{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9999194392975107,
  "eval_steps": 500,
  "global_step": 6206,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 122.5660216815097,
      "learning_rate": 5.3475935828877005e-08,
      "loss": 1.4579,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 96.20238517935834,
      "learning_rate": 1.0695187165775401e-07,
      "loss": 1.524,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 849.0350587509935,
      "learning_rate": 1.6042780748663104e-07,
      "loss": 1.4637,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 182.20744858900673,
      "learning_rate": 2.1390374331550802e-07,
      "loss": 1.5894,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 231.05838823335316,
      "learning_rate": 2.6737967914438503e-07,
      "loss": 1.5317,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 202.13167007470753,
      "learning_rate": 3.208556149732621e-07,
      "loss": 1.503,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 118.1413889257339,
      "learning_rate": 3.7433155080213904e-07,
      "loss": 1.4591,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.38572531457369,
      "learning_rate": 4.2780748663101604e-07,
      "loss": 0.9505,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 82.05875043350517,
      "learning_rate": 4.812834224598931e-07,
      "loss": 1.4157,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 70.05842179949386,
      "learning_rate": 5.347593582887701e-07,
      "loss": 1.2957,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 56.513707608857935,
      "learning_rate": 5.882352941176471e-07,
      "loss": 1.2736,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.69388740050222,
      "learning_rate": 6.417112299465242e-07,
      "loss": 0.8761,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 49.36372707295016,
      "learning_rate": 6.951871657754011e-07,
      "loss": 1.2075,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 54.19809693130065,
      "learning_rate": 7.486631016042781e-07,
      "loss": 1.1013,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 27.337284203855997,
      "learning_rate": 8.021390374331551e-07,
      "loss": 1.0114,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 50.81692104494523,
      "learning_rate": 8.556149732620321e-07,
      "loss": 1.0453,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 29.485661099962932,
      "learning_rate": 9.090909090909091e-07,
      "loss": 0.8666,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 22.564586972051583,
      "learning_rate": 9.625668449197862e-07,
      "loss": 0.8693,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 24.12789244295064,
      "learning_rate": 1.0160427807486633e-06,
      "loss": 0.8764,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 57.57569204864652,
      "learning_rate": 1.0695187165775401e-06,
      "loss": 0.891,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 22.184258097094585,
      "learning_rate": 1.1229946524064172e-06,
      "loss": 0.8579,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 28.40191743303267,
      "learning_rate": 1.1764705882352942e-06,
      "loss": 0.9162,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.798274748644812,
      "learning_rate": 1.2299465240641713e-06,
      "loss": 0.7764,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 34.09266117254502,
      "learning_rate": 1.2834224598930483e-06,
      "loss": 0.9927,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 22.695569160117596,
      "learning_rate": 1.3368983957219254e-06,
      "loss": 0.911,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.139679518356331,
      "learning_rate": 1.3903743315508022e-06,
      "loss": 0.7543,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 22.45203333256374,
      "learning_rate": 1.4438502673796793e-06,
      "loss": 0.7663,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 29.435173416448855,
      "learning_rate": 1.4973262032085562e-06,
      "loss": 0.9151,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 48.716233763851676,
      "learning_rate": 1.5508021390374334e-06,
      "loss": 0.898,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 34.64254333954248,
      "learning_rate": 1.6042780748663103e-06,
      "loss": 0.854,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 17.356025431171073,
      "learning_rate": 1.6577540106951873e-06,
      "loss": 0.7167,
      "step": 31
    },
    {
      "epoch": 0.01,
      "grad_norm": 23.525924335048796,
      "learning_rate": 1.7112299465240642e-06,
      "loss": 0.8054,
      "step": 32
    },
    {
      "epoch": 0.01,
      "grad_norm": 20.999975435460883,
      "learning_rate": 1.7647058823529414e-06,
      "loss": 0.8317,
      "step": 33
    },
    {
      "epoch": 0.01,
      "grad_norm": 116.68779213216546,
      "learning_rate": 1.8181818181818183e-06,
      "loss": 0.7837,
      "step": 34
    },
    {
      "epoch": 0.01,
      "grad_norm": 26.071062158998576,
      "learning_rate": 1.8716577540106954e-06,
      "loss": 0.6666,
      "step": 35
    },
    {
      "epoch": 0.01,
      "grad_norm": 38.38066714403462,
      "learning_rate": 1.9251336898395724e-06,
      "loss": 0.7705,
      "step": 36
    },
    {
      "epoch": 0.01,
      "grad_norm": 22.450538903347848,
      "learning_rate": 1.9786096256684497e-06,
      "loss": 0.7418,
      "step": 37
    },
    {
      "epoch": 0.01,
      "grad_norm": 30.30609757613985,
      "learning_rate": 2.0320855614973265e-06,
      "loss": 0.7849,
      "step": 38
    },
    {
      "epoch": 0.01,
      "grad_norm": 33.965207939999864,
      "learning_rate": 2.0855614973262034e-06,
      "loss": 0.801,
      "step": 39
    },
    {
      "epoch": 0.01,
      "grad_norm": 27.47207670640874,
      "learning_rate": 2.1390374331550802e-06,
      "loss": 0.7227,
      "step": 40
    },
    {
      "epoch": 0.01,
      "grad_norm": 79.25148574388751,
      "learning_rate": 2.1925133689839575e-06,
      "loss": 0.7811,
      "step": 41
    },
    {
      "epoch": 0.01,
      "grad_norm": 29.200339339053983,
      "learning_rate": 2.2459893048128343e-06,
      "loss": 0.7301,
      "step": 42
    },
    {
      "epoch": 0.01,
      "grad_norm": 31.11611622077372,
      "learning_rate": 2.2994652406417116e-06,
      "loss": 0.6627,
      "step": 43
    },
    {
      "epoch": 0.01,
      "grad_norm": 31.134151427915988,
      "learning_rate": 2.3529411764705885e-06,
      "loss": 0.7633,
      "step": 44
    },
    {
      "epoch": 0.01,
      "grad_norm": 53.41608383517845,
      "learning_rate": 2.4064171122994653e-06,
      "loss": 0.8341,
      "step": 45
    },
    {
      "epoch": 0.01,
      "grad_norm": 44.86140085240313,
      "learning_rate": 2.4598930481283426e-06,
      "loss": 0.7442,
      "step": 46
    },
    {
      "epoch": 0.01,
      "grad_norm": 22.250899081589598,
      "learning_rate": 2.5133689839572194e-06,
      "loss": 0.7348,
      "step": 47
    },
    {
      "epoch": 0.01,
      "grad_norm": 22.921556463732713,
      "learning_rate": 2.5668449197860967e-06,
      "loss": 0.6637,
      "step": 48
    },
    {
      "epoch": 0.01,
      "grad_norm": 24.09751015733259,
      "learning_rate": 2.6203208556149735e-06,
      "loss": 0.695,
      "step": 49
    },
    {
      "epoch": 0.01,
      "grad_norm": 135.24250227202447,
      "learning_rate": 2.673796791443851e-06,
      "loss": 0.626,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 24.999289116754852,
      "learning_rate": 2.7272727272727272e-06,
      "loss": 0.759,
      "step": 51
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.29332324334594,
      "learning_rate": 2.7807486631016045e-06,
      "loss": 0.6928,
      "step": 52
    },
    {
      "epoch": 0.01,
      "grad_norm": 19.952618467661395,
      "learning_rate": 2.8342245989304818e-06,
      "loss": 0.762,
      "step": 53
    },
    {
      "epoch": 0.01,
      "grad_norm": 47.66698831670102,
      "learning_rate": 2.8877005347593586e-06,
      "loss": 0.6553,
      "step": 54
    },
    {
      "epoch": 0.01,
      "grad_norm": 19.87890828382766,
      "learning_rate": 2.9411764705882355e-06,
      "loss": 0.677,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 22.021060921393875,
      "learning_rate": 2.9946524064171123e-06,
      "loss": 0.6556,
      "step": 56
    },
    {
      "epoch": 0.01,
      "grad_norm": 51.313613097860795,
      "learning_rate": 3.0481283422459896e-06,
      "loss": 0.6911,
      "step": 57
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.232961668505869,
      "learning_rate": 3.101604278074867e-06,
      "loss": 0.6446,
      "step": 58
    },
    {
      "epoch": 0.01,
      "grad_norm": 17.793694324941825,
      "learning_rate": 3.1550802139037433e-06,
      "loss": 0.7448,
      "step": 59
    },
    {
      "epoch": 0.01,
      "grad_norm": 17.99288737861374,
      "learning_rate": 3.2085561497326205e-06,
      "loss": 0.5912,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.122313155923214,
      "learning_rate": 3.262032085561498e-06,
      "loss": 0.7101,
      "step": 61
    },
    {
      "epoch": 0.01,
      "grad_norm": 17.17160324873029,
      "learning_rate": 3.3155080213903747e-06,
      "loss": 0.7451,
      "step": 62
    },
    {
      "epoch": 0.01,
      "grad_norm": 29.498395656556355,
      "learning_rate": 3.368983957219252e-06,
      "loss": 0.7817,
      "step": 63
    },
    {
      "epoch": 0.01,
      "grad_norm": 44.811745296161426,
      "learning_rate": 3.4224598930481284e-06,
      "loss": 0.6402,
      "step": 64
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.25018370181595,
      "learning_rate": 3.4759358288770056e-06,
      "loss": 0.5297,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 85.16994505724112,
      "learning_rate": 3.529411764705883e-06,
      "loss": 0.7363,
      "step": 66
    },
    {
      "epoch": 0.01,
      "grad_norm": 20.3349676620617,
      "learning_rate": 3.5828877005347597e-06,
      "loss": 0.6979,
      "step": 67
    },
    {
      "epoch": 0.01,
      "grad_norm": 20.41634192789583,
      "learning_rate": 3.6363636363636366e-06,
      "loss": 0.639,
      "step": 68
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.61051722348918,
      "learning_rate": 3.6898395721925134e-06,
      "loss": 0.6495,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 23.737252354348907,
      "learning_rate": 3.7433155080213907e-06,
      "loss": 0.6276,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 30.61292490845703,
      "learning_rate": 3.796791443850268e-06,
      "loss": 0.6799,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 18.93007032813426,
      "learning_rate": 3.850267379679145e-06,
      "loss": 0.6498,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 28.244661419400103,
      "learning_rate": 3.903743315508022e-06,
      "loss": 0.6814,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.863467310016713,
      "learning_rate": 3.957219251336899e-06,
      "loss": 0.6161,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 42.089666891351996,
      "learning_rate": 4.010695187165775e-06,
      "loss": 0.6478,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 23.754336196286815,
      "learning_rate": 4.064171122994653e-06,
      "loss": 0.6172,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 18.767807519935523,
      "learning_rate": 4.11764705882353e-06,
      "loss": 0.6644,
      "step": 77
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.954630485400536,
      "learning_rate": 4.171122994652407e-06,
      "loss": 0.6248,
      "step": 78
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.921577083561282,
      "learning_rate": 4.224598930481284e-06,
      "loss": 0.5517,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 26.780388977711148,
      "learning_rate": 4.2780748663101604e-06,
      "loss": 0.7048,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 22.546568332881034,
      "learning_rate": 4.331550802139038e-06,
      "loss": 0.6402,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 24.135059031162864,
      "learning_rate": 4.385026737967915e-06,
      "loss": 0.5676,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 18.366818375625567,
      "learning_rate": 4.438502673796792e-06,
      "loss": 0.5909,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 34.29237119670637,
      "learning_rate": 4.491978609625669e-06,
      "loss": 0.6756,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 38.921358940231755,
      "learning_rate": 4.5454545454545455e-06,
      "loss": 0.6855,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 21.573474684677027,
      "learning_rate": 4.598930481283423e-06,
      "loss": 0.5491,
      "step": 86
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.27110683432754,
      "learning_rate": 4.6524064171123e-06,
      "loss": 0.6556,
      "step": 87
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.517023376309627,
      "learning_rate": 4.705882352941177e-06,
      "loss": 0.7152,
      "step": 88
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.485806438586652,
      "learning_rate": 4.759358288770054e-06,
      "loss": 0.5804,
      "step": 89
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.101339416206482,
      "learning_rate": 4.812834224598931e-06,
      "loss": 0.6072,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 18.75766830436097,
      "learning_rate": 4.866310160427808e-06,
      "loss": 0.5972,
      "step": 91
    },
    {
      "epoch": 0.01,
      "grad_norm": 42.48051171926769,
      "learning_rate": 4.919786096256685e-06,
      "loss": 0.6329,
      "step": 92
    },
    {
      "epoch": 0.01,
      "grad_norm": 58.99865852419644,
      "learning_rate": 4.973262032085562e-06,
      "loss": 0.659,
      "step": 93
    },
    {
      "epoch": 0.02,
      "grad_norm": 20.583671652291816,
      "learning_rate": 5.026737967914439e-06,
      "loss": 0.5816,
      "step": 94
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.163742040869481,
      "learning_rate": 5.0802139037433165e-06,
      "loss": 0.6186,
      "step": 95
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.42481259047122,
      "learning_rate": 5.133689839572193e-06,
      "loss": 0.6531,
      "step": 96
