "learning_rate": 1.1911491280002906e-06,
      "loss": 0.1201,
      "step": 14761
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3690286015071917,
      "learning_rate": 1.1896366660467173e-06,
      "loss": 0.0439,
      "step": 14762
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4384374309254217,
      "learning_rate": 1.18812514152421e-06,
      "loss": 0.0632,
      "step": 14763
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6916562054927202,
      "learning_rate": 1.1866145544922852e-06,
      "loss": 0.2004,
      "step": 14764
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5567121091404642,
      "learning_rate": 1.1851049050104062e-06,
      "loss": 0.0724,
      "step": 14765
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4199980655340543,
      "learning_rate": 1.1835961931380203e-06,
      "loss": 0.3561,
      "step": 14766
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7417273379715585,
      "learning_rate": 1.1820884189345156e-06,
      "loss": 0.223,
      "step": 14767
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5083757000496552,
      "learning_rate": 1.1805815824592643e-06,
      "loss": 0.0934,
      "step": 14768
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6673217170571106,
      "learning_rate": 1.17907568377158e-06,
      "loss": 0.1727,
      "step": 14769
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.45383523657061164,
      "learning_rate": 1.1775707229307654e-06,
      "loss": 0.0705,
      "step": 14770
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6130900914067552,
      "learning_rate": 1.1760666999960617e-06,
      "loss": 0.1478,
      "step": 14771
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9159739841590241,
      "learning_rate": 1.1745636150266887e-06,
      "loss": 0.1626,
      "step": 14772
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.182838718447469,
      "learning_rate": 1.1730614680818181e-06,
      "loss": 0.1308,
      "step": 14773
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5273164410604633,
      "learning_rate": 1.1715602592205916e-06,
      "loss": 0.135,
      "step": 14774
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.39703548992189297,
      "learning_rate": 1.1700599885021175e-06,
      "loss": 0.0601,
      "step": 14775
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7582458259251045,
      "learning_rate": 1.1685606559854623e-06,
      "loss": 0.2384,
      "step": 14776
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4257603146262292,
      "learning_rate": 1.1670622617296484e-06,
      "loss": 0.0671,
      "step": 14777
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6190012196619188,
      "learning_rate": 1.1655648057936758e-06,
      "loss": 0.1242,
      "step": 14778
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4238257306170363,
      "learning_rate": 1.1640682882365028e-06,
      "loss": 0.0691,
      "step": 14779
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6359870389140017,
      "learning_rate": 1.162572709117038e-06,
      "loss": 0.1306,
      "step": 14780
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8977641651919875,
      "learning_rate": 1.1610780684941703e-06,
      "loss": 0.2327,
      "step": 14781
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5139919888997038,
      "learning_rate": 1.1595843664267442e-06,
      "loss": 0.0794,
      "step": 14782
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3565986372749968,
      "learning_rate": 1.1580916029735682e-06,
      "loss": 0.0633,
      "step": 14783
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5529905557466671,
      "learning_rate": 1.1565997781934125e-06,
      "loss": 0.0793,
      "step": 14784
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4445974434170897,
      "learning_rate": 1.1551088921450097e-06,
      "loss": 0.0746,
      "step": 14785
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.541018849160299,
      "learning_rate": 1.153618944887061e-06,
      "loss": 0.0973,
      "step": 14786
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.536179914495247,
      "learning_rate": 1.1521299364782273e-06,
      "loss": 0.1859,
      "step": 14787
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.31468649712364327,
      "learning_rate": 1.1506418669771229e-06,
      "loss": 0.0636,
      "step": 14788
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.528508952309201,
      "learning_rate": 1.1491547364423427e-06,
      "loss": 0.1908,
      "step": 14789
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4807906387398429,
      "learning_rate": 1.1476685449324343e-06,
      "loss": 0.0758,
      "step": 14790
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7791938211413001,
      "learning_rate": 1.146183292505912e-06,
      "loss": 0.0939,
      "step": 14791
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5407704977577498,
      "learning_rate": 1.1446989792212482e-06,
      "loss": 0.0997,
      "step": 14792
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5663919485349911,
      "learning_rate": 1.1432156051368798e-06,
      "loss": 0.1297,
      "step": 14793
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4427939615066595,
      "learning_rate": 1.14173317031121e-06,
      "loss": 0.0838,
      "step": 14794
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4324698005283171,
      "learning_rate": 1.140251674802606e-06,
      "loss": 0.0635,
      "step": 14795
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8328963484315574,
      "learning_rate": 1.1387711186693933e-06,
      "loss": 0.19,
      "step": 14796
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4683768873571354,
      "learning_rate": 1.1372915019698615e-06,
      "loss": 0.1074,
      "step": 14797
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5569687987240026,
      "learning_rate": 1.135812824762264e-06,
      "loss": 0.0799,
      "step": 14798
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4580314176955938,
      "learning_rate": 1.1343350871048207e-06,
      "loss": 0.0932,
      "step": 14799
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7870640657288133,
      "learning_rate": 1.1328582890557076e-06,
      "loss": 0.1684,
      "step": 14800
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0281588880345114,
      "learning_rate": 1.1313824306730696e-06,
      "loss": 0.2925,
      "step": 14801
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6299779700196521,
      "learning_rate": 1.1299075120150076e-06,
      "loss": 0.124,
      "step": 14802
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8342350331018327,
      "learning_rate": 1.1284335331396002e-06,
      "loss": 0.1471,
      "step": 14803
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.48087964257096155,
      "learning_rate": 1.1269604941048701e-06,
      "loss": 0.1418,
      "step": 14804
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5643089949306425,
      "learning_rate": 1.1254883949688127e-06,
      "loss": 0.0986,
      "step": 14805
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8017377007604076,
      "learning_rate": 1.1240172357893874e-06,
      "loss": 0.123,
      "step": 14806
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8168339046926818,
      "learning_rate": 1.12254701662452e-06,
      "loss": 0.2946,
      "step": 14807
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8416030158776174,
      "learning_rate": 1.1210777375320836e-06,
      "loss": 0.2001,
      "step": 14808
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.423422234810529,
      "learning_rate": 1.1196093985699291e-06,
      "loss": 0.0861,
      "step": 14809
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6781185869645463,
      "learning_rate": 1.1181419997958742e-06,
      "loss": 0.1273,
      "step": 14810
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3421194264237914,
      "learning_rate": 1.1166755412676782e-06,
      "loss": 0.0447,
      "step": 14811
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5402108997849951,
      "learning_rate": 1.1152100230430839e-06,
      "loss": 0.0597,
      "step": 14812
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.48428965976800387,
      "learning_rate": 1.1137454451797868e-06,
      "loss": 0.0862,
      "step": 14813
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5026648121476724,
      "learning_rate": 1.1122818077354547e-06,
      "loss": 0.085,
      "step": 14814
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.34136081451621475,
      "learning_rate": 1.1108191107677052e-06,
      "loss": 0.0409,
      "step": 14815
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.552654421384032,
      "learning_rate": 1.1093573543341312e-06,
      "loss": 0.1488,
      "step": 14816
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.37838622928101473,
      "learning_rate": 1.10789653849227e-06,
      "loss": 0.0528,
      "step": 14817
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4263281060939712,
      "learning_rate": 1.1064366632996563e-06,
      "loss": 0.0737,
      "step": 14818
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.368298433600194,
      "learning_rate": 1.1049777288137497e-06,
      "loss": 0.0569,
      "step": 14819
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4411315868439136,
      "learning_rate": 1.1035197350919984e-06,
      "loss": 0.0507,
      "step": 14820
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.42221766127259736,
      "learning_rate": 1.1020626821917928e-06,
      "loss": 0.0785,
      "step": 14821
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.717806621885921,
      "learning_rate": 1.1006065701705147e-06,
      "loss": 0.1311,
      "step": 14822
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3898601963382641,
      "learning_rate": 1.0991513990854796e-06,
      "loss": 0.0435,
      "step": 14823
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5325555276433532,
      "learning_rate": 1.0976971689939885e-06,
      "loss": 0.1024,
      "step": 14824
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.34627347767186006,
      "learning_rate": 1.096243879953282e-06,
      "loss": 0.0384,
      "step": 14825
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4815337451966359,
      "learning_rate": 1.094791532020592e-06,
      "loss": 0.0751,
      "step": 14826
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8149121507282573,
      "learning_rate": 1.0933401252530896e-06,
      "loss": 0.2223,
      "step": 14827
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3455265651241502,
      "learning_rate": 1.0918896597079204e-06,
      "loss": 0.059,
      "step": 14828
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6138966944674982,
      "learning_rate": 1.0904401354421861e-06,
      "loss": 0.098,
      "step": 14829
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5172037526610274,
      "learning_rate": 1.0889915525129635e-06,
      "loss": 0.0826,
      "step": 14830
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7878161345052384,
      "learning_rate": 1.0875439109772761e-06,
      "loss": 0.425,
      "step": 14831
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7417345134290777,
      "learning_rate": 1.0860972108921257e-06,
      "loss": 0.1599,
      "step": 14832
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3364839719988757,
      "learning_rate": 1.0846514523144612e-06,
      "loss": 0.0523,
      "step": 14833
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.40886503980007854,
      "learning_rate": 1.083206635301215e-06,
      "loss": 0.0555,
      "step": 14834
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7723306254143146,
      "learning_rate": 1.0817627599092605e-06,
      "loss": 0.0973,
      "step": 14835
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3451530469944439,
      "learning_rate": 1.0803198261954528e-06,
      "loss": 0.0404,
      "step": 14836
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9212931242402903,
      "learning_rate": 1.0788778342165878e-06,
      "loss": 0.1525,
      "step": 14837
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3586447599207013,
      "learning_rate": 1.0774367840294535e-06,
      "loss": 0.0757,
      "step": 14838
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8499587190025936,
      "learning_rate": 1.075996675690774e-06,
      "loss": 0.1496,
      "step": 14839
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5618043435882806,
      "learning_rate": 1.074557509257254e-06,
      "loss": 0.0876,
      "step": 14840
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5668398381656667,
      "learning_rate": 1.0731192847855481e-06,
      "loss": 0.1029,
      "step": 14841
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3795311990851467,
      "learning_rate": 1.0716820023322832e-06,
      "loss": 0.0421,
      "step": 14842
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3592862189685558,
      "learning_rate": 1.0702456619540474e-06,
      "loss": 0.0702,
      "step": 14843
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4693275473312454,
      "learning_rate": 1.0688102637073927e-06,
      "loss": 0.0933,
      "step": 14844
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5042532364205927,
      "learning_rate": 1.0673758076488266e-06,
      "loss": 0.1512,
      "step": 14845
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8306231008393047,
      "learning_rate": 1.0659422938348262e-06,
      "loss": 0.1633,
      "step": 14846
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.44843140633971607,
      "learning_rate": 1.0645097223218353e-06,
      "loss": 0.0805,
      "step": 14847
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4432805597133776,
      "learning_rate": 1.063078093166245e-06,
      "loss": 0.0597,
      "step": 14848
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8175454621323203,
      "learning_rate": 1.0616474064244263e-06,
      "loss": 0.2012,
      "step": 14849
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5096957971997242,
      "learning_rate": 1.060217662152707e-06,
      "loss": 0.144,
      "step": 14850
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.35257149586192144,
      "learning_rate": 1.0587888604073754e-06,
      "loss": 0.0538,
      "step": 14851
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6918404162524991,
      "learning_rate": 1.0573610012446834e-06,
      "loss": 0.1513,
      "step": 14852
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.37632203906301265,
      "learning_rate": 1.0559340847208472e-06,
      "loss": 0.0561,
      "step": 14853
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8722760811783351,
      "learning_rate": 1.0545081108920468e-06,
      "loss": 0.2414,
      "step": 14854
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5710760628174648,
      "learning_rate": 1.0530830798144237e-06,
      "loss": 0.1305,
      "step": 14855
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4010999656003445,
      "learning_rate": 1.05165899154408e-06,
      "loss": 0.0607,
      "step": 14856
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5449876304912322,
      "learning_rate": 1.0502358461370875e-06,
      "loss": 0.0733,
      "step": 14857
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3403813617366787,
      "learning_rate": 1.048813643649471e-06,
      "loss": 0.0694,
      "step": 14858
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3659387631673954,
      "learning_rate": 1.0473923841372302e-06,
      "loss": 0.0479,
      "step": 14859
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.46081203492445943,
      "learning_rate": 1.0459720676563116e-06,
      "loss": 0.0987,
      "step": 14860
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7274166533949675,
      "learning_rate": 1.0445526942626489e-06,
      "loss": 0.1354,
      "step": 14861
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6353254222377452,
      "learning_rate": 1.0431342640121077e-06,
      "loss": 0.1247,
      "step": 14862
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7022344315714772,
      "learning_rate": 1.041716776960544e-06,
      "loss": 0.083,
      "step": 14863
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4485338688434794,
      "learning_rate": 1.0403002331637547e-06,
      "loss": 0.0555,
      "step": 14864
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5098579341360812,
      "learning_rate": 1.0388846326775254e-06,
      "loss": 0.0789,
      "step": 14865
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4078475358330599,
      "learning_rate": 1.0374699755575756e-06,
      "loss": 0.0569,
      "step": 14866
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7096097057728401,
      "learning_rate": 1.0360562618596082e-06,
      "loss": 0.1297,
      "step": 14867
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5226344620589882,
      "learning_rate": 1.0346434916392783e-06,
      "loss": 0.0626,
      "step": 14868
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5305107167261326,
      "learning_rate": 1.0332316649522084e-06,
      "loss": 0.0603,
      "step": 14869
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4556318134281494,
      "learning_rate": 1.0318207818539842e-06,
      "loss": 0.0848,
      "step": 14870
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3577163818716593,
      "learning_rate": 1.0304108424001562e-06,
      "loss": 0.0352,
      "step": 14871
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9736961010906815,
      "learning_rate": 1.0290018466462297e-06,
      "loss": 0.2614,
      "step": 14872
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5497302879166109,
      "learning_rate": 1.02759379464768e-06,
      "loss": 0.1317,
      "step": 14873
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9532528551468475,
      "learning_rate": 1.026186686459943e-06,
      "loss": 0.4113,
      "step": 14874
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.39960003231741287,
      "learning_rate": 1.0247805221384187e-06,
      "loss": 0.0495,
      "step": 14875
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6864041504050745,
      "learning_rate": 1.023375301738466e-06,
      "loss": 0.1011,
      "step": 14876
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3728380144465383,
      "learning_rate": 1.021971025315413e-06,
      "loss": 0.0724,
      "step": 14877
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.49059932318793226,
      "learning_rate": 1.0205676929245455e-06,
      "loss": 0.0881,
      "step": 14878
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5424310158363077,
      "learning_rate": 1.0191653046211113e-06,
      "loss": 0.4178,
      "step": 14879
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.42187025170789927,
      "learning_rate": 1.0177638604603245e-06,
      "loss": 0.0568,
      "step": 14880
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6488920598400452,
      "learning_rate": 1.0163633604973633e-06,
      "loss": 0.1228,
      "step": 14881
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5258000642020987,
      "learning_rate": 1.0149638047873694e-06,
      "loss": 0.173,
      "step": 14882
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8390712028184081,
      "learning_rate": 1.0135651933854351e-06,
      "loss": 0.5352,
      "step": 14883
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.544905158901862,
      "learning_rate": 1.0121675263466302e-06,
      "loss": 0.0986,
      "step": 14884
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9234625567550231,
      "learning_rate": 1.01077080372598e-06,
      "loss": 0.2764,
      "step": 14885
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6016940711152357,
      "learning_rate": 1.0093750255784794e-06,
      "loss": 0.1048,
      "step": 14886
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4391005683499052,
      "learning_rate": 1.0079801919590765e-06,
      "loss": 0.0852,
      "step": 14887
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5201294565035517,
      "learning_rate": 1.0065863029226852e-06,
      "loss": 0.0811,
      "step": 14888
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6539172711272564,
      "learning_rate": 1.0051933585241867e-06,
      "loss": 0.1127,
      "step": 14889
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.44835657552375807,
      "learning_rate": 1.0038013588184264e-06,
      "loss": 0.0714,
      "step": 14890
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.620638736591517,
      "learning_rate": 1.0024103038602017e-06,
      "loss": 0.0762,
      "step": 14891
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5729950356528983,
      "learning_rate": 1.00102019370428e-06,
      "loss": 0.1269,
      "step": 14892
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3845912222040592,
      "learning_rate": 9.996310284053928e-07,
      "loss": 0.0607,
      "step": 14893
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8248645580331766,
      "learning_rate": 9.982428080182376e-07,
      "loss": 0.1793,
      "step": 14894
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3958042775205276,
      "learning_rate": 9.9685553259746e-07,
      "loss": 0.0508,
      "step": 14895
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5890304388175556,
      "learning_rate": 9.954692021976824e-07,
      "loss": 0.1371,
      "step": 14896
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5424090673655321,
      "learning_rate": 9.94083816873484e-07,
      "loss": 0.0867,
      "step": 14897
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.45178300392211085,
      "learning_rate": 9.926993766794152e-07,
      "loss": 0.056,
      "step": 14898
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5061796465422567,
      "learning_rate": 9.913158816699769e-07,
      "loss": 0.0989,
      "step": 14899
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1899101067150535,
      "learning_rate": 9.89933331899634e-07,
      "loss": 0.1498,
      "step": 14900
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.479663889486702,
      "learning_rate": 9.885517274228262e-07,
      "loss": 0.106,
      "step": 14901
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.362626160041988,
      "learning_rate": 9.871710682939462e-07,
      "loss": 0.0344,
      "step": 14902
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7543184786289326,
      "learning_rate": 9.857913545673502e-07,
      "loss": 0.0798,
      "step": 14903
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7481120655619189,
      "learning_rate": 9.844125862973564e-07,
      "loss": 0.1826,
      "step": 14904
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.62049795957535,
      "learning_rate": 9.830347635382515e-07,
      "loss": 0.0791,
      "step": 14905
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.507430150899149,
      "learning_rate": 9.816578863442837e-07,
      "loss": 0.0768,
      "step": 14906
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4374701128273734,
      "learning_rate": 9.802819547696573e-07,
      "loss": 0.0995,
      "step": 14907
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.408575780774961,
      "learning_rate": 9.789069688685399e-07,
      "loss": 0.076,
      "step": 14908
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5278061289719159,
      "learning_rate": 9.77532928695077e-07,
      "loss": 0.0726,
      "step": 14909
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3047695731504132,
      "learning_rate": 9.761598343033556e-07,
      "loss": 0.0416,
      "step": 14910
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5233710468885949,
      "learning_rate": 9.747876857474387e-07,
      "loss": 0.0741,
      "step": 14911
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.34578321106511345,
      "learning_rate": 9.734164830813463e-07,
      "loss": 0.0784,
      "step": 14912
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.28996837299229883,
      "learning_rate": 9.720462263590718e-07,
      "loss": 0.0426,
      "step": 14913
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5546971410492825,
      "learning_rate": 9.706769156345552e-07,
      "loss": 0.1246,
      "step": 14914
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.375816931400009,
      "learning_rate": 9.693085509617118e-07,
      "loss": 0.0836,
      "step": 14915
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4657562503981607,
      "learning_rate": 9.679411323944037e-07,
      "loss": 0.0721,
      "step": 14916
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9342557467013726,
      "learning_rate": 9.665746599864856e-07,
      "loss": 0.2297,
      "step": 14917
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4446915833599604,
      "learning_rate": 9.652091337917446e-07,
      "loss": 0.0624,
      "step": 14918
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7671352224783368,
      "learning_rate": 9.63844553863949e-07,
      "loss": 0.2867,
      "step": 14919
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.617858123069472,
      "learning_rate": 9.624809202568114e-07,
      "loss": 0.0909,
      "step": 14920
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.39127351215753414,
      "learning_rate": 9.61118233024033e-07,
      "loss": 0.0689,
      "step": 14921
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4592430239045101,
      "learning_rate": 9.597564922192547e-07,
      "loss": 0.11,
      "step": 14922
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.35871147961248384,
      "learning_rate": 9.583956978960973e-07,
      "loss": 0.041,
      "step": 14923
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6382371428637437,
      "learning_rate": 9.570358501081233e-07,
      "loss": 0.1892,
      "step": 14924
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6972951418165378,
      "learning_rate": 9.556769489088846e-07,
      "loss": 0.0847,
      "step": 14925
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.36228588108648463,
      "learning_rate": 9.543189943518744e-07,
      "loss": 0.0444,
      "step": 14926
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5650869446584662,
      "learning_rate": 9.529619864905609e-07,
      "loss": 0.1006,
      "step": 14927
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.33212586324681187,
      "learning_rate": 9.516059253783627e-07,
      "loss": 0.0384,
      "step": 14928
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0452048880576836,
      "learning_rate": 9.502508110686786e-07,
      "loss": 0.3152,
      "step": 14929
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.42050889305992395,
      "learning_rate": 9.48896643614855e-07,
      "loss": 0.0956,
      "step": 14930
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.49832556335902334,
      "learning_rate": 9.475434230702101e-07,
      "loss": 0.1205,
      "step": 14931
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.44016986385727036,
      "learning_rate": 9.461911494880099e-07,
      "loss": 0.0743,
      "step": 14932
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9839109988077985,
      "learning_rate": 9.448398229215117e-07,
      "loss": 0.4116,
      "step": 14933
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6185657900990162,
      "learning_rate": 9.434894434239061e-07,
      "loss": 0.1409,
      "step": 14934
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5131064315217408,
      "learning_rate": 9.421400110483619e-07,
      "loss": 0.0754,
      "step": 14935
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4304177036323659,
      "learning_rate": 9.40791525848006e-07,
      "loss": 0.0723,
      "step": 14936
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.348088036367005,
      "learning_rate": 9.39443987875932e-07,
      "loss": 0.6078,
      "step": 14937
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4042628242637235,
      "learning_rate": 9.380973971851892e-07,
      "loss": 0.0479,
      "step": 14938
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.736102073798761,
      "learning_rate": 9.367517538288018e-07,
      "loss": 0.1204,
      "step": 14939
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6297608457510462,
      "learning_rate": 9.354070578597385e-07,
      "loss": 0.1803,
      "step": 14940
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5449584076000775,
      "learning_rate": 9.340633093309459e-07,
      "loss": 0.0758,
      "step": 14941
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5973458526699136,
      "learning_rate": 9.327205082953289e-07,
      "loss": 0.0885,
      "step": 14942
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8021505778099259,
      "learning_rate": 9.313786548057562e-07,
      "loss": 0.1431,
      "step": 14943
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5668906397761386,
      "learning_rate": 9.300377489150552e-07,
      "loss": 0.1365,
      "step": 14944
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3559233572760417,
      "learning_rate": 9.286977906760142e-07,
      "loss": 0.0431,
      "step": 14945
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5631736398354186,
      "learning_rate": 9.273587801413991e-07,
      "loss": 0.1286,
      "step": 14946
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8440491602563679,
      "learning_rate": 9.260207173639152e-07,
      "loss": 0.3347,
      "step": 14947
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5338054249583201,
      "learning_rate": 9.246836023962507e-07,
      "loss": 0.1845,
      "step": 14948
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6624059244840043,
      "learning_rate": 9.233474352910443e-07,
      "loss": 0.1684,
      "step": 14949
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7322545440151123,
      "learning_rate": 9.220122161009092e-07,
      "loss": 0.243,
      "step": 14950
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5697154771041436,
      "learning_rate": 9.206779448784064e-07,
      "loss": 0.1057,
      "step": 14951
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.48439117636449464,
      "learning_rate": 9.193446216760715e-07,
      "loss": 0.1111,
      "step": 14952
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4893755916888932,
      "learning_rate": 9.180122465463959e-07,
      "loss": 0.0875,
      "step": 14953
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3661122214026589,
      "learning_rate": 9.166808195418375e-07,
      "loss": 0.0359,
      "step": 14954
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4169637199429676,
      "learning_rate": 9.153503407148157e-07,
      "loss": 0.1046,
      "step": 14955
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7778412066684677,
      "learning_rate": 9.140208101177105e-07,
      "loss": 0.2306,
      "step": 14956
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.41245688435359296,
      "learning_rate": 9.126922278028666e-07,
      "loss": 0.0963,
      "step": 14957
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5689151202682893,
      "learning_rate": 9.113645938225973e-07,
      "loss": 0.0963,
      "step": 14958
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.38882104654117766,
      "learning_rate": 9.100379082291638e-07,
      "loss": 0.0566,
      "step": 14959
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7404125452382389,
      "learning_rate": 9.08712171074802e-07,
      "loss": 0.1971,
      "step": 14960
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8724327585294853,
      "learning_rate": 9.073873824117091e-07,
      "loss": 0.2255,
      "step": 14961
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6674735032616077,
      "learning_rate": 9.060635422920433e-07,
      "loss": 0.4774,
      "step": 14962
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5813467887774096,
      "learning_rate": 9.047406507679184e-07,
      "loss": 0.0687,
      "step": 14963
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3741858136700171,
      "learning_rate": 9.034187078914264e-07,
      "loss": 0.0523,
      "step": 14964
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6058213657352743,
      "learning_rate": 9.020977137146058e-07,
      "loss": 0.1634,
      "step": 14965
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5791035872005542,
      "learning_rate": 9.007776682894736e-07,
      "loss": 0.0712,
      "step": 14966
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.34450953422300146,
      "learning_rate": 8.994585716679938e-07,
      "loss": 0.0549,
      "step": 14967
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5239250882980562,
      "learning_rate": 8.981404239021024e-07,
      "loss": 0.0884,
      "step": 14968
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.726825492543408,
      "learning_rate": 8.968232250436942e-07,
      "loss": 0.2111,
      "step": 14969
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5119568569650673,
      "learning_rate": 8.955069751446332e-07,
      "loss": 0.122,
      "step": 14970
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6835555677951833,
      "learning_rate": 8.941916742567336e-07,
      "loss": 0.1145,
      "step": 14971
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.38403145953455536,
      "learning_rate": 8.928773224317843e-07,
      "loss": 0.0548,
      "step": 14972
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5073230263860131,
      "learning_rate": 8.915639197215359e-07,
      "loss": 0.0925,
      "step": 14973
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5774926652413477,
      "learning_rate": 8.902514661776884e-07,
      "loss": 0.1643,
      "step": 14974
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.985474280118007,
      "learning_rate": 8.889399618519229e-07,
      "loss": 0.2679,
      "step": 14975
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.51740846832688,
      "learning_rate": 8.876294067958673e-07,
      "loss": 0.081,
      "step": 14976
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5305268184372907,
      "learning_rate": 8.863198010611279e-07,
      "loss": 0.1732,
      "step": 14977
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6528621755167707,
      "learning_rate": 8.850111446992576e-07,
      "loss": 0.1975,
      "step": 14978
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5681717005312186,
      "learning_rate": 8.837034377617792e-07,
      "loss": 0.0737,
      "step": 14979
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5702358756015115,
      "learning_rate": 8.823966803001821e-07,
      "loss": 0.1196,
      "step": 14980
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.43815710095125865,
      "learning_rate": 8.810908723659139e-07,
      "loss": 0.084,
      "step": 14981
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4443676549596026,
      "learning_rate": 8.797860140103781e-07,
      "loss": 0.0683,
      "step": 14982
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4450521719812892,
      "learning_rate": 8.784821052849584e-07,
      "loss": 0.0632,
      "step": 14983
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4713458467310934,
      "learning_rate": 8.771791462409807e-07,
      "loss": 0.0833,
      "step": 14984
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9553159005438403,
      "learning_rate": 8.758771369297536e-07,
      "loss": 0.2465,
      "step": 14985
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4295757796062766,
      "learning_rate": 8.745760774025281e-07,
      "loss": 0.1019,
      "step": 14986
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6657558331150222,
      "learning_rate": 8.732759677105379e-07,
      "loss": 0.1539,
      "step": 14987
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.28671306510614925,
      "learning_rate": 8.719768079049561e-07,
      "loss": 0.0768,
      "step": 14988
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3497398341884177,
      "learning_rate": 8.706785980369448e-07,
      "loss": 0.3903,
      "step": 14989
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.43813178222361066,
      "learning_rate": 8.693813381576099e-07,
      "loss": 0.1074,
      "step": 14990
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5429267418412514,
      "learning_rate": 8.680850283180248e-07,
      "loss": 0.1028,
      "step": 14991
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.699671118157193,
      "learning_rate": 8.667896685692234e-07,
      "loss": 0.0912,
      "step": 14992
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4007550229681726,
      "learning_rate": 8.654952589622151e-07,
      "loss": 0.0632,
      "step": 14993
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6481720449215662,
      "learning_rate": 8.642017995479507e-07,
      "loss": 0.1492,
      "step": 14994
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6706644878750765,
      "learning_rate": 8.62909290377359e-07,
      "loss": 0.1436,
      "step": 14995
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5713158863033384,
      "learning_rate": 8.616177315013268e-07,
      "loss": 0.0619,
      "step": 14996
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5658728352411316,
      "learning_rate": 8.603271229707055e-07,
      "loss": 0.0812,
      "step": 14997
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9911186949413329,
      "learning_rate": 8.590374648363042e-07,
      "loss": 0.1623,
      "step": 14998
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3902119742296417,
      "learning_rate": 8.57748757148899e-07,
      "loss": 0.0454,
      "step": 14999
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.546299131427702,
      "learning_rate": 8.564609999592272e-07,
      "loss": 0.116,
      "step": 15000
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5837800859582019,
      "learning_rate": 8.551741933179924e-07,
      "loss": 0.0777,
      "step": 15001
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9768223365676008,
      "learning_rate": 8.538883372758516e-07,
      "loss": 0.1742,
      "step": 15002
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4886623465933471,
      "learning_rate": 8.526034318834309e-07,
      "loss": 0.0774,
      "step": 15003
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5202385981064814,
      "learning_rate": 8.513194771913202e-07,
      "loss": 0.0881,
      "step": 15004
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8930243093723387,
      "learning_rate": 8.500364732500682e-07,
      "loss": 0.1757,
      "step": 15005
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5645054097084662,
      "learning_rate": 8.487544201101843e-07,
      "loss": 0.0838,
      "step": 15006
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3179221035059152,
      "learning_rate": 8.474733178221505e-07,
      "loss": 0.0539,
      "step": 15007
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5690658623805082,
      "learning_rate": 8.461931664364042e-07,
      "loss": 0.1033,
      "step": 15008
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4650497754325987,
      "learning_rate": 8.449139660033411e-07,
      "loss": 0.0592,
      "step": 15009
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5561983020295861,
      "learning_rate": 8.436357165733238e-07,
      "loss": 0.1393,
      "step": 15010
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4442107205881482,
      "learning_rate": 8.423584181966815e-07,
      "loss": 0.0447,
      "step": 15011
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4106003442873629,
      "learning_rate": 8.410820709237072e-07,
      "loss": 0.0697,
      "step": 15012
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5747123339323053,
      "learning_rate": 8.398066748046412e-07,
      "loss": 0.1541,
      "step": 15013
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4147270168975559,
      "learning_rate": 8.385322298897019e-07,
      "loss": 0.0579,
      "step": 15014
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.363080362182723,
      "learning_rate": 8.37258736229063e-07,
      "loss": 0.0618,
      "step": 15015
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.34638092335046367,
      "learning_rate": 8.359861938728675e-07,
      "loss": 0.0342,
      "step": 15016
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.38188152311496576,
      "learning_rate": 8.347146028712117e-07,
      "loss": 0.0499,
      "step": 15017
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.33253919333312615,
      "learning_rate": 8.334439632741609e-07,
      "loss": 0.0469,
      "step": 15018
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7079191932831883,
      "learning_rate": 8.321742751317363e-07,
      "loss": 0.1534,
      "step": 15019
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.34792341632722484,
      "learning_rate": 8.309055384939396e-07,
      "loss": 0.0403,
      "step": 15020
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7566459612830848,
      "learning_rate": 8.296377534107086e-07,
      "loss": 0.1826,
      "step": 15021
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6254897465049033,
      "learning_rate": 8.283709199319617e-07,
      "loss": 0.1191,
      "step": 15022
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6905085219262083,
      "learning_rate": 8.271050381075729e-07,
      "loss": 0.1228,
      "step": 15023
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8099602608780402,
      "learning_rate": 8.258401079873857e-07,
      "loss": 0.23,
      "step": 15024
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1096921163859363,
      "learning_rate": 8.245761296211991e-07,
      "loss": 0.5367,
      "step": 15025
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.40230779590882854,
      "learning_rate": 8.233131030587759e-07,
      "loss": 0.0442,
      "step": 15026
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3688243372673318,
      "learning_rate": 8.220510283498378e-07,
      "loss": 0.0735,
      "step": 15027
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5363474745802413,
      "learning_rate": 8.207899055440837e-07,
      "loss": 0.0884,
      "step": 15028
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9517371857094322,
      "learning_rate": 8.195297346911573e-07,
      "loss": 0.2954,
      "step": 15029
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3171068425831853,
      "learning_rate": 8.182705158406773e-07,
      "loss": 0.0664,
      "step": 15030
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4437768617498894,
      "learning_rate": 8.170122490422121e-07,
      "loss": 0.0818,
      "step": 15031
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5691256110733283,
      "learning_rate": 8.157549343453141e-07,
      "loss": 0.1268,
      "step": 15032
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5017417950452513,
      "learning_rate": 8.144985717994713e-07,
      "loss": 0.1281,
      "step": 15033
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.23364946411305262,
      "learning_rate": 8.132431614541552e-07,
      "loss": 0.0314,
      "step": 15034
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6961123294680172,
      "learning_rate": 8.119887033587875e-07,
      "loss": 0.5137,
      "step": 15035
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4291638752800655,
      "learning_rate": 8.107351975627592e-07,
      "loss": 0.0603,
      "step": 15036
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.405412200766539,
      "learning_rate": 8.094826441154224e-07,
      "loss": 0.0887,
      "step": 15037
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9455394289652499,
      "learning_rate": 8.082310430660933e-07,
      "loss": 0.3203,
      "step": 15038
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4797498535316998,
      "learning_rate": 8.069803944640436e-07,
      "loss": 0.0865,
      "step": 15039
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3114984999852257,
      "learning_rate": 8.057306983585145e-07,
      "loss": 0.0433,
      "step": 15040
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.38245674924865164,
      "learning_rate": 8.044819547987081e-07,
      "loss": 0.066,
      "step": 15041
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3263911015758575,
      "learning_rate": 8.032341638337881e-07,
      "loss": 0.0423,
      "step": 15042
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5067225812288699,
      "learning_rate": 8.019873255128762e-07,
      "loss": 0.132,
      "step": 15043
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4114394982561369,
      "learning_rate": 8.007414398850665e-07,
      "loss": 0.0736,
      "step": 15044
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.488184641516581,
      "learning_rate": 7.994965069994142e-07,
      "loss": 0.1227,
      "step": 15045
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.386764493382797,
      "learning_rate": 7.982525269049246e-07,
      "loss": 0.7728,
      "step": 15046
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9415920916682771,
      "learning_rate": 7.970094996505778e-07,
      "loss": 0.2254,
      "step": 15047
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4414265260119798,
      "learning_rate": 7.957674252853125e-07,
      "loss": 0.1071,
      "step": 15048
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9446748450395721,
      "learning_rate": 7.945263038580313e-07,
      "loss": 0.329,
      "step": 15049
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3582649429940202,
      "learning_rate": 7.932861354175952e-07,
      "loss": 0.0767,
      "step": 15050
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7401604950901682,
      "learning_rate": 7.920469200128344e-07,
      "loss": 0.1756,
      "step": 15051
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4635230276135766,
      "learning_rate": 7.908086576925322e-07,
      "loss": 0.1216,
      "step": 15052
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6659654614119459,
      "learning_rate": 7.895713485054495e-07,
      "loss": 0.2158,
      "step": 15053
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0397690098545307,
      "learning_rate": 7.883349925002892e-07,
      "loss": 0.2052,
      "step": 15054
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5793888972230704,
      "learning_rate": 7.870995897257316e-07,
      "loss": 0.1472,
      "step": 15055
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4334703851974497,
      "learning_rate": 7.858651402304157e-07,
      "loss": 0.135,
      "step": 15056
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6897938163736512,
      "learning_rate": 7.84631644062947e-07,
      "loss": 0.1502,
      "step": 15057
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.41133697495301796,
      "learning_rate": 7.833991012718811e-07,
      "loss": 0.0762,
      "step": 15058
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5680916493337189,
      "learning_rate": 7.821675119057486e-07,
      "loss": 0.0914,
      "step": 15059
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1269781586631953,
      "learning_rate": 7.809368760130386e-07,
      "loss": 0.1057,
      "step": 15060
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6407934629573089,
      "learning_rate": 7.797071936422013e-07,
      "loss": 0.1975,
      "step": 15061
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5827950515118325,
      "learning_rate": 7.784784648416477e-07,
      "loss": 0.1032,
      "step": 15062
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5468470483460195,
      "learning_rate": 7.772506896597586e-07,
      "loss": 0.1751,
      "step": 15063
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.42409515789858104,
      "learning_rate": 7.760238681448679e-07,
      "loss": 0.1011,
      "step": 15064
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2958607815844196,
      "learning_rate": 7.747980003452809e-07,
      "loss": 0.0329,
      "step": 15065
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.39506894338236304,
      "learning_rate": 7.735730863092538e-07,
      "loss": 0.062,
      "step": 15066
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6107946187454094,
      "learning_rate": 7.723491260850174e-07,
      "loss": 0.0934,
      "step": 15067
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3227178708993708,
      "learning_rate": 7.711261197207609e-07,
      "loss": 0.0571,
      "step": 15068
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5567339907323176,
      "learning_rate": 7.699040672646345e-07,
      "loss": 0.1177,
      "step": 15069
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.41318295610381145,
      "learning_rate": 7.686829687647501e-07,
      "loss": 0.0739,
      "step": 15070
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.43057224915577347,
      "learning_rate": 7.674628242691801e-07,
      "loss": 0.0608,
      "step": 15071
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3013799178745583,
      "learning_rate": 7.662436338259693e-07,
      "loss": 0.0548,
      "step": 15072
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.48059337537446434,
      "learning_rate": 7.650253974831129e-07,
      "loss": 0.0738,
      "step": 15073
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.39681766879964075,
      "learning_rate": 7.638081152885724e-07,
      "loss": 0.0435,
      "step": 15074
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8395548810286294,
      "learning_rate": 7.625917872902788e-07,
      "loss": 0.257,
      "step": 15075
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.616562639934245,
      "learning_rate": 7.613764135361162e-07,
      "loss": 0.1509,
      "step": 15076
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5670395983857971,
      "learning_rate": 7.601619940739352e-07,
      "loss": 0.1953,
      "step": 15077
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.25710249690984127,
      "learning_rate": 7.589485289515503e-07,
      "loss": 0.0303,
      "step": 15078
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5052582969474826,
      "learning_rate": 7.577360182167314e-07,
      "loss": 0.121,
      "step": 15079
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.083466539181271,
      "learning_rate": 7.565244619172213e-07,
      "loss": 0.4173,
      "step": 15080
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6241745019532886,
      "learning_rate": 7.553138601007176e-07,
      "loss": 0.1153,
      "step": 15081
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5724019905202484,
      "learning_rate": 7.541042128148851e-07,
      "loss": 0.0788,
      "step": 15082
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4388499820029803,
      "learning_rate": 7.528955201073412e-07,
      "loss": 0.5703,
      "step": 15083
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4026651588982162,
      "learning_rate": 7.516877820256813e-07,
      "loss": 0.0517,
      "step": 15084
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5943657077745529,
      "learning_rate": 7.504809986174504e-07,
      "loss": 0.0846,
      "step": 15085
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5191015900028191,
      "learning_rate": 7.492751699301636e-07,
      "loss": 0.0944,
      "step": 15086
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.40248903621515136,
      "learning_rate": 7.480702960112884e-07,
      "loss": 0.0683,
      "step": 15087
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5483419568837568,
      "learning_rate": 7.468663769082729e-07,
      "loss": 0.0915,
      "step": 15088
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.42097689980331415,
      "learning_rate": 7.456634126685069e-07,
      "loss": 0.0425,
      "step": 15089
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6030963474261904,
      "learning_rate": 7.444614033393555e-07,
      "loss": 0.2038,
      "step": 15090
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7910437115368765,
      "learning_rate": 7.432603489681361e-07,
      "loss": 0.1903,
      "step": 15091
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6458407624328412,
      "learning_rate": 7.420602496021501e-07,
      "loss": 0.1569,
      "step": 15092
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5202926152236717,
      "learning_rate": 7.408611052886316e-07,
      "loss": 0.1513,
      "step": 15093
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8075469054450687,
      "learning_rate": 7.396629160747987e-07,
      "loss": 0.1203,
      "step": 15094
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2747123347356982,
      "learning_rate": 7.384656820078217e-07,
      "loss": 0.0363,
      "step": 15095
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1115173944967978,
      "learning_rate": 7.37269403134841e-07,
      "loss": 0.2775,
      "step": 15096
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5671444042315164,
      "learning_rate": 7.360740795029519e-07,
      "loss": 0.1161,
      "step": 15097
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5197473898966368,
      "learning_rate": 7.34879711159217e-07,
      "loss": 0.0779,
      "step": 15098
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3139140545922989,
      "learning_rate": 7.336862981506515e-07,
      "loss": 0.0328,
      "step": 15099
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5101698124356759,
      "learning_rate": 7.324938405242565e-07,
      "loss": 0.1806,
      "step": 15100
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6150895155606789,
      "learning_rate": 7.31302338326964e-07,
      "loss": 0.1129,
      "step": 15101
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5335994283157537,
      "learning_rate": 7.301117916056977e-07,
      "loss": 0.6064,
      "step": 15102
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.31050263113734927,
      "learning_rate": 7.28922200407317e-07,
      "loss": 0.0617,
      "step": 15103
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4621306924636885,
      "learning_rate": 7.277335647786682e-07,
      "loss": 0.0994,
      "step": 15104
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4168972597312778,
      "learning_rate": 7.265458847665413e-07,
      "loss": 0.0507,
      "step": 15105
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4221619498556871,
      "learning_rate": 7.253591604177018e-07,
      "loss": 0.0993,
      "step": 15106
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5628689459355012,
      "learning_rate": 7.241733917788679e-07,
      "loss": 0.0682,
      "step": 15107
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5274085979634594,
      "learning_rate": 7.229885788967244e-07,
      "loss": 0.1656,
      "step": 15108
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.322609844540527,
      "learning_rate": 7.218047218179231e-07,
      "loss": 0.2709,
      "step": 15109
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1528323573511858,
      "learning_rate": 7.206218205890652e-07,
      "loss": 0.3842,
      "step": 15110
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6597274024945609,
      "learning_rate": 7.194398752567277e-07,
      "loss": 0.1932,
      "step": 15111
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.557031425370441,
      "learning_rate": 7.182588858674427e-07,
      "loss": 0.1268,
      "step": 15112
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5363915601934682,
      "learning_rate": 7.170788524677119e-07,
      "loss": 0.0766,
      "step": 15113
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.32993777496025606,
      "learning_rate": 7.158997751039842e-07,
      "loss": 0.0821,
      "step": 15114
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5171258411781553,
      "learning_rate": 7.147216538226892e-07,
      "loss": 0.0782,
      "step": 15115
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6297383069072799,
      "learning_rate": 7.135444886702064e-07,
      "loss": 0.1695,
      "step": 15116
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5093626728778328,
      "learning_rate": 7.123682796928849e-07,
      "loss": 0.1261,
      "step": 15117
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.38997917943778077,
      "learning_rate": 7.111930269370265e-07,
      "loss": 0.0414,
      "step": 15118
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5907266167004549,
      "learning_rate": 7.100187304489081e-07,
      "loss": 0.1068,
      "step": 15119
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5925379010653384,
      "learning_rate": 7.088453902747594e-07,
      "loss": 0.1118,
      "step": 15120
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4327076999183405,
      "learning_rate": 7.076730064607795e-07,
      "loss": 0.0674,
      "step": 15121
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7679291877644056,
      "learning_rate": 7.065015790531177e-07,
      "loss": 0.2662,
      "step": 15122
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4621430397339221,
      "learning_rate": 7.053311080979036e-07,
      "loss": 0.0838,
      "step": 15123
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4227816737951814,
      "learning_rate": 7.041615936412115e-07,
      "loss": 0.0832,
      "step": 15124
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.40531675906687853,
      "learning_rate": 7.029930357290937e-07,
      "loss": 0.0565,
      "step": 15125
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.537175906269103,
      "learning_rate": 7.018254344075492e-07,
      "loss": 0.0512,
      "step": 15126
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4851703204066147,
      "learning_rate": 7.006587897225552e-07,
      "loss": 0.0811,
      "step": 15127
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5560286009008587,
      "learning_rate": 6.994931017200362e-07,
      "loss": 0.1703,
      "step": 15128
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5116031091596003,
      "learning_rate": 6.983283704458915e-07,
      "loss": 0.1915,
      "step": 15129
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6818349889744749,
      "learning_rate": 6.971645959459732e-07,
      "loss": 0.1384,
      "step": 15130
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4423971609434029,
      "learning_rate": 6.960017782661032e-07,
      "loss": 0.0819,
      "step": 15131
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3507881792283839,
      "learning_rate": 6.948399174520614e-07,
      "loss": 0.0791,
      "step": 15132
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3679587240340167,
      "learning_rate": 6.936790135495918e-07,
      "loss": 0.0476,
      "step": 15133
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3501113260810365,
      "learning_rate": 6.925190666043968e-07,
      "loss": 0.0748,
      "step": 15134
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7198109480330378,
      "learning_rate": 6.913600766621481e-07,
      "loss": 0.1745,
      "step": 15135
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7313539265362802,
      "learning_rate": 6.90202043768473e-07,
      "loss": 0.1711,
      "step": 15136
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5048603325500286,
      "learning_rate": 6.890449679689714e-07,
      "loss": 0.074,
      "step": 15137
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3623537614680402,
      "learning_rate": 6.878888493091873e-07,
      "loss": 0.0463,
      "step": 15138
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6848417303916854,
      "learning_rate": 6.867336878346425e-07,
      "loss": 0.2547,
      "step": 15139
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.48688112832554487,
      "learning_rate": 6.855794835908203e-07,
      "loss": 0.0905,
      "step": 15140
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5266062081565828,
      "learning_rate": 6.844262366231563e-07,
      "loss": 0.1111,
      "step": 15141
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5359064831197297,
      "learning_rate": 6.832739469770588e-07,
      "loss": 0.0999,
      "step": 15142
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.41578668143240444,
      "learning_rate": 6.821226146978942e-07,
      "loss": 0.0659,
      "step": 15143
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3003925222426145,
      "learning_rate": 6.809722398309904e-07,
      "loss": 0.0437,
      "step": 15144
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.43189040301107967,
      "learning_rate": 6.798228224216358e-07,
      "loss": 0.085,
      "step": 15145
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.51368140718001,
      "learning_rate": 6.786743625150888e-07,
      "loss": 0.1306,
      "step": 15146
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.45061332555829886,
      "learning_rate": 6.775268601565604e-07,
      "loss": 0.0469,
      "step": 15147
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.404937259459876,
      "learning_rate": 6.76380315391234e-07,
      "loss": 0.059,
      "step": 15148
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5129721490884428,
      "learning_rate": 6.752347282642457e-07,
      "loss": 0.0977,
      "step": 15149
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4568325063218121,
      "learning_rate": 6.740900988206955e-07,
      "loss": 0.0896,
      "step": 15150
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5770189128720231,
      "learning_rate": 6.729464271056557e-07,
      "loss": 0.086,
      "step": 15151
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.534119819583988,
      "learning_rate": 6.718037131641514e-07,
      "loss": 0.094,
      "step": 15152