    },
    {
      "epoch": 0.02,
      "grad_norm": 68.4171386846051,
      "learning_rate": 5.187165775401069e-06,
      "loss": 0.6845,
      "step": 97
    },
    {
      "epoch": 0.02,
      "grad_norm": 32.42262668051302,
      "learning_rate": 5.240641711229947e-06,
      "loss": 0.6458,
      "step": 98
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6575772208807753,
      "learning_rate": 5.294117647058824e-06,
      "loss": 0.5741,
      "step": 99
    },
    {
      "epoch": 0.02,
      "grad_norm": 26.97516445099372,
      "learning_rate": 5.347593582887702e-06,
      "loss": 0.6071,
      "step": 100
    },
    {
      "epoch": 0.02,
      "grad_norm": 19.268147547208816,
      "learning_rate": 5.4010695187165785e-06,
      "loss": 0.6456,
      "step": 101
    },
    {
      "epoch": 0.02,
      "grad_norm": 22.009344652004238,
      "learning_rate": 5.4545454545454545e-06,
      "loss": 0.6481,
      "step": 102
    },
    {
      "epoch": 0.02,
      "grad_norm": 37.86598942391542,
      "learning_rate": 5.508021390374332e-06,
      "loss": 0.6428,
      "step": 103
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.220507916056364,
      "learning_rate": 5.561497326203209e-06,
      "loss": 0.5524,
      "step": 104
    },
    {
      "epoch": 0.02,
      "grad_norm": 31.882822163137917,
      "learning_rate": 5.614973262032086e-06,
      "loss": 0.6773,
      "step": 105
    },
    {
      "epoch": 0.02,
      "grad_norm": 21.066689453711753,
      "learning_rate": 5.6684491978609635e-06,
      "loss": 0.5308,
      "step": 106
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.810074093176306,
      "learning_rate": 5.7219251336898395e-06,
      "loss": 0.5844,
      "step": 107
    },
    {
      "epoch": 0.02,
      "grad_norm": 37.64071299743144,
      "learning_rate": 5.775401069518717e-06,
      "loss": 0.645,
      "step": 108
    },
    {
      "epoch": 0.02,
      "grad_norm": 42.61317343790438,
      "learning_rate": 5.828877005347594e-06,
      "loss": 0.6186,
      "step": 109
    },
    {
      "epoch": 0.02,
      "grad_norm": 20.320797351695944,
      "learning_rate": 5.882352941176471e-06,
      "loss": 0.6261,
      "step": 110
    },
    {
      "epoch": 0.02,
      "grad_norm": 41.82428402340237,
      "learning_rate": 5.935828877005349e-06,
      "loss": 0.6811,
      "step": 111
    },
    {
      "epoch": 0.02,
      "grad_norm": 18.05125983518669,
      "learning_rate": 5.989304812834225e-06,
      "loss": 0.5138,
      "step": 112
    },
    {
      "epoch": 0.02,
      "grad_norm": 30.54162507624327,
      "learning_rate": 6.0427807486631015e-06,
      "loss": 0.6124,
      "step": 113
    },
    {
      "epoch": 0.02,
      "grad_norm": 27.222009084195935,
      "learning_rate": 6.096256684491979e-06,
      "loss": 0.5959,
      "step": 114
    },
    {
      "epoch": 0.02,
      "grad_norm": 19.57620562665309,
      "learning_rate": 6.149732620320856e-06,
      "loss": 0.6031,
      "step": 115
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.855726387451963,
      "learning_rate": 6.203208556149734e-06,
      "loss": 0.5891,
      "step": 116
    },
    {
      "epoch": 0.02,
      "grad_norm": 24.941928283806412,
      "learning_rate": 6.25668449197861e-06,
      "loss": 0.7241,
      "step": 117
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.537936109313009,
      "learning_rate": 6.3101604278074865e-06,
      "loss": 0.5994,
      "step": 118
    },
    {
      "epoch": 0.02,
      "grad_norm": 19.81397782910211,
      "learning_rate": 6.363636363636364e-06,
      "loss": 0.6061,
      "step": 119
    },
    {
      "epoch": 0.02,
      "grad_norm": 26.976056572828575,
      "learning_rate": 6.417112299465241e-06,
      "loss": 0.5641,
      "step": 120
    },
    {
      "epoch": 0.02,
      "grad_norm": 20.92909696757465,
      "learning_rate": 6.470588235294119e-06,
      "loss": 0.5742,
      "step": 121
    },
    {
      "epoch": 0.02,
      "grad_norm": 59.960917768114115,
      "learning_rate": 6.524064171122996e-06,
      "loss": 0.5995,
      "step": 122
    },
    {
      "epoch": 0.02,
      "grad_norm": 21.486053491694033,
      "learning_rate": 6.577540106951872e-06,
      "loss": 0.6093,
      "step": 123
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.968826538969584,
      "learning_rate": 6.631016042780749e-06,
      "loss": 0.596,
      "step": 124
    },
    {
      "epoch": 0.02,
      "grad_norm": 25.91913743513616,
      "learning_rate": 6.684491978609626e-06,
      "loss": 0.6512,
      "step": 125
    },
    {
      "epoch": 0.02,
      "grad_norm": 20.615735313978416,
      "learning_rate": 6.737967914438504e-06,
      "loss": 0.5516,
      "step": 126
    },
    {
      "epoch": 0.02,
      "grad_norm": 20.31173641758594,
      "learning_rate": 6.791443850267381e-06,
      "loss": 0.5893,
      "step": 127
    },
    {
      "epoch": 0.02,
      "grad_norm": 18.18932198046168,
      "learning_rate": 6.844919786096257e-06,
      "loss": 0.6299,
      "step": 128
    },
    {
      "epoch": 0.02,
      "grad_norm": 18.325853887283998,
      "learning_rate": 6.898395721925134e-06,
      "loss": 0.6586,
      "step": 129
    },
    {
      "epoch": 0.02,
      "grad_norm": 22.484889889241355,
      "learning_rate": 6.951871657754011e-06,
      "loss": 0.574,
      "step": 130
    },
    {
      "epoch": 0.02,
      "grad_norm": 22.084320699756343,
      "learning_rate": 7.005347593582889e-06,
      "loss": 0.6587,
      "step": 131
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2480209551400834,
      "learning_rate": 7.058823529411766e-06,
      "loss": 0.4398,
      "step": 132
    },
    {
      "epoch": 0.02,
      "grad_norm": 18.00151064391816,
      "learning_rate": 7.112299465240642e-06,
      "loss": 0.6409,
      "step": 133
    },
    {
      "epoch": 0.02,
      "grad_norm": 21.052998309067092,
      "learning_rate": 7.1657754010695195e-06,
      "loss": 0.5988,
      "step": 134
    },
    {
      "epoch": 0.02,
      "grad_norm": 17.284082776274925,
      "learning_rate": 7.219251336898396e-06,
      "loss": 0.5568,
      "step": 135
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.783320304280428,
      "learning_rate": 7.272727272727273e-06,
      "loss": 0.6072,
      "step": 136
    },
    {
      "epoch": 0.02,
      "grad_norm": 22.41423729347594,
      "learning_rate": 7.326203208556151e-06,
      "loss": 0.5824,
      "step": 137
    },
    {
      "epoch": 0.02,
      "grad_norm": 20.34357850447982,
      "learning_rate": 7.379679144385027e-06,
      "loss": 0.5785,
      "step": 138
    },
    {
      "epoch": 0.02,
      "grad_norm": 91.10070873339994,
      "learning_rate": 7.433155080213904e-06,
      "loss": 0.6492,
      "step": 139
    },
    {
      "epoch": 0.02,
      "grad_norm": 28.06425496642058,
      "learning_rate": 7.486631016042781e-06,
      "loss": 0.6717,
      "step": 140
    },
    {
      "epoch": 0.02,
      "grad_norm": 19.17170518135643,
      "learning_rate": 7.540106951871658e-06,
      "loss": 0.6571,
      "step": 141
    },
    {
      "epoch": 0.02,
      "grad_norm": 24.877663676776276,
      "learning_rate": 7.593582887700536e-06,
      "loss": 0.5846,
      "step": 142
    },
    {
      "epoch": 0.02,
      "grad_norm": 22.753318407414667,
      "learning_rate": 7.647058823529411e-06,
      "loss": 0.5865,
      "step": 143
    },
    {
      "epoch": 0.02,
      "grad_norm": 69.6797782080603,
      "learning_rate": 7.70053475935829e-06,
      "loss": 0.6192,
      "step": 144
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.339715015322948,
      "learning_rate": 7.754010695187166e-06,
      "loss": 0.5302,
      "step": 145
    },
    {
      "epoch": 0.02,
      "grad_norm": 25.125502174512487,
      "learning_rate": 7.807486631016043e-06,
      "loss": 0.5945,
      "step": 146
    },
    {
      "epoch": 0.02,
      "grad_norm": 31.807966557250012,
      "learning_rate": 7.86096256684492e-06,
      "loss": 0.5701,
      "step": 147
    },
    {
      "epoch": 0.02,
      "grad_norm": 16.756136730927324,
      "learning_rate": 7.914438502673799e-06,
      "loss": 0.6455,
      "step": 148
    },
    {
      "epoch": 0.02,
      "grad_norm": 21.694097248619133,
      "learning_rate": 7.967914438502674e-06,
      "loss": 0.6354,
      "step": 149
    },
    {
      "epoch": 0.02,
      "grad_norm": 24.988217091215123,
      "learning_rate": 8.02139037433155e-06,
      "loss": 0.5974,
      "step": 150
    },
    {
      "epoch": 0.02,
      "grad_norm": 48.94818189142812,
      "learning_rate": 8.07486631016043e-06,
      "loss": 0.5621,
      "step": 151
    },
    {
      "epoch": 0.02,
      "grad_norm": 16.786693796458923,
      "learning_rate": 8.128342245989306e-06,
      "loss": 0.5029,
      "step": 152
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2132368709705952,
      "learning_rate": 8.181818181818183e-06,
      "loss": 0.4582,
      "step": 153
    },
    {
      "epoch": 0.02,
      "grad_norm": 17.838014549382752,
      "learning_rate": 8.23529411764706e-06,
      "loss": 0.6081,
      "step": 154
    },
    {
      "epoch": 0.02,
      "grad_norm": 23.174949309393295,
      "learning_rate": 8.288770053475937e-06,
      "loss": 0.5913,
      "step": 155
    },
    {
      "epoch": 0.03,
      "grad_norm": 58.10277685435587,
      "learning_rate": 8.342245989304813e-06,
      "loss": 0.62,
      "step": 156
    },
    {
      "epoch": 0.03,
      "grad_norm": 35.11148524301774,
      "learning_rate": 8.39572192513369e-06,
      "loss": 0.6108,
      "step": 157
    },
    {
      "epoch": 0.03,
      "grad_norm": 26.345809275562033,
      "learning_rate": 8.449197860962567e-06,
      "loss": 0.6028,
      "step": 158
    },
    {
      "epoch": 0.03,
      "grad_norm": 28.757447702452716,
      "learning_rate": 8.502673796791444e-06,
      "loss": 0.6476,
      "step": 159
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2057898552029815,
      "learning_rate": 8.556149732620321e-06,
      "loss": 0.4502,
      "step": 160
    },
    {
      "epoch": 0.03,
      "grad_norm": 28.7196085975342,
      "learning_rate": 8.609625668449198e-06,
      "loss": 0.5792,
      "step": 161
    },
    {
      "epoch": 0.03,
      "grad_norm": 51.55763623930404,
      "learning_rate": 8.663101604278076e-06,
      "loss": 0.5271,
      "step": 162
    },
    {
      "epoch": 0.03,
      "grad_norm": 29.457758212672534,
      "learning_rate": 8.716577540106953e-06,
      "loss": 0.6488,
      "step": 163
    },
    {
      "epoch": 0.03,
      "grad_norm": 64.51326898401457,
      "learning_rate": 8.77005347593583e-06,
      "loss": 0.5934,
      "step": 164
    },
    {
      "epoch": 0.03,
      "grad_norm": 27.709961183521948,
      "learning_rate": 8.823529411764707e-06,
      "loss": 0.5677,
      "step": 165
    },
    {
      "epoch": 0.03,
      "grad_norm": 32.89692578423412,
      "learning_rate": 8.877005347593584e-06,
      "loss": 0.5662,
      "step": 166
    },
    {
      "epoch": 0.03,
      "grad_norm": 30.36290733478553,
      "learning_rate": 8.93048128342246e-06,
      "loss": 0.6224,
      "step": 167
    },
    {
      "epoch": 0.03,
      "grad_norm": 23.382179698659886,
      "learning_rate": 8.983957219251337e-06,
      "loss": 0.5562,
      "step": 168
    },
    {
      "epoch": 0.03,
      "grad_norm": 42.860233528227255,
      "learning_rate": 9.037433155080214e-06,
      "loss": 0.5669,
      "step": 169
    },
    {
      "epoch": 0.03,
      "grad_norm": 49.17982754033979,
      "learning_rate": 9.090909090909091e-06,
      "loss": 0.4837,
      "step": 170
    },
    {
      "epoch": 0.03,
      "grad_norm": 29.229854648173912,
      "learning_rate": 9.144385026737968e-06,
      "loss": 0.5601,
      "step": 171
    },
    {
      "epoch": 0.03,
      "grad_norm": 30.51681488275903,
      "learning_rate": 9.197860962566846e-06,
      "loss": 0.5652,
      "step": 172
    },
    {
      "epoch": 0.03,
      "grad_norm": 40.12952587187656,
      "learning_rate": 9.251336898395723e-06,
      "loss": 0.566,
      "step": 173
    },
    {
      "epoch": 0.03,
      "grad_norm": 32.46229786172875,
      "learning_rate": 9.3048128342246e-06,
      "loss": 0.5716,
      "step": 174
    },
    {
      "epoch": 0.03,
      "grad_norm": 44.679071142403664,
      "learning_rate": 9.358288770053477e-06,
      "loss": 0.5181,
      "step": 175
    },
    {
      "epoch": 0.03,
      "grad_norm": 40.71036783457782,
      "learning_rate": 9.411764705882354e-06,
      "loss": 0.6452,
      "step": 176
    },
    {
      "epoch": 0.03,
      "grad_norm": 78.8545624812088,
      "learning_rate": 9.46524064171123e-06,
      "loss": 0.6516,
      "step": 177
    },
    {
      "epoch": 0.03,
      "grad_norm": 21.921839800170993,
      "learning_rate": 9.518716577540108e-06,
      "loss": 0.5642,
      "step": 178
    },
    {
      "epoch": 0.03,
      "grad_norm": 67.10835639297564,
      "learning_rate": 9.572192513368986e-06,
      "loss": 0.5199,
      "step": 179
    },
    {
      "epoch": 0.03,
      "grad_norm": 35.91729067083796,
      "learning_rate": 9.625668449197861e-06,
      "loss": 0.6715,
      "step": 180
    },
    {
      "epoch": 0.03,
      "grad_norm": 30.53398989521768,
      "learning_rate": 9.679144385026738e-06,
      "loss": 0.5827,
      "step": 181
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.225844987409763,
      "learning_rate": 9.732620320855617e-06,
      "loss": 0.4322,
      "step": 182
    },
    {
      "epoch": 0.03,
      "grad_norm": 42.32698934436769,
      "learning_rate": 9.786096256684493e-06,
      "loss": 0.5656,
      "step": 183
    },
    {
      "epoch": 0.03,
      "grad_norm": 25.03027998026712,
      "learning_rate": 9.83957219251337e-06,
      "loss": 0.5824,
      "step": 184
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.317887942116737,
      "learning_rate": 9.893048128342247e-06,
      "loss": 0.4833,
      "step": 185
    },
    {
      "epoch": 0.03,
      "grad_norm": 31.93535496217499,
      "learning_rate": 9.946524064171124e-06,
      "loss": 0.5946,
      "step": 186
    },
    {
      "epoch": 0.03,
      "grad_norm": 21.01641343569306,
      "learning_rate": 1e-05,
      "loss": 0.5958,
      "step": 187
    },
    {