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.40941366828003256,
      "learning_rate": 6.70661957041166e-07,
      "loss": 0.05,
      "step": 15153
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3813345989872998,
      "learning_rate": 6.695211587816552e-07,
      "loss": 0.0474,
      "step": 15154
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.43413329627312525,
      "learning_rate": 6.68381318430536e-07,
      "loss": 0.0921,
      "step": 15155
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.525763929737318,
      "learning_rate": 6.672424360326834e-07,
      "loss": 0.1251,
      "step": 15156
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6763339580656151,
      "learning_rate": 6.661045116329312e-07,
      "loss": 0.2256,
      "step": 15157
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6456956211877692,
      "learning_rate": 6.649675452760851e-07,
      "loss": 0.1702,
      "step": 15158
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5175392705466925,
      "learning_rate": 6.638315370069064e-07,
      "loss": 0.0897,
      "step": 15159
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5132986859642032,
      "learning_rate": 6.626964868701236e-07,
      "loss": 0.1364,
      "step": 15160
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5645780097228028,
      "learning_rate": 6.615623949104199e-07,
      "loss": 0.1507,
      "step": 15161
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.41940238915913475,
      "learning_rate": 6.604292611724488e-07,
      "loss": 0.0653,
      "step": 15162
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5953094116895867,
      "learning_rate": 6.592970857008191e-07,
      "loss": 0.0749,
      "step": 15163
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5023024454993271,
      "learning_rate": 6.581658685401116e-07,
      "loss": 0.1159,
      "step": 15164
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7192936045375229,
      "learning_rate": 6.570356097348573e-07,
      "loss": 0.13,
      "step": 15165
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8794043974440837,
      "learning_rate": 6.559063093295542e-07,
      "loss": 0.3246,
      "step": 15166
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6312716615571957,
      "learning_rate": 6.547779673686694e-07,
      "loss": 0.1744,
      "step": 15167
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4039064452322088,
      "learning_rate": 6.536505838966228e-07,
      "loss": 0.0722,
      "step": 15168
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4971411296913122,
      "learning_rate": 6.525241589578013e-07,
      "loss": 0.1323,
      "step": 15169
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.41802792440349634,
      "learning_rate": 6.513986925965526e-07,
      "loss": 0.0432,
      "step": 15170
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7419036982105697,
      "learning_rate": 6.502741848571887e-07,
      "loss": 0.1527,
      "step": 15171
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6611594474038619,
      "learning_rate": 6.491506357839794e-07,
      "loss": 0.2012,
      "step": 15172
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3274877079076909,
      "learning_rate": 6.48028045421159e-07,
      "loss": 0.0816,
      "step": 15173
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3770627755853752,
      "learning_rate": 6.469064138129311e-07,
      "loss": 0.041,
      "step": 15174
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.39955977392734643,
      "learning_rate": 6.45785741003449e-07,
      "loss": 0.0414,
      "step": 15175
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7314417150257877,
      "learning_rate": 6.44666027036836e-07,
      "loss": 0.2532,
      "step": 15176
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6971876738858375,
      "learning_rate": 6.435472719571789e-07,
      "loss": 0.1356,
      "step": 15177
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5126872514923029,
      "learning_rate": 6.424294758085176e-07,
      "loss": 0.1582,
      "step": 15178
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.33295428519656906,
      "learning_rate": 6.413126386348667e-07,
      "loss": 0.0707,
      "step": 15179
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8046375649384525,
      "learning_rate": 6.401967604801939e-07,
      "loss": 0.1963,
      "step": 15180
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5297720621098633,
      "learning_rate": 6.390818413884337e-07,
      "loss": 0.0934,
      "step": 15181
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6119668141844994,
      "learning_rate": 6.379678814034785e-07,
      "loss": 0.1516,
      "step": 15182
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5933588490998591,
      "learning_rate": 6.368548805691876e-07,
      "loss": 0.1612,
      "step": 15183
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0297560980380087,
      "learning_rate": 6.357428389293818e-07,
      "loss": 0.3825,
      "step": 15184
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.019343873184767,
      "learning_rate": 6.346317565278426e-07,
      "loss": 0.3125,
      "step": 15185
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4373313848360457,
      "learning_rate": 6.335216334083072e-07,
      "loss": 0.0535,
      "step": 15186
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6248028331622671,
      "learning_rate": 6.324124696144961e-07,
      "loss": 0.1373,
      "step": 15187
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6019517988093853,
      "learning_rate": 6.313042651900636e-07,
      "loss": 0.1797,
      "step": 15188
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.49661939884029643,
      "learning_rate": 6.301970201786523e-07,
      "loss": 0.0751,
      "step": 15189
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4641222523426539,
      "learning_rate": 6.290907346238411e-07,
      "loss": 0.0682,
      "step": 15190
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.49762819094967137,
      "learning_rate": 6.27985408569201e-07,
      "loss": 0.1615,
      "step": 15191
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.61480356273211,
      "learning_rate": 6.268810420582388e-07,
      "loss": 0.0909,
      "step": 15192
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5351978652528081,
      "learning_rate": 6.257776351344419e-07,
      "loss": 0.1128,
      "step": 15193
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5524783404604802,
      "learning_rate": 6.246751878412421e-07,
      "loss": 0.143,
      "step": 15194
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.676152460145413,
      "learning_rate": 6.235737002220521e-07,
      "loss": 0.1327,
      "step": 15195
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.47426711185493586,
      "learning_rate": 6.224731723202343e-07,
      "loss": 0.1019,
      "step": 15196
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6232570878232352,
      "learning_rate": 6.213736041791207e-07,
      "loss": 0.1083,
      "step": 15197
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8940539077722418,
      "learning_rate": 6.202749958419962e-07,
      "loss": 0.4293,
      "step": 15198
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6152644400233293,
      "learning_rate": 6.191773473521234e-07,
      "loss": 0.137,
      "step": 15199
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3519360684788253,
      "learning_rate": 6.180806587527066e-07,
      "loss": 0.0738,
      "step": 15200
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.776539726817521,
      "learning_rate": 6.169849300869307e-07,
      "loss": 0.1494,
      "step": 15201
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5669384385300055,
      "learning_rate": 6.158901613979307e-07,
      "loss": 0.0886,
      "step": 15202
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0719270812105621,
      "learning_rate": 6.147963527288109e-07,
      "loss": 0.0669,
      "step": 15203
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.48404976719203885,
      "learning_rate": 6.137035041226341e-07,
      "loss": 0.0976,
      "step": 15204
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4339906707202422,
      "learning_rate": 6.126116156224326e-07,
      "loss": 0.076,
      "step": 15205
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.0466972771289083,
      "learning_rate": 6.115206872711832e-07,
      "loss": 0.2917,
      "step": 15206
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5806840072808975,
      "learning_rate": 6.104307191118458e-07,
      "loss": 0.091,
      "step": 15207
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3830372870447395,
      "learning_rate": 6.093417111873306e-07,
      "loss": 0.0622,
      "step": 15208
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4853003803713905,
      "learning_rate": 6.082536635405117e-07,
      "loss": 0.0761,
      "step": 15209
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7183028826391854,
      "learning_rate": 6.07166576214227e-07,
      "loss": 0.1291,
      "step": 15210
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.35575918391405276,
      "learning_rate": 6.060804492512756e-07,
      "loss": 0.0517,
      "step": 15211
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5327007009506439,
      "learning_rate": 6.049952826944233e-07,
      "loss": 0.1517,
      "step": 15212
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0496512485282483,
      "learning_rate": 6.039110765863859e-07,
      "loss": 0.5311,
      "step": 15213
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4744446168333567,
      "learning_rate": 6.028278309698515e-07,
      "loss": 0.1001,
      "step": 15214
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.48671801169254003,
      "learning_rate": 6.017455458874721e-07,
      "loss": 0.0425,
      "step": 15215
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.37571528583907904,
      "learning_rate": 6.006642213818608e-07,
      "loss": 0.0412,
      "step": 15216
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6122569459055585,
      "learning_rate": 5.995838574955809e-07,
      "loss": 0.1118,
      "step": 15217
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.45709815009526,
      "learning_rate": 5.985044542711704e-07,
      "loss": 0.0885,
      "step": 15218
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8778604712761054,
      "learning_rate": 5.974260117511288e-07,
      "loss": 0.2314,
      "step": 15219
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.308566238716119,
      "learning_rate": 5.963485299779165e-07,
      "loss": 0.0486,
      "step": 15220
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7395662875826193,
      "learning_rate": 5.952720089939495e-07,
      "loss": 0.1962,
      "step": 15221
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5720545189130407,
      "learning_rate": 5.941964488416135e-07,
      "loss": 0.1279,
      "step": 15222
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3325218194579358,
      "learning_rate": 5.931218495632551e-07,
      "loss": 0.0597,
      "step": 15223
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4820901285433758,
      "learning_rate": 5.920482112011849e-07,
      "loss": 0.1309,
      "step": 15224
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5543838603279218,
      "learning_rate": 5.909755337976635e-07,
      "loss": 0.1335,
      "step": 15225
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6390676974697693,
      "learning_rate": 5.899038173949295e-07,
      "loss": 0.1278,
      "step": 15226
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5195888536251042,
      "learning_rate": 5.888330620351767e-07,
      "loss": 0.1064,
      "step": 15227
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.45611668953231144,
      "learning_rate": 5.87763267760566e-07,
      "loss": 0.0598,
      "step": 15228
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.50458827589681,
      "learning_rate": 5.866944346132053e-07,
      "loss": 0.1103,
      "step": 15229
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.49797220399618114,
      "learning_rate": 5.856265626351831e-07,
      "loss": 0.0944,
      "step": 15230
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.759159845693791,
      "learning_rate": 5.845596518685381e-07,
      "loss": 0.1963,
      "step": 15231
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5225649735708373,
      "learning_rate": 5.834937023552811e-07,
      "loss": 0.086,
      "step": 15232
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5416169229340781,
      "learning_rate": 5.824287141373757e-07,
      "loss": 0.155,
      "step": 15233
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5528992977277964,
      "learning_rate": 5.813646872567469e-07,
      "loss": 0.1279,
      "step": 15234
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.27290104879008,
      "learning_rate": 5.803016217552942e-07,
      "loss": 0.3229,
      "step": 15235
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7853963658659936,
      "learning_rate": 5.792395176748678e-07,
      "loss": 0.1181,
      "step": 15236
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4801154677634488,
      "learning_rate": 5.781783750572839e-07,
      "loss": 0.099,
      "step": 15237
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4116917607052098,
      "learning_rate": 5.771181939443176e-07,
      "loss": 0.0803,
      "step": 15238
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5008412260849524,
      "learning_rate": 5.760589743777134e-07,
      "loss": 0.1017,
      "step": 15239
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6432011384227355,
      "learning_rate": 5.750007163991684e-07,
      "loss": 0.172,
      "step": 15240
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4022881487529671,
      "learning_rate": 5.739434200503519e-07,
      "loss": 0.0644,
      "step": 15241
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4377887362280416,
      "learning_rate": 5.728870853728862e-07,
      "loss": 0.0964,
      "step": 15242
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5404877272618988,
      "learning_rate": 5.71831712408366e-07,
      "loss": 0.1175,
      "step": 15243
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6561324445616763,
      "learning_rate": 5.707773011983359e-07,
      "loss": 0.1086,
      "step": 15244
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3490022250647781,
      "learning_rate": 5.697238517843151e-07,
      "loss": 0.0571,
      "step": 15245
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4877769089130684,
      "learning_rate": 5.686713642077679e-07,
      "loss": 0.0677,
      "step": 15246
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4481279624347568,
      "learning_rate": 5.676198385101445e-07,
      "loss": 0.0947,
      "step": 15247
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.45230434915863127,
      "learning_rate": 5.665692747328339e-07,
      "loss": 0.0822,
      "step": 15248
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.41832002576147254,
      "learning_rate": 5.655196729172057e-07,
      "loss": 0.093,
      "step": 15249
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.790782689939778,
      "learning_rate": 5.644710331045744e-07,
      "loss": 0.1192,
      "step": 15250
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3882788140699131,
      "learning_rate": 5.634233553362372e-07,
      "loss": 0.0787,
      "step": 15251
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3568276927216494,
      "learning_rate": 5.623766396534308e-07,
      "loss": 0.056,
      "step": 15252
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3348906271680568,
      "learning_rate": 5.613308860973721e-07,
      "loss": 0.0507,
      "step": 15253
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2984912885542476,
      "learning_rate": 5.602860947092281e-07,
      "loss": 0.0359,
      "step": 15254
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.391248386908719,
      "learning_rate": 5.592422655301411e-07,
      "loss": 0.0571,
      "step": 15255
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4865564319055943,
      "learning_rate": 5.581993986011975e-07,
      "loss": 0.0613,
      "step": 15256
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7693337961282354,
      "learning_rate": 5.571574939634644e-07,
      "loss": 0.1575,
      "step": 15257
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4499549072920071,
      "learning_rate": 5.561165516579537e-07,
      "loss": 0.0805,
      "step": 15258
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6932540283688529,
      "learning_rate": 5.550765717256545e-07,
      "loss": 0.1533,
      "step": 15259
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5252146449873917,
      "learning_rate": 5.540375542075094e-07,
      "loss": 0.0829,
      "step": 15260
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.37456604162362317,
      "learning_rate": 5.529994991444271e-07,
      "loss": 0.0627,
      "step": 15261
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4511220428187284,
      "learning_rate": 5.519624065772722e-07,
      "loss": 0.5962,
      "step": 15262
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6300150238610971,
      "learning_rate": 5.509262765468842e-07,
      "loss": 0.1572,
      "step": 15263
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6888579507120063,
      "learning_rate": 5.498911090940473e-07,
      "loss": 0.1973,
      "step": 15264
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7747121693813158,
      "learning_rate": 5.488569042595177e-07,
      "loss": 0.1823,
      "step": 15265
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4691486457500124,
      "learning_rate": 5.478236620840183e-07,
      "loss": 0.0727,
      "step": 15266
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6261275535519577,
      "learning_rate": 5.467913826082278e-07,
      "loss": 0.1933,
      "step": 15267
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.624232684600329,
      "learning_rate": 5.457600658727802e-07,
      "loss": 0.0778,
      "step": 15268
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5606891089379046,
      "learning_rate": 5.447297119182876e-07,
      "loss": 0.0862,
      "step": 15269
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4216814621622448,
      "learning_rate": 5.437003207853147e-07,
      "loss": 0.0791,
      "step": 15270
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6451443932966072,
      "learning_rate": 5.42671892514382e-07,
      "loss": 0.1105,
      "step": 15271
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.45469276779553425,
      "learning_rate": 5.416444271459875e-07,
      "loss": 0.0815,
      "step": 15272
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5921783111787811,
      "learning_rate": 5.406179247205822e-07,
      "loss": 0.0559,
      "step": 15273
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5177114512713195,
      "learning_rate": 5.395923852785783e-07,
      "loss": 0.112,
      "step": 15274
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4520163635681971,
      "learning_rate": 5.385678088603518e-07,
      "loss": 0.0562,
      "step": 15275
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3796476697036257,
      "learning_rate": 5.375441955062426e-07,
      "loss": 0.0352,
      "step": 15276
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4414498685781731,
      "learning_rate": 5.365215452565436e-07,
      "loss": 0.1167,
      "step": 15277
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.44880885504390194,
      "learning_rate": 5.354998581515308e-07,
      "loss": 0.0829,
      "step": 15278
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.42954661153476725,
      "learning_rate": 5.344791342314193e-07,
      "loss": 0.0905,
      "step": 15279
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5065277327622524,
      "learning_rate": 5.33459373536399e-07,
      "loss": 0.0615,
      "step": 15280
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.423148761272863,
      "learning_rate": 5.324405761066159e-07,
      "loss": 0.0811,
      "step": 15281
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5041264810080524,
      "learning_rate": 5.314227419821849e-07,
      "loss": 0.0796,
      "step": 15282
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4702692579717444,
      "learning_rate": 5.304058712031739e-07,
      "loss": 0.0722,
      "step": 15283
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6376260292522313,
      "learning_rate": 5.293899638096261e-07,
      "loss": 0.0883,
      "step": 15284
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3564156760795552,
      "learning_rate": 5.283750198415232e-07,
      "loss": 0.0492,
      "step": 15285
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7793793027778773,
      "learning_rate": 5.273610393388417e-07,
      "loss": 0.1174,
      "step": 15286
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.47762666413528815,
      "learning_rate": 5.26348022341494e-07,
      "loss": 0.0975,
      "step": 15287
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6381123110731567,
      "learning_rate": 5.25335968889365e-07,
      "loss": 0.1072,
      "step": 15288
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.43256277177723607,
      "learning_rate": 5.243248790222949e-07,
      "loss": 0.0809,
      "step": 15289
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2929373898584112,
      "learning_rate": 5.233147527800991e-07,
      "loss": 0.0581,
      "step": 15290
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7153221908209895,
      "learning_rate": 5.223055902025403e-07,
      "loss": 0.1479,
      "step": 15291
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6441055079540057,
      "learning_rate": 5.212973913293562e-07,
      "loss": 0.1965,
      "step": 15292
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.49621584105926453,
      "learning_rate": 5.202901562002343e-07,
      "loss": 0.1149,
      "step": 15293
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5467312779933516,
      "learning_rate": 5.192838848548348e-07,
      "loss": 0.0993,
      "step": 15294
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.45082704985012134,
      "learning_rate": 5.182785773327731e-07,
      "loss": 0.054,
      "step": 15295
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5087570123917792,
      "learning_rate": 5.172742336736314e-07,
      "loss": 0.0915,
      "step": 15296
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.44940678948050233,
      "learning_rate": 5.162708539169447e-07,
      "loss": 0.084,
      "step": 15297
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6549764530812419,
      "learning_rate": 5.152684381022233e-07,
      "loss": 0.1218,
      "step": 15298
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5149425807210787,
      "learning_rate": 5.142669862689325e-07,
      "loss": 0.1014,
      "step": 15299
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5142764690947695,
      "learning_rate": 5.132664984564995e-07,
      "loss": 0.107,
      "step": 15300
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5761766027725955,
      "learning_rate": 5.122669747043091e-07,
      "loss": 0.1125,
      "step": 15301
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6850778304962294,
      "learning_rate": 5.112684150517188e-07,
      "loss": 0.1087,
      "step": 15302
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.021512872706107,
      "learning_rate": 5.102708195380446e-07,
      "loss": 0.2864,
      "step": 15303
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4730072285529127,
      "learning_rate": 5.092741882025576e-07,
      "loss": 0.1028,
      "step": 15304
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8979423131369513,
      "learning_rate": 5.082785210844987e-07,
      "loss": 0.2246,
      "step": 15305
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7670320346030628,
      "learning_rate": 5.072838182230644e-07,
      "loss": 0.1583,
      "step": 15306
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4207838234679563,
      "learning_rate": 5.062900796574233e-07,
      "loss": 0.0755,
      "step": 15307
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3605808304313821,
      "learning_rate": 5.05297305426694e-07,
      "loss": 0.0478,
      "step": 15308
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.912329300305093,
      "learning_rate": 5.043054955699622e-07,
      "loss": 0.197,
      "step": 15309
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8619838953113466,
      "learning_rate": 5.033146501262798e-07,
      "loss": 0.2534,
      "step": 15310
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6154614740695026,
      "learning_rate": 5.023247691346572e-07,
      "loss": 0.1701,
      "step": 15311
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5579117679982749,
      "learning_rate": 5.013358526340661e-07,
      "loss": 0.1599,
      "step": 15312
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7484844639067865,
      "learning_rate": 5.003479006634366e-07,
      "loss": 0.274,
      "step": 15313
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.45445480820998,
      "learning_rate": 4.993609132616705e-07,
      "loss": 0.0668,
      "step": 15314
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3615776169602662,
      "learning_rate": 4.983748904676261e-07,
      "loss": 0.0508,
      "step": 15315
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6655806036781192,
      "learning_rate": 4.973898323201192e-07,
      "loss": 0.1086,
      "step": 15316
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.38886696711160257,
      "learning_rate": 4.96405738857933e-07,
      "loss": 0.0732,
      "step": 15317
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4686398645708974,
      "learning_rate": 4.954226101198167e-07,
      "loss": 0.089,
      "step": 15318
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4362717507545218,
      "learning_rate": 4.944404461444757e-07,
      "loss": 0.105,
      "step": 15319
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.41192597872664427,
      "learning_rate": 4.934592469705762e-07,
      "loss": 0.1109,
      "step": 15320
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1401976968403085,
      "learning_rate": 4.924790126367456e-07,
      "loss": 0.3769,
      "step": 15321
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.41995146956659957,
      "learning_rate": 4.914997431815837e-07,
      "loss": 0.0496,
      "step": 15322
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.24166331239555836,
      "learning_rate": 4.905214386436429e-07,
      "loss": 0.0382,
      "step": 15323
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.49538656674727577,
      "learning_rate": 4.895440990614342e-07,
      "loss": 0.0776,
      "step": 15324
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5765302235323947,
      "learning_rate": 4.885677244734432e-07,
      "loss": 0.1961,
      "step": 15325
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6101015627753555,
      "learning_rate": 4.875923149181061e-07,
      "loss": 0.1541,
      "step": 15326
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8967392524715552,
      "learning_rate": 4.866178704338281e-07,
      "loss": 0.1973,
      "step": 15327
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4701043019197404,
      "learning_rate": 4.856443910589731e-07,
      "loss": 0.075,
      "step": 15328
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5437996994308414,
      "learning_rate": 4.846718768318659e-07,
      "loss": 0.1312,
      "step": 15329
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2331064216667638,
      "learning_rate": 4.837003277907953e-07,
      "loss": 0.3311,
      "step": 15330
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9244108982158656,
      "learning_rate": 4.827297439740169e-07,
      "loss": 0.3999,
      "step": 15331
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.27016424527244887,
      "learning_rate": 4.817601254197362e-07,
      "loss": 0.0237,
      "step": 15332
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5766126498280785,
      "learning_rate": 4.807914721661311e-07,
      "loss": 0.136,
      "step": 15333
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6466028524594029,
      "learning_rate": 4.798237842513403e-07,
      "loss": 0.1524,
      "step": 15334
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5660002668499357,
      "learning_rate": 4.788570617134586e-07,
      "loss": 0.1299,
      "step": 15335
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.2777753121335365,
      "learning_rate": 4.778913045905469e-07,
      "loss": 0.0817,
      "step": 15336
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4488373565901901,
      "learning_rate": 4.769265129206308e-07,
      "loss": 0.0527,
      "step": 15337
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.36517719390003744,
      "learning_rate": 4.759626867416933e-07,
      "loss": 0.0674,
      "step": 15338
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.44910163581958507,
      "learning_rate": 4.7499982609167936e-07,
      "loss": 0.0842,
      "step": 15339
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5820667991525152,
      "learning_rate": 4.740379310084975e-07,
      "loss": 0.1196,
      "step": 15340
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5267302851739114,
      "learning_rate": 4.730770015300201e-07,
      "loss": 0.1738,
      "step": 15341
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.346373209766561,
      "learning_rate": 4.721170376940809e-07,
      "loss": 0.0509,
      "step": 15342
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8632650048706392,
      "learning_rate": 4.7115803953847184e-07,
      "loss": 0.2444,
      "step": 15343
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2962192828057242,
      "learning_rate": 4.702000071009516e-07,
      "loss": 0.325,
      "step": 15344
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3900313144092534,
      "learning_rate": 4.6924294041923166e-07,
      "loss": 0.0417,
      "step": 15345
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4526675649215686,
      "learning_rate": 4.682868395310014e-07,
      "loss": 0.0692,
      "step": 15346
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.641586730223111,
      "learning_rate": 4.673317044738973e-07,
      "loss": 0.1299,
      "step": 15347
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4616399209531493,
      "learning_rate": 4.6637753528553097e-07,
      "loss": 0.0828,
      "step": 15348
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6902548427308931,