      "epoch": 0.03,
      "grad_norm": 84.57102601338673,
      "learning_rate": 9.999999318931088e-06,
      "loss": 0.576,
      "step": 188
    },
    {
      "epoch": 0.03,
      "grad_norm": 25.58380639550186,
      "learning_rate": 9.999997275724535e-06,
      "loss": 0.5934,
      "step": 189
    },
    {
      "epoch": 0.03,
      "grad_norm": 43.93180382709066,
      "learning_rate": 9.999993870380897e-06,
      "loss": 0.5593,
      "step": 190
    },
    {
      "epoch": 0.03,
      "grad_norm": 23.349849368866,
      "learning_rate": 9.999989102901105e-06,
      "loss": 0.5626,
      "step": 191
    },
    {
      "epoch": 0.03,
      "grad_norm": 20.888299437307587,
      "learning_rate": 9.999982973286455e-06,
      "loss": 0.5749,
      "step": 192
    },
    {
      "epoch": 0.03,
      "grad_norm": 20.0977512771481,
      "learning_rate": 9.999975481538618e-06,
      "loss": 0.5075,
      "step": 193
    },
    {
      "epoch": 0.03,
      "grad_norm": 38.11821401436929,
      "learning_rate": 9.999966627659635e-06,
      "loss": 0.5046,
      "step": 194
    },
    {
      "epoch": 0.03,
      "grad_norm": 22.473927278749628,
      "learning_rate": 9.999956411651916e-06,
      "loss": 0.5435,
      "step": 195
    },
    {
      "epoch": 0.03,
      "grad_norm": 18.120374263358624,
      "learning_rate": 9.999944833518248e-06,
      "loss": 0.5539,
      "step": 196
    },
    {
      "epoch": 0.03,
      "grad_norm": 18.818380286558053,
      "learning_rate": 9.999931893261783e-06,
      "loss": 0.6751,
      "step": 197
    },
    {
      "epoch": 0.03,
      "grad_norm": 22.25554843026403,
      "learning_rate": 9.999917590886046e-06,
      "loss": 0.5946,
      "step": 198
    },
    {
      "epoch": 0.03,
      "grad_norm": 22.97019807220915,
      "learning_rate": 9.999901926394932e-06,
      "loss": 0.5891,
      "step": 199
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.706462387191223,
      "learning_rate": 9.99988489979271e-06,
      "loss": 0.4632,
      "step": 200
    },
    {
      "epoch": 0.03,
      "grad_norm": 24.061854873862444,
      "learning_rate": 9.999866511084021e-06,
      "loss": 0.5995,
      "step": 201
    },
    {
      "epoch": 0.03,
      "grad_norm": 30.882464693094764,
      "learning_rate": 9.999846760273873e-06,
      "loss": 0.5464,
      "step": 202
    },
    {
      "epoch": 0.03,
      "grad_norm": 26.51172217095793,
      "learning_rate": 9.999825647367643e-06,
      "loss": 0.5838,
      "step": 203
    },
    {
      "epoch": 0.03,
      "grad_norm": 37.093160416079336,
      "learning_rate": 9.999803172371088e-06,
      "loss": 0.5999,
      "step": 204
    },
    {
      "epoch": 0.03,
      "grad_norm": 25.646550754862893,
      "learning_rate": 9.999779335290328e-06,
      "loss": 0.5762,
      "step": 205
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.564564908728737,
      "learning_rate": 9.999754136131855e-06,
      "loss": 0.4914,
      "step": 206
    },
    {
      "epoch": 0.03,
      "grad_norm": 23.616305190856394,
      "learning_rate": 9.999727574902538e-06,
      "loss": 0.5531,
      "step": 207
    },
    {
      "epoch": 0.03,
      "grad_norm": 19.747613211173068,
      "learning_rate": 9.999699651609611e-06,
      "loss": 0.5821,
      "step": 208
    },
    {
      "epoch": 0.03,
      "grad_norm": 27.09812594656937,
      "learning_rate": 9.999670366260682e-06,
      "loss": 0.5854,
      "step": 209
    },
    {
      "epoch": 0.03,
      "grad_norm": 29.071729394812532,
      "learning_rate": 9.999639718863728e-06,
      "loss": 0.5739,
      "step": 210
    },
    {
      "epoch": 0.03,
      "grad_norm": 16.119811914476216,
      "learning_rate": 9.9996077094271e-06,
      "loss": 0.5815,
      "step": 211
    },
    {
      "epoch": 0.03,
      "grad_norm": 34.84882929176075,
      "learning_rate": 9.999574337959514e-06,
      "loss": 0.5799,
      "step": 212
    },
    {
      "epoch": 0.03,
      "grad_norm": 27.849279396445613,
      "learning_rate": 9.999539604470068e-06,
      "loss": 0.5737,
      "step": 213
    },
    {
      "epoch": 0.03,
      "grad_norm": 24.952346926553595,
      "learning_rate": 9.999503508968218e-06,
      "loss": 0.5676,
      "step": 214
    },
    {
      "epoch": 0.03,
      "grad_norm": 26.946784765670863,
      "learning_rate": 9.9994660514638e-06,
      "loss": 0.5265,
      "step": 215
    },
    {
      "epoch": 0.03,
      "grad_norm": 39.515364664534594,
      "learning_rate": 9.99942723196702e-06,
      "loss": 0.5673,
      "step": 216
    },
    {
      "epoch": 0.03,
      "grad_norm": 21.597549155378786,
      "learning_rate": 9.999387050488451e-06,
      "loss": 0.4892,
      "step": 217
    },
    {
      "epoch": 0.04,
      "grad_norm": 36.36299545430016,
      "learning_rate": 9.99934550703904e-06,
      "loss": 0.5446,
      "step": 218
    },
    {
      "epoch": 0.04,
      "grad_norm": 21.538372120559256,
      "learning_rate": 9.999302601630106e-06,
      "loss": 0.5604,
      "step": 219
    },
    {
      "epoch": 0.04,
      "grad_norm": 27.13417033785625,
      "learning_rate": 9.999258334273338e-06,
      "loss": 0.5454,
      "step": 220
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.306466726987151,
      "learning_rate": 9.999212704980792e-06,
      "loss": 0.4934,
      "step": 221
    },
    {
      "epoch": 0.04,
      "grad_norm": 52.223827376076414,
      "learning_rate": 9.999165713764902e-06,
      "loss": 0.5634,
      "step": 222
    },
    {
      "epoch": 0.04,
      "grad_norm": 23.92417627579469,
      "learning_rate": 9.999117360638469e-06,
      "loss": 0.5653,
      "step": 223
    },
    {
      "epoch": 0.04,
      "grad_norm": 26.25244952786487,
      "learning_rate": 9.999067645614666e-06,
      "loss": 0.6201,
      "step": 224
    },
    {
      "epoch": 0.04,
      "grad_norm": 52.99497059344627,
      "learning_rate": 9.999016568707036e-06,
      "loss": 0.649,
      "step": 225
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2354855596202785,
      "learning_rate": 9.998964129929493e-06,
      "loss": 0.4768,
      "step": 226
    },
    {
      "epoch": 0.04,
      "grad_norm": 22.988882270251633,
      "learning_rate": 9.998910329296322e-06,
      "loss": 0.5786,
      "step": 227
    },
    {
      "epoch": 0.04,
      "grad_norm": 24.13374025934291,
      "learning_rate": 9.998855166822186e-06,
      "loss": 0.5448,
      "step": 228
    },
    {
      "epoch": 0.04,
      "grad_norm": 19.478786990599023,
      "learning_rate": 9.998798642522105e-06,
      "loss": 0.5741,
      "step": 229
    },
    {
      "epoch": 0.04,
      "grad_norm": 26.826941026382045,
      "learning_rate": 9.998740756411483e-06,
      "loss": 0.5658,
      "step": 230
    },
    {
      "epoch": 0.04,
      "grad_norm": 14.563016385572427,
      "learning_rate": 9.998681508506087e-06,
      "loss": 0.4788,
      "step": 231
    },
    {
      "epoch": 0.04,
      "grad_norm": 15.526395978081707,
      "learning_rate": 9.998620898822059e-06,
      "loss": 0.5559,
      "step": 232
    },
    {
      "epoch": 0.04,
      "grad_norm": 22.194535950164592,
      "learning_rate": 9.998558927375909e-06,
      "loss": 0.5206,
      "step": 233
    },
    {
      "epoch": 0.04,
      "grad_norm": 33.49532012075608,
      "learning_rate": 9.998495594184523e-06,
      "loss": 0.5189,
      "step": 234
    },
    {
      "epoch": 0.04,
      "grad_norm": 57.31725056280748,
      "learning_rate": 9.998430899265152e-06,
      "loss": 0.534,
      "step": 235
    },
    {
      "epoch": 0.04,
      "grad_norm": 44.49557379578555,
      "learning_rate": 9.998364842635422e-06,
      "loss": 0.5228,
      "step": 236
    },
    {
      "epoch": 0.04,
      "grad_norm": 26.558282803659548,
      "learning_rate": 9.998297424313327e-06,
      "loss": 0.4949,
      "step": 237
    },
    {
      "epoch": 0.04,
      "grad_norm": 21.077766343461054,
      "learning_rate": 9.998228644317235e-06,
      "loss": 0.6122,
      "step": 238
    },
    {
      "epoch": 0.04,
      "grad_norm": 27.299226322814544,
      "learning_rate": 9.998158502665884e-06,
      "loss": 0.56,
      "step": 239
    },
    {
      "epoch": 0.04,
      "grad_norm": 17.952229427066612,
      "learning_rate": 9.99808699937838e-06,
      "loss": 0.5543,
      "step": 240
    },
    {
      "epoch": 0.04,
      "grad_norm": 49.61385356334192,
      "learning_rate": 9.998014134474207e-06,
      "loss": 0.5375,
      "step": 241
    },
    {
      "epoch": 0.04,
      "grad_norm": 18.227289147125834,
      "learning_rate": 9.997939907973212e-06,
      "loss": 0.5632,
      "step": 242
    },
    {
      "epoch": 0.04,
      "grad_norm": 22.92871168209305,
      "learning_rate": 9.997864319895616e-06,
      "loss": 0.5781,
      "step": 243
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2574590512697594,
      "learning_rate": 9.997787370262012e-06,
      "loss": 0.4581,
      "step": 244
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2005917968713267,
      "learning_rate": 9.997709059093364e-06,
      "loss": 0.4531,
      "step": 245
    },
    {
      "epoch": 0.04,
      "grad_norm": 23.441321088762436,
      "learning_rate": 9.997629386411006e-06,
      "loss": 0.5519,
      "step": 246
    },
    {
      "epoch": 0.04,
      "grad_norm": 34.888071097660706,
      "learning_rate": 9.997548352236644e-06,
      "loss": 0.5391,
      "step": 247
    },
    {
      "epoch": 0.04,
      "grad_norm": 14.904388776270206,
      "learning_rate": 9.99746595659235e-06,
      "loss": 0.5634,
      "step": 248
    },
    {
      "epoch": 0.04,
      "grad_norm": 29.944998619314298,
      "learning_rate": 9.997382199500577e-06,
      "loss": 0.611,
      "step": 249
    },
    {
      "epoch": 0.04,
      "grad_norm": 29.277373234281534,
      "learning_rate": 9.997297080984136e-06,
      "loss": 0.5593,
      "step": 250
    },
    {
      "epoch": 0.04,
      "grad_norm": 29.870702777628296,
      "learning_rate": 9.997210601066218e-06,
      "loss": 0.4689,
      "step": 251
    },
    {
      "epoch": 0.04,
      "grad_norm": 18.661590600953268,
      "learning_rate": 9.997122759770386e-06,
      "loss": 0.5765,
      "step": 252
    },
    {
      "epoch": 0.04,
      "grad_norm": 18.172031386490158,
      "learning_rate": 9.997033557120567e-06,
      "loss": 0.5655,
      "step": 253
    },
    {
      "epoch": 0.04,
      "grad_norm": 14.495604024380304,
      "learning_rate": 9.996942993141063e-06,
      "loss": 0.5272,
      "step": 254
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4875376276342165,
      "learning_rate": 9.996851067856546e-06,
      "loss": 0.4763,
      "step": 255
    },
    {
      "epoch": 0.04,
      "grad_norm": 18.066879056490265,
      "learning_rate": 9.996757781292058e-06,
      "loss": 0.535,
      "step": 256
    },
    {
      "epoch": 0.04,
      "grad_norm": 23.907625456763284,
      "learning_rate": 9.996663133473017e-06,
      "loss": 0.5069,
      "step": 257
    },
    {
      "epoch": 0.04,
      "grad_norm": 22.68612313214401,
      "learning_rate": 9.996567124425201e-06,
      "loss": 0.5409,
      "step": 258
    },
    {
      "epoch": 0.04,
      "grad_norm": 28.580822847531774,
      "learning_rate": 9.996469754174772e-06,
      "loss": 0.5293,
      "step": 259
    },
    {
      "epoch": 0.04,
      "grad_norm": 17.87811428048285,
      "learning_rate": 9.996371022748251e-06,
      "loss": 0.5215,
      "step": 260
    },
    {
      "epoch": 0.04,
      "grad_norm": 20.25414039382917,
      "learning_rate": 9.99627093017254e-06,
      "loss": 0.527,
      "step": 261
    },
    {
      "epoch": 0.04,
      "grad_norm": 24.734183136640993,
      "learning_rate": 9.996169476474902e-06,
      "loss": 0.5709,
      "step": 262
    },
    {
      "epoch": 0.04,
      "grad_norm": 18.01084560508422,
      "learning_rate": 9.996066661682981e-06,
      "loss": 0.5359,
      "step": 263
    },
    {
      "epoch": 0.04,
      "grad_norm": 24.46686164993537,
      "learning_rate": 9.995962485824783e-06,
      "loss": 0.5931,
      "step": 264
    },
    {
      "epoch": 0.04,
      "grad_norm": 18.35433198798991,
      "learning_rate": 9.995856948928688e-06,
      "loss": 0.5721,
      "step": 265
    },
    {
      "epoch": 0.04,
      "grad_norm": 15.213726242595348,
      "learning_rate": 9.99575005102345e-06,
      "loss": 0.4902,
      "step": 266
    },
    {
      "epoch": 0.04,
      "grad_norm": 18.392656248707723,
      "learning_rate": 9.99564179213819e-06,
      "loss": 0.5915,
      "step": 267
    },
    {
      "epoch": 0.04,
      "grad_norm": 15.04115075164745,
      "learning_rate": 9.995532172302399e-06,
      "loss": 0.5435,
      "step": 268
    },
    {
      "epoch": 0.04,
      "grad_norm": 24.847149451890505,
      "learning_rate": 9.995421191545942e-06,
      "loss": 0.5364,
      "step": 269
    },
    {
      "epoch": 0.04,
      "grad_norm": 18.766567633854013,
      "learning_rate": 9.995308849899052e-06,
      "loss": 0.6014,
      "step": 270
    },
    {
      "epoch": 0.04,
      "grad_norm": 18.934027444938955,
      "learning_rate": 9.995195147392335e-06,
      "loss": 0.5001,
      "step": 271
    },
    {
      "epoch": 0.04,
      "grad_norm": 23.29224964131107,
      "learning_rate": 9.995080084056767e-06,
      "loss": 0.5458,
      "step": 272
    },
    {
      "epoch": 0.04,
      "grad_norm": 116.0395013873673,
      "learning_rate": 9.994963659923695e-06,
      "loss": 0.4876,
      "step": 273
    },
    {
      "epoch": 0.04,
      "grad_norm": 20.550561641908725,
      "learning_rate": 9.994845875024834e-06,
      "loss": 0.5308,
      "step": 274
    },
    {
      "epoch": 0.04,
      "grad_norm": 17.723653697155687,
      "learning_rate": 9.994726729392272e-06,
      "loss": 0.5316,
      "step": 275
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2152913079936658,
      "learning_rate": 9.99460622305847e-06,
      "loss": 0.4777,
      "step": 276
    },
    {
      "epoch": 0.04,
      "grad_norm": 35.263337400817974,
      "learning_rate": 9.994484356056256e-06,
      "loss": 0.5959,
      "step": 277
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1001898209957774,
      "learning_rate": 9.994361128418828e-06,
      "loss": 0.4649,
      "step": 278
    },
    {