      "learning_rate": 4.654243320034557e-07,
      "loss": 0.2399,
      "step": 15349
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5249331929902151,
      "learning_rate": 4.644720946652165e-07,
      "loss": 0.0779,
      "step": 15350
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.42376424196252777,
      "learning_rate": 4.6352082330828884e-07,
      "loss": 0.0764,
      "step": 15351
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6123107159606023,
      "learning_rate": 4.625705179701345e-07,
      "loss": 0.1321,
      "step": 15352
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.33612828625160696,
      "learning_rate": 4.616211786881597e-07,
      "loss": 0.048,
      "step": 15353
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8106763163856052,
      "learning_rate": 4.606728054997511e-07,
      "loss": 0.1608,
      "step": 15354
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.37647598708257257,
      "learning_rate": 4.5972539844223726e-07,
      "loss": 0.0607,
      "step": 15355
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5444198216188313,
      "learning_rate": 4.587789575529217e-07,
      "loss": 0.1811,
      "step": 15356
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6051348574342523,
      "learning_rate": 4.578334828690661e-07,
      "loss": 0.1373,
      "step": 15357
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3941545997277137,
      "learning_rate": 4.5688897442789923e-07,
      "loss": 0.0645,
      "step": 15358
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4048413458208841,
      "learning_rate": 4.559454322665968e-07,
      "loss": 0.0635,
      "step": 15359
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3863426069076205,
      "learning_rate": 4.55002856422318e-07,
      "loss": 0.0506,
      "step": 15360
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7586161942257159,
      "learning_rate": 4.540612469321637e-07,
      "loss": 0.1895,
      "step": 15361
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4493619571815115,
      "learning_rate": 4.5312060383321253e-07,
      "loss": 0.0668,
      "step": 15362
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.29074353859840246,
      "learning_rate": 4.5218092716249326e-07,
      "loss": 0.0439,
      "step": 15363
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3896057766850114,
      "learning_rate": 4.5124221695700684e-07,
      "loss": 0.1068,
      "step": 15364
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0411671135254312,
      "learning_rate": 4.503044732537043e-07,
      "loss": 0.3354,
      "step": 15365
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6621918142540025,
      "learning_rate": 4.4936769608951154e-07,
      "loss": 0.1645,
      "step": 15366
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.695246658528935,
      "learning_rate": 4.4843188550130476e-07,
      "loss": 0.1127,
      "step": 15367
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8649529438690097,
      "learning_rate": 4.474970415259322e-07,
      "loss": 0.2701,
      "step": 15368
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6460055253904269,
      "learning_rate": 4.4656316420019495e-07,
      "loss": 0.0721,
      "step": 15369
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2574912230751429,
      "learning_rate": 4.456302535608636e-07,
      "loss": 0.0181,
      "step": 15370
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7173571794263993,
      "learning_rate": 4.446983096446672e-07,
      "loss": 0.2179,
      "step": 15371
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4799188421937678,
      "learning_rate": 4.4376733248829847e-07,
      "loss": 0.0783,
      "step": 15372
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.683369575280559,
      "learning_rate": 4.42837322128406e-07,
      "loss": 0.0885,
      "step": 15373
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.521567913473053,
      "learning_rate": 4.4190827860160766e-07,
      "loss": 0.1392,
      "step": 15374
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.524830808929452,
      "learning_rate": 4.409802019444825e-07,
      "loss": 0.071,
      "step": 15375
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5079968411484952,
      "learning_rate": 4.4005309219356526e-07,
      "loss": 0.3772,
      "step": 15376
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5028136145867447,
      "learning_rate": 4.3912694938536e-07,
      "loss": 0.1094,
      "step": 15377
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.719702666488082,
      "learning_rate": 4.382017735563293e-07,
      "loss": 0.0903,
      "step": 15378
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.37459555087856317,
      "learning_rate": 4.3727756474289947e-07,
      "loss": 0.0438,
      "step": 15379
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.39012315239124223,
      "learning_rate": 4.3635432298145265e-07,
      "loss": 0.0515,
      "step": 15380
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4815030833395449,
      "learning_rate": 4.3543204830834016e-07,
      "loss": 0.0887,
      "step": 15381
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4692171603097667,
      "learning_rate": 4.3451074075987477e-07,
      "loss": 0.0741,
      "step": 15382
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4537921239467923,
      "learning_rate": 4.3359040037232734e-07,
      "loss": 0.0735,
      "step": 15383
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.48368441741070756,
      "learning_rate": 4.3267102718193007e-07,
      "loss": 0.1653,
      "step": 15384
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7221560853821604,
      "learning_rate": 4.317526212248818e-07,
      "loss": 0.1189,
      "step": 15385
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6426461261383919,
      "learning_rate": 4.3083518253733965e-07,
      "loss": 0.1931,
      "step": 15386
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.34490824745821147,
      "learning_rate": 4.2991871115542484e-07,
      "loss": 0.0775,
      "step": 15387
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7577289829795313,
      "learning_rate": 4.290032071152195e-07,
      "loss": 0.0962,
      "step": 15388
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4356957890931087,
      "learning_rate": 4.28088670452767e-07,
      "loss": 0.078,
      "step": 15389
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4219381330920857,
      "learning_rate": 4.27175101204072e-07,
      "loss": 0.0497,
      "step": 15390
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.439412266378332,
      "learning_rate": 4.2626249940510846e-07,
      "loss": 0.0502,
      "step": 15391
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.45405999651749807,
      "learning_rate": 4.2535086509179476e-07,
      "loss": 0.0828,
      "step": 15392
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6531717690320024,
      "learning_rate": 4.244401983000357e-07,
      "loss": 0.1602,
      "step": 15393
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4030134542720136,
      "learning_rate": 4.2353049906567744e-07,
      "loss": 0.0771,
      "step": 15394
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5495761279735925,
      "learning_rate": 4.2262176742453584e-07,
      "loss": 0.0527,
      "step": 15395
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3846265879461876,
      "learning_rate": 4.2171400341239064e-07,
      "loss": 0.055,
      "step": 15396
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5622152589834811,
      "learning_rate": 4.208072070649771e-07,
      "loss": 0.0992,
      "step": 15397
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.41434822545960187,
      "learning_rate": 4.1990137841800003e-07,
      "loss": 0.0782,
      "step": 15398
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.38073270488592315,
      "learning_rate": 4.1899651750712254e-07,
      "loss": 0.0505,
      "step": 15399
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.39871938589796296,
      "learning_rate": 4.180926243679689e-07,
      "loss": 0.0624,
      "step": 15400
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.43544948944941475,
      "learning_rate": 4.171896990361246e-07,
      "loss": 0.0546,
      "step": 15401
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.43344423874224725,
      "learning_rate": 4.1628774154714166e-07,
      "loss": 0.0566,
      "step": 15402
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.45017597066965426,
      "learning_rate": 4.153867519365279e-07,
      "loss": 0.063,
      "step": 15403
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8644196226512134,
      "learning_rate": 4.144867302397548e-07,
      "loss": 0.3398,
      "step": 15404
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5733799755408646,
      "learning_rate": 4.1358767649226085e-07,
      "loss": 0.0779,
      "step": 15405
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6169113888293927,
      "learning_rate": 4.1268959072944256e-07,
      "loss": 0.1012,
      "step": 15406
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.48803901110701964,
      "learning_rate": 4.1179247298665233e-07,
      "loss": 0.1046,
      "step": 15407
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.42673128076498973,
      "learning_rate": 4.1089632329921747e-07,
      "loss": 0.0746,
      "step": 15408
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.333374444749921,
      "learning_rate": 4.1000114170241245e-07,
      "loss": 0.0591,
      "step": 15409
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7906112917849032,
      "learning_rate": 4.0910692823148967e-07,
      "loss": 0.1781,
      "step": 15410
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.44774291855291865,
      "learning_rate": 4.0821368292164876e-07,
      "loss": 0.1159,
      "step": 15411
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3559617472660215,
      "learning_rate": 4.0732140580806153e-07,
      "loss": 0.0314,
      "step": 15412
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3785204402452197,
      "learning_rate": 4.0643009692585275e-07,
      "loss": 0.0802,
      "step": 15413
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4884128825946422,
      "learning_rate": 4.0553975631011933e-07,
      "loss": 0.0683,
      "step": 15414
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7317771068001079,
      "learning_rate": 4.046503839959109e-07,
      "loss": 0.2298,
      "step": 15415
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4566311457779005,
      "learning_rate": 4.0376198001824127e-07,
      "loss": 0.0781,
      "step": 15416
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4575524408708837,
      "learning_rate": 4.0287454441209347e-07,
      "loss": 0.0953,
      "step": 15417
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.37372027178226946,
      "learning_rate": 4.0198807721240074e-07,
      "loss": 0.044,
      "step": 15418
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.46439201184857304,
      "learning_rate": 4.011025784540684e-07,
      "loss": 0.0473,
      "step": 15419
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5704002861324657,
      "learning_rate": 4.0021804817195476e-07,
      "loss": 0.0953,
      "step": 15420
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4400301948975566,
      "learning_rate": 3.9933448640088744e-07,
      "loss": 0.0662,
      "step": 15421
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.40319011197651505,
      "learning_rate": 3.9845189317565257e-07,
      "loss": 0.0508,
      "step": 15422
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5072896120083016,
      "learning_rate": 3.9757026853099733e-07,
      "loss": 0.0776,
      "step": 15423
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3978833468237589,
      "learning_rate": 3.966896125016328e-07,
      "loss": 0.0399,
      "step": 15424
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.46697205740510184,
      "learning_rate": 3.958099251222286e-07,
      "loss": 0.0435,
      "step": 15425
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3697376648466651,
      "learning_rate": 3.9493120642742356e-07,
      "loss": 0.0472,
      "step": 15426
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7851258605061645,
      "learning_rate": 3.9405345645181225e-07,
      "loss": 0.2655,
      "step": 15427
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.45417193299087044,
      "learning_rate": 3.931766752299476e-07,
      "loss": 0.0669,
      "step": 15428
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.43315505684589006,
      "learning_rate": 3.9230086279635204e-07,
      "loss": 0.1273,
      "step": 15429
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.319812463294227,
      "learning_rate": 3.91426019185509e-07,
      "loss": 0.04,
      "step": 15430
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4483554953466593,
      "learning_rate": 3.905521444318605e-07,
      "loss": 0.0792,
      "step": 15431
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6165467373046104,
      "learning_rate": 3.8967923856980946e-07,
      "loss": 0.1514,
      "step": 15432
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7846693259040216,
      "learning_rate": 3.888073016337257e-07,
      "loss": 0.1908,
      "step": 15433
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1040921103025823,
      "learning_rate": 3.8793633365793724e-07,
      "loss": 0.5286,
      "step": 15434
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3000726874190878,
      "learning_rate": 3.870663346767306e-07,
      "loss": 0.0523,
      "step": 15435
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7030078643250298,
      "learning_rate": 3.8619730472436444e-07,
      "loss": 0.1123,
      "step": 15436
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5255560084884444,
      "learning_rate": 3.853292438350503e-07,
      "loss": 0.0575,
      "step": 15437
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7682413376562531,
      "learning_rate": 3.8446215204296364e-07,
      "loss": 0.1476,
      "step": 15438
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9120885030625666,
      "learning_rate": 3.835960293822438e-07,
      "loss": 0.212,
      "step": 15439
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.39896665842321316,
      "learning_rate": 3.827308758869913e-07,
      "loss": 0.0544,
      "step": 15440
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.35104285708955457,
      "learning_rate": 3.818666915912678e-07,
      "loss": 0.0283,
      "step": 15441
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9718080572583944,
      "learning_rate": 3.810034765290932e-07,
      "loss": 0.2839,
      "step": 15442
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.33108213290186395,
      "learning_rate": 3.8014123073445706e-07,
      "loss": 0.0554,
      "step": 15443
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0237577675493634,
      "learning_rate": 3.7927995424130167e-07,
      "loss": 0.2495,
      "step": 15444
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.236890300563056,
      "learning_rate": 3.784196470835444e-07,
      "loss": 0.333,
      "step": 15445
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4785258377868741,
      "learning_rate": 3.775603092950497e-07,
      "loss": 0.059,
      "step": 15446
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4570481856798454,
      "learning_rate": 3.7670194090965174e-07,
      "loss": 0.7711,
      "step": 15447
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.47109065361007285,
      "learning_rate": 3.758445419611428e-07,
      "loss": 0.1038,
      "step": 15448
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4320052749510805,
      "learning_rate": 3.749881124832849e-07,
      "loss": 0.069,
      "step": 15449
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4690523744614874,
      "learning_rate": 3.7413265250979267e-07,
      "loss": 0.1043,
      "step": 15450
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.41564439266448566,
      "learning_rate": 3.732781620743475e-07,
      "loss": 0.1108,
      "step": 15451
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.43295834595561916,
      "learning_rate": 3.724246412105864e-07,
      "loss": 0.135,
      "step": 15452
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.45714797650517613,
      "learning_rate": 3.7157208995212135e-07,
      "loss": 0.129,
      "step": 15453
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1663803447222592,
      "learning_rate": 3.707205083325116e-07,
      "loss": 0.5444,
      "step": 15454
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3683759757335412,
      "learning_rate": 3.6986989638528867e-07,
      "loss": 0.0686,
      "step": 15455
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.34027638186363124,
      "learning_rate": 3.690202541439369e-07,
      "loss": 0.0616,
      "step": 15456
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6739585023678399,
      "learning_rate": 3.681715816419129e-07,
      "loss": 0.1627,
      "step": 15457
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6670881992481055,
      "learning_rate": 3.6732387891262323e-07,
      "loss": 0.1304,
      "step": 15458
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8190817129665705,
      "learning_rate": 3.6647714598944947e-07,
      "loss": 0.2461,
      "step": 15459
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4812623150413126,
      "learning_rate": 3.6563138290572064e-07,
      "loss": 0.0879,
      "step": 15460
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6538478134872313,
      "learning_rate": 3.6478658969474335e-07,
      "loss": 0.1077,
      "step": 15461
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6322266185131984,
      "learning_rate": 3.639427663897715e-07,
      "loss": 0.3941,
      "step": 15462
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4792289034319503,
      "learning_rate": 3.6309991302403145e-07,
      "loss": 0.3697,
      "step": 15463
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.47002998331918194,
      "learning_rate": 3.622580296307021e-07,
      "loss": 0.0632,
      "step": 15464
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4854247093275862,
      "learning_rate": 3.614171162429292e-07,
      "loss": 0.075,
      "step": 15465
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.42837860358323565,
      "learning_rate": 3.605771728938251e-07,
      "loss": 0.0814,
      "step": 15466
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4944394976651838,
      "learning_rate": 3.597381996164578e-07,
      "loss": 0.0868,
      "step": 15467
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6162662171975718,
      "learning_rate": 3.589001964438538e-07,
      "loss": 0.1265,
      "step": 15468
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9919359414209945,
      "learning_rate": 3.580631634090087e-07,
      "loss": 0.2568,
      "step": 15469
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7598217818261334,
      "learning_rate": 3.572271005448796e-07,
      "loss": 0.1718,
      "step": 15470
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4095036873799596,
      "learning_rate": 3.563920078843791e-07,
      "loss": 0.0582,
      "step": 15471
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4263409656752157,
      "learning_rate": 3.555578854603836e-07,
      "loss": 0.1083,
      "step": 15472
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5457298271773907,
      "learning_rate": 3.547247333057391e-07,
      "loss": 0.0956,
      "step": 15473
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3468567667273361,
      "learning_rate": 3.5389255145324706e-07,
      "loss": 0.0274,
      "step": 15474
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.42757387492366333,
      "learning_rate": 3.5306133993566473e-07,
      "loss": 0.0509,
      "step": 15475
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.46108245001735104,
      "learning_rate": 3.5223109878572135e-07,
      "loss": 0.1249,
      "step": 15476
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5143282749444907,
      "learning_rate": 3.514018280361048e-07,
      "loss": 0.1307,
      "step": 15477
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.43006493931449297,
      "learning_rate": 3.5057352771946383e-07,
      "loss": 0.0573,
      "step": 15478
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5114266403318688,
      "learning_rate": 3.4974619786840855e-07,
      "loss": 0.0653,
      "step": 15479
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4570379739022523,
      "learning_rate": 3.4891983851551015e-07,
      "loss": 0.052,
      "step": 15480
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7444935992993793,
      "learning_rate": 3.480944496933064e-07,
      "loss": 0.2568,
      "step": 15481
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.47895482784985466,
      "learning_rate": 3.472700314342908e-07,
      "loss": 0.1091,
      "step": 15482
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.906010221955748,
      "learning_rate": 3.464465837709235e-07,
      "loss": 0.2861,
      "step": 15483
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5530038213115306,
      "learning_rate": 3.456241067356203e-07,
      "loss": 0.1339,
      "step": 15484
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.45258610735367705,
      "learning_rate": 3.448026003607663e-07,
      "loss": 0.1117,
      "step": 15485
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4897785067983981,
      "learning_rate": 3.4398206467870506e-07,
      "loss": 0.0886,
      "step": 15486
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.37943755058060713,
      "learning_rate": 3.4316249972173864e-07,
      "loss": 0.0602,
      "step": 15487
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5940195905381419,
      "learning_rate": 3.423439055221356e-07,
      "loss": 0.0874,
      "step": 15488
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3566559386716335,
      "learning_rate": 3.415262821121229e-07,
      "loss": 0.0751,
      "step": 15489
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.091604550274914,
      "learning_rate": 3.407096295238971e-07,
      "loss": 0.3305,
      "step": 15490
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9504142888725468,
      "learning_rate": 3.398939477896018e-07,
      "loss": 0.2999,
      "step": 15491
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4487335384122968,
      "learning_rate": 3.390792369413559e-07,
      "loss": 0.1188,
      "step": 15492
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4399381605389433,
      "learning_rate": 3.3826549701123366e-07,
      "loss": 0.1172,
      "step": 15493
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5302137761598191,
      "learning_rate": 3.3745272803127615e-07,
      "loss": 0.1164,
      "step": 15494
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3462958041947737,
      "learning_rate": 3.3664093003347994e-07,
      "loss": 0.0742,
      "step": 15495
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.574013175477633,
      "learning_rate": 3.358301030498029e-07,
      "loss": 0.0671,
      "step": 15496
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4475002102230002,
      "learning_rate": 3.350202471121722e-07,
      "loss": 0.0434,
      "step": 15497
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.414836541352508,
      "learning_rate": 3.3421136225247065e-07,
      "loss": 0.0509,
      "step": 15498
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.44018962660129657,
      "learning_rate": 3.33403448502545e-07,
      "loss": 0.0813,
      "step": 15499
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3865354700958151,
      "learning_rate": 3.3259650589420323e-07,
      "loss": 0.0682,
      "step": 15500
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.41627613596076396,
      "learning_rate": 3.31790534459217e-07,
      "loss": 0.0439,
      "step": 15501
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3896808623055871,
      "learning_rate": 3.309855342293167e-07,
      "loss": 0.0441,
      "step": 15502
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6559206583559957,
      "learning_rate": 3.301815052361934e-07,
      "loss": 0.2738,
      "step": 15503
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.38960708082098017,
      "learning_rate": 3.293784475115053e-07,
      "loss": 0.0697,
      "step": 15504
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6010420435357339,
      "learning_rate": 3.285763610868714e-07,
      "loss": 0.1282,
      "step": 15505
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6039737948639333,
      "learning_rate": 3.277752459938638e-07,
      "loss": 0.1531,
      "step": 15506
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7928733722666772,
      "learning_rate": 3.269751022640294e-07,
      "loss": 0.1267,
      "step": 15507
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.413506433507576,
      "learning_rate": 3.2617592992886525e-07,
      "loss": 0.0352,
      "step": 15508
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3882078036860944,
      "learning_rate": 3.2537772901983787e-07,
      "loss": 0.0482,
      "step": 15509
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5075823216605796,
      "learning_rate": 3.245804995683749e-07,
      "loss": 0.1065,
      "step": 15510
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3674720797355186,
      "learning_rate": 3.2378424160586227e-07,
      "loss": 0.0562,
      "step": 15511
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.41182237788109655,
      "learning_rate": 3.2298895516364447e-07,
      "loss": 0.0637,
      "step": 15512
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5082912779815167,
      "learning_rate": 3.221946402730408e-07,
      "loss": 0.0876,
      "step": 15513
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4044401938170957,
      "learning_rate": 3.214012969653207e-07,
      "loss": 0.113,
      "step": 15514
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.38737413455016584,
      "learning_rate": 3.206089252717176e-07,
      "loss": 0.0533,
      "step": 15515
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4127195832143209,
      "learning_rate": 3.198175252234259e-07,
      "loss": 0.0597,
      "step": 15516
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5070534864590106,
      "learning_rate": 3.190270968516068e-07,
      "loss": 0.1139,
      "step": 15517
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8015691150419948,
      "learning_rate": 3.1823764018737987e-07,
      "loss": 0.1233,
      "step": 15518
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.49606582714779607,
      "learning_rate": 3.174491552618286e-07,
      "loss": 0.0741,
      "step": 15519
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7739267653247236,
      "learning_rate": 3.166616421059865e-07,
      "loss": 0.05,
      "step": 15520
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2994934711728739,
      "learning_rate": 3.1587510075087036e-07,
      "loss": 0.043,
      "step": 15521
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0843083881457454,
      "learning_rate": 3.1508953122744157e-07,
      "loss": 0.3756,
      "step": 15522
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4347086861740869,
      "learning_rate": 3.1430493356662815e-07,
      "loss": 0.051,
      "step": 15523
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4117047056938775,
      "learning_rate": 3.135213077993193e-07,
      "loss": 0.0875,
      "step": 15524
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9303225704116836,
      "learning_rate": 3.127386539563737e-07,
      "loss": 0.2654,
      "step": 15525
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4185709620706234,
      "learning_rate": 3.119569720685972e-07,
      "loss": 0.0834,
      "step": 15526
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.99726847824254,
      "learning_rate": 3.11176262166768e-07,
      "loss": 0.2908,
      "step": 15527
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8706522995609681,
      "learning_rate": 3.103965242816226e-07,
      "loss": 0.2128,
      "step": 15528
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6095543185773891,
      "learning_rate": 3.096177584438614e-07,
      "loss": 0.0878,
      "step": 15529
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8525412316068455,
      "learning_rate": 3.0883996468414336e-07,
      "loss": 0.3304,
      "step": 15530
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5024570678938994,
      "learning_rate": 3.08063143033091e-07,
      "loss": 0.0718,
      "step": 15531
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6093387971340037,
      "learning_rate": 3.0728729352129115e-07,
      "loss": 0.1659,
      "step": 15532
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5337825756741224,
      "learning_rate": 3.0651241617928594e-07,
      "loss": 0.1759,
      "step": 15533
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.49885017254244146,
      "learning_rate": 3.0573851103758156e-07,
      "loss": 0.0621,
      "step": 15534
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3925498877650388,
      "learning_rate": 3.049655781266508e-07,
      "loss": 0.0527,
      "step": 15535
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7310197496432845,
      "learning_rate": 3.0419361747692774e-07,
      "loss": 0.1564,
      "step": 15536
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9442681665402068,
      "learning_rate": 3.034226291187964e-07,
      "loss": 0.2206,
      "step": 15537
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3578820887482297,
      "learning_rate": 3.026526130826185e-07,
      "loss": 0.0434,
      "step": 15538
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6429415595726469,
      "learning_rate": 3.018835693987032e-07,
      "loss": 0.186,
      "step": 15539
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5944373146059548,
      "learning_rate": 3.011154980973374e-07,
      "loss": 0.0962,
      "step": 15540
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5681773653094908,
      "learning_rate": 3.0034839920875246e-07,
      "loss": 0.0986,
      "step": 15541
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6466037719365557,
      "learning_rate": 2.99582272763152e-07,
      "loss": 0.1498,
      "step": 15542
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6109188766342712,
      "learning_rate": 2.9881711879070076e-07,
      "loss": 0.0551,
      "step": 15543
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2322656633778986,
      "learning_rate": 2.9805293732152186e-07,
      "loss": 0.2585,
      "step": 15544
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3675758352444546,
      "learning_rate": 2.9728972838570233e-07,
      "loss": 0.0665,
      "step": 15545
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2361959841533217,
      "learning_rate": 2.9652749201329045e-07,
      "loss": 0.349,
      "step": 15546
    },
    {
      "epoch": 0.95,
      "grad_norm": 3.1516968828134466,
      "learning_rate": 2.957662282342927e-07,
      "loss": 1.1414,
      "step": 15547
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5220426941307899,
      "learning_rate": 2.9500593707868785e-07,
      "loss": 0.1198,
      "step": 15548
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5091436005303991,
      "learning_rate": 2.94246618576402e-07,
      "loss": 0.1221,
      "step": 15549
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5918164774442416,
      "learning_rate": 2.934882727573363e-07,
      "loss": 0.1119,
      "step": 15550
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7151278803776296,
      "learning_rate": 2.9273089965133894e-07,
      "loss": 0.0685,
      "step": 15551
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5425963811613769,
      "learning_rate": 2.9197449928823615e-07,
      "loss": 0.1211,
      "step": 15552
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6461419261373109,
      "learning_rate": 2.9121907169780136e-07,
      "loss": 0.0636,
      "step": 15553
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.45721549595629923,
      "learning_rate": 2.904646169097858e-07,
      "loss": 0.0495,
      "step": 15554
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3109794999292821,
      "learning_rate": 2.897111349538795e-07,
      "loss": 0.0529,
      "step": 15555
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.515209928551027,
      "learning_rate": 2.889586258597615e-07,
      "loss": 0.1508,
      "step": 15556
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9130657502874397,
      "learning_rate": 2.882070896570499e-07,
      "loss": 0.2153,
      "step": 15557
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.31884555810261794,
      "learning_rate": 2.874565263753348e-07,
      "loss": 0.0417,
      "step": 15558
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4050743494647879,
      "learning_rate": 2.8670693604416764e-07,
      "loss": 0.0556,
      "step": 15559
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4498624216174185,
      "learning_rate": 2.859583186930609e-07,
      "loss": 0.0667,
      "step": 15560
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.136293413826899,
      "learning_rate": 2.852106743514854e-07,
      "loss": 0.2262,
      "step": 15561
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4437583600925708,
      "learning_rate": 2.844640030488788e-07,
      "loss": 0.0997,
      "step": 15562
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.46038883763482563,
      "learning_rate": 2.837183048146369e-07,
      "loss": 0.104,
      "step": 15563
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4990663464883532,
      "learning_rate": 2.8297357967811966e-07,
      "loss": 0.0734,
      "step": 15564
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5493397265781912,
      "learning_rate": 2.8222982766864523e-07,
      "loss": 0.0887,
      "step": 15565
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.40039817161321795,
      "learning_rate": 2.8148704881550135e-07,
      "loss": 0.036,
      "step": 15566
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.46467522528591154,
      "learning_rate": 2.8074524314792295e-07,
      "loss": 0.0834,
      "step": 15567
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5975202494566025,
      "learning_rate": 2.8000441069512275e-07,
      "loss": 0.0767,
      "step": 15568
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3926580088707217,
      "learning_rate": 2.7926455148626353e-07,
      "loss": 0.0582,
      "step": 15569
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4369537945162571,
      "learning_rate": 2.785256655504748e-07,
      "loss": 0.0949,
      "step": 15570
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6708913907796945,
      "learning_rate": 2.777877529168499e-07,
      "loss": 0.1307,
      "step": 15571
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5301075686690635,
      "learning_rate": 2.7705081361443785e-07,
      "loss": 0.0822,
      "step": 15572
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.620367583868941,
      "learning_rate": 2.7631484767225434e-07,
      "loss": 0.0857,
      "step": 15573
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4904610881556805,
      "learning_rate": 2.755798551192734e-07,
      "loss": 0.0762,
      "step": 15574
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7752793340959911,
      "learning_rate": 2.7484583598443306e-07,
      "loss": 0.1325,
      "step": 15575
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.44485613670638136,
      "learning_rate": 2.7411279029662953e-07,
      "loss": 0.1204,
      "step": 15576
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3769090114972836,
      "learning_rate": 2.7338071808473145e-07,
      "loss": 0.0625,
      "step": 15577
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5292843275387653,
      "learning_rate": 2.7264961937755185e-07,
      "loss": 0.361,
      "step": 15578
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7892840832876412,
      "learning_rate": 2.719194942038761e-07,
      "loss": 0.1314,
      "step": 15579
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4733636339150137,
      "learning_rate": 2.7119034259245614e-07,
      "loss": 0.0717,
      "step": 15580
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4004089249893264,
      "learning_rate": 2.70462164571994e-07,
      "loss": 0.0561,
      "step": 15581
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5626703783469789,
      "learning_rate": 2.6973496017115853e-07,
      "loss": 0.1079,
      "step": 15582
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5522818327331592,
      "learning_rate": 2.690087294185795e-07,
      "loss": 0.0659,
      "step": 15583
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4196303007034548,
      "learning_rate": 2.6828347234285343e-07,
      "loss": 0.0442,
      "step": 15584
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5814028264680686,
      "learning_rate": 2.6755918897253265e-07,
      "loss": 0.1414,
      "step": 15585
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5163991184476645,
      "learning_rate": 2.668358793361275e-07,
      "loss": 0.1117,
      "step": 15586
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.363251590801656,
      "learning_rate": 2.661135434621237e-07,
      "loss": 0.0502,
      "step": 15587
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.39102961409574166,
      "learning_rate": 2.6539218137895396e-07,
      "loss": 0.0344,
      "step": 15588
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.47796023089094586,
      "learning_rate": 2.6467179311502056e-07,
      "loss": 0.1256,
      "step": 15589
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.36491386152559624,
      "learning_rate": 2.6395237869868696e-07,
      "loss": 0.038,
      "step": 15590
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.48038922707571724,
      "learning_rate": 2.632339381582749e-07,
      "loss": 0.0643,
      "step": 15591
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5673992853230599,
      "learning_rate": 2.6251647152207014e-07,
      "loss": 0.0657,
      "step": 15592
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5250779965003788,
      "learning_rate": 2.617999788183223e-07,
      "loss": 0.0763,
      "step": 15593
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4209990644044442,
      "learning_rate": 2.610844600752366e-07,
      "loss": 0.05,
      "step": 15594
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.488811846635984,
      "learning_rate": 2.6036991532098764e-07,
      "loss": 0.0877,
      "step": 15595
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6415563531895071,
      "learning_rate": 2.59656344583703e-07,
      "loss": 0.0919,
      "step": 15596
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.461476992514987,
      "learning_rate": 2.5894374789148243e-07,
      "loss": 0.0951,
      "step": 15597
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7043632638204383,
      "learning_rate": 2.5823212527237296e-07,
      "loss": 0.2163,
      "step": 15598
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4373733320609404,
      "learning_rate": 2.575214767543993e-07,
      "loss": 0.129,
      "step": 15599
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4958896405380964,
      "learning_rate": 2.5681180236553914e-07,
      "loss": 0.1821,
      "step": 15600
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5508834803621717,
      "learning_rate": 2.5610310213372845e-07,
      "loss": 0.0827,
      "step": 15601
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.296853635810513,
      "learning_rate": 2.553953760868727e-07,
      "loss": 0.0417,
      "step": 15602
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3984797334801975,
      "learning_rate": 2.5468862425283293e-07,
      "loss": 0.3053,
      "step": 15603
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.42200951136458315,
      "learning_rate": 2.539828466594396e-07,
      "loss": 0.0794,
      "step": 15604
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6594878017749224,
      "learning_rate": 2.5327804333447616e-07,
      "loss": 0.1617,
      "step": 15605
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3972110514080986,
      "learning_rate": 2.5257421430569253e-07,
      "loss": 0.0906,
      "step": 15606
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.44250342088439154,
      "learning_rate": 2.518713596007916e-07,
      "loss": 0.0497,
      "step": 15607
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.45924143798753503,
      "learning_rate": 2.5116947924745947e-07,
      "loss": 0.1129,
      "step": 15608
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8193423923237527,
      "learning_rate": 2.5046857327331583e-07,
      "loss": 0.1334,
      "step": 15609
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5126154701066877,
      "learning_rate": 2.4976864170596637e-07,
      "loss": 0.1974,
      "step": 15610
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.37632513983381066,
      "learning_rate": 2.490696845729584e-07,
      "loss": 0.0831,
      "step": 15611
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6915915046731674,
      "learning_rate": 2.4837170190182e-07,
      "loss": 0.1687,
      "step": 15612
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4842162373672349,
      "learning_rate": 2.476746937200236e-07,
      "loss": 0.0721,
      "step": 15613
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8128632401637079,
      "learning_rate": 2.469786600550167e-07,
      "loss": 0.1575,
      "step": 15614
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4738381548942932,
      "learning_rate": 2.4628360093419676e-07,
      "loss": 0.0905,
      "step": 15615
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4042135194131594,
      "learning_rate": 2.4558951638493366e-07,
      "loss": 0.084,
      "step": 15616
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.46685631869835176,
      "learning_rate": 2.4489640643454994e-07,
      "loss": 0.1306,
      "step": 15617
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4200018425134721,
      "learning_rate": 2.442042711103404e-07,
      "loss": 0.0857,
      "step": 15618
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6487017007383877,
      "learning_rate": 2.435131104395444e-07,
      "loss": 0.2083,
      "step": 15619
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6594664166574493,
      "learning_rate": 2.428229244493846e-07,
      "loss": 0.1272,
      "step": 15620
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3926241551170777,
      "learning_rate": 2.4213371316702815e-07,
      "loss": 0.0659,
      "step": 15621
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5079924997693036,
      "learning_rate": 2.414454766196117e-07,
      "loss": 0.0678,
      "step": 15622
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6377110074831591,
      "learning_rate": 2.4075821483422467e-07,
      "loss": 0.1366,
      "step": 15623
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3644614846866738,
      "learning_rate": 2.400719278379371e-07,
      "loss": 0.0957,
      "step": 15624
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.504883450783172,
      "learning_rate": 2.3938661565776066e-07,
      "loss": 0.0873,
      "step": 15625
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.766743144486915,
      "learning_rate": 2.3870227832067936e-07,
      "loss": 0.209,
      "step": 15626
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4197163570454367,
      "learning_rate": 2.3801891585363278e-07,
      "loss": 0.0719,
      "step": 15627
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5983229064381052,
      "learning_rate": 2.373365282835327e-07,
      "loss": 0.0908,
      "step": 15628
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5559509217560268,
      "learning_rate": 2.3665511563723542e-07,
      "loss": 0.0865,
      "step": 15629
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7281984945702195,
      "learning_rate": 2.359746779415778e-07,
      "loss": 0.1383,
      "step": 15630
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8190992940594335,
      "learning_rate": 2.352952152233412e-07,
      "loss": 0.255,
      "step": 15631
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1532021505458911,
      "learning_rate": 2.346167275092792e-07,
      "loss": 0.4047,
      "step": 15632
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.44090120707153513,
      "learning_rate": 2.339392148261066e-07,
      "loss": 0.1004,
      "step": 15633
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6133065014990601,
      "learning_rate": 2.3326267720049922e-07,
      "loss": 0.1179,
      "step": 15634
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.938679552590905,
      "learning_rate": 2.3258711465908855e-07,
      "loss": 0.2598,
      "step": 15635
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5313009843972651,
      "learning_rate": 2.319125272284728e-07,
      "loss": 0.0692,
      "step": 15636
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4181453224468861,
      "learning_rate": 2.3123891493521122e-07,
      "loss": 0.0533,
      "step": 15637
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.32310476156175794,
      "learning_rate": 2.305662778058243e-07,
      "loss": 0.0312,
      "step": 15638
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4406890028305796,
      "learning_rate": 2.298946158667936e-07,
      "loss": 0.0471,
      "step": 15639
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.45607360170880684,
      "learning_rate": 2.2922392914456748e-07,
      "loss": 0.0742,
      "step": 15640
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.42179485047985044,
      "learning_rate": 2.2855421766554697e-07,
      "loss": 0.0854,
      "step": 15641
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8811609172404412,
      "learning_rate": 2.2788548145609712e-07,
      "loss": 0.1466,
      "step": 15642
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6335025033765408,
      "learning_rate": 2.272177205425524e-07,
      "loss": 0.1417,
      "step": 15643
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5742621763392216,
      "learning_rate": 2.2655093495119738e-07,
      "loss": 0.0379,
      "step": 15644
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0811244939781532,
      "learning_rate": 2.2588512470828883e-07,
      "loss": 0.2493,
      "step": 15645
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5693286336404222,
      "learning_rate": 2.252202898400363e-07,
      "loss": 0.0875,
      "step": 15646
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.418741947497144,
      "learning_rate": 2.2455643037261608e-07,
      "loss": 0.0853,
      "step": 15647
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.37554446600029323,
      "learning_rate": 2.2389354633216842e-07,
      "loss": 0.041,
      "step": 15648
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.35001932123076296,
      "learning_rate": 2.2323163774478628e-07,
      "loss": 0.0547,
      "step": 15649
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3452682416859117,
      "learning_rate": 2.225707046365294e-07,
      "loss": 0.0614,
      "step": 15650
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3829829146264905,
      "learning_rate": 2.2191074703342417e-07,
      "loss": 0.0616,
      "step": 15651
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.613825270920084,
      "learning_rate": 2.2125176496144983e-07,
      "loss": 0.1566,
      "step": 15652
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4314771624279936,
      "learning_rate": 2.2059375844655227e-07,
      "loss": 0.0566,
      "step": 15653
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4146911550732529,
      "learning_rate": 2.1993672751463579e-07,
      "loss": 0.0634,
      "step": 15654
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4286066397939758,
      "learning_rate": 2.1928067219156856e-07,
      "loss": 0.0998,
      "step": 15655
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.46500056488023567,
      "learning_rate": 2.1862559250318272e-07,
      "loss": 0.0571,
      "step": 15656
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5175761878750599,
      "learning_rate": 2.1797148847526871e-07,
      "loss": 0.058,
      "step": 15657
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7037804163046424,
      "learning_rate": 2.1731836013357542e-07,
      "loss": 0.2506,
      "step": 15658
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.44614893003884776,
      "learning_rate": 2.1666620750382115e-07,
      "loss": 0.0753,
      "step": 15659
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3556548613443742,
      "learning_rate": 2.1601503061167984e-07,
      "loss": 0.0655,
      "step": 15660
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0686762810044166,
      "learning_rate": 2.1536482948278923e-07,
      "loss": 0.4778,
      "step": 15661
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4443443051423737,
      "learning_rate": 2.1471560414274561e-07,
      "loss": 0.0609,
      "step": 15662
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6719683347575771,
      "learning_rate": 2.140673546171118e-07,
      "loss": 0.0775,
      "step": 15663
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7353860502331292,
      "learning_rate": 2.1342008093141187e-07,
      "loss": 0.2442,
      "step": 15664
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5252864676650066,
      "learning_rate": 2.1277378311112816e-07,
      "loss": 0.0739,
      "step": 15665
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.39727126557710646,
      "learning_rate": 2.121284611817015e-07,
      "loss": 0.0546,
      "step": 15666
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.39084002042185945,
      "learning_rate": 2.1148411516854204e-07,
      "loss": 0.0795,
      "step": 15667
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7412693394226373,
      "learning_rate": 2.1084074509702122e-07,
      "loss": 0.1311,
      "step": 15668
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3384359230842902,
      "learning_rate": 2.1019835099246598e-07,
      "loss": 0.0618,
      "step": 15669
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.45743882132843156,
      "learning_rate": 2.0955693288016442e-07,
      "loss": 0.0785,
      "step": 15670
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3806051291010356,
      "learning_rate": 2.0891649078537688e-07,
      "loss": 0.0723,
      "step": 15671
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6400316634751744,
      "learning_rate": 2.0827702473331379e-07,
      "loss": 0.1176,
      "step": 15672
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.684795328048062,
      "learning_rate": 2.076385347491494e-07,
      "loss": 0.1566,
      "step": 15673
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4486272705053058,
      "learning_rate": 2.0700102085802474e-07,
      "loss": 0.0272,
      "step": 15674
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0010454326956841,
      "learning_rate": 2.0636448308503918e-07,
      "loss": 0.2555,
      "step": 15675
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4398202458327705,
      "learning_rate": 2.0572892145525324e-07,
      "loss": 0.0712,
      "step": 15676
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5097470099968621,
      "learning_rate": 2.0509433599368854e-07,
      "loss": 0.131,
      "step": 15677
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9574648598073239,
      "learning_rate": 2.044607267253279e-07,
      "loss": 0.2911,
      "step": 15678
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4180253716125843,
      "learning_rate": 2.038280936751208e-07,
      "loss": 0.0595,
      "step": 15679
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.610419064508702,
      "learning_rate": 2.031964368679723e-07,
      "loss": 0.1343,
      "step": 15680
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.38366361481692535,
      "learning_rate": 2.025657563287514e-07,
      "loss": 0.0603,
      "step": 15681
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.41307181320211905,
      "learning_rate": 2.0193605208228826e-07,
      "loss": 0.0543,
      "step": 15682
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.42832494554652933,
      "learning_rate": 2.0130732415337416e-07,
      "loss": 0.0541,
      "step": 15683
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5907603272884132,
      "learning_rate": 2.0067957256676428e-07,
      "loss": 0.1142,
      "step": 15684
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5798984380098396,
      "learning_rate": 2.000527973471722e-07,
      "loss": 0.0964,
      "step": 15685
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.38790492841448687,
      "learning_rate": 1.9942699851927537e-07,
      "loss": 0.0832,
      "step": 15686
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.41000800677460725,
      "learning_rate": 1.9880217610770967e-07,
      "loss": 0.0439,
      "step": 15687
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5855189326327698,
      "learning_rate": 1.9817833013708042e-07,
      "loss": 0.1714,
      "step": 15688
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4129931554077157,
      "learning_rate": 1.9755546063194296e-07,
      "loss": 0.0732,
      "step": 15689
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5574910191392739,
      "learning_rate": 1.9693356761682212e-07,
      "loss": 0.0761,
      "step": 15690
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3908642001812269,
      "learning_rate": 1.963126511162011e-07,
      "loss": 0.0668,
      "step": 15691
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5178790151493525,
      "learning_rate": 1.956927111545298e-07,
      "loss": 0.0518,
      "step": 15692
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4130247291972238,
      "learning_rate": 1.9507374775621368e-07,
      "loss": 0.0538,
      "step": 15693
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.39293975963346023,
      "learning_rate": 1.9445576094561936e-07,
      "loss": 0.0735,
      "step": 15694
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5380896506396365,
      "learning_rate": 1.9383875074707735e-07,
      "loss": 0.1519,
      "step": 15695
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5572326687530423,
      "learning_rate": 1.9322271718488494e-07,
      "loss": 0.1038,
      "step": 15696
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4295066977530494,
      "learning_rate": 1.9260766028329214e-07,
      "loss": 0.0641,
      "step": 15697
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.38896788863266046,
      "learning_rate": 1.9199358006651292e-07,
      "loss": 0.0835,
      "step": 15698
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4262770701881532,
      "learning_rate": 1.9138047655872794e-07,
      "loss": 0.0609,
      "step": 15699
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.35468471782026434,
      "learning_rate": 1.9076834978407066e-07,
      "loss": 0.0525,
      "step": 15700
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5696821906074432,
      "learning_rate": 1.9015719976664402e-07,
      "loss": 0.1064,
      "step": 15701
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9584153370444181,
      "learning_rate": 1.8954702653050937e-07,
      "loss": 0.1345,
      "step": 15702
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5841879851346203,
      "learning_rate": 1.889378300996919e-07,
      "loss": 0.1213,
      "step": 15703
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4082354909823486,
      "learning_rate": 1.8832961049816965e-07,
      "loss": 0.0436,
      "step": 15704
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8686253165829342,
      "learning_rate": 1.877223677498957e-07,
      "loss": 0.1275,
      "step": 15705
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1098217932666934,
      "learning_rate": 1.8711610187877317e-07,
      "loss": 0.3984,
      "step": 15706
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5037672630439616,
      "learning_rate": 1.8651081290867177e-07,
      "loss": 0.0692,
      "step": 15707
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8236595427675533,
      "learning_rate": 1.8590650086342254e-07,
      "loss": 0.205,
      "step": 15708
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8665031636386419,
      "learning_rate": 1.8530316576682027e-07,
      "loss": 0.1656,
      "step": 15709
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7148137838136163,
      "learning_rate": 1.8470080764261544e-07,
      "loss": 0.0973,
      "step": 15710
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4125133674518903,
      "learning_rate": 1.8409942651452238e-07,
      "loss": 0.0705,
      "step": 15711
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6247865348416661,
      "learning_rate": 1.834990224062222e-07,
      "loss": 0.213,
      "step": 15712
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6022270022651531,
      "learning_rate": 1.828995953413515e-07,
      "loss": 0.1175,
      "step": 15713
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3571461924002964,
      "learning_rate": 1.8230114534350527e-07,
      "loss": 0.029,
      "step": 15714
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5992602206066558,
      "learning_rate": 1.8170367243625362e-07,
      "loss": 0.182,
      "step": 15715
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8200093930719351,
      "learning_rate": 1.8110717664311383e-07,
      "loss": 0.1061,
      "step": 15716
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7747483311739558,
      "learning_rate": 1.8051165798757262e-07,
      "loss": 0.2256,
      "step": 15717
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3406932901544003,
      "learning_rate": 1.7991711649306964e-07,
      "loss": 0.0519,
      "step": 15718
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.43071787158739394,
      "learning_rate": 1.793235521830222e-07,
      "loss": 0.0946,
      "step": 15719
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3758938198367459,
      "learning_rate": 1.78730965080795e-07,
      "loss": 0.0722,
      "step": 15720
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3397371801170936,
      "learning_rate": 1.7813935520971937e-07,
      "loss": 0.0531,
      "step": 15721
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3583222210113249,
      "learning_rate": 1.775487225930822e-07,
      "loss": 0.0384,
      "step": 15722
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6186491808433853,
      "learning_rate": 1.7695906725414545e-07,
      "loss": 0.2198,
      "step": 15723
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5700049562124408,
      "learning_rate": 1.7637038921611836e-07,
      "loss": 0.137,
      "step": 15724
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6756337167710835,
      "learning_rate": 1.7578268850217962e-07,
      "loss": 0.1875,
      "step": 15725
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3399248746400404,
      "learning_rate": 1.7519596513546898e-07,
      "loss": 0.0436,
      "step": 15726
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3487263213239714,
      "learning_rate": 1.7461021913908194e-07,
      "loss": 0.0474,
      "step": 15727
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5426608065284269,
      "learning_rate": 1.7402545053608333e-07,
      "loss": 0.1117,
      "step": 15728
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4727905831164244,
      "learning_rate": 1.7344165934949641e-07,
      "loss": 0.0927,
      "step": 15729
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3810617131439113,
      "learning_rate": 1.728588456023028e-07,
      "loss": 0.121,
      "step": 15730
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6756743474467214,
      "learning_rate": 1.7227700931744805e-07,
      "loss": 0.1989,
      "step": 15731
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3913638584651113,
      "learning_rate": 1.716961505178416e-07,
      "loss": 0.057,
      "step": 15732
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9475839503165404,
      "learning_rate": 1.7111626922635405e-07,
      "loss": 0.2703,
      "step": 15733
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.40048528794450433,
      "learning_rate": 1.705373654658088e-07,
      "loss": 0.0485,
      "step": 15734
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9552434050131838,
      "learning_rate": 1.6995943925900425e-07,
      "loss": 0.2755,
      "step": 15735
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.48360676669434577,
      "learning_rate": 1.693824906286945e-07,
      "loss": 0.1279,
      "step": 15736
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5331855579152508,
      "learning_rate": 1.688065195975863e-07,
      "loss": 0.0744,
      "step": 15737
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3876262695073282,
      "learning_rate": 1.6823152618836435e-07,
      "loss": 0.0678,
      "step": 15738
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.22732285670075125,
      "learning_rate": 1.6765751042366328e-07,