      "epoch": 0.04,
      "grad_norm": 24.12068182949776,
      "learning_rate": 9.99423654017976e-06,
      "loss": 0.5232,
      "step": 279
    },
    {
      "epoch": 0.05,
      "grad_norm": 50.66715646834188,
      "learning_rate": 9.99411059137299e-06,
      "loss": 0.5588,
      "step": 280
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.650007219125833,
      "learning_rate": 9.993983282032831e-06,
      "loss": 0.5284,
      "step": 281
    },
    {
      "epoch": 0.05,
      "grad_norm": 34.57803254152516,
      "learning_rate": 9.993854612193967e-06,
      "loss": 0.5828,
      "step": 282
    },
    {
      "epoch": 0.05,
      "grad_norm": 40.86973153855904,
      "learning_rate": 9.993724581891451e-06,
      "loss": 0.5499,
      "step": 283
    },
    {
      "epoch": 0.05,
      "grad_norm": 17.63516792646248,
      "learning_rate": 9.993593191160704e-06,
      "loss": 0.5242,
      "step": 284
    },
    {
      "epoch": 0.05,
      "grad_norm": 22.431439466588927,
      "learning_rate": 9.993460440037525e-06,
      "loss": 0.5672,
      "step": 285
    },
    {
      "epoch": 0.05,
      "grad_norm": 28.27244449728826,
      "learning_rate": 9.993326328558076e-06,
      "loss": 0.5573,
      "step": 286
    },
    {
      "epoch": 0.05,
      "grad_norm": 69.61736822464846,
      "learning_rate": 9.993190856758892e-06,
      "loss": 0.525,
      "step": 287
    },
    {
      "epoch": 0.05,
      "grad_norm": 19.913915733585295,
      "learning_rate": 9.99305402467688e-06,
      "loss": 0.5368,
      "step": 288
    },
    {
      "epoch": 0.05,
      "grad_norm": 43.833892898165445,
      "learning_rate": 9.99291583234932e-06,
      "loss": 0.5604,
      "step": 289
    },
    {
      "epoch": 0.05,
      "grad_norm": 15.592956349232821,
      "learning_rate": 9.992776279813854e-06,
      "loss": 0.5595,
      "step": 290
    },
    {
      "epoch": 0.05,
      "grad_norm": 14.496479439748734,
      "learning_rate": 9.992635367108505e-06,
      "loss": 0.54,
      "step": 291
    },
    {
      "epoch": 0.05,
      "grad_norm": 21.801186293712377,
      "learning_rate": 9.992493094271657e-06,
      "loss": 0.5456,
      "step": 292
    },
    {
      "epoch": 0.05,
      "grad_norm": 15.521213994153893,
      "learning_rate": 9.992349461342073e-06,
      "loss": 0.508,
      "step": 293
    },
    {
      "epoch": 0.05,
      "grad_norm": 61.73552245107981,
      "learning_rate": 9.992204468358879e-06,
      "loss": 0.5279,
      "step": 294
    },
    {
      "epoch": 0.05,
      "grad_norm": 21.689513247406182,
      "learning_rate": 9.992058115361578e-06,
      "loss": 0.5112,
      "step": 295
    },
    {
      "epoch": 0.05,
      "grad_norm": 25.61535919933569,
      "learning_rate": 9.991910402390041e-06,
      "loss": 0.515,
      "step": 296
    },
    {
      "epoch": 0.05,
      "grad_norm": 118.51700796085427,
      "learning_rate": 9.991761329484505e-06,
      "loss": 0.4884,
      "step": 297
    },
    {
      "epoch": 0.05,
      "grad_norm": 18.761074282300584,
      "learning_rate": 9.991610896685587e-06,
      "loss": 0.5407,
      "step": 298
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4482620210295873,
      "learning_rate": 9.991459104034262e-06,
      "loss": 0.5206,
      "step": 299
    },
    {
      "epoch": 0.05,
      "grad_norm": 23.295496932705312,
      "learning_rate": 9.991305951571891e-06,
      "loss": 0.5356,
      "step": 300
    },
    {
      "epoch": 0.05,
      "grad_norm": 12.217357741647287,
      "learning_rate": 9.99115143934019e-06,
      "loss": 0.5142,
      "step": 301
    },
    {
      "epoch": 0.05,
      "grad_norm": 39.894290936272284,
      "learning_rate": 9.990995567381255e-06,
      "loss": 0.5593,
      "step": 302
    },
    {
      "epoch": 0.05,
      "grad_norm": 22.34063424998089,
      "learning_rate": 9.990838335737551e-06,
      "loss": 0.5281,
      "step": 303
    },
    {
      "epoch": 0.05,
      "grad_norm": 29.1087149830401,
      "learning_rate": 9.990679744451909e-06,
      "loss": 0.5024,
      "step": 304
    },
    {
      "epoch": 0.05,
      "grad_norm": 39.04360417713495,
      "learning_rate": 9.990519793567539e-06,
      "loss": 0.4991,
      "step": 305
    },
    {
      "epoch": 0.05,
      "grad_norm": 37.967141610223415,
      "learning_rate": 9.990358483128012e-06,
      "loss": 0.6056,
      "step": 306
    },
    {
      "epoch": 0.05,
      "grad_norm": 17.3353241881632,
      "learning_rate": 9.990195813177272e-06,
      "loss": 0.5398,
      "step": 307
    },
    {
      "epoch": 0.05,
      "grad_norm": 20.07522917913113,
      "learning_rate": 9.99003178375964e-06,
      "loss": 0.5157,
      "step": 308
    },
    {
      "epoch": 0.05,
      "grad_norm": 21.32991615089147,
      "learning_rate": 9.989866394919795e-06,
      "loss": 0.5543,
      "step": 309
    },
    {
      "epoch": 0.05,
      "grad_norm": 72.44284433249531,
      "learning_rate": 9.9896996467028e-06,
      "loss": 0.4953,
      "step": 310
    },
    {
      "epoch": 0.05,
      "grad_norm": 21.319769035482324,
      "learning_rate": 9.98953153915408e-06,
      "loss": 0.4953,
      "step": 311
    },
    {
      "epoch": 0.05,
      "grad_norm": 80.55786410346872,
      "learning_rate": 9.989362072319431e-06,
      "loss": 0.4916,
      "step": 312
    },
    {
      "epoch": 0.05,
      "grad_norm": 23.81783195210229,
      "learning_rate": 9.98919124624502e-06,
      "loss": 0.5682,
      "step": 313
    },
    {
      "epoch": 0.05,
      "grad_norm": 33.330639064743856,
      "learning_rate": 9.989019060977388e-06,
      "loss": 0.5809,
      "step": 314
    },
    {
      "epoch": 0.05,
      "grad_norm": 35.23341737570225,
      "learning_rate": 9.988845516563437e-06,
      "loss": 0.5576,
      "step": 315
    },
    {
      "epoch": 0.05,
      "grad_norm": 31.59904410655793,
      "learning_rate": 9.988670613050452e-06,
      "loss": 0.5942,
      "step": 316
    },
    {
      "epoch": 0.05,
      "grad_norm": 19.303374005522244,
      "learning_rate": 9.988494350486077e-06,
      "loss": 0.5701,
      "step": 317
    },
    {
      "epoch": 0.05,
      "grad_norm": 33.51499512469324,
      "learning_rate": 9.988316728918332e-06,
      "loss": 0.5293,
      "step": 318
    },
    {
      "epoch": 0.05,
      "grad_norm": 43.43297427047658,
      "learning_rate": 9.988137748395608e-06,
      "loss": 0.4729,
      "step": 319
    },
    {
      "epoch": 0.05,
      "grad_norm": 39.685120874705795,
      "learning_rate": 9.987957408966662e-06,
      "loss": 0.5238,
      "step": 320
    },
    {
      "epoch": 0.05,
      "grad_norm": 50.81876616983587,
      "learning_rate": 9.987775710680622e-06,
      "loss": 0.5382,
      "step": 321
    },
    {
      "epoch": 0.05,
      "grad_norm": 23.443865897767214,
      "learning_rate": 9.987592653586992e-06,
      "loss": 0.5383,
      "step": 322
    },
    {
      "epoch": 0.05,
      "grad_norm": 26.676038890179647,
      "learning_rate": 9.987408237735638e-06,
      "loss": 0.5777,
      "step": 323
    },
    {
      "epoch": 0.05,
      "grad_norm": 35.95779328482772,
      "learning_rate": 9.987222463176803e-06,
      "loss": 0.5146,
      "step": 324
    },
    {
      "epoch": 0.05,
      "grad_norm": 49.413121870741655,
      "learning_rate": 9.987035329961092e-06,
      "loss": 0.5714,
      "step": 325
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2661824546367435,
      "learning_rate": 9.986846838139492e-06,
      "loss": 0.4628,
      "step": 326
    },
    {
      "epoch": 0.05,
      "grad_norm": 22.909845685667744,
      "learning_rate": 9.986656987763348e-06,
      "loss": 0.5652,
      "step": 327
    },
    {
      "epoch": 0.05,
      "grad_norm": 44.75717727170816,
      "learning_rate": 9.986465778884384e-06,
      "loss": 0.6211,
      "step": 328
    },
    {
      "epoch": 0.05,
      "grad_norm": 26.470796711705937,
      "learning_rate": 9.986273211554689e-06,
      "loss": 0.5085,
      "step": 329
    },
    {
      "epoch": 0.05,
      "grad_norm": 70.9588388675342,
      "learning_rate": 9.986079285826721e-06,
      "loss": 0.5153,
      "step": 330
    },
    {
      "epoch": 0.05,
      "grad_norm": 29.68828398855501,
      "learning_rate": 9.985884001753317e-06,
      "loss": 0.4738,
      "step": 331
    },
    {
      "epoch": 0.05,
      "grad_norm": 30.295885616262016,
      "learning_rate": 9.985687359387673e-06,
      "loss": 0.52,
      "step": 332
    },
    {
      "epoch": 0.05,
      "grad_norm": 30.940617235406467,
      "learning_rate": 9.98548935878336e-06,
      "loss": 0.5988,
      "step": 333
    },
    {
      "epoch": 0.05,
      "grad_norm": 27.67229063242171,
      "learning_rate": 9.985289999994322e-06,
      "loss": 0.4904,
      "step": 334
    },
    {
      "epoch": 0.05,
      "grad_norm": 20.829003469840004,
      "learning_rate": 9.985089283074867e-06,
      "loss": 0.5243,
      "step": 335
    },
    {
      "epoch": 0.05,
      "grad_norm": 38.44824909771365,
      "learning_rate": 9.984887208079675e-06,
      "loss": 0.5253,
      "step": 336
    },
    {
      "epoch": 0.05,
      "grad_norm": 49.4598365701943,
      "learning_rate": 9.9846837750638e-06,
      "loss": 0.5281,
      "step": 337
    },
    {
      "epoch": 0.05,
      "grad_norm": 72.11861428752515,
      "learning_rate": 9.98447898408266e-06,
      "loss": 0.4996,
      "step": 338
    },
    {
      "epoch": 0.05,
      "grad_norm": 22.212622132797467,
      "learning_rate": 9.984272835192047e-06,
      "loss": 0.5199,
      "step": 339
    },
    {
      "epoch": 0.05,
      "grad_norm": 37.45339726520185,
      "learning_rate": 9.984065328448122e-06,
      "loss": 0.5341,
      "step": 340
    },
    {
      "epoch": 0.05,
      "grad_norm": 91.32610103206306,
      "learning_rate": 9.983856463907415e-06,
      "loss": 0.5574,
      "step": 341
    },
    {
      "epoch": 0.06,
      "grad_norm": 65.31059173141998,
      "learning_rate": 9.983646241626825e-06,
      "loss": 0.4818,
      "step": 342
    },
    {
      "epoch": 0.06,
      "grad_norm": 40.552655348204645,
      "learning_rate": 9.983434661663625e-06,
      "loss": 0.4886,
      "step": 343
    },
    {
      "epoch": 0.06,
      "grad_norm": 58.39408005261295,
      "learning_rate": 9.983221724075453e-06,
      "loss": 0.5584,
      "step": 344
    },
    {
      "epoch": 0.06,
      "grad_norm": 144.2902028408597,
      "learning_rate": 9.983007428920322e-06,
      "loss": 0.5377,
      "step": 345
    },
    {
      "epoch": 0.06,
      "grad_norm": 26.933321668542007,
      "learning_rate": 9.982791776256608e-06,
      "loss": 0.4838,
      "step": 346
    },
    {
      "epoch": 0.06,
      "grad_norm": 23.027458811876798,
      "learning_rate": 9.982574766143063e-06,
      "loss": 0.5939,
      "step": 347
    },
    {
      "epoch": 0.06,
      "grad_norm": 24.863361562216706,
      "learning_rate": 9.982356398638807e-06,
      "loss": 0.5593,
      "step": 348
    },
    {
      "epoch": 0.06,
      "grad_norm": 33.83998134688836,
      "learning_rate": 9.982136673803328e-06,
      "loss": 0.5112,
      "step": 349
    },
    {
      "epoch": 0.06,
      "grad_norm": 15.899535320294381,
      "learning_rate": 9.981915591696484e-06,
      "loss": 0.5451,
      "step": 350
    },
    {
      "epoch": 0.06,
      "grad_norm": 40.47621905200634,
      "learning_rate": 9.981693152378509e-06,
      "loss": 0.5137,
      "step": 351
    },
    {
      "epoch": 0.06,
      "grad_norm": 22.347207755338726,
      "learning_rate": 9.981469355909996e-06,
      "loss": 0.5799,
      "step": 352
    },
    {
      "epoch": 0.06,
      "grad_norm": 21.8418243578854,
      "learning_rate": 9.981244202351916e-06,
      "loss": 0.5997,
      "step": 353
    },
    {
      "epoch": 0.06,
      "grad_norm": 26.85124037083119,
      "learning_rate": 9.981017691765606e-06,
      "loss": 0.5601,
      "step": 354
    },
    {
      "epoch": 0.06,
      "grad_norm": 18.2564746566136,
      "learning_rate": 9.980789824212776e-06,
      "loss": 0.5069,
      "step": 355
    },
    {
      "epoch": 0.06,
      "grad_norm": 27.07945100773563,
      "learning_rate": 9.980560599755498e-06,
      "loss": 0.4989,
      "step": 356
    },
    {
      "epoch": 0.06,
      "grad_norm": 21.05545669655485,
      "learning_rate": 9.980330018456227e-06,
      "loss": 0.5536,
      "step": 357
    },
    {
      "epoch": 0.06,
      "grad_norm": 30.021464652348506,
      "learning_rate": 9.980098080377771e-06,
      "loss": 0.4864,
      "step": 358
    },
    {
      "epoch": 0.06,
      "grad_norm": 21.133571780039833,
      "learning_rate": 9.979864785583325e-06,
      "loss": 0.4633,
      "step": 359
    },
    {
      "epoch": 0.06,
      "grad_norm": 22.096160915361523,
      "learning_rate": 9.979630134136438e-06,
      "loss": 0.5231,
      "step": 360
    },
    {
      "epoch": 0.06,
      "grad_norm": 22.209467276723977,
      "learning_rate": 9.979394126101039e-06,
      "loss": 0.5541,
      "step": 361
    },
    {
      "epoch": 0.06,
      "grad_norm": 51.07370030596758,
      "learning_rate": 9.979156761541421e-06,
      "loss": 0.5364,
      "step": 362
    },
    {
      "epoch": 0.06,
      "grad_norm": 23.764647011805877,
      "learning_rate": 9.978918040522249e-06,
      "loss": 0.5244,
      "step": 363
    },
    {
      "epoch": 0.06,
      "grad_norm": 26.210163478876417,
      "learning_rate": 9.97867796310856e-06,
      "loss": 0.4787,
      "step": 364
    },
    {
      "epoch": 0.06,
      "grad_norm": 28.770515093485404,
      "learning_rate": 9.978436529365757e-06,
      "loss": 0.593,
      "step": 365
    },
    {
      "epoch": 0.06,
      "grad_norm": 34.681586124336654,
      "learning_rate": 9.978193739359611e-06,
      "loss": 0.4687,
      "step": 366
    },
    {
      "epoch": 0.06,
      "grad_norm": 23.057910576495072,
      "learning_rate": 9.977949593156264e-06,
      "loss": 0.4882,
      "step": 367
    },
    {
      "epoch": 0.06,
      "grad_norm": 55.649207961095534,
      "learning_rate": 9.977704090822232e-06,
      "loss": 0.5808,
      "step": 368
    },
    {
      "epoch": 0.06,
      "grad_norm": 16.620917117475674,
      "learning_rate": 9.977457232424394e-06,
      "loss": 0.4904,
      "step": 369
    },