      "loss": 0.023,
      "step": 15739
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4381091868062963,
      "learning_rate": 1.6708447232608448e-07,
      "loss": 0.0677,
      "step": 15740
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7625010594499386,
      "learning_rate": 1.665124119181849e-07,
      "loss": 0.2509,
      "step": 15741
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5498478398177808,
      "learning_rate": 1.6594132922248818e-07,
      "loss": 0.1134,
      "step": 15742
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.44570073514038827,
      "learning_rate": 1.653712242614791e-07,
      "loss": 0.085,
      "step": 15743
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5623640206582844,
      "learning_rate": 1.648020970576064e-07,
      "loss": 0.112,
      "step": 15744
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.34159619191662743,
      "learning_rate": 1.642339476332716e-07,
      "loss": 0.0422,
      "step": 15745
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.47694095136066583,
      "learning_rate": 1.636667760108457e-07,
      "loss": 0.0758,
      "step": 15746
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3380565094207437,
      "learning_rate": 1.6310058221265802e-07,
      "loss": 0.067,
      "step": 15747
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.48179142251490426,
      "learning_rate": 1.6253536626100184e-07,
      "loss": 0.0607,
      "step": 15748
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8647321047328326,
      "learning_rate": 1.6197112817812886e-07,
      "loss": 0.2515,
      "step": 15749
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3710448123268943,
      "learning_rate": 1.6140786798625184e-07,
      "loss": 0.0704,
      "step": 15750
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6630606839301783,
      "learning_rate": 1.608455857075475e-07,
      "loss": 0.0825,
      "step": 15751
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8686691545966154,
      "learning_rate": 1.602842813641564e-07,
      "loss": 0.3795,
      "step": 15752
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5488581833814982,
      "learning_rate": 1.5972395497817482e-07,
      "loss": 0.0603,
      "step": 15753
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5558030984016132,
      "learning_rate": 1.591646065716629e-07,
      "loss": 0.2014,
      "step": 15754
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.47673054271936544,
      "learning_rate": 1.5860623616664184e-07,
      "loss": 0.0652,
      "step": 15755
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3199174213664345,
      "learning_rate": 1.5804884378509965e-07,
      "loss": 0.0392,
      "step": 15756
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8768538066996067,
      "learning_rate": 1.5749242944897435e-07,
      "loss": 0.2031,
      "step": 15757
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3879008571330153,
      "learning_rate": 1.5693699318017897e-07,
      "loss": 0.052,
      "step": 15758
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5402706399519123,
      "learning_rate": 1.5638253500057652e-07,
      "loss": 0.075,
      "step": 15759
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4105774041952366,
      "learning_rate": 1.558290549319996e-07,
      "loss": 0.0572,
      "step": 15760
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3167586570560746,
      "learning_rate": 1.552765529962391e-07,
      "loss": 0.0448,
      "step": 15761
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.34918691894186676,
      "learning_rate": 1.5472502921504428e-07,
      "loss": 0.0374,
      "step": 15762
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.42260801679349996,
      "learning_rate": 1.5417448361013387e-07,
      "loss": 0.0626,
      "step": 15763
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8322002506678846,
      "learning_rate": 1.5362491620317664e-07,
      "loss": 0.1567,
      "step": 15764
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4000939719829005,
      "learning_rate": 1.5307632701581364e-07,
      "loss": 0.056,
      "step": 15765
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5523756002544372,
      "learning_rate": 1.5252871606964424e-07,
      "loss": 0.0875,
      "step": 15766
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.28823101710062976,
      "learning_rate": 1.51982083386229e-07,
      "loss": 0.0304,
      "step": 15767
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.352689207479238,
      "learning_rate": 1.51436428987084e-07,
      "loss": 0.4507,
      "step": 15768
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5997366292992637,
      "learning_rate": 1.5089175289369483e-07,
      "loss": 0.0788,
      "step": 15769
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.48218698116640757,
      "learning_rate": 1.5034805512750827e-07,
      "loss": 0.0841,
      "step": 15770
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3521786494227853,
      "learning_rate": 1.4980533570992662e-07,
      "loss": 0.0353,
      "step": 15771
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4876380551427026,
      "learning_rate": 1.4926359466231887e-07,
      "loss": 0.0552,
      "step": 15772
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5424166469879285,
      "learning_rate": 1.4872283200601245e-07,
      "loss": 0.0617,
      "step": 15773
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5036476535508797,
      "learning_rate": 1.4818304776229864e-07,
      "loss": 0.0847,
      "step": 15774
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5069778262653475,
      "learning_rate": 1.4764424195242986e-07,
      "loss": 0.1235,
      "step": 15775
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4317354592666966,
      "learning_rate": 1.47106414597617e-07,
      "loss": 0.0656,
      "step": 15776
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7497172191490505,
      "learning_rate": 1.4656956571903746e-07,
      "loss": 0.1862,
      "step": 15777
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8088278202732985,
      "learning_rate": 1.460336953378244e-07,
      "loss": 0.3346,
      "step": 15778
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9922149948168999,
      "learning_rate": 1.4549880347507762e-07,
      "loss": 0.2346,
      "step": 15779
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5656249774763275,
      "learning_rate": 1.4496489015185522e-07,
      "loss": 0.0831,
      "step": 15780
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.38160768510866383,
      "learning_rate": 1.444319553891793e-07,
      "loss": 0.0354,
      "step": 15781
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3427852819554761,
      "learning_rate": 1.4389999920802754e-07,
      "loss": 0.0745,
      "step": 15782
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5523618839790795,
      "learning_rate": 1.4336902162934984e-07,
      "loss": 0.1185,
      "step": 15783
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.45523052979369827,
      "learning_rate": 1.4283902267404891e-07,
      "loss": 0.0599,
      "step": 15784
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5930288396648032,
      "learning_rate": 1.4231000236298863e-07,
      "loss": 0.108,
      "step": 15785
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3830318419087345,
      "learning_rate": 1.417819607169968e-07,
      "loss": 0.0861,
      "step": 15786
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5295194621473827,
      "learning_rate": 1.4125489775686507e-07,
      "loss": 0.1627,
      "step": 15787
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.46360896048608485,
      "learning_rate": 1.4072881350334356e-07,
      "loss": 0.1189,
      "step": 15788
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3332214740450619,
      "learning_rate": 1.402037079771462e-07,
      "loss": 0.0804,
      "step": 15789
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.42125859187564024,
      "learning_rate": 1.3967958119894263e-07,
      "loss": 0.0519,
      "step": 15790
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6751965957768968,
      "learning_rate": 1.3915643318937466e-07,
      "loss": 0.0841,
      "step": 15791
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7320234339745233,
      "learning_rate": 1.3863426396903134e-07,
      "loss": 0.1603,
      "step": 15792
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7970835675488863,
      "learning_rate": 1.3811307355847681e-07,
      "loss": 0.1915,
      "step": 15793
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4878526168224064,
      "learning_rate": 1.3759286197822795e-07,
      "loss": 0.5288,
      "step": 15794
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4747007397348793,
      "learning_rate": 1.3707362924876567e-07,
      "loss": 0.0943,
      "step": 15795
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8919878211152856,
      "learning_rate": 1.3655537539053188e-07,
      "loss": 0.2175,
      "step": 15796
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.33498261450803346,
      "learning_rate": 1.360381004239353e-07,
      "loss": 0.0454,
      "step": 15797
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5915596315172043,
      "learning_rate": 1.3552180436933738e-07,
      "loss": 0.0812,
      "step": 15798
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6943001665727332,
      "learning_rate": 1.3500648724706355e-07,
      "loss": 0.1254,
      "step": 15799
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.31264357976726104,
      "learning_rate": 1.3449214907740592e-07,
      "loss": 0.0345,
      "step": 15800
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4298919921752915,
      "learning_rate": 1.3397878988061218e-07,
      "loss": 0.0549,
      "step": 15801
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6133561028876008,
      "learning_rate": 1.3346640967689673e-07,
      "loss": 0.1344,
      "step": 15802
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5437489853599728,
      "learning_rate": 1.3295500848642671e-07,
      "loss": 0.169,
      "step": 15803
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4255887827886462,
      "learning_rate": 1.3244458632934166e-07,
      "loss": 0.0553,
      "step": 15804
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9653756545539012,
      "learning_rate": 1.3193514322573374e-07,
      "loss": 0.143,
      "step": 15805
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.26822468876569827,
      "learning_rate": 1.31426679195662e-07,
      "loss": 0.0392,
      "step": 15806
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4077160085318314,
      "learning_rate": 1.3091919425914367e-07,
      "loss": 0.0383,
      "step": 15807
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.45243981600285094,
      "learning_rate": 1.3041268843616283e-07,
      "loss": 0.0812,
      "step": 15808
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5224906865919173,
      "learning_rate": 1.299071617466535e-07,
      "loss": 0.1032,
      "step": 15809
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.434566451041951,
      "learning_rate": 1.2940261421052756e-07,
      "loss": 0.0482,
      "step": 15810
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.29072667591358925,
      "learning_rate": 1.2889904584764135e-07,
      "loss": 0.04,
      "step": 15811
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.46894305389305546,
      "learning_rate": 1.2839645667782617e-07,
      "loss": 0.0598,
      "step": 15812
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5015502372019598,
      "learning_rate": 1.2789484672086627e-07,
      "loss": 0.0716,
      "step": 15813
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4787688253906058,
      "learning_rate": 1.2739421599651247e-07,
      "loss": 0.1246,
      "step": 15814
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3328809755506702,
      "learning_rate": 1.2689456452447124e-07,
      "loss": 0.0507,
      "step": 15815
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5795538243849981,
      "learning_rate": 1.263958923244213e-07,
      "loss": 0.1482,
      "step": 15816
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.48466810630512536,
      "learning_rate": 1.2589819941598856e-07,
      "loss": 0.1034,
      "step": 15817
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.40102174477840824,
      "learning_rate": 1.254014858187713e-07,
      "loss": 0.0453,
      "step": 15818
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5877955924141154,
      "learning_rate": 1.249057515523233e-07,
      "loss": 0.0636,
      "step": 15819
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.272318541149813,
      "learning_rate": 1.2441099663616507e-07,
      "loss": 0.3664,
      "step": 15820
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6702285877112056,
      "learning_rate": 1.239172210897699e-07,
      "loss": 0.1242,
      "step": 15821
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7220990823481364,
      "learning_rate": 1.234244249325861e-07,
      "loss": 0.1877,
      "step": 15822
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6369249151943785,
      "learning_rate": 1.2293260818400932e-07,
      "loss": 0.1354,
      "step": 15823
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4686430673019108,
      "learning_rate": 1.2244177086340458e-07,
      "loss": 0.0781,
      "step": 15824
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.47424325432059133,
      "learning_rate": 1.2195191299009533e-07,
      "loss": 0.0976,
      "step": 15825
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5853128906462096,
      "learning_rate": 1.214630345833717e-07,
      "loss": 0.1236,
      "step": 15826
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7688830537018433,
      "learning_rate": 1.2097513566247386e-07,
      "loss": 0.1556,
      "step": 15827
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.41220831847501116,
      "learning_rate": 1.2048821624661976e-07,
      "loss": 0.0407,
      "step": 15828
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8256532552313703,
      "learning_rate": 1.2000227635497185e-07,
      "loss": 0.191,
      "step": 15829
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.527285128943477,
      "learning_rate": 1.195173160066676e-07,
      "loss": 0.0696,
      "step": 15830
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5328028823134345,
      "learning_rate": 1.190333352207973e-07,
      "loss": 0.0801,
      "step": 15831
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5521127958279803,
      "learning_rate": 1.1855033401641513e-07,
      "loss": 0.093,
      "step": 15832
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4781263863697509,
      "learning_rate": 1.1806831241253647e-07,
      "loss": 0.1441,
      "step": 15833
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.49694700677914383,
      "learning_rate": 1.1758727042814333e-07,
      "loss": 0.1464,
      "step": 15834
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9579735195577881,
      "learning_rate": 1.1710720808217058e-07,
      "loss": 0.1641,
      "step": 15835
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7679479627243696,
      "learning_rate": 1.1662812539351975e-07,
      "loss": 0.1563,
      "step": 15836
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6950052535275767,
      "learning_rate": 1.1615002238105355e-07,
      "loss": 0.0484,
      "step": 15837
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5137525604059104,
      "learning_rate": 1.156728990635958e-07,
      "loss": 0.092,
      "step": 15838
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5178650255903675,
      "learning_rate": 1.1519675545992869e-07,
      "loss": 0.0785,
      "step": 15839
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5968284239453995,
      "learning_rate": 1.1472159158880113e-07,
      "loss": 0.17,
      "step": 15840
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.575361826507174,
      "learning_rate": 1.142474074689176e-07,
      "loss": 0.1569,
      "step": 15841
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5282226767473601,
      "learning_rate": 1.1377420311894927e-07,
      "loss": 0.0997,
      "step": 15842
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.37088768089792606,
      "learning_rate": 1.1330197855752567e-07,
      "loss": 0.0773,
      "step": 15843
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6353609531679854,
      "learning_rate": 1.128307338032375e-07,
      "loss": 0.138,
      "step": 15844
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5650836250488852,
      "learning_rate": 1.1236046887464213e-07,
      "loss": 0.0789,
      "step": 15845
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2271614258436506,
      "learning_rate": 1.1189118379024699e-07,
      "loss": 0.024,
      "step": 15846
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7324418121869316,
      "learning_rate": 1.1142287856853729e-07,
      "loss": 0.1643,
      "step": 15847
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5819772794186114,
      "learning_rate": 1.109555532279427e-07,
      "loss": 0.1194,
      "step": 15848
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5389787895113703,
      "learning_rate": 1.10489207786868e-07,
      "loss": 0.1202,
      "step": 15849
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4448590069680101,
      "learning_rate": 1.1002384226366791e-07,
      "loss": 0.0748,
      "step": 15850
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0177363312655214,
      "learning_rate": 1.0955945667666944e-07,
      "loss": 0.307,
      "step": 15851
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.41848754180205516,
      "learning_rate": 1.0909605104415522e-07,
      "loss": 0.0568,
      "step": 15852
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4810190260225399,
      "learning_rate": 1.0863362538436339e-07,
      "loss": 0.1443,
      "step": 15853
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4837721833872442,
      "learning_rate": 1.0817217971550719e-07,
      "loss": 0.0926,
      "step": 15854
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6075201379047728,
      "learning_rate": 1.0771171405575265e-07,
      "loss": 0.1775,
      "step": 15855
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6292725076653763,
      "learning_rate": 1.0725222842322691e-07,
      "loss": 0.2512,
      "step": 15856
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5080227203595787,
      "learning_rate": 1.0679372283602108e-07,
      "loss": 0.1315,
      "step": 15857
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3562067662531874,
      "learning_rate": 1.0633619731218459e-07,
      "loss": 0.0567,
      "step": 15858
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3513244001080973,
      "learning_rate": 1.0587965186973637e-07,
      "loss": 0.0467,
      "step": 15859
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7070641398656078,
      "learning_rate": 1.054240865266426e-07,
      "loss": 0.1122,
      "step": 15860
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.659644975104564,
      "learning_rate": 1.0496950130084727e-07,
      "loss": 0.1649,
      "step": 15861
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.41431136220284387,
      "learning_rate": 1.0451589621024438e-07,
      "loss": 0.0726,
      "step": 15862
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.530024412858691,
      "learning_rate": 1.0406327127269188e-07,
      "loss": 0.1757,
      "step": 15863
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4896117361427048,
      "learning_rate": 1.0361162650600886e-07,
      "loss": 0.0684,
      "step": 15864
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9124139175180915,
      "learning_rate": 1.0316096192798108e-07,
      "loss": 0.192,
      "step": 15865
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5070270427209034,
      "learning_rate": 1.027112775563499e-07,
      "loss": 0.1376,
      "step": 15866
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8246916661915925,
      "learning_rate": 1.0226257340881784e-07,
      "loss": 0.1861,
      "step": 15867
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6571455634010751,
      "learning_rate": 1.0181484950305132e-07,
      "loss": 0.2189,
      "step": 15868
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.418937187386679,
      "learning_rate": 1.0136810585667788e-07,
      "loss": 0.0576,
      "step": 15869
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8490925995898057,
      "learning_rate": 1.0092234248728627e-07,
      "loss": 0.2111,
      "step": 15870
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5673461106910922,
      "learning_rate": 1.0047755941242632e-07,
      "loss": 0.11,
      "step": 15871
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.415946652306177,
      "learning_rate": 1.0003375664961178e-07,
      "loss": 0.056,
      "step": 15872
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5829728983701377,
      "learning_rate": 9.959093421630927e-08,
      "loss": 0.1191,
      "step": 15873
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3874552045638902,
      "learning_rate": 9.914909212996315e-08,
      "loss": 0.4582,
      "step": 15874
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.85596887113647,
      "learning_rate": 9.870823040795951e-08,
      "loss": 0.254,
      "step": 15875
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3553141048459087,
      "learning_rate": 9.826834906765947e-08,
      "loss": 0.0411,
      "step": 15876
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.447399633646008,
      "learning_rate": 9.782944812637973e-08,
      "loss": 0.0517,
      "step": 15877
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7110857440349088,
      "learning_rate": 9.739152760140368e-08,
      "loss": 0.1644,
      "step": 15878
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8312231118298474,
      "learning_rate": 9.695458750996755e-08,
      "loss": 0.1402,
      "step": 15879
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.962911558334013,
      "learning_rate": 9.65186278692798e-08,
      "loss": 0.2783,
      "step": 15880
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.34862657684027015,
      "learning_rate": 9.608364869649888e-08,
      "loss": 0.048,
      "step": 15881
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3997840975949186,
      "learning_rate": 9.564965000875282e-08,
      "loss": 0.0417,
      "step": 15882
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.11858338489701,
      "learning_rate": 9.52166318231279e-08,
      "loss": 0.4058,
      "step": 15883
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.41147979085201464,
      "learning_rate": 9.478459415667717e-08,
      "loss": 0.0452,
      "step": 15884
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.39728014216224616,
      "learning_rate": 9.435353702640093e-08,
      "loss": 0.0696,
      "step": 15885
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5348392435351211,
      "learning_rate": 9.392346044927725e-08,
      "loss": 0.1034,
      "step": 15886
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9278335916761191,
      "learning_rate": 9.349436444223703e-08,
      "loss": 0.2122,
      "step": 15887
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5780060209232744,
      "learning_rate": 9.306624902217232e-08,
      "loss": 0.134,
      "step": 15888
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5724079945169734,
      "learning_rate": 9.263911420593907e-08,
      "loss": 0.1383,
      "step": 15889
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4733114723115034,
      "learning_rate": 9.22129600103544e-08,
      "loss": 0.0636,
      "step": 15890
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5371091106250746,
      "learning_rate": 9.178778645219932e-08,
      "loss": 0.1603,
      "step": 15891
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1433681197814063,
      "learning_rate": 9.136359354820767e-08,
      "loss": 0.2824,
      "step": 15892
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7554459604948833,
      "learning_rate": 9.094038131508275e-08,
      "loss": 0.1329,
      "step": 15893
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2869891059432807,
      "learning_rate": 9.051814976948625e-08,
      "loss": 0.0466,
      "step": 15894
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5008305778346812,
      "learning_rate": 9.009689892804374e-08,
      "loss": 0.0605,
      "step": 15895
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.39815457484416467,
      "learning_rate": 8.967662880733641e-08,
      "loss": 0.0624,
      "step": 15896
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5589695615782233,
      "learning_rate": 8.92573394239149e-08,
      "loss": 0.0856,
      "step": 15897
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6379094717904505,
      "learning_rate": 8.883903079428546e-08,
      "loss": 0.1313,
      "step": 15898
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4092343140898503,
      "learning_rate": 8.842170293491547e-08,
      "loss": 0.0452,
      "step": 15899
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6876407936626712,
      "learning_rate": 8.800535586223624e-08,
      "loss": 0.1414,
      "step": 15900
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.708152064900774,
      "learning_rate": 8.75899895926402e-08,
      "loss": 0.1694,
      "step": 15901
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4045080903433912,
      "learning_rate": 8.717560414247816e-08,
      "loss": 0.0471,
      "step": 15902
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.44148275463477166,
      "learning_rate": 8.676219952807041e-08,
      "loss": 0.0686,
      "step": 15903
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9294775671510519,
      "learning_rate": 8.634977576568726e-08,
      "loss": 0.3707,
      "step": 15904
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7531037238426384,
      "learning_rate": 8.59383328715685e-08,
      "loss": 0.0931,
      "step": 15905
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.42966381751049393,
      "learning_rate": 8.552787086191505e-08,
      "loss": 0.1494,
      "step": 15906
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1208453427691207,
      "learning_rate": 8.511838975288344e-08,
      "loss": 0.1728,
      "step": 15907
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7154218236947266,
      "learning_rate": 8.470988956059689e-08,
      "loss": 0.1826,
      "step": 15908
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5610824250191625,
      "learning_rate": 8.430237030113697e-08,
      "loss": 0.046,
      "step": 15909
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4859961711759355,
      "learning_rate": 8.389583199055195e-08,
      "loss": 0.0682,
      "step": 15910
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.32061026229528683,
      "learning_rate": 8.349027464484293e-08,
      "loss": 0.052,
      "step": 15911
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.44661941558993257,
      "learning_rate": 8.30856982799777e-08,
      "loss": 0.0827,
      "step": 15912
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5849037337038985,
      "learning_rate": 8.268210291188794e-08,
      "loss": 0.0787,
      "step": 15913
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.740827817372661,
      "learning_rate": 8.227948855646094e-08,
      "loss": 0.1952,
      "step": 15914
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.40825167821304215,
      "learning_rate": 8.187785522954794e-08,
      "loss": 0.0543,
      "step": 15915
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4726162076469941,
      "learning_rate": 8.147720294696126e-08,
      "loss": 0.1123,
      "step": 15916
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.26997140632314826,
      "learning_rate": 8.107753172447718e-08,
      "loss": 0.0351,
      "step": 15917
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6096787636481429,
      "learning_rate": 8.067884157782757e-08,
      "loss": 0.1102,
      "step": 15918
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6877621371458476,
      "learning_rate": 8.028113252271097e-08,
      "loss": 0.1104,
      "step": 15919
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4440602907862109,
      "learning_rate": 7.988440457478708e-08,
      "loss": 0.079,
      "step": 15920
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.35797388313052547,
      "learning_rate": 7.948865774967395e-08,
      "loss": 0.0842,
      "step": 15921
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6088555147149932,
      "learning_rate": 7.909389206295081e-08,
      "loss": 0.0848,
      "step": 15922
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3958556866350884,
      "learning_rate": 7.870010753016355e-08,
      "loss": 0.0512,
      "step": 15923
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.33209922109968415,
      "learning_rate": 7.830730416681086e-08,
      "loss": 0.0394,
      "step": 15924
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5504776775057799,
      "learning_rate": 7.791548198836095e-08,
      "loss": 0.1413,
      "step": 15925
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5127423431181182,
      "learning_rate": 7.752464101024037e-08,
      "loss": 0.1755,
      "step": 15926
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3199116738537275,
      "learning_rate": 7.713478124783679e-08,
      "loss": 0.0288,
      "step": 15927
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5422709316615433,
      "learning_rate": 7.674590271649628e-08,
      "loss": 0.124,
      "step": 15928
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5484949948762139,
      "learning_rate": 7.63580054315316e-08,
      "loss": 0.1571,
      "step": 15929
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.44830607446883564,
      "learning_rate": 7.597108940821663e-08,
      "loss": 0.1008,
      "step": 15930
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.40346044362838734,
      "learning_rate": 7.558515466178085e-08,
      "loss": 0.0447,
      "step": 15931
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.559888956472675,
      "learning_rate": 7.520020120742044e-08,
      "loss": 0.1115,
      "step": 15932
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4948011078988056,
      "learning_rate": 7.481622906028995e-08,
      "loss": 0.1094,
      "step": 15933
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6264542911068517,
      "learning_rate": 7.443323823551062e-08,
      "loss": 0.186,
      "step": 15934
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4838977745600733,
      "learning_rate": 7.405122874815651e-08,
      "loss": 0.0877,
      "step": 15935
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.833328963999267,
      "learning_rate": 7.367020061326835e-08,
      "loss": 0.2209,
      "step": 15936
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3990682437010345,
      "learning_rate": 7.329015384585081e-08,
      "loss": 0.0647,
      "step": 15937
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4365588842699204,
      "learning_rate": 7.291108846086136e-08,
      "loss": 0.0747,
      "step": 15938
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.601603968996605,
      "learning_rate": 7.253300447322974e-08,
      "loss": 0.1595,
      "step": 15939
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.31400510217524563,
      "learning_rate": 7.215590189783572e-08,
      "loss": 0.0635,
      "step": 15940
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4880987859123654,
      "learning_rate": 7.177978074953128e-08,
      "loss": 0.0684,
      "step": 15941
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4826732343001734,
      "learning_rate": 7.140464104312127e-08,
      "loss": 0.4064,
      "step": 15942
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7846616256186354,
      "learning_rate": 7.10304827933772e-08,
      "loss": 0.2176,
      "step": 15943
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.37799970449921444,
      "learning_rate": 7.065730601502895e-08,
      "loss": 0.0523,
      "step": 15944
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3743055909451265,
      "learning_rate": 7.028511072276756e-08,
      "loss": 0.0736,
      "step": 15945
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.31957492588489117,
      "learning_rate": 6.991389693124795e-08,
      "loss": 0.0349,
      "step": 15946
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.46558383200097564,
      "learning_rate": 6.954366465508622e-08,
      "loss": 0.0808,
      "step": 15947
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.560982991787857,
      "learning_rate": 6.917441390885682e-08,
      "loss": 0.1198,
      "step": 15948
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.33262157005156284,
      "learning_rate": 6.880614470709812e-08,
      "loss": 0.074,
      "step": 15949
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5287283563105567,
      "learning_rate": 6.843885706430687e-08,
      "loss": 0.0909,
      "step": 15950
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.40897809698137505,
      "learning_rate": 6.807255099494925e-08,
      "loss": 0.0737,
      "step": 15951
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4229279659950788,
      "learning_rate": 6.770722651344153e-08,
      "loss": 0.1172,
      "step": 15952
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.42486083478680803,
      "learning_rate": 6.734288363416941e-08,
      "loss": 0.0813,
      "step": 15953
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.44716337474438983,
      "learning_rate": 6.69795223714742e-08,
      "loss": 0.0646,
      "step": 15954
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6314891035786387,
      "learning_rate": 6.661714273966668e-08,
      "loss": 0.1142,
      "step": 15955
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5349388973134206,
      "learning_rate": 6.625574475301044e-08,
      "loss": 0.097,
      "step": 15956
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6337802945603553,
      "learning_rate": 6.589532842573576e-08,
      "loss": 0.1704,
      "step": 15957
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5033211160031635,
      "learning_rate": 6.553589377203128e-08,
      "loss": 0.0756,
      "step": 15958
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.41663196886184295,
      "learning_rate": 6.517744080604682e-08,
      "loss": 0.0591,
      "step": 15959
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6182491372102438,
      "learning_rate": 6.481996954189884e-08,
      "loss": 0.1287,
      "step": 15960
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6427042350849215,
      "learning_rate": 6.446347999365942e-08,
      "loss": 0.1282,
      "step": 15961
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.41805031077304833,
      "learning_rate": 6.41079721753618e-08,
      "loss": 0.0948,
      "step": 15962
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5826113648585409,
      "learning_rate": 6.375344610100309e-08,
      "loss": 0.1975,
      "step": 15963
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4497581141312521,
      "learning_rate": 6.339990178454436e-08,
      "loss": 0.103,
      "step": 15964
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.34925052437843984,
      "learning_rate": 6.304733923990225e-08,
      "loss": 0.0522,
      "step": 15965
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4556299883477697,
      "learning_rate": 6.26957584809601e-08,
      "loss": 0.0608,
      "step": 15966
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5898758406159844,
      "learning_rate": 6.234515952155685e-08,
      "loss": 0.1061,
      "step": 15967
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7168249643251363,
      "learning_rate": 6.199554237549532e-08,
      "loss": 0.1748,
      "step": 15968
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5180150489147559,
      "learning_rate": 6.16469070565423e-08,
      "loss": 0.1289,
      "step": 15969
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5019999133063294,
      "learning_rate": 6.129925357842292e-08,
      "loss": 0.1549,
      "step": 15970
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5997722327811705,
      "learning_rate": 6.095258195482623e-08,
      "loss": 0.1294,
      "step": 15971
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3045879355363636,
      "learning_rate": 6.060689219939686e-08,
      "loss": 0.0345,
      "step": 15972
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6161179702955312,
      "learning_rate": 6.02621843257517e-08,
      "loss": 0.1231,
      "step": 15973
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5021328377143802,
      "learning_rate": 5.991845834745491e-08,
      "loss": 0.0644,
      "step": 15974
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5886060065946436,
      "learning_rate": 5.957571427804287e-08,
      "loss": 0.052,
      "step": 15975
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.584939004564765,
      "learning_rate": 5.923395213101035e-08,
      "loss": 0.1386,
      "step": 15976
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3896177514855861,
      "learning_rate": 5.8893171919810476e-08,
      "loss": 0.0775,
      "step": 15977
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.54256315185201,
      "learning_rate": 5.8553373657860286e-08,
      "loss": 0.178,
      "step": 15978
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.49686422353140136,
      "learning_rate": 5.821455735854076e-08,
      "loss": 0.1049,
      "step": 15979
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4794367445282085,
      "learning_rate": 5.787672303518843e-08,
      "loss": 0.0824,
      "step": 15980
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5301697418495441,
      "learning_rate": 5.7539870701103784e-08,
      "loss": 0.0777,
      "step": 15981
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.34654669087218976,
      "learning_rate": 5.7204000369551204e-08,
      "loss": 0.0478,
      "step": 15982
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.42888322028168036,
      "learning_rate": 5.6869112053753446e-08,
      "loss": 0.0701,
      "step": 15983
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3493225579096635,
      "learning_rate": 5.653520576689442e-08,
      "loss": 0.0652,
      "step": 15984
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.800010186205007,
      "learning_rate": 5.620228152211915e-08,
      "loss": 0.1397,
      "step": 15985
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.665396302413347,
      "learning_rate": 5.587033933253938e-08,
      "loss": 0.124,
      "step": 15986
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4412656675637663,
      "learning_rate": 5.553937921121965e-08,
      "loss": 0.0488,
      "step": 15987
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.47804737539642217,
      "learning_rate": 5.520940117119122e-08,
      "loss": 0.075,
      "step": 15988
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3726853556024811,
      "learning_rate": 5.488040522544646e-08,
      "loss": 0.0886,
      "step": 15989
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5411101709712095,
      "learning_rate": 5.4552391386938904e-08,
      "loss": 0.1032,
      "step": 15990
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6018502812409486,
      "learning_rate": 5.4225359668580446e-08,
      "loss": 0.1639,
      "step": 15991
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8514606229612965,
      "learning_rate": 5.389931008324689e-08,
      "loss": 0.2395,
      "step": 15992
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9393811395002927,
      "learning_rate": 5.3574242643775194e-08,
      "loss": 0.2683,
      "step": 15993
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3426152560967141,
      "learning_rate": 5.3250157362966235e-08,
      "loss": 0.05,
      "step": 15994
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2630615284496029,
      "learning_rate": 5.292705425357647e-08,
      "loss": 0.0424,
      "step": 15995
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8925083788769973,
      "learning_rate": 5.260493332832628e-08,
      "loss": 0.1745,
      "step": 15996
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0215002813058447,
      "learning_rate": 5.228379459989996e-08,
      "loss": 0.3016,
      "step": 15997
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.45654242091714775,
      "learning_rate": 5.196363808094018e-08,
      "loss": 0.0817,
      "step": 15998
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4279497076101462,
      "learning_rate": 5.164446378405075e-08,
      "loss": 0.0503,
      "step": 15999
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.734229915240307,
      "learning_rate": 5.132627172179938e-08,
      "loss": 0.199,
      "step": 16000
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3312541328160198,
      "learning_rate": 5.1009061906712176e-08,
      "loss": 0.0382,
      "step": 16001
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4296243628494794,
      "learning_rate": 5.069283435127914e-08,
      "loss": 0.0938,
      "step": 16002
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.606939725638042,
      "learning_rate": 5.0377589067951426e-08,
      "loss": 0.1103,
      "step": 16003
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6621367753887722,
      "learning_rate": 5.0063326069135776e-08,
      "loss": 0.1418,
      "step": 16004
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.42775822262672225,
      "learning_rate": 4.97500453672084e-08,
      "loss": 0.08,
      "step": 16005
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.37394630456287636,
      "learning_rate": 4.9437746974503876e-08,
      "loss": 0.0641,
      "step": 16006
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3939951345810713,
      "learning_rate": 4.912643090331792e-08,
      "loss": 0.0634,
      "step": 16007
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5503840653166517,
      "learning_rate": 4.881609716590463e-08,
      "loss": 0.1011,
      "step": 16008
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5780321758421624,
      "learning_rate": 4.8506745774482e-08,
      "loss": 0.1253,
      "step": 16009
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6178609845902255,
      "learning_rate": 4.8198376741231954e-08,
      "loss": 0.2584,
      "step": 16010
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.394130214653409,
      "learning_rate": 4.789099007829478e-08,
      "loss": 0.0545,
      "step": 16011
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.36727244244349516,
      "learning_rate": 4.758458579777192e-08,
      "loss": 0.0492,
      "step": 16012
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.46448080212871223,
      "learning_rate": 4.7279163911725934e-08,
      "loss": 0.0755,
      "step": 16013
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4176721072578472,
      "learning_rate": 4.6974724432183315e-08,
      "loss": 0.0489,
      "step": 16014
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.43137881703512343,
      "learning_rate": 4.6671267371128926e-08,
      "loss": 0.0699,
      "step": 16015
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.504086285976388,
      "learning_rate": 4.636879274050876e-08,
      "loss": 0.1135,
      "step": 16016
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1858260984418407,
      "learning_rate": 4.606730055223274e-08,
      "loss": 0.1306,
      "step": 16017
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.44668382049219996,
      "learning_rate": 4.57667908181747e-08,
      "loss": 0.12,
      "step": 16018
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.40838928942636316,
      "learning_rate": 4.5467263550158514e-08,
      "loss": 0.0901,
      "step": 16019
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8590456402118717,
      "learning_rate": 4.51687187599803e-08,
      "loss": 0.1167,
      "step": 16020
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7372362118828408,
      "learning_rate": 4.487115645939455e-08,
      "loss": 0.2279,
      "step": 16021
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5313010010977933,
      "learning_rate": 4.457457666011689e-08,
      "loss": 0.0739,
      "step": 16022
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.607148228807398,
      "learning_rate": 4.427897937382408e-08,
      "loss": 0.1132,
      "step": 16023
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6561360104902814,
      "learning_rate": 4.3984364612151276e-08,
      "loss": 0.1596,
      "step": 16024
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6373387607995715,
      "learning_rate": 4.36907323867003e-08,
      "loss": 0.1447,
      "step": 16025
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.45070810898433517,
      "learning_rate": 4.339808270903134e-08,
      "loss": 0.0755,
      "step": 16026
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6548004497048553,
      "learning_rate": 4.310641559066297e-08,
      "loss": 0.1296,
      "step": 16027
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5528127176529817,
      "learning_rate": 4.281573104308323e-08,
      "loss": 0.1184,
      "step": 16028
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.48502230080914943,
      "learning_rate": 4.2526029077732954e-08,
      "loss": 0.0996,
      "step": 16029
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4213414587972537,
      "learning_rate": 4.223730970601969e-08,
      "loss": 0.4961,
      "step": 16030
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.49231492086391143,
      "learning_rate": 4.194957293931212e-08,
      "loss": 0.0956,
      "step": 16031
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.42500720205864834,
      "learning_rate": 4.166281878893452e-08,
      "loss": 0.0868,
      "step": 16032
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3242393755616346,
      "learning_rate": 4.137704726618063e-08,
      "loss": 0.0388,
      "step": 16033
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4518241222515496,
      "learning_rate": 4.109225838229702e-08,
      "loss": 0.1029,
      "step": 16034
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5007187459547242,
      "learning_rate": 4.080845214849971e-08,
      "loss": 0.0638,
      "step": 16035
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.39853840699909915,
      "learning_rate": 4.052562857596032e-08,
      "loss": 0.0469,
      "step": 16036
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1809938318246878,
      "learning_rate": 4.0243787675817156e-08,
      "loss": 0.3658,
      "step": 16037
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.570834999971601,
      "learning_rate": 3.996292945916413e-08,
      "loss": 0.1175,
      "step": 16038
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.44150062000383766,
      "learning_rate": 3.968305393705629e-08,
      "loss": 0.0793,
      "step": 16039
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.39951943011248664,
      "learning_rate": 3.940416112051537e-08,
      "loss": 0.0461,
      "step": 16040
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.32141884913560886,
      "learning_rate": 3.912625102052425e-08,
      "loss": 0.0464,
      "step": 16041
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3518227030159834,
      "learning_rate": 3.884932364801863e-08,
      "loss": 0.0516,
      "step": 16042
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4467692678763523,
      "learning_rate": 3.857337901390368e-08,
      "loss": 0.0608,
      "step": 16043
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.35875136291751114,
      "learning_rate": 3.8298417129045695e-08,
      "loss": 0.0507,
      "step": 16044
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6501959915486476,
      "learning_rate": 3.802443800426658e-08,
      "loss": 0.1304,
      "step": 16045
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.46047997603743446,
      "learning_rate": 3.775144165035771e-08,
      "loss": 0.0725,
      "step": 16046
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3400541736943729,
      "learning_rate": 3.747942807806326e-08,
      "loss": 0.0381,
      "step": 16047
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0977185166081784,
      "learning_rate": 3.720839729809134e-08,
      "loss": 0.3975,
      "step": 16048
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5970466771537487,
      "learning_rate": 3.6938349321119506e-08,
      "loss": 0.1298,
      "step": 16049
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4227160805130841,
      "learning_rate": 3.66692841577726e-08,
      "loss": 0.0769,
      "step": 16050
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.35876813952064046,
      "learning_rate": 3.64012018186477e-08,
      "loss": 0.0622,
      "step": 16051
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5543844756395286,
      "learning_rate": 3.61341023142947e-08,
      "loss": 0.2077,
      "step": 16052
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.390137466940876,
      "learning_rate": 3.586798565523575e-08,
      "loss": 0.0818,
      "step": 16053
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6153585986400429,
      "learning_rate": 3.5602851851945785e-08,
      "loss": 0.188,
      "step": 16054
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.688901206947001,
      "learning_rate": 3.533870091486369e-08,
      "loss": 0.2485,
      "step": 16055
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9351161149111494,
      "learning_rate": 3.5075532854386694e-08,
      "loss": 0.2777,
      "step": 16056
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8257823327421052,
      "learning_rate": 3.4813347680878736e-08,
      "loss": 0.3356,
      "step": 16057
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3852355541816665,
      "learning_rate": 3.4552145404659344e-08,
      "loss": 0.0713,
      "step": 16058
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3235918012224595,
      "learning_rate": 3.42919260360175e-08,
      "loss": 0.0452,
      "step": 16059
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4342353449974484,
      "learning_rate": 3.4032689585192234e-08,
      "loss": 0.0455,
      "step": 16060
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4130926943628626,
      "learning_rate": 3.377443606239206e-08,
      "loss": 0.4255,
      "step": 16061
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7828292179657738,
      "learning_rate": 3.351716547778661e-08,
      "loss": 0.2102,
      "step": 16062
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2907609003922335,
      "learning_rate": 3.3260877841501114e-08,
      "loss": 0.0301,
      "step": 16063
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.343839689866053,
      "learning_rate": 3.3005573163630285e-08,
      "loss": 0.0546,
      "step": 16064
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5076641057532876,
      "learning_rate": 3.275125145422164e-08,
      "loss": 0.08,
      "step": 16065
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.587466779340513,
      "learning_rate": 3.249791272329217e-08,
      "loss": 0.1438,
      "step": 16066
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4511278368438604,
      "learning_rate": 3.224555698080889e-08,
      "loss": 0.0712,
      "step": 16067
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4098056705693678,
      "learning_rate": 3.1994184236716626e-08,
      "loss": 0.0809,
      "step": 16068
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.48838988097275243,
      "learning_rate": 3.17437945009047e-08,
      "loss": 0.1437,
      "step": 16069
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.36680525444569506,
      "learning_rate": 3.1494387783234656e-08,
      "loss": 0.0594,
      "step": 16070
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7500489712142698,
      "learning_rate": 3.124596409352365e-08,
      "loss": 0.2101,
      "step": 16071
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5439392786378865,
      "learning_rate": 3.099852344155274e-08,
      "loss": 0.1346,
      "step": 16072
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7175347075618562,
      "learning_rate": 3.075206583706691e-08,
      "loss": 0.1869,
      "step": 16073
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5319039607560504,
      "learning_rate": 3.050659128976674e-08,
      "loss": 0.0631,
      "step": 16074
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6483498176525989,
      "learning_rate": 3.026209980931671e-08,
      "loss": 0.1391,
      "step": 16075
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7323125374721459,
      "learning_rate": 3.0018591405342466e-08,
      "loss": 0.1872,
      "step": 16076
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.47321672479091825,
      "learning_rate": 2.9776066087430776e-08,
      "loss": 0.104,
      "step": 16077
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4447337243350544,
      "learning_rate": 2.9534523865132334e-08,
      "loss": 0.0801,
      "step": 16078
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3836999258120955,
      "learning_rate": 2.9293964747956204e-08,
      "loss": 0.0853,
      "step": 16079
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.33725699389727254,
      "learning_rate": 2.905438874536981e-08,
      "loss": 0.1014,
      "step": 16080
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4330869468106587,
      "learning_rate": 2.8815795866810047e-08,
      "loss": 0.1055,
      "step": 16081
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6271447572539095,
      "learning_rate": 2.8578186121669403e-08,
      "loss": 0.0826,
      "step": 16082
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.42739117415633165,
      "learning_rate": 2.834155951930151e-08,
      "loss": 0.0546,
      "step": 16083
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.28317690092619924,
      "learning_rate": 2.810591606902113e-08,
      "loss": 0.045,
      "step": 16084
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.42400978992072025,
      "learning_rate": 2.7871255780109738e-08,
      "loss": 0.0892,
      "step": 16085
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5314620406884567,
      "learning_rate": 2.7637578661804385e-08,
      "loss": 0.1074,
      "step": 16086
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.39027642921035877,
      "learning_rate": 2.7404884723300494e-08,
      "loss": 0.0708,
      "step": 16087
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3997366569916552,
      "learning_rate": 2.7173173973768505e-08,
      "loss": 0.0443,
      "step": 16088
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5703190455069397,
      "learning_rate": 2.6942446422323352e-08,
      "loss": 0.106,
      "step": 16089
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.40570012203146366,
      "learning_rate": 2.671270207805221e-08,
      "loss": 0.0458,
      "step": 16090
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4579840138812791,
      "learning_rate": 2.6483940949997844e-08,
      "loss": 0.0665,
      "step": 16091
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5406553289281523,
      "learning_rate": 2.625616304717249e-08,
      "loss": 0.1324,
      "step": 16092
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.44810563949010385,
      "learning_rate": 2.6029368378535645e-08,
      "loss": 0.086,
      "step": 16093
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3158024280255848,
      "learning_rate": 2.580355695302461e-08,
      "loss": 0.0618,
      "step": 16094
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5021317451088286,
      "learning_rate": 2.557872877952394e-08,
      "loss": 0.0869,
      "step": 16095
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5333250526552651,
      "learning_rate": 2.5354883866887668e-08,
      "loss": 0.1882,
      "step": 16096
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7672844864278932,
      "learning_rate": 2.5132022223928188e-08,
      "loss": 0.1387,
      "step": 16097
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6600141725959762,
      "learning_rate": 2.4910143859419033e-08,
      "loss": 0.1098,
      "step": 16098
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0881452278430948,
      "learning_rate": 2.468924878209766e-08,
      "loss": 0.252,
      "step": 16099
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5147673544297093,
      "learning_rate": 2.44693370006599e-08,
      "loss": 0.078,
      "step": 16100
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6336047731167238,
      "learning_rate": 2.4250408523762704e-08,
      "loss": 0.0858,
      "step": 16101
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.28429380102309304,
      "learning_rate": 2.403246336002696e-08,
      "loss": 0.0362,
      "step": 16102
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5819393626787237,
      "learning_rate": 2.381550151803469e-08,
      "loss": 0.0878,
      "step": 16103
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6743567427822748,
      "learning_rate": 2.359952300632351e-08,
      "loss": 0.2462,
      "step": 16104
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6441971950347374,
      "learning_rate": 2.338452783339773e-08,
      "loss": 0.1179,
      "step": 16105
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6071749965326877,
      "learning_rate": 2.3170516007725573e-08,
      "loss": 0.0885,
      "step": 16106
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5279254895882406,
      "learning_rate": 2.295748753773086e-08,
      "loss": 0.0967,
      "step": 16107
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4650527137804008,
      "learning_rate": 2.2745442431798548e-08,
      "loss": 0.0839,
      "step": 16108
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5231968213847884,
      "learning_rate": 2.2534380698280287e-08,
      "loss": 0.1015,
      "step": 16109
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.41886901611760746,
      "learning_rate": 2.2324302345483327e-08,
      "loss": 0.0632,
      "step": 16110
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5049031604734795,
      "learning_rate": 2.2115207381678826e-08,
      "loss": 0.0647,
      "step": 16111
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.43501467659790266,
      "learning_rate": 2.1907095815101863e-08,
      "loss": 0.0616,
      "step": 16112
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.42312946814765573,
      "learning_rate": 2.1699967653940334e-08,
      "loss": 0.0827,
      "step": 16113
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.586206967298559,
      "learning_rate": 2.1493822906354377e-08,
      "loss": 0.1143,
      "step": 16114
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4378081275914786,
      "learning_rate": 2.1288661580459722e-08,
      "loss": 0.0931,
      "step": 16115
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4130749031939478,
      "learning_rate": 2.1084483684327693e-08,
      "loss": 0.0547,
      "step": 16116
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5289215942773169,
      "learning_rate": 2.088128922600463e-08,
      "loss": 0.0968,
      "step": 16117
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.35459548483548353,
      "learning_rate": 2.0679078213486913e-08,
      "loss": 0.0541,
      "step": 16118
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6226868433371602,
      "learning_rate": 2.0477850654734844e-08,
      "loss": 0.1215,
      "step": 16119
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6686751336165937,
      "learning_rate": 2.0277606557672634e-08,
      "loss": 0.1569,
      "step": 16120
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6340864385564032,
      "learning_rate": 2.0078345930182873e-08,
      "loss": 0.147,
      "step": 16121
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6041008490457104,
      "learning_rate": 1.988006878011206e-08,
      "loss": 0.1145,
      "step": 16122
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.40968504246336424,
      "learning_rate": 1.9682775115265063e-08,
      "loss": 0.0437,
      "step": 16123
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2341629126280346,
      "learning_rate": 1.9486464943410664e-08,
      "loss": 0.0237,
      "step": 16124
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3921671537928308,
      "learning_rate": 1.9291138272276023e-08,
      "loss": 0.0729,
      "step": 16125
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9530101298563223,
      "learning_rate": 1.90967951095522e-08,
      "loss": 0.1737,
      "step": 16126
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4365491772940728,
      "learning_rate": 1.8903435462891416e-08,
      "loss": 0.3235,
      "step": 16127
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6183769915494208,
      "learning_rate": 1.8711059339904247e-08,
      "loss": 0.1631,
      "step": 16128
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4576415901429498,
      "learning_rate": 1.8519666748167962e-08,
      "loss": 0.3356,
      "step": 16129
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8177890279278394,
      "learning_rate": 1.8329257695215428e-08,