    {
      "epoch": 0.06,
      "grad_norm": 24.19347034531309,
      "learning_rate": 9.97720901803e-06,
      "loss": 0.4757,
      "step": 370
    },
    {
      "epoch": 0.06,
      "grad_norm": 32.05019191077451,
      "learning_rate": 9.976959447706673e-06,
      "loss": 0.4838,
      "step": 371
    },
    {
      "epoch": 0.06,
      "grad_norm": 56.0939360157241,
      "learning_rate": 9.976708521522403e-06,
      "loss": 0.5237,
      "step": 372
    },
    {
      "epoch": 0.06,
      "grad_norm": 26.584647344869083,
      "learning_rate": 9.976456239545547e-06,
      "loss": 0.4829,
      "step": 373
    },
    {
      "epoch": 0.06,
      "grad_norm": 27.571158085912383,
      "learning_rate": 9.976202601844834e-06,
      "loss": 0.4916,
      "step": 374
    },
    {
      "epoch": 0.06,
      "grad_norm": 49.97319046051048,
      "learning_rate": 9.975947608489363e-06,
      "loss": 0.4987,
      "step": 375
    },
    {
      "epoch": 0.06,
      "grad_norm": 67.18675397070747,
      "learning_rate": 9.975691259548598e-06,
      "loss": 0.6057,
      "step": 376
    },
    {
      "epoch": 0.06,
      "grad_norm": 28.003749912193964,
      "learning_rate": 9.975433555092383e-06,
      "loss": 0.5284,
      "step": 377
    },
    {
      "epoch": 0.06,
      "grad_norm": 49.46711209658895,
      "learning_rate": 9.975174495190915e-06,
      "loss": 0.4835,
      "step": 378
    },
    {
      "epoch": 0.06,
      "grad_norm": 24.435425595843665,
      "learning_rate": 9.974914079914775e-06,
      "loss": 0.4766,
      "step": 379
    },
    {
      "epoch": 0.06,
      "grad_norm": 23.408063889696827,
      "learning_rate": 9.974652309334904e-06,
      "loss": 0.5403,
      "step": 380
    },
    {
      "epoch": 0.06,
      "grad_norm": 32.54455940449194,
      "learning_rate": 9.974389183522618e-06,
      "loss": 0.5166,
      "step": 381
    },
    {
      "epoch": 0.06,
      "grad_norm": 35.884292004176885,
      "learning_rate": 9.9741247025496e-06,
      "loss": 0.5509,
      "step": 382
    },
    {
      "epoch": 0.06,
      "grad_norm": 36.60357293360651,
      "learning_rate": 9.973858866487898e-06,
      "loss": 0.514,
      "step": 383
    },
    {
      "epoch": 0.06,
      "grad_norm": 48.61984072630844,
      "learning_rate": 9.973591675409934e-06,
      "loss": 0.542,
      "step": 384
    },
    {
      "epoch": 0.06,
      "grad_norm": 27.110096291137786,
      "learning_rate": 9.973323129388504e-06,
      "loss": 0.5647,
      "step": 385
    },
    {
      "epoch": 0.06,
      "grad_norm": 93.34157823078688,
      "learning_rate": 9.97305322849676e-06,
      "loss": 0.5082,
      "step": 386
    },
    {
      "epoch": 0.06,
      "grad_norm": 27.100647256232424,
      "learning_rate": 9.972781972808234e-06,
      "loss": 0.5348,
      "step": 387
    },
    {
      "epoch": 0.06,
      "grad_norm": 34.74180137614563,
      "learning_rate": 9.972509362396825e-06,
      "loss": 0.47,
      "step": 388
    },
    {
      "epoch": 0.06,
      "grad_norm": 32.81135544139614,
      "learning_rate": 9.972235397336796e-06,
      "loss": 0.4373,
      "step": 389
    },
    {
      "epoch": 0.06,
      "grad_norm": 50.96121732700845,
      "learning_rate": 9.971960077702785e-06,
      "loss": 0.5756,
      "step": 390
    },
    {
      "epoch": 0.06,
      "grad_norm": 24.153800490986022,
      "learning_rate": 9.971683403569795e-06,
      "loss": 0.5628,
      "step": 391
    },
    {
      "epoch": 0.06,
      "grad_norm": 22.339641623232918,
      "learning_rate": 9.971405375013201e-06,
      "loss": 0.4971,
      "step": 392
    },
    {
      "epoch": 0.06,
      "grad_norm": 38.07131761187898,
      "learning_rate": 9.971125992108746e-06,
      "loss": 0.4576,
      "step": 393
    },
    {
      "epoch": 0.06,
      "grad_norm": 29.705080223259554,
      "learning_rate": 9.97084525493254e-06,
      "loss": 0.4428,
      "step": 394
    },
    {
      "epoch": 0.06,
      "grad_norm": 32.76539459169689,
      "learning_rate": 9.970563163561064e-06,
      "loss": 0.5592,
      "step": 395
    },
    {
      "epoch": 0.06,
      "grad_norm": 32.92321553073833,
      "learning_rate": 9.970279718071169e-06,
      "loss": 0.5312,
      "step": 396
    },
    {
      "epoch": 0.06,
      "grad_norm": 25.99961677338992,
      "learning_rate": 9.969994918540071e-06,
      "loss": 0.5416,
      "step": 397
    },
    {
      "epoch": 0.06,
      "grad_norm": 23.912075970658957,
      "learning_rate": 9.969708765045361e-06,
      "loss": 0.4807,
      "step": 398
    },
    {
      "epoch": 0.06,
      "grad_norm": 32.496952791203256,
      "learning_rate": 9.96942125766499e-06,
      "loss": 0.5429,
      "step": 399
    },
    {
      "epoch": 0.06,
      "grad_norm": 34.37941225536387,
      "learning_rate": 9.969132396477286e-06,
      "loss": 0.5443,
      "step": 400
    },
    {
      "epoch": 0.06,
      "grad_norm": 21.76009615829801,
      "learning_rate": 9.968842181560943e-06,
      "loss": 0.4996,
      "step": 401
    },
    {
      "epoch": 0.06,
      "grad_norm": 30.417509061826994,
      "learning_rate": 9.968550612995023e-06,
      "loss": 0.4605,
      "step": 402
    },
    {
      "epoch": 0.06,
      "grad_norm": 25.11449780546829,
      "learning_rate": 9.968257690858955e-06,
      "loss": 0.517,
      "step": 403
    },
    {
      "epoch": 0.07,
      "grad_norm": 28.43934247717701,
      "learning_rate": 9.967963415232544e-06,
      "loss": 0.591,
      "step": 404
    },
    {
      "epoch": 0.07,
      "grad_norm": 33.299709043872284,
      "learning_rate": 9.967667786195955e-06,
      "loss": 0.4226,
      "step": 405
    },
    {
      "epoch": 0.07,
      "grad_norm": 30.647604254977566,
      "learning_rate": 9.967370803829725e-06,
      "loss": 0.4932,
      "step": 406
    },
    {
      "epoch": 0.07,
      "grad_norm": 29.23689961238255,
      "learning_rate": 9.967072468214763e-06,
      "loss": 0.5449,
      "step": 407
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.152256864473235,
      "learning_rate": 9.966772779432343e-06,
      "loss": 0.4798,
      "step": 408
    },
    {
      "epoch": 0.07,
      "grad_norm": 25.93754312059422,
      "learning_rate": 9.966471737564107e-06,
      "loss": 0.4526,
      "step": 409
    },
    {
      "epoch": 0.07,
      "grad_norm": 31.5058520624137,
      "learning_rate": 9.96616934269207e-06,
      "loss": 0.5186,
      "step": 410
    },
    {
      "epoch": 0.07,
      "grad_norm": 33.40071467600732,
      "learning_rate": 9.965865594898608e-06,
      "loss": 0.5213,
      "step": 411
    },
    {
      "epoch": 0.07,
      "grad_norm": 45.87345368097057,
      "learning_rate": 9.965560494266475e-06,
      "loss": 0.6058,
      "step": 412
    },
    {
      "epoch": 0.07,
      "grad_norm": 17.9022278175797,
      "learning_rate": 9.965254040878786e-06,
      "loss": 0.5054,
      "step": 413
    },
    {
      "epoch": 0.07,
      "grad_norm": 19.032318318820096,
      "learning_rate": 9.964946234819028e-06,
      "loss": 0.5522,
      "step": 414
    },
    {
      "epoch": 0.07,
      "grad_norm": 32.094765870079655,
      "learning_rate": 9.964637076171056e-06,
      "loss": 0.5628,
      "step": 415
    },
    {
      "epoch": 0.07,
      "grad_norm": 26.350475094482974,
      "learning_rate": 9.964326565019094e-06,
      "loss": 0.5878,
      "step": 416
    },
    {
      "epoch": 0.07,
      "grad_norm": 39.22819349584317,
      "learning_rate": 9.964014701447733e-06,
      "loss": 0.4956,
      "step": 417
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8518828840626311,
      "learning_rate": 9.963701485541935e-06,
      "loss": 0.5035,
      "step": 418
    },
    {
      "epoch": 0.07,
      "grad_norm": 24.044781875480382,
      "learning_rate": 9.963386917387025e-06,
      "loss": 0.5318,
      "step": 419
    },
    {
      "epoch": 0.07,
      "grad_norm": 37.146033781379785,
      "learning_rate": 9.963070997068702e-06,
      "loss": 0.5467,
      "step": 420
    },
    {
      "epoch": 0.07,
      "grad_norm": 31.711666229701752,
      "learning_rate": 9.962753724673033e-06,
      "loss": 0.4822,
      "step": 421
    },
    {
      "epoch": 0.07,
      "grad_norm": 51.637175864837644,
      "learning_rate": 9.96243510028645e-06,
      "loss": 0.5468,
      "step": 422
    },
    {
      "epoch": 0.07,
      "grad_norm": 32.61322001761153,
      "learning_rate": 9.962115123995754e-06,
      "loss": 0.4857,
      "step": 423
    },
    {
      "epoch": 0.07,
      "grad_norm": 23.183350056129182,
      "learning_rate": 9.961793795888118e-06,
      "loss": 0.5071,
      "step": 424
    },
    {
      "epoch": 0.07,
      "grad_norm": 17.49179420014295,
      "learning_rate": 9.96147111605108e-06,
      "loss": 0.506,
      "step": 425
    },
    {
      "epoch": 0.07,
      "grad_norm": 25.104310791925805,
      "learning_rate": 9.961147084572544e-06,
      "loss": 0.5302,
      "step": 426
    },
    {
      "epoch": 0.07,
      "grad_norm": 22.80102658283887,
      "learning_rate": 9.96082170154079e-06,
      "loss": 0.5345,
      "step": 427
    },
    {
      "epoch": 0.07,
      "grad_norm": 25.702194501266998,
      "learning_rate": 9.960494967044457e-06,
      "loss": 0.5415,
      "step": 428
    },
    {
      "epoch": 0.07,
      "grad_norm": 32.649951165165845,
      "learning_rate": 9.960166881172558e-06,
      "loss": 0.473,
      "step": 429
    },
    {
      "epoch": 0.07,
      "grad_norm": 21.075870770267354,
      "learning_rate": 9.959837444014473e-06,
      "loss": 0.4941,
      "step": 430
    },
    {
      "epoch": 0.07,
      "grad_norm": 24.642503203066365,
      "learning_rate": 9.95950665565995e-06,
      "loss": 0.5225,
      "step": 431
    },
    {
      "epoch": 0.07,
      "grad_norm": 20.451268840433812,
      "learning_rate": 9.959174516199105e-06,
      "loss": 0.4903,
      "step": 432
    },
    {
      "epoch": 0.07,
      "grad_norm": 41.82872156073963,
      "learning_rate": 9.95884102572242e-06,
      "loss": 0.5129,
      "step": 433
    },
    {
      "epoch": 0.07,
      "grad_norm": 15.76760743144387,
      "learning_rate": 9.958506184320749e-06,
      "loss": 0.4902,
      "step": 434
    },
    {
      "epoch": 0.07,
      "grad_norm": 35.96038019174613,
      "learning_rate": 9.95816999208531e-06,
      "loss": 0.5312,
      "step": 435
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2998482128820028,
      "learning_rate": 9.957832449107694e-06,
      "loss": 0.4963,
      "step": 436
    },
    {
      "epoch": 0.07,
      "grad_norm": 24.39134148388489,
      "learning_rate": 9.957493555479856e-06,
      "loss": 0.4961,
      "step": 437
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.886915339037717,
      "learning_rate": 9.957153311294119e-06,
      "loss": 0.5326,
      "step": 438
    },
    {
      "epoch": 0.07,
      "grad_norm": 17.96802745745171,
      "learning_rate": 9.956811716643173e-06,
      "loss": 0.4957,
      "step": 439
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.993686561358084,
      "learning_rate": 9.956468771620082e-06,
      "loss": 0.4385,
      "step": 440
    },
    {
      "epoch": 0.07,
      "grad_norm": 18.14101545828282,
      "learning_rate": 9.956124476318271e-06,
      "loss": 0.4956,
      "step": 441
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.773021169052909,
      "learning_rate": 9.955778830831537e-06,
      "loss": 0.4576,
      "step": 442
    },
    {
      "epoch": 0.07,
      "grad_norm": 16.97778938081952,
      "learning_rate": 9.955431835254044e-06,
      "loss": 0.5322,
      "step": 443
    },
    {
      "epoch": 0.07,
      "grad_norm": 17.945285245376137,
      "learning_rate": 9.95508348968032e-06,
      "loss": 0.5379,
      "step": 444
    },
    {
      "epoch": 0.07,
      "grad_norm": 38.47737829536334,
      "learning_rate": 9.954733794205264e-06,
      "loss": 0.5317,
      "step": 445
    },
    {
      "epoch": 0.07,
      "grad_norm": 24.438213751992887,
      "learning_rate": 9.954382748924148e-06,
      "loss": 0.499,
      "step": 446
    },
    {
      "epoch": 0.07,
      "grad_norm": 15.170834029802196,
      "learning_rate": 9.9540303539326e-06,
      "loss": 0.5723,
      "step": 447
    },
    {
      "epoch": 0.07,
      "grad_norm": 33.862426250772906,
      "learning_rate": 9.953676609326627e-06,
      "loss": 0.4964,
      "step": 448
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.112890994802129,
      "learning_rate": 9.953321515202597e-06,
      "loss": 0.4915,
      "step": 449
    },
    {
      "epoch": 0.07,
      "grad_norm": 32.85664456727147,
      "learning_rate": 9.952965071657244e-06,
      "loss": 0.5366,
      "step": 450
    },
    {
      "epoch": 0.07,
      "grad_norm": 21.9750949204955,
      "learning_rate": 9.952607278787679e-06,
      "loss": 0.4266,
      "step": 451
    },
    {
      "epoch": 0.07,
      "grad_norm": 74.81021995694343,
      "learning_rate": 9.95224813669137e-06,
      "loss": 0.4706,
      "step": 452
    },
    {
      "epoch": 0.07,
      "grad_norm": 45.53066585967331,
      "learning_rate": 9.95188764546616e-06,
      "loss": 0.5144,
      "step": 453
    },
    {
      "epoch": 0.07,
      "grad_norm": 36.2506933342848,
      "learning_rate": 9.951525805210256e-06,
      "loss": 0.5315,
      "step": 454
    },
    {
      "epoch": 0.07,
      "grad_norm": 34.822025071845225,
      "learning_rate": 9.951162616022234e-06,
      "loss": 0.5531,
      "step": 455
    },
    {
      "epoch": 0.07,
      "grad_norm": 28.245875605121736,
      "learning_rate": 9.950798078001034e-06,
      "loss": 0.5798,
      "step": 456
    },
    {
      "epoch": 0.07,
      "grad_norm": 16.650132236007526,
      "learning_rate": 9.950432191245968e-06,
      "loss": 0.5383,
      "step": 457
    },
    {
      "epoch": 0.07,
      "grad_norm": 17.354218528978308,
      "learning_rate": 9.950064955856716e-06,
      "loss": 0.5129,
      "step": 458
    },
    {
      "epoch": 0.07,
      "grad_norm": 15.983261598181162,
      "learning_rate": 9.949696371933319e-06,
      "loss": 0.4969,
      "step": 459
    },
    {
      "epoch": 0.07,
      "grad_norm": 19.455593519089483,
      "learning_rate": 9.94932643957619e-06,
      "loss": 0.4953,
      "step": 460
    },
    {