      "loss": 0.2327,
      "step": 16130
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4348901875387391,
      "learning_rate": 1.8139832188543428e-08,
      "loss": 0.067,
      "step": 16131
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8270479757728436,
      "learning_rate": 1.795139023560988e-08,
      "loss": 0.3098,
      "step": 16132
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6341994080160429,
      "learning_rate": 1.7763931843836623e-08,
      "loss": 0.1277,
      "step": 16133
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6734611533740662,
      "learning_rate": 1.7577457020598322e-08,
      "loss": 0.1535,
      "step": 16134
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2828080957921811,
      "learning_rate": 1.7391965773239093e-08,
      "loss": 0.0519,
      "step": 16135
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8336256508320451,
      "learning_rate": 1.7207458109066987e-08,
      "loss": 0.2223,
      "step": 16136
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.942260556284959,
      "learning_rate": 1.7023934035337308e-08,
      "loss": 0.3069,
      "step": 16137
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.649457289107267,
      "learning_rate": 1.6841393559283157e-08,
      "loss": 0.1132,
      "step": 16138
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3534663378350149,
      "learning_rate": 1.6659836688084907e-08,
      "loss": 0.0366,
      "step": 16139
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3499271724849072,
      "learning_rate": 1.647926342889794e-08,
      "loss": 0.0673,
      "step": 16140
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6589612992654332,
      "learning_rate": 1.6299673788824908e-08,
      "loss": 0.1517,
      "step": 16141
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5250452615024288,
      "learning_rate": 1.6121067774937937e-08,
      "loss": 0.1382,
      "step": 16142
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.42749501883859864,
      "learning_rate": 1.5943445394270284e-08,
      "loss": 0.094,
      "step": 16143
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4911012009379398,
      "learning_rate": 1.5766806653816357e-08,
      "loss": 0.0718,
      "step": 16144
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.29763102370226463,
      "learning_rate": 1.5591151560528926e-08,
      "loss": 0.0332,
      "step": 16145
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5442933959853475,
      "learning_rate": 1.5416480121321907e-08,
      "loss": 0.1371,
      "step": 16146
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.372423022174858,
      "learning_rate": 1.5242792343075908e-08,
      "loss": 0.0364,
      "step": 16147
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8247044405698905,
      "learning_rate": 1.5070088232624348e-08,
      "loss": 0.0629,
      "step": 16148
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5299866434967407,
      "learning_rate": 1.4898367796770118e-08,
      "loss": 0.0992,
      "step": 16149
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5364695339519773,
      "learning_rate": 1.472763104227448e-08,
      "loss": 0.1124,
      "step": 16150
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6093026239563898,
      "learning_rate": 1.4557877975857059e-08,
      "loss": 0.18,
      "step": 16151
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8280509002638885,
      "learning_rate": 1.438910860420417e-08,
      "loss": 0.1575,
      "step": 16152
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3339221073824688,
      "learning_rate": 1.4221322933957725e-08,
      "loss": 0.066,
      "step": 16153
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5902063060744153,
      "learning_rate": 1.4054520971726327e-08,
      "loss": 0.1593,
      "step": 16154
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.40586344829118054,
      "learning_rate": 1.388870272407139e-08,
      "loss": 0.0451,
      "step": 16155
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4330488281496723,
      "learning_rate": 1.372386819752658e-08,
      "loss": 0.1143,
      "step": 16156
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.38763602126561497,
      "learning_rate": 1.3560017398581148e-08,
      "loss": 0.0717,
      "step": 16157
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.41146473674993067,
      "learning_rate": 1.3397150333685494e-08,
      "loss": 0.0603,
      "step": 16158
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.447233498175711,
      "learning_rate": 1.3235267009251151e-08,
      "loss": 0.097,
      "step": 16159
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.572557647894951,
      "learning_rate": 1.3074367431650802e-08,
      "loss": 0.1056,
      "step": 16160
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7042786705718532,
      "learning_rate": 1.291445160721827e-08,
      "loss": 0.1635,
      "step": 16161
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4148539425716393,
      "learning_rate": 1.2755519542254069e-08,
      "loss": 0.0582,
      "step": 16162
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.42422992565779694,
      "learning_rate": 1.259757124301153e-08,
      "loss": 0.0604,
      "step": 16163
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4057317612384175,
      "learning_rate": 1.2440606715710678e-08,
      "loss": 0.0564,
      "step": 16164
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6544029926289175,
      "learning_rate": 1.2284625966532681e-08,
      "loss": 0.1233,
      "step": 16165
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5400841940041711,
      "learning_rate": 1.2129629001614295e-08,
      "loss": 0.0985,
      "step": 16166
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3693804614795122,
      "learning_rate": 1.1975615827061747e-08,
      "loss": 0.0636,
      "step": 16167
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4583059678141413,
      "learning_rate": 1.1822586448939631e-08,
      "loss": 0.0769,
      "step": 16168
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3907006866100224,
      "learning_rate": 1.1670540873268132e-08,
      "loss": 0.0563,
      "step": 16169
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5539830353993248,
      "learning_rate": 1.15194791060369e-08,
      "loss": 0.1225,
      "step": 16170
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4463793548962417,
      "learning_rate": 1.1369401153191184e-08,
      "loss": 0.0645,
      "step": 16171
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4686647931949205,
      "learning_rate": 1.1220307020642917e-08,
      "loss": 0.0655,
      "step": 16172
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.37631836289146314,
      "learning_rate": 1.1072196714256855e-08,
      "loss": 0.0521,
      "step": 16173
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7502693470343362,
      "learning_rate": 1.0925070239869994e-08,
      "loss": 0.1135,
      "step": 16174
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7995112935252056,
      "learning_rate": 1.0778927603272148e-08,
      "loss": 0.2158,
      "step": 16175
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4510956222752506,
      "learning_rate": 1.0633768810214273e-08,
      "loss": 0.0834,
      "step": 16176
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.47200052411768,
      "learning_rate": 1.0489593866416791e-08,
      "loss": 0.081,
      "step": 16177
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3838350297912305,
      "learning_rate": 1.0346402777552943e-08,
      "loss": 0.0479,
      "step": 16178
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.640671930556749,
      "learning_rate": 1.0204195549259887e-08,
      "loss": 0.1239,
      "step": 16179
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5385097391760805,
      "learning_rate": 1.0062972187138698e-08,
      "loss": 0.1439,
      "step": 16180
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.43517033313583875,
      "learning_rate": 9.922732696748816e-09,
      "loss": 0.0617,
      "step": 16181
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6190339333441074,
      "learning_rate": 9.78347708360805e-09,
      "loss": 0.1025,
      "step": 16182
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4290342869541161,
      "learning_rate": 9.645205353200904e-09,
      "loss": 0.0622,
      "step": 16183
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4897464513143023,
      "learning_rate": 9.507917510975795e-09,
      "loss": 0.137,
      "step": 16184
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4907100103664823,
      "learning_rate": 9.371613562331183e-09,
      "loss": 0.0462,
      "step": 16185
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4297024054698963,
      "learning_rate": 9.236293512637773e-09,
      "loss": 0.0907,
      "step": 16186
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.38908266644342787,
      "learning_rate": 9.101957367219083e-09,
      "loss": 0.0632,
      "step": 16187
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5232039932961251,
      "learning_rate": 8.968605131368102e-09,
      "loss": 0.1448,
      "step": 16188
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4000566523186992,
      "learning_rate": 8.836236810336184e-09,
      "loss": 0.0749,
      "step": 16189
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.40790744395402284,
      "learning_rate": 8.704852409330278e-09,
      "loss": 0.055,
      "step": 16190
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.44142773572369665,
      "learning_rate": 8.574451933524019e-09,
      "loss": 0.0482,
      "step": 16191
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.36771009272752897,
      "learning_rate": 8.445035388052192e-09,
      "loss": 0.1235,
      "step": 16192
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4962552737771642,
      "learning_rate": 8.316602778010719e-09,
      "loss": 0.0768,
      "step": 16193
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5033518252271789,
      "learning_rate": 8.189154108456665e-09,
      "loss": 0.1059,
      "step": 16194
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.36577207651142524,
      "learning_rate": 8.062689384405464e-09,
      "loss": 0.086,
      "step": 16195
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7640521959363403,
      "learning_rate": 7.937208610836467e-09,
      "loss": 0.2512,
      "step": 16196
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4038354175567589,
      "learning_rate": 7.812711792692939e-09,
      "loss": 0.0601,
      "step": 16197
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.39211964372358366,
      "learning_rate": 7.689198934870967e-09,
      "loss": 0.0445,
      "step": 16198
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6543380343110617,
      "learning_rate": 7.566670042238878e-09,
      "loss": 0.1892,
      "step": 16199
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5626512006263199,
      "learning_rate": 7.445125119617813e-09,
      "loss": 0.0933,
      "step": 16200
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6393144811221501,
      "learning_rate": 7.324564171795611e-09,
      "loss": 0.0996,
      "step": 16201
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.33527743239438923,
      "learning_rate": 7.2049872035157e-09,
      "loss": 0.0511,
      "step": 16202
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6692669696343566,
      "learning_rate": 7.086394219485426e-09,
      "loss": 0.1031,
      "step": 16203
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5227630351436366,
      "learning_rate": 6.968785224378826e-09,
      "loss": 0.0797,
      "step": 16204
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5314200073072517,
      "learning_rate": 6.852160222819981e-09,
      "loss": 0.1054,
      "step": 16205
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.38538903568152477,
      "learning_rate": 6.736519219405213e-09,
      "loss": 0.0544,
      "step": 16206
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8547335926543844,
      "learning_rate": 6.621862218683661e-09,
      "loss": 0.1287,
      "step": 16207
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3718736158429465,
      "learning_rate": 6.508189225173933e-09,
      "loss": 0.0466,
      "step": 16208
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.43737562839848015,
      "learning_rate": 6.395500243347452e-09,
      "loss": 0.1089,
      "step": 16209
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.25692035708880334,
      "learning_rate": 6.28379527764511e-09,
      "loss": 0.0345,
      "step": 16210
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.46131442165672937,
      "learning_rate": 6.17307433245784e-09,
      "loss": 0.0616,
      "step": 16211
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.46962671205062007,
      "learning_rate": 6.063337412151593e-09,
      "loss": 0.0959,
      "step": 16212
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.597011803901714,
      "learning_rate": 5.954584521045137e-09,
      "loss": 0.1079,
      "step": 16213
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3199423369499446,
      "learning_rate": 5.846815663418381e-09,
      "loss": 0.0327,
      "step": 16214
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7244987010480157,
      "learning_rate": 5.740030843515154e-09,
      "loss": 0.1532,
      "step": 16215
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.728784441321603,
      "learning_rate": 5.634230065537649e-09,
      "loss": 0.1472,
      "step": 16216
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5235114239048658,
      "learning_rate": 5.529413333654754e-09,
      "loss": 0.1019,
      "step": 16217
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.41522367484458506,
      "learning_rate": 5.425580651993722e-09,
      "loss": 0.1056,
      "step": 16218
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7571524936542364,
      "learning_rate": 5.3227320246374005e-09,
      "loss": 0.2182,
      "step": 16219
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4931231491083345,
      "learning_rate": 5.220867455640876e-09,
      "loss": 0.0978,
      "step": 16220
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.952136186683439,
      "learning_rate": 5.1199869490120565e-09,
      "loss": 0.3556,
      "step": 16221
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.40402339835660256,
      "learning_rate": 5.020090508719988e-09,
      "loss": 0.0855,
      "step": 16222
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5164138205490344,
      "learning_rate": 4.9211781387031865e-09,
      "loss": 0.0575,
      "step": 16223
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2869801838067473,
      "learning_rate": 4.823249842850208e-09,
      "loss": 0.03,
      "step": 16224
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4267269020950202,
      "learning_rate": 4.726305625019079e-09,
      "loss": 0.0767,
      "step": 16225
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.42021863973563267,
      "learning_rate": 4.630345489028965e-09,
      "loss": 0.0477,
      "step": 16226
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.178662665766016,
      "learning_rate": 4.535369438654624e-09,
      "loss": 0.1437,
      "step": 16227
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9285657826934881,
      "learning_rate": 4.441377477637509e-09,
      "loss": 0.2435,
      "step": 16228
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.39318643589388885,
      "learning_rate": 4.348369609674663e-09,
      "loss": 0.0622,
      "step": 16229
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5509112220230121,
      "learning_rate": 4.256345838432596e-09,
      "loss": 0.1098,
      "step": 16230
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.37770349990808966,
      "learning_rate": 4.165306167530636e-09,
      "loss": 0.0508,
      "step": 16231
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.25024170466096457,
      "learning_rate": 4.075250600554803e-09,
      "loss": 0.0273,
      "step": 16232
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1439018488084807,
      "learning_rate": 3.98617914105226e-09,
      "loss": 0.247,
      "step": 16233
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.39758573704289896,
      "learning_rate": 3.89809179252576e-09,
      "loss": 0.0401,
      "step": 16234
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.42985646185894494,
      "learning_rate": 3.81098855844475e-09,
      "loss": 0.058,
      "step": 16235
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.37595501454496966,
      "learning_rate": 3.72486944223982e-09,
      "loss": 0.0489,
      "step": 16236
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.420208198993576,
      "learning_rate": 3.6397344473027005e-09,
      "loss": 0.0832,
      "step": 16237
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5820908509583074,
      "learning_rate": 3.5555835769807143e-09,
      "loss": 0.1356,
      "step": 16238
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6924363484842325,
      "learning_rate": 3.4724168345906527e-09,
      "loss": 0.1453,
      "step": 16239
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.31988992125162974,
      "learning_rate": 3.3902342234076735e-09,
      "loss": 0.0389,
      "step": 16240
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.383857408349966,
      "learning_rate": 3.309035746665301e-09,
      "loss": 0.0568,
      "step": 16241
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5434271464688369,
      "learning_rate": 3.2288214075582024e-09,
      "loss": 0.1318,
      "step": 16242
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5638859914340635,
      "learning_rate": 3.1495912092477374e-09,
      "loss": 0.1007,
      "step": 16243
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6438199026547466,
      "learning_rate": 3.071345154853633e-09,
      "loss": 0.1552,
      "step": 16244
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5815684163322917,
      "learning_rate": 2.9940832474567582e-09,
      "loss": 0.1423,
      "step": 16245
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5213467856002733,
      "learning_rate": 2.9178054900935726e-09,
      "loss": 0.0758,
      "step": 16246
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.37658053039155204,
      "learning_rate": 2.8425118857755562e-09,
      "loss": 0.0624,
      "step": 16247
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.587853101368485,
      "learning_rate": 2.768202437458678e-09,
      "loss": 0.1097,
      "step": 16248
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.43503750168163274,
      "learning_rate": 2.6948771480739265e-09,
      "loss": 0.124,
      "step": 16249
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5568296710409719,
      "learning_rate": 2.622536020507882e-09,
      "loss": 0.0941,
      "step": 16250
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1303932868593083,
      "learning_rate": 2.5511790576082662e-09,
      "loss": 0.3514,
      "step": 16251
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6955385087359739,
      "learning_rate": 2.480806262181168e-09,
      "loss": 0.1712,
      "step": 16252
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3950026159869179,
      "learning_rate": 2.4114176370021446e-09,
      "loss": 0.0451,
      "step": 16253
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4431511515794177,
      "learning_rate": 2.3430131847995694e-09,
      "loss": 0.0669,
      "step": 16254
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3779663909959796,
      "learning_rate": 2.2755929082657335e-09,
      "loss": 0.4592,
      "step": 16255
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.922684897157736,
      "learning_rate": 2.2091568100596206e-09,
      "loss": 0.3423,
      "step": 16256
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7746595054308194,
      "learning_rate": 2.1437048927930304e-09,
      "loss": 0.1197,
      "step": 16257
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4882991140309518,
      "learning_rate": 2.0792371590472316e-09,
      "loss": 0.0964,
      "step": 16258
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4914491165478556,
      "learning_rate": 2.0157536113535325e-09,
      "loss": 0.2929,
      "step": 16259
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7435160471992632,
      "learning_rate": 1.9532542522154864e-09,
      "loss": 0.0907,
      "step": 16260
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7841712850666342,
      "learning_rate": 1.891739084095012e-09,
      "loss": 0.1451,
      "step": 16261
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2896274711851831,
      "learning_rate": 1.8312081094123967e-09,
      "loss": 0.0352,
      "step": 16262
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.49726715844532454,
      "learning_rate": 1.7716613305490682e-09,
      "loss": 0.0811,
      "step": 16263
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4448246824288858,
      "learning_rate": 1.7130987498503726e-09,
      "loss": 0.0988,
      "step": 16264
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.29392400010455927,
      "learning_rate": 1.6555203696255739e-09,
      "loss": 0.04,
      "step": 16265
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5765304185021084,
      "learning_rate": 1.5989261921367516e-09,
      "loss": 0.2076,
      "step": 16266
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.27191806833539905,
      "learning_rate": 1.5433162196126782e-09,
      "loss": 0.0249,
      "step": 16267
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.43267782936245985,
      "learning_rate": 1.4886904542460445e-09,
      "loss": 0.0532,
      "step": 16268
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3614665282243373,
      "learning_rate": 1.4350488981851318e-09,
      "loss": 0.0484,
      "step": 16269
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4224196149920677,
      "learning_rate": 1.3823915535421395e-09,
      "loss": 0.0705,
      "step": 16270
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4657558463477983,
      "learning_rate": 1.330718422390409e-09,
      "loss": 0.0792,
      "step": 16271
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.42282580342512244,
      "learning_rate": 1.2800295067644242e-09,
      "loss": 0.0658,
      "step": 16272
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8319989804832266,
      "learning_rate": 1.2303248086570351e-09,
      "loss": 0.2292,
      "step": 16273
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.38679321000005207,
      "learning_rate": 1.1816043300305612e-09,
      "loss": 0.0748,
      "step": 16274
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3445823181560556,
      "learning_rate": 1.133868072800137e-09,
      "loss": 0.0502,
      "step": 16275
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8408953203659095,
      "learning_rate": 1.0871160388448154e-09,
      "loss": 0.2176,
      "step": 16276
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.48150951123701063,
      "learning_rate": 1.0413482300047905e-09,
      "loss": 0.091,
      "step": 16277
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6563869731788812,
      "learning_rate": 9.96564648084175e-10,
      "loss": 0.1076,
      "step": 16278
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3131271245472773,
      "learning_rate": 9.52765294845448e-10,
      "loss": 0.0482,
      "step": 16279
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7420987474620109,
      "learning_rate": 9.099501720122306e-10,
      "loss": 0.1316,
      "step": 16280
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8891445212978497,
      "learning_rate": 8.681192812692862e-10,
      "loss": 0.2145,
      "step": 16281
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3418173078375629,
      "learning_rate": 8.272726242652962e-10,
      "loss": 0.0567,
      "step": 16282
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5157513074454589,
      "learning_rate": 7.874102026073083e-10,
      "loss": 0.0977,
      "step": 16283
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4618391671786376,
      "learning_rate": 7.48532017866288e-10,
      "loss": 0.0986,
      "step": 16284
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.46710068399880394,
      "learning_rate": 7.106380715715677e-10,
      "loss": 0.1513,
      "step": 16285
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.35849284084791616,
      "learning_rate": 6.737283652163972e-10,
      "loss": 0.054,
      "step": 16286
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7100364470879208,
      "learning_rate": 6.378029002523933e-10,
      "loss": 0.1513,
      "step": 16287
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.26049535929518364,
      "learning_rate": 6.028616780950902e-10,
      "loss": 0.0324,
      "step": 16288
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4773475225084832,
      "learning_rate": 5.689047001183889e-10,
      "loss": 0.1189,
      "step": 16289
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.49355756822673774,
      "learning_rate": 5.359319676601082e-10,
      "loss": 0.088,
      "step": 16290
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5920968165824135,
      "learning_rate": 5.03943482019209e-10,
      "loss": 0.1073,
      "step": 16291
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6960498472716484,
      "learning_rate": 4.729392444530189e-10,
      "loss": 0.1376,
      "step": 16292
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7334173273676191,
      "learning_rate": 4.4291925618555883e-10,
      "loss": 0.1197,
      "step": 16293
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8769991932064926,
      "learning_rate": 4.138835183964407e-10,
      "loss": 0.296,
      "step": 16294
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.36676087599905477,
      "learning_rate": 3.858320322319697e-10,
      "loss": 0.0528,
      "step": 16295
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.596582918511335,
      "learning_rate": 3.587647987912668e-10,
      "loss": 0.1031,
      "step": 16296
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.42314151417669854,
      "learning_rate": 3.3268181914014593e-10,
      "loss": 0.0555,
      "step": 16297
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.36653152861877203,
      "learning_rate": 3.0758309431111465e-10,
      "loss": 0.0648,
      "step": 16298
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.29616262758182244,
      "learning_rate": 2.834686252894958e-10,
      "loss": 0.0569,
      "step": 16299
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.677337209942941,
      "learning_rate": 2.6033841302453013e-10,
      "loss": 0.1504,
      "step": 16300
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5569071127327132,
      "learning_rate": 2.3819245842382485e-10,
      "loss": 0.1327,
      "step": 16301
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4067816564491886,
      "learning_rate": 2.170307623644563e-10,
      "loss": 0.0666,
      "step": 16302
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3740430588968175,
      "learning_rate": 1.968533256763161e-10,
      "loss": 0.0795,
      "step": 16303
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.39822149362630965,
      "learning_rate": 1.7766014915598927e-10,
      "loss": 0.0483,
      "step": 16304
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.739031686839175,
      "learning_rate": 1.594512335584275e-10,
      "loss": 0.1472,
      "step": 16305
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2469460349582098,
      "learning_rate": 1.4222657959694906e-10,
      "loss": 0.2968,
      "step": 16306
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5091203537479027,
      "learning_rate": 1.2598618795434115e-10,
      "loss": 0.1269,
      "step": 16307
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5367106543586053,
      "learning_rate": 1.1073005926898195e-10,
      "loss": 0.0771,
      "step": 16308
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6953036026082383,
      "learning_rate": 9.645819414039192e-11,
      "loss": 0.2019,
      "step": 16309
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.40139991626919075,
      "learning_rate": 8.317059313200926e-11,
      "loss": 0.0607,
      "step": 16310
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.38048183522438017,
      "learning_rate": 7.086725676563877e-11,
      "loss": 0.072,
      "step": 16311
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3744875036082268,
      "learning_rate": 5.954818552700303e-11,
      "loss": 0.0525,
      "step": 16312
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4737527568408738,
      "learning_rate": 4.9213379857415695e-11,
      "loss": 0.037,
      "step": 16313
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6775530513912578,
      "learning_rate": 3.986284017043484e-11,
      "loss": 0.1729,
      "step": 16314
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.42507823828078156,
      "learning_rate": 3.149656682965851e-11,
      "loss": 0.0698,
      "step": 16315
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.34488988626470435,
      "learning_rate": 2.4114560165378052e-11,
      "loss": 0.1061,
      "step": 16316
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7694987466062279,
      "learning_rate": 1.7716820469027008e-11,
      "loss": 0.1668,
      "step": 16317
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.48040728275195366,
      "learning_rate": 1.2303347993181113e-11,
      "loss": 0.0428,
      "step": 16318
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6336816816922578,
      "learning_rate": 7.874142948782747e-12,
      "loss": 0.1099,
      "step": 16319
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3914366584094326,
      "learning_rate": 4.429205510692036e-12,
      "loss": 0.0825,
      "step": 16320
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.33366128936590417,
      "learning_rate": 1.9685358149112985e-12,
      "loss": 0.0462,
      "step": 16321
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3933917520313606,
      "learning_rate": 4.921339585850504e-13,
      "loss": 0.0907,
      "step": 16322
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.49707289206058153,
      "learning_rate": 0.0,
      "loss": 0.0931,
      "step": 16323
    },
    {
      "epoch": 1.0,
      "step": 16323,
      "total_flos": 0.0,
      "train_loss": 0.06980257563588302,
      "train_runtime": 50663.685,
      "train_samples_per_second": 5.155,
      "train_steps_per_second": 0.322
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 16323,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}