      "epoch": 0.07,
      "grad_norm": 20.147914724490157,
      "learning_rate": 9.948955158886113e-06,
      "loss": 0.5159,
      "step": 461
    },
    {
      "epoch": 0.07,
      "grad_norm": 26.59561655932231,
      "learning_rate": 9.94858252996423e-06,
      "loss": 0.4919,
      "step": 462
    },
    {
      "epoch": 0.07,
      "grad_norm": 18.080481206147777,
      "learning_rate": 9.948208552912057e-06,
      "loss": 0.4196,
      "step": 463
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.151717867852996,
      "learning_rate": 9.947833227831477e-06,
      "loss": 0.5,
      "step": 464
    },
    {
      "epoch": 0.07,
      "grad_norm": 81.61403177628088,
      "learning_rate": 9.947456554824736e-06,
      "loss": 0.5375,
      "step": 465
    },
    {
      "epoch": 0.08,
      "grad_norm": 16.824867867931683,
      "learning_rate": 9.947078533994454e-06,
      "loss": 0.4652,
      "step": 466
    },
    {
      "epoch": 0.08,
      "grad_norm": 15.050256485325868,
      "learning_rate": 9.94669916544361e-06,
      "loss": 0.5241,
      "step": 467
    },
    {
      "epoch": 0.08,
      "grad_norm": 63.02157767426345,
      "learning_rate": 9.94631844927556e-06,
      "loss": 0.514,
      "step": 468
    },
    {
      "epoch": 0.08,
      "grad_norm": 68.19499595231524,
      "learning_rate": 9.945936385594017e-06,
      "loss": 0.4761,
      "step": 469
    },
    {
      "epoch": 0.08,
      "grad_norm": 27.273817847327035,
      "learning_rate": 9.945552974503065e-06,
      "loss": 0.5537,
      "step": 470
    },
    {
      "epoch": 0.08,
      "grad_norm": 23.9123170228459,
      "learning_rate": 9.94516821610716e-06,
      "loss": 0.5655,
      "step": 471
    },
    {
      "epoch": 0.08,
      "grad_norm": 21.401173626886173,
      "learning_rate": 9.944782110511119e-06,
      "loss": 0.5059,
      "step": 472
    },
    {
      "epoch": 0.08,
      "grad_norm": 61.47661594360003,
      "learning_rate": 9.944394657820127e-06,
      "loss": 0.4985,
      "step": 473
    },
    {
      "epoch": 0.08,
      "grad_norm": 37.26648989157921,
      "learning_rate": 9.944005858139737e-06,
      "loss": 0.4723,
      "step": 474
    },
    {
      "epoch": 0.08,
      "grad_norm": 25.466351867268152,
      "learning_rate": 9.943615711575867e-06,
      "loss": 0.528,
      "step": 475
    },
    {
      "epoch": 0.08,
      "grad_norm": 21.341293850335937,
      "learning_rate": 9.943224218234809e-06,
      "loss": 0.5433,
      "step": 476
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.202135558227354,
      "learning_rate": 9.94283137822321e-06,
      "loss": 0.4733,
      "step": 477
    },
    {
      "epoch": 0.08,
      "grad_norm": 24.972381702990607,
      "learning_rate": 9.942437191648095e-06,
      "loss": 0.5491,
      "step": 478
    },
    {
      "epoch": 0.08,
      "grad_norm": 22.390358102787957,
      "learning_rate": 9.94204165861685e-06,
      "loss": 0.496,
      "step": 479
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.404522545439457,
      "learning_rate": 9.94164477923723e-06,
      "loss": 0.5141,
      "step": 480
    },
    {
      "epoch": 0.08,
      "grad_norm": 15.025411452689683,
      "learning_rate": 9.941246553617353e-06,
      "loss": 0.4987,
      "step": 481
    },
    {
      "epoch": 0.08,
      "grad_norm": 69.67307670682064,
      "learning_rate": 9.940846981865709e-06,
      "loss": 0.5341,
      "step": 482
    },
    {
      "epoch": 0.08,
      "grad_norm": 20.15731549368307,
      "learning_rate": 9.940446064091151e-06,
      "loss": 0.589,
      "step": 483
    },
    {
      "epoch": 0.08,
      "grad_norm": 21.460104371931862,
      "learning_rate": 9.940043800402903e-06,
      "loss": 0.5487,
      "step": 484
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.176050738192115,
      "learning_rate": 9.939640190910552e-06,
      "loss": 0.5719,
      "step": 485
    },
    {
      "epoch": 0.08,
      "grad_norm": 21.164347680110982,
      "learning_rate": 9.939235235724047e-06,
      "loss": 0.5964,
      "step": 486
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.430828246057462,
      "learning_rate": 9.938828934953716e-06,
      "loss": 0.4923,
      "step": 487
    },
    {
      "epoch": 0.08,
      "grad_norm": 18.61747113368015,
      "learning_rate": 9.938421288710242e-06,
      "loss": 0.5749,
      "step": 488
    },
    {
      "epoch": 0.08,
      "grad_norm": 22.535523152962813,
      "learning_rate": 9.93801229710468e-06,
      "loss": 0.5214,
      "step": 489
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1597154503489981,
      "learning_rate": 9.937601960248452e-06,
      "loss": 0.4547,
      "step": 490
    },
    {
      "epoch": 0.08,
      "grad_norm": 57.693816484905334,
      "learning_rate": 9.937190278253345e-06,
      "loss": 0.5041,
      "step": 491
    },
    {
      "epoch": 0.08,
      "grad_norm": 13.484892748239167,
      "learning_rate": 9.93677725123151e-06,
      "loss": 0.5483,
      "step": 492
    },
    {
      "epoch": 0.08,
      "grad_norm": 13.583313174375064,
      "learning_rate": 9.936362879295471e-06,
      "loss": 0.5042,
      "step": 493
    },
    {
      "epoch": 0.08,
      "grad_norm": 22.288730700915963,
      "learning_rate": 9.93594716255811e-06,
      "loss": 0.5434,
      "step": 494
    },
    {
      "epoch": 0.08,
      "grad_norm": 35.04838780673886,
      "learning_rate": 9.935530101132683e-06,
      "loss": 0.507,
      "step": 495
    },
    {
      "epoch": 0.08,
      "grad_norm": 20.625097478047156,
      "learning_rate": 9.935111695132807e-06,
      "loss": 0.5119,
      "step": 496
    },
    {
      "epoch": 0.08,
      "grad_norm": 86.06254025668136,
      "learning_rate": 9.934691944672468e-06,
      "loss": 0.4172,
      "step": 497
    },
    {
      "epoch": 0.08,
      "grad_norm": 25.48809362354762,
      "learning_rate": 9.934270849866019e-06,
      "loss": 0.4966,
      "step": 498
    },
    {
      "epoch": 0.08,
      "grad_norm": 23.410966011072528,
      "learning_rate": 9.933848410828175e-06,
      "loss": 0.4952,
      "step": 499
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.350562452813703,
      "learning_rate": 9.933424627674022e-06,
      "loss": 0.554,
      "step": 500
    },
    {
      "epoch": 0.08,
      "grad_norm": 13.027639221252468,
      "learning_rate": 9.932999500519011e-06,
      "loss": 0.4475,
      "step": 501
    },
    {
      "epoch": 0.08,
      "grad_norm": 19.041522097836694,
      "learning_rate": 9.932573029478956e-06,
      "loss": 0.4994,
      "step": 502
    },
    {
      "epoch": 0.08,
      "grad_norm": 16.027027086172566,
      "learning_rate": 9.932145214670042e-06,
      "loss": 0.5044,
      "step": 503
    },
    {
      "epoch": 0.08,
      "grad_norm": 64.66387108030902,
      "learning_rate": 9.931716056208814e-06,
      "loss": 0.5528,
      "step": 504
    },
    {
      "epoch": 0.08,
      "grad_norm": 24.000625605576143,
      "learning_rate": 9.931285554212189e-06,
      "loss": 0.4812,
      "step": 505
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.960465458234653,
      "learning_rate": 9.930853708797448e-06,
      "loss": 0.546,
      "step": 506
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.035419882416011,
      "learning_rate": 9.930420520082238e-06,
      "loss": 0.4314,
      "step": 507
    },
    {
      "epoch": 0.08,
      "grad_norm": 32.5275889505669,
      "learning_rate": 9.929985988184568e-06,
      "loss": 0.5419,
      "step": 508
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.00641954233929,
      "learning_rate": 9.929550113222822e-06,
      "loss": 0.5335,
      "step": 509
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.146765223423792,
      "learning_rate": 9.929112895315739e-06,
      "loss": 0.4833,
      "step": 510
    },
    {
      "epoch": 0.08,
      "grad_norm": 17.155008550667898,
      "learning_rate": 9.928674334582431e-06,
      "loss": 0.5453,
      "step": 511
    },
    {
      "epoch": 0.08,
      "grad_norm": 13.954164077546192,
      "learning_rate": 9.928234431142376e-06,
      "loss": 0.4691,
      "step": 512
    },
    {
      "epoch": 0.08,
      "grad_norm": 13.152318314094025,
      "learning_rate": 9.927793185115416e-06,
      "loss": 0.5193,
      "step": 513
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2858546313823787,
      "learning_rate": 9.927350596621753e-06,
      "loss": 0.4962,
      "step": 514
    },
    {
      "epoch": 0.08,
      "grad_norm": 19.581461298857505,
      "learning_rate": 9.926906665781967e-06,
      "loss": 0.5338,
      "step": 515
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.46761132147807,
      "learning_rate": 9.926461392716993e-06,
      "loss": 0.5111,
      "step": 516
    },
    {
      "epoch": 0.08,
      "grad_norm": 23.159004436177707,
      "learning_rate": 9.926014777548139e-06,
      "loss": 0.5462,
      "step": 517
    },
    {
      "epoch": 0.08,
      "grad_norm": 13.498808288714008,
      "learning_rate": 9.92556682039707e-06,
      "loss": 0.4943,
      "step": 518
    },
    {
      "epoch": 0.08,
      "grad_norm": 15.320359291926477,
      "learning_rate": 9.925117521385827e-06,
      "loss": 0.5232,
      "step": 519
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.714600184919549,
      "learning_rate": 9.92466688063681e-06,
      "loss": 0.4788,
      "step": 520
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2266657596263884,
      "learning_rate": 9.924214898272783e-06,
      "loss": 0.4692,
      "step": 521
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.280608901628383,
      "learning_rate": 9.923761574416884e-06,
      "loss": 0.5105,
      "step": 522
    },
    {
      "epoch": 0.08,
      "grad_norm": 13.36639863504587,
      "learning_rate": 9.923306909192608e-06,
      "loss": 0.4934,
      "step": 523
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2740002298563655,
      "learning_rate": 9.922850902723815e-06,
      "loss": 0.4731,
      "step": 524
    },
    {
      "epoch": 0.08,
      "grad_norm": 15.338048540651087,
      "learning_rate": 9.922393555134739e-06,
      "loss": 0.5108,
      "step": 525
    },
    {
      "epoch": 0.08,
      "grad_norm": 15.656241275884797,
      "learning_rate": 9.92193486654997e-06,
      "loss": 0.4961,
      "step": 526
    },
    {
      "epoch": 0.08,
      "grad_norm": 16.13458761927888,
      "learning_rate": 9.921474837094473e-06,
      "loss": 0.5769,
      "step": 527
    },
    {
      "epoch": 0.09,
      "grad_norm": 21.537727390519827,
      "learning_rate": 9.921013466893567e-06,
      "loss": 0.4841,
      "step": 528
    },
    {
      "epoch": 0.09,
      "grad_norm": 13.895990491642683,
      "learning_rate": 9.920550756072945e-06,
      "loss": 0.5524,
      "step": 529
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.921325160411415,
      "learning_rate": 9.92008670475866e-06,
      "loss": 0.497,
      "step": 530
    },
    {
      "epoch": 0.09,
      "grad_norm": 15.56181054602899,
      "learning_rate": 9.919621313077135e-06,
      "loss": 0.5261,
      "step": 531
    },
    {
      "epoch": 0.09,
      "grad_norm": 20.49411978735034,
      "learning_rate": 9.919154581155152e-06,
      "loss": 0.4794,
      "step": 532
    },
    {
      "epoch": 0.09,
      "grad_norm": 18.480975845296932,
      "learning_rate": 9.918686509119867e-06,
      "loss": 0.5356,
      "step": 533
    },
    {
      "epoch": 0.09,
      "grad_norm": 13.073752984602349,
      "learning_rate": 9.91821709709879e-06,
      "loss": 0.552,
      "step": 534
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.350033040569631,
      "learning_rate": 9.917746345219805e-06,
      "loss": 0.5218,
      "step": 535
    },
    {
      "epoch": 0.09,
      "grad_norm": 19.471724327339587,
      "learning_rate": 9.917274253611156e-06,
      "loss": 0.4993,
      "step": 536
    },
    {
      "epoch": 0.09,
      "grad_norm": 18.51573328978897,
      "learning_rate": 9.916800822401457e-06,
      "loss": 0.4766,
      "step": 537
    },
    {
      "epoch": 0.09,
      "grad_norm": 15.081392424003008,
      "learning_rate": 9.91632605171968e-06,
      "loss": 0.5776,
      "step": 538
    },
    {
      "epoch": 0.09,
      "grad_norm": 15.624456193197261,
      "learning_rate": 9.915849941695167e-06,
      "loss": 0.4756,
      "step": 539
    },
    {
      "epoch": 0.09,
      "grad_norm": 25.571778920649678,
      "learning_rate": 9.915372492457624e-06,
      "loss": 0.5568,
      "step": 540
    },
    {
      "epoch": 0.09,
      "grad_norm": 28.210014449808007,
      "learning_rate": 9.914893704137123e-06,
      "loss": 0.4193,
      "step": 541
    },
    {
      "epoch": 0.09,
      "grad_norm": 21.436324578551755,
      "learning_rate": 9.914413576864094e-06,
      "loss": 0.5066,
      "step": 542
    },
    {
      "epoch": 0.09,
      "grad_norm": 62.02139390483134,
      "learning_rate": 9.913932110769342e-06,
      "loss": 0.4802,
      "step": 543
    },
    {
      "epoch": 0.09,
      "grad_norm": 21.891837004744193,
      "learning_rate": 9.91344930598403e-06,
      "loss": 0.4749,
      "step": 544
    },
    {
      "epoch": 0.09,
      "grad_norm": 16.37938273771956,
      "learning_rate": 9.912965162639686e-06,
      "loss": 0.5342,
      "step": 545
    },
    {
      "epoch": 0.09,
      "grad_norm": 13.084183962994475,
      "learning_rate": 9.912479680868205e-06,
      "loss": 0.527,
      "step": 546
    },
    {
      "epoch": 0.09,
      "grad_norm": 19.76839280081624,
      "learning_rate": 9.911992860801846e-06,
      "loss": 0.5004,
      "step": 547
    },
    {
      "epoch": 0.09,
      "grad_norm": 13.805090440603099,
      "learning_rate": 9.911504702573232e-06,
      "loss": 0.5424,
      "step": 548
    },
    {
      "epoch": 0.09,
      "grad_norm": 13.46879663480542,
      "learning_rate": 9.91101520631535e-06,
      "loss": 0.4667,
      "step": 549
    },
    {
      "epoch": 0.09,
      "grad_norm": 24.261619553669544,
      "learning_rate": 9.910524372161554e-06,
      "loss": 0.5283,
      "step": 550
    },
    {
      "epoch": 0.09,
      "grad_norm": 17.686624891732816,
      "learning_rate": 9.910032200245559e-06,
      "loss": 0.4712,
      "step": 551
    },
    {
      "epoch": 0.09,
      "grad_norm": 24.0586317736002,
      "learning_rate": 9.909538690701448e-06,
      "loss": 0.5809,
      "step": 552
    },
    {
      "epoch": 0.09,
      "grad_norm": 15.201467138321378,
      "learning_rate": 9.909043843663664e-06,
      "loss": 0.4624,
      "step": 553
    },
    {
      "epoch": 0.09,
      "grad_norm": 21.430824461404985,
      "learning_rate": 9.908547659267019e-06,
      "loss": 0.5516,
      "step": 554
    },
    {
      "epoch": 0.09,
      "grad_norm": 21.342766524121007,
      "learning_rate": 9.908050137646685e-06,
      "loss": 0.4976,
      "step": 555
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.304421935314497,
      "learning_rate": 9.907551278938203e-06,
      "loss": 0.5097,
      "step": 556
    },
    {
      "epoch": 0.09,
      "grad_norm": 25.06287670697131,
      "learning_rate": 9.907051083277477e-06,
      "loss": 0.532,
      "step": 557
    },
    {
      "epoch": 0.09,
      "grad_norm": 19.133356596383926,
      "learning_rate": 9.90654955080077e-06,
      "loss": 0.4935,
      "step": 558
    },
    {
      "epoch": 0.09,
      "grad_norm": 15.586515916242897,
      "learning_rate": 9.906046681644717e-06,
      "loss": 0.551,
      "step": 559
    },
    {
      "epoch": 0.09,
      "grad_norm": 18.851700225091328,
      "learning_rate": 9.905542475946312e-06,
      "loss": 0.4755,
      "step": 560
    },
    {
      "epoch": 0.09,
      "grad_norm": 25.686780346664317,
      "learning_rate": 9.905036933842914e-06,
      "loss": 0.4676,
      "step": 561
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.444961722378883,
      "learning_rate": 9.904530055472246e-06,
      "loss": 0.4954,
      "step": 562
    },
    {
      "epoch": 0.09,
      "grad_norm": 16.291664276457745,
      "learning_rate": 9.904021840972397e-06,
      "loss": 0.5171,
      "step": 563
    },
    {
      "epoch": 0.09,
      "grad_norm": 17.460842707843682,
      "learning_rate": 9.90351229048182e-06,
      "loss": 0.556,
      "step": 564
    },
    {
      "epoch": 0.09,
      "grad_norm": 51.26147783962857,
      "learning_rate": 9.903001404139329e-06,
      "loss": 0.5271,
      "step": 565
    },
    {
      "epoch": 0.09,
      "grad_norm": 20.873330973764407,
      "learning_rate": 9.902489182084101e-06,
      "loss": 0.519,
      "step": 566
    },
    {
      "epoch": 0.09,
      "grad_norm": 15.087833043249425,
      "learning_rate": 9.901975624455685e-06,
      "loss": 0.5394,
      "step": 567
    },
    {
      "epoch": 0.09,
      "grad_norm": 17.038618637013016,
      "learning_rate": 9.901460731393984e-06,
      "loss": 0.469,
      "step": 568
    },
    {
      "epoch": 0.09,
      "grad_norm": 23.25601760477705,
      "learning_rate": 9.90094450303927e-06,
      "loss": 0.4479,
      "step": 569
    },
    {
      "epoch": 0.09,
      "grad_norm": 24.280666091088996,
      "learning_rate": 9.900426939532178e-06,
      "loss": 0.4584,
      "step": 570
    },
    {
      "epoch": 0.09,
      "grad_norm": 38.76345732975971,
      "learning_rate": 9.899908041013706e-06,
      "loss": 0.56,
      "step": 571
    },
    {
      "epoch": 0.09,
      "grad_norm": 15.615505893819986,
      "learning_rate": 9.899387807625217e-06,
      "loss": 0.4932,
      "step": 572
    },
    {
      "epoch": 0.09,
      "grad_norm": 18.937921332598645,
      "learning_rate": 9.898866239508437e-06,
      "loss": 0.5296,
      "step": 573
    },
    {
      "epoch": 0.09,
      "grad_norm": 19.36456621559494,
      "learning_rate": 9.898343336805456e-06,
      "loss": 0.5151,
      "step": 574
    },
    {
      "epoch": 0.09,
      "grad_norm": 25.640873109264188,
      "learning_rate": 9.897819099658726e-06,
      "loss": 0.5045,
      "step": 575
    },
    {
      "epoch": 0.09,
      "grad_norm": 17.443202463664413,
      "learning_rate": 9.897293528211062e-06,
      "loss": 0.4804,
      "step": 576
    },
    {
      "epoch": 0.09,
      "grad_norm": 19.67824539218894,
      "learning_rate": 9.896766622605649e-06,
      "loss": 0.5653,
      "step": 577
    },
    {
      "epoch": 0.09,
      "grad_norm": 19.553427435820534,
      "learning_rate": 9.896238382986028e-06,
      "loss": 0.564,
      "step": 578
    },
    {
      "epoch": 0.09,
      "grad_norm": 13.010337660887183,
      "learning_rate": 9.895708809496106e-06,
      "loss": 0.5158,
      "step": 579
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3807963752760515,
      "learning_rate": 9.895177902280151e-06,
      "loss": 0.4787,
      "step": 580
    },
    {
      "epoch": 0.09,
      "grad_norm": 17.235160096487327,
      "learning_rate": 9.8946456614828e-06,
      "loss": 0.5379,
      "step": 581
    },
    {
      "epoch": 0.09,
      "grad_norm": 15.332992515366278,
      "learning_rate": 9.894112087249049e-06,
      "loss": 0.508,
      "step": 582
    },
    {
      "epoch": 0.09,
      "grad_norm": 28.486584242033622,
      "learning_rate": 9.893577179724259e-06,
      "loss": 0.47,
      "step": 583
    },
    {
      "epoch": 0.09,
      "grad_norm": 38.77331446611641,
      "learning_rate": 9.893040939054152e-06,
      "loss": 0.5041,
      "step": 584
    },
    {
      "epoch": 0.09,
      "grad_norm": 21.207054655645923,
      "learning_rate": 9.892503365384818e-06,
      "loss": 0.4968,
      "step": 585
    },
    {
      "epoch": 0.09,
      "grad_norm": 22.202255991886688,
      "learning_rate": 9.891964458862702e-06,
      "loss": 0.5485,
      "step": 586
    },
    {
      "epoch": 0.09,
      "grad_norm": 13.730282123072067,
      "learning_rate": 9.891424219634622e-06,
      "loss": 0.497,
      "step": 587
    },
    {
      "epoch": 0.09,
      "grad_norm": 14.780413534325842,
      "learning_rate": 9.89088264784775e-06,
      "loss": 0.413,
      "step": 588
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.560090180842033,
      "learning_rate": 9.890339743649628e-06,
      "loss": 0.5148,
      "step": 589
    },
    {
      "epoch": 0.1,
      "grad_norm": 57.951380138492446,
      "learning_rate": 9.889795507188156e-06,
      "loss": 0.513,
      "step": 590
    },
    {
      "epoch": 0.1,
      "grad_norm": 23.839207559460178,
      "learning_rate": 9.889249938611599e-06,
      "loss": 0.542,
      "step": 591
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3100884104163806,
      "learning_rate": 9.888703038068587e-06,
      "loss": 0.4517,
      "step": 592
    },
    {
      "epoch": 0.1,
      "grad_norm": 22.611797131759342,
      "learning_rate": 9.88815480570811e-06,
      "loss": 0.5044,
      "step": 593
    },
    {
      "epoch": 0.1,
      "grad_norm": 17.975168232540277,
      "learning_rate": 9.887605241679518e-06,
      "loss": 0.5426,
      "step": 594
    },
    {
      "epoch": 0.1,
      "grad_norm": 21.6299249713566,
      "learning_rate": 9.887054346132534e-06,
      "loss": 0.606,
      "step": 595
    },
    {
      "epoch": 0.1,
      "grad_norm": 21.433210431466645,
      "learning_rate": 9.886502119217232e-06,
      "loss": 0.4522,
      "step": 596
    },
    {
      "epoch": 0.1,
      "grad_norm": 16.8744986725075,
      "learning_rate": 9.885948561084056e-06,
      "loss": 0.5388,
      "step": 597
    },
    {
      "epoch": 0.1,
      "grad_norm": 17.774643117427285,
      "learning_rate": 9.885393671883811e-06,
      "loss": 0.4213,
      "step": 598
    },
    {
      "epoch": 0.1,
      "grad_norm": 27.177227386358386,
      "learning_rate": 9.884837451767662e-06,
      "loss": 0.4277,
      "step": 599
    },
    {
      "epoch": 0.1,
      "grad_norm": 51.94739877270996,
      "learning_rate": 9.88427990088714e-06,
      "loss": 0.5516,
      "step": 600
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.48021643779853,
      "learning_rate": 9.883721019394136e-06,
      "loss": 0.468,
      "step": 601
    },
    {
      "epoch": 0.1,
      "grad_norm": 16.97300760191231,
      "learning_rate": 9.883160807440907e-06,
      "loss": 0.4785,
      "step": 602
    },
    {
      "epoch": 0.1,
      "grad_norm": 19.219778185051,
      "learning_rate": 9.88259926518007e-06,
      "loss": 0.4626,
      "step": 603
    },
    {
      "epoch": 0.1,
      "grad_norm": 19.837493866912144,
      "learning_rate": 9.882036392764602e-06,
      "loss": 0.538,
      "step": 604
    },
    {
      "epoch": 0.1,
      "grad_norm": 13.111250667867075,
      "learning_rate": 9.881472190347846e-06,
      "loss": 0.5101,
      "step": 605
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.154732540806226,
      "learning_rate": 9.880906658083507e-06,
      "loss": 0.4959,
      "step": 606
    },
    {
      "epoch": 0.1,
      "grad_norm": 18.974950073827124,
      "learning_rate": 9.880339796125653e-06,
      "loss": 0.4927,
      "step": 607
    },
    {
      "epoch": 0.1,
      "grad_norm": 16.008869701656987,
      "learning_rate": 9.87977160462871e-06,
      "loss": 0.3937,
      "step": 608
    },
    {
      "epoch": 0.1,
      "grad_norm": 11.973939111385997,
      "learning_rate": 9.87920208374747e-06,
      "loss": 0.4829,
      "step": 609
    },
    {
      "epoch": 0.1,
      "grad_norm": 19.07527390670295,
      "learning_rate": 9.878631233637086e-06,
      "loss": 0.4573,
      "step": 610
    },
    {
      "epoch": 0.1,
      "grad_norm": 37.3900957125809,
      "learning_rate": 9.878059054453076e-06,
      "loss": 0.4768,
      "step": 611
    },
    {
      "epoch": 0.1,
      "grad_norm": 26.026006237698343,
      "learning_rate": 9.877485546351312e-06,
      "loss": 0.479,
      "step": 612
    },
    {
      "epoch": 0.1,
      "grad_norm": 18.584637638478732,
      "learning_rate": 9.876910709488038e-06,
      "loss": 0.5385,
      "step": 613
    },
    {
      "epoch": 0.1,
      "grad_norm": 19.707979074227556,
      "learning_rate": 9.876334544019855e-06,
      "loss": 0.4924,
      "step": 614
    },
    {
      "epoch": 0.1,
      "grad_norm": 18.95593368007674,
      "learning_rate": 9.875757050103722e-06,
      "loss": 0.4882,
      "step": 615
    },
    {
      "epoch": 0.1,
      "grad_norm": 21.66833157168632,
      "learning_rate": 9.87517822789697e-06,
      "loss": 0.4594,
      "step": 616
    },
    {
      "epoch": 0.1,
      "grad_norm": 17.8796696852202,
      "learning_rate": 9.874598077557284e-06,
      "loss": 0.4469,
      "step": 617
    },
    {
      "epoch": 0.1,
      "grad_norm": 24.646881349237603,
      "learning_rate": 9.874016599242711e-06,
      "loss": 0.4439,
      "step": 618
    },
    {
      "epoch": 0.1,
      "grad_norm": 15.796450595747965,
      "learning_rate": 9.873433793111665e-06,
      "loss": 0.4118,
      "step": 619
    },
    {
      "epoch": 0.1,
      "grad_norm": 24.956213659004316,
      "learning_rate": 9.872849659322916e-06,
      "loss": 0.4529,
      "step": 620
    },
    {
      "epoch": 0.1,
      "grad_norm": 26.042736685819456,
      "learning_rate": 9.872264198035599e-06,
      "loss": 0.5503,
      "step": 621
    },
    {
      "epoch": 0.1,
      "grad_norm": 21.845539964652424,
      "learning_rate": 9.871677409409209e-06,
      "loss": 0.5688,
      "step": 622
    },
    {
      "epoch": 0.1,
      "grad_norm": 21.452662831951695,
      "learning_rate": 9.871089293603604e-06,
      "loss": 0.5397,
      "step": 623
    },
    {
      "epoch": 0.1,
      "grad_norm": 43.34961729679202,
      "learning_rate": 9.870499850779003e-06,
      "loss": 0.4833,
      "step": 624
    },
    {
      "epoch": 0.1,
      "grad_norm": 20.337429337999552,
      "learning_rate": 9.869909081095988e-06,
      "loss": 0.4573,
      "step": 625
    },
    {
      "epoch": 0.1,
      "grad_norm": 26.97196410088741,
      "learning_rate": 9.8693169847155e-06,
      "loss": 0.4408,
      "step": 626
    },
    {
      "epoch": 0.1,
      "grad_norm": 36.297996472117006,
      "learning_rate": 9.86872356179884e-06,
      "loss": 0.4758,
      "step": 627
    },
    {
      "epoch": 0.1,
      "grad_norm": 40.79420099801528,
      "learning_rate": 9.868128812507675e-06,
      "loss": 0.4362,
      "step": 628
    },
    {
      "epoch": 0.1,
      "grad_norm": 36.893820811217815,
      "learning_rate": 9.86753273700403e-06,
      "loss": 0.452,
      "step": 629
    },
    {
      "epoch": 0.1,
      "grad_norm": 42.23198273548744,
      "learning_rate": 9.866935335450293e-06,
      "loss": 0.5667,
      "step": 630
    },
    {
      "epoch": 0.1,
      "grad_norm": 35.32524614457651,
      "learning_rate": 9.866336608009215e-06,
      "loss": 0.4725,
      "step": 631
    },
    {
      "epoch": 0.1,
      "grad_norm": 22.064314699962953,
      "learning_rate": 9.865736554843904e-06,
      "loss": 0.4945,
      "step": 632
    },
    {
      "epoch": 0.1,
      "grad_norm": 22.90459776756543,
      "learning_rate": 9.865135176117829e-06,
      "loss": 0.5529,
      "step": 633
    },
    {
      "epoch": 0.1,
      "grad_norm": 25.79532954683492,
      "learning_rate": 9.864532471994823e-06,
      "loss": 0.4612,
      "step": 634
    },
    {
      "epoch": 0.1,
      "grad_norm": 58.04159874681187,
      "learning_rate": 9.86392844263908e-06,
      "loss": 0.489,
      "step": 635
    },
    {
      "epoch": 0.1,
      "grad_norm": 132.27078165539208,
      "learning_rate": 9.863323088215155e-06,
      "loss": 0.5218,
      "step": 636
    },
    {
      "epoch": 0.1,
      "grad_norm": 28.432883357996204,
      "learning_rate": 9.862716408887962e-06,
      "loss": 0.4941,
      "step": 637
    },
    {
      "epoch": 0.1,
      "grad_norm": 31.952343496665208,
      "learning_rate": 9.862108404822779e-06,
      "loss": 0.4648,
      "step": 638
    },
    {
      "epoch": 0.1,
      "grad_norm": 27.63591118434334,
      "learning_rate": 9.861499076185241e-06,
      "loss": 0.5342,
      "step": 639
    },
    {
      "epoch": 0.1,
      "grad_norm": 18.844582797634676,
      "learning_rate": 9.860888423141346e-06,
      "loss": 0.5314,
      "step": 640
    },
    {
      "epoch": 0.1,
      "grad_norm": 51.4965595435129,
      "learning_rate": 9.860276445857456e-06,
      "loss": 0.5417,
      "step": 641
    },
    {
      "epoch": 0.1,
      "grad_norm": 45.4082197802291,
      "learning_rate": 9.859663144500284e-06,
      "loss": 0.5674,
      "step": 642
    },
    {
      "epoch": 0.1,
      "grad_norm": 134.49260996660362,
      "learning_rate": 9.859048519236915e-06,
      "loss": 0.4253,
      "step": 643
    },
    {
      "epoch": 0.1,
      "grad_norm": 32.943868056721065,
      "learning_rate": 9.85843257023479e-06,
      "loss": 0.4867,
      "step": 644
    },
    {
      "epoch": 0.1,
      "grad_norm": 246.0130793456897,
      "learning_rate": 9.85781529766171e-06,
      "loss": 0.5137,
      "step": 645
    },
    {
      "epoch": 0.1,
      "grad_norm": 35.67741187273675,
      "learning_rate": 9.857196701685836e-06,
      "loss": 0.5042,
      "step": 646
    },
    {
      "epoch": 0.1,
      "grad_norm": 30.118258654122457,
      "learning_rate": 9.85657678247569e-06,
      "loss": 0.4454,
      "step": 647
    },
    {
      "epoch": 0.1,
      "grad_norm": 41.90626237438905,
      "learning_rate": 9.855955540200157e-06,
      "loss": 0.5067,
      "step": 648
    },
    {
      "epoch": 0.1,
      "grad_norm": 36.34444784455322,
      "learning_rate": 9.85533297502848e-06,
      "loss": 0.4962,
      "step": 649
    },
    {
      "epoch": 0.1,
      "grad_norm": 34.78305457061313,
      "learning_rate": 9.854709087130261e-06,
      "loss": 0.5105,
      "step": 650
    },
    {
      "epoch": 0.1,
      "grad_norm": 38.952759762501486,
      "learning_rate": 9.854083876675466e-06,
      "loss": 0.4625,
      "step": 651
    },
    {
      "epoch": 0.11,
      "grad_norm": 22.529771829953194,
      "learning_rate": 9.853457343834421e-06,
      "loss": 0.5859,
      "step": 652
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2709564754567324,
      "learning_rate": 9.852829488777808e-06,
      "loss": 0.4506,
      "step": 653
    },
    {
      "epoch": 0.11,
      "grad_norm": 21.212993863384746,
      "learning_rate": 9.852200311676675e-06,
      "loss": 0.451,
      "step": 654
    },
    {
      "epoch": 0.11,
      "grad_norm": 25.56247381096788,
      "learning_rate": 9.851569812702423e-06,
      "loss": 0.5056,
      "step": 655
    },
    {
      "epoch": 0.11,
      "grad_norm": 21.084801633821954,
      "learning_rate": 9.850937992026821e-06,
      "loss": 0.5108,
      "step": 656
    },
    {
      "epoch": 0.11,
      "grad_norm": 25.00414206163789,
      "learning_rate": 9.850304849821992e-06,
      "loss": 0.4934,
      "step": 657
    },
    {
      "epoch": 0.11,
      "grad_norm": 42.71270858955537,
      "learning_rate": 9.849670386260424e-06,
      "loss": 0.5476,
      "step": 658
    },
    {
      "epoch": 0.11,
      "grad_norm": 31.07559909689798,
      "learning_rate": 9.84903460151496e-06,
      "loss": 0.4741,
      "step": 659
    },
    {
      "epoch": 0.11,
      "grad_norm": 33.013418555197035,
      "learning_rate": 9.848397495758806e-06,
      "loss": 0.4722,
      "step": 660
    },
    {
      "epoch": 0.11,
      "grad_norm": 27.5653662709689,
      "learning_rate": 9.847759069165528e-06,
      "loss": 0.5184,
      "step": 661
    },
    {
      "epoch": 0.11,
      "grad_norm": 70.36658088058428,
      "learning_rate": 9.847119321909049e-06,
      "loss": 0.5245,
      "step": 662
    },
    {
      "epoch": 0.11,
      "grad_norm": 27.000916080366956,
      "learning_rate": 9.846478254163657e-06,
      "loss": 0.5482,
      "step": 663
    },
    {
      "epoch": 0.11,
      "grad_norm": 25.355126591060326,
      "learning_rate": 9.845835866103993e-06,
      "loss": 0.5133,
      "step": 664
    },
    {
      "epoch": 0.11,
      "grad_norm": 18.154023877963184,
      "learning_rate": 9.845192157905063e-06,
      "loss": 0.5096,
      "step": 665
    },
    {
      "epoch": 0.11,
      "grad_norm": 24.44797918382641,
      "learning_rate": 9.844547129742231e-06,
      "loss": 0.5231,
      "step": 666
    },
    {
      "epoch": 0.11,
      "grad_norm": 18.450134762094933,
      "learning_rate": 9.84390078179122e-06,
      "loss": 0.5063,
      "step": 667
    },
    {
      "epoch": 0.11,
      "grad_norm": 15.528500908678737,
      "learning_rate": 9.843253114228113e-06,
      "loss": 0.5097,
      "step": 668
    },
    {
      "epoch": 0.11,
      "grad_norm": 20.689411881792772,
      "learning_rate": 9.842604127229353e-06,
      "loss": 0.4861,
      "step": 669
    },
    {
      "epoch": 0.11,
      "grad_norm": 31.949361452678566,
      "learning_rate": 9.841953820971742e-06,
      "loss": 0.5113,
      "step": 670
    },
    {
      "epoch": 0.11,
      "grad_norm": 16.50430020339359,
      "learning_rate": 9.841302195632441e-06,
      "loss": 0.4773,
      "step": 671
    },
    {
      "epoch": 0.11,
      "grad_norm": 19.874510449106946,
      "learning_rate": 9.84064925138897e-06,
      "loss": 0.4891,
      "step": 672
    },
    {
      "epoch": 0.11,
      "grad_norm": 17.795942870847803,
      "learning_rate": 9.839994988419208e-06,
      "loss": 0.4424,
      "step": 673
    },
    {
      "epoch": 0.11,
      "grad_norm": 18.373444821743597,
      "learning_rate": 9.839339406901398e-06,
      "loss": 0.505,
      "step": 674
    },
    {
      "epoch": 0.11,
      "grad_norm": 47.6997226648378,
      "learning_rate": 9.838682507014138e-06,
      "loss": 0.6177,
      "step": 675
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.629724275034955,
      "learning_rate": 9.838024288936381e-06,
      "loss": 0.4528,
      "step": 676
    },
    {
      "epoch": 0.11,
      "grad_norm": 97.13753108723606,
      "learning_rate": 9.837364752847447e-06,
      "loss": 0.4399,
      "step": 677
    },
    {
      "epoch": 0.11,
      "grad_norm": 22.615554189219147,
      "learning_rate": 9.836703898927012e-06,
      "loss": 0.5192,
      "step": 678
    },
    {
      "epoch": 0.11,
      "grad_norm": 17.612337043071484,
      "learning_rate": 9.83604172735511e-06,
      "loss": 0.468,
      "step": 679
    },
    {
      "epoch": 0.11,
      "grad_norm": 15.20704373579663,
      "learning_rate": 9.835378238312136e-06,
      "loss": 0.47,
      "step": 680
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.752093832860716,
      "learning_rate": 9.834713431978842e-06,
      "loss": 0.4284,
      "step": 681
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4054445478548272,
      "learning_rate": 9.834047308536342e-06,
      "loss": 0.4502,
      "step": 682
    },
    {
      "epoch": 0.11,
      "grad_norm": 13.562943089696912,
      "learning_rate": 9.8333798681661e-06,
      "loss": 0.4852,
      "step": 683
    },
    {
      "epoch": 0.11,
      "grad_norm": 24.384289530626003,
      "learning_rate": 9.83271111104995e-06,
      "loss": 0.5033,
      "step": 684
    },
    {
      "epoch": 0.11,
      "grad_norm": 18.980872074508046,
      "learning_rate": 9.83204103737008e-06,
      "loss": 0.4998,
      "step": 685
    },
    {
      "epoch": 0.11,
      "grad_norm": 15.838423874686844,
      "learning_rate": 9.831369647309033e-06,
      "loss": 0.5644,
      "step": 686
    },
    {
      "epoch": 0.11,
      "grad_norm": 21.18435459921517,
      "learning_rate": 9.830696941049718e-06,
      "loss": 0.5675,
      "step": 687
    },
    {
      "epoch": 0.11,
      "grad_norm": 23.130773855942554,
      "learning_rate": 9.830022918775399e-06,
      "loss": 0.4905,
      "step": 688
    },
    {
      "epoch": 0.11,
      "grad_norm": 15.76408715108176,
      "learning_rate": 9.829347580669694e-06,
      "loss": 0.4969,
      "step": 689
    },
    {
      "epoch": 0.11,
      "grad_norm": 27.70651280220986,
      "learning_rate": 9.828670926916588e-06,
      "loss": 0.5273,
      "step": 690
    },
    {
      "epoch": 0.11,
      "grad_norm": 44.96128793236422,
      "learning_rate": 9.827992957700417e-06,
      "loss": 0.4954,
      "step": 691
    },
    {
      "epoch": 0.11,
      "grad_norm": 28.94951983207798,
      "learning_rate": 9.82731367320588e-06,
      "loss": 0.4722,
      "step": 692
    },
    {
      "epoch": 0.11,
      "grad_norm": 24.60375973401854,
      "learning_rate": 9.826633073618034e-06,
      "loss": 0.5158,
      "step": 693
    },
    {
      "epoch": 0.11,
      "grad_norm": 15.61526604575308,
      "learning_rate": 9.825951159122291e-06,
      "loss": 0.4808,
      "step": 694
    },
    {
      "epoch": 0.11,
      "grad_norm": 20.029970377581847,
      "learning_rate": 9.825267929904423e-06,
      "loss": 0.479,
      "step": 695
    },
    {
      "epoch": 0.11,
      "grad_norm": 18.80831135800364,
      "learning_rate": 9.824583386150563e-06,
      "loss": 0.4689,
      "step": 696
    },
    {
      "epoch": 0.11,
      "grad_norm": 25.130011668809697,
      "learning_rate": 9.8238975280472e-06,
      "loss": 0.4587,
      "step": 697
    },
    {
      "epoch": 0.11,
      "grad_norm": 19.66733424673074,
      "learning_rate": 9.823210355781177e-06,
      "loss": 0.4644,
      "step": 698
    },
    {
      "epoch": 0.11,
      "grad_norm": 12.520007016067943,
      "learning_rate": 9.822521869539699e-06,
      "loss": 0.5742,
      "step": 699
    },
    {
      "epoch": 0.11,
      "grad_norm": 28.13264344536298,
      "learning_rate": 9.821832069510333e-06,
      "loss": 0.546,
      "step": 700
    },
    {
      "epoch": 0.11,
      "grad_norm": 16.692464746770355,
      "learning_rate": 9.821140955880995e-06,
      "loss": 0.4861,
      "step": 701
    },
    {
      "epoch": 0.11,
      "grad_norm": 19.615348003479365,
      "learning_rate": 9.820448528839965e-06,
      "loss": 0.517,
      "step": 702
    },
    {
      "epoch": 0.11,
      "grad_norm": 24.007174766752577,
      "learning_rate": 9.81975478857588e-06,
      "loss": 0.4822,
      "step": 703
    },
    {
      "epoch": 0.11,
      "grad_norm": 20.007041817194377,
      "learning_rate": 9.819059735277731e-06,
      "loss": 0.4692,
      "step": 704
    },
    {
      "epoch": 0.11,
      "grad_norm": 16.558918496167855,
      "learning_rate": 9.818363369134876e-06,
      "loss": 0.44,
      "step": 705
    },
    {
      "epoch": 0.11,
      "grad_norm": 19.801961439717147,
      "learning_rate": 9.817665690337017e-06,
      "loss": 0.5109,
      "step": 706
    },
    {
      "epoch": 0.11,
      "grad_norm": 27.136501245619463,
      "learning_rate": 9.816966699074226e-06,
      "loss": 0.571,
      "step": 707
    },
    {
      "epoch": 0.11,
      "grad_norm": 18.310853629813945,
      "learning_rate": 9.816266395536924e-06,
      "loss": 0.4353,
      "step": 708
    },
    {
      "epoch": 0.11,
      "grad_norm": 16.678283092310608,
      "learning_rate": 9.815564779915896e-06,
      "loss": 0.4831,
      "step": 709
    },
    {
      "epoch": 0.11,
      "grad_norm": 19.480562073306537,
      "learning_rate": 9.81486185240228e-06,
      "loss": 0.4973,
      "step": 710
    },
    {
      "epoch": 0.11,
      "grad_norm": 14.852937165804487,
      "learning_rate": 9.814157613187573e-06,
      "loss": 0.4689,
      "step": 711
    },
    {
      "epoch": 0.11,
      "grad_norm": 18.060259669713776,
      "learning_rate": 9.81345206246363e-06,
      "loss": 0.5666,
      "step": 712
    },
    {
      "epoch": 0.11,
      "grad_norm": 16.267854879660476,
      "learning_rate": 9.812745200422659e-06,
      "loss": 0.5042,
      "step": 713
    },
    {
      "epoch": 0.12,
      "grad_norm": 15.279511206182923,
      "learning_rate": 9.812037027257234e-06,
      "loss": 0.4792,
      "step": 714
    },
    {
      "epoch": 0.12,
      "grad_norm": 18.077932595783246,
      "learning_rate": 9.811327543160276e-06,
      "loss": 0.5004,
      "step": 715
    },
    {
      "epoch": 0.12,
      "grad_norm": 11.85758983948475,
      "learning_rate": 9.810616748325072e-06,
      "loss": 0.4722,
      "step": 716
    },
    {
      "epoch": 0.12,
      "grad_norm": 19.08561775926478,
      "learning_rate": 9.809904642945261e-06,
      "loss": 0.3992,
      "step": 717
    },
    {
      "epoch": 0.12,
      "grad_norm": 18.989014440011633,
      "learning_rate": 9.809191227214838e-06,
      "loss": 0.5258,
      "step": 718
    },
    {
      "epoch": 0.12,
      "grad_norm": 21.899839398838346,
      "learning_rate": 9.808476501328158e-06,
      "loss": 0.5198,
      "step": 719
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2860451595562876,
      "learning_rate": 9.807760465479934e-06,
      "loss": 0.4561,
      "step": 720
    },
    {
      "epoch": 0.12,
      "grad_norm": 22.612184072437053,
      "learning_rate": 9.807043119865232e-06,
      "loss": 0.4657,
      "step": 721
    },
    {
      "epoch": 0.12,
      "grad_norm": 39.008201714932646,
      "learning_rate": 9.806324464679477e-06,
      "loss": 0.488,
      "step": 722
    },
    {
      "epoch": 0.12,
      "grad_norm": 70.3485836861302,
      "learning_rate": 9.80560450011845e-06,
      "loss": 0.5922,
      "step": 723
    },
    {
      "epoch": 0.12,
      "grad_norm": 68.34010661053567,
      "learning_rate": 9.80488322637829e-06,
      "loss": 0.4226,
      "step": 724
    },
    {
      "epoch": 0.12,
      "grad_norm": 23.69579046044365,
      "learning_rate": 9.804160643655494e-06,
      "loss": 0.5183,
      "step": 725
    },
    {
      "epoch": 0.12,
      "grad_norm": 66.22987903247811,
      "learning_rate": 9.803436752146909e-06,
      "loss": 0.4681,
      "step": 726
    },
    {
      "epoch": 0.12,
      "grad_norm": 22.13576625594009,
      "learning_rate": 9.802711552049746e-06,
      "loss": 0.4655,
      "step": 727
    },
    {
      "epoch": 0.12,
      "grad_norm": 29.5295051982629,
      "learning_rate": 9.801985043561569e-06,
      "loss": 0.4876,
      "step": 728
    },
    {
      "epoch": 0.12,
      "grad_norm": 23.514806650665207,
      "learning_rate": 9.801257226880297e-06,
      "loss": 0.5135,
      "step": 729
    },
    {
      "epoch": 0.12,
      "grad_norm": 30.136030534841247,
      "learning_rate": 9.80052810220421e-06,
      "loss": 0.5332,
      "step": 730
    },
    {
      "epoch": 0.12,
      "grad_norm": 21.830147411374625,
      "learning_rate": 9.79979766973194e-06,
      "loss": 0.538,
      "step": 731
    },
    {
      "epoch": 0.12,
      "grad_norm": 49.81259418507051,
      "learning_rate": 9.799065929662478e-06,
      "loss": 0.5068,
      "step": 732
    },
    {
      "epoch": 0.12,
      "grad_norm": 30.120344600958013,
      "learning_rate": 9.798332882195172e-06,
      "loss": 0.4602,
      "step": 733
    },
    {
      "epoch": 0.12,
      "grad_norm": 24.56373477936228,
      "learning_rate": 9.79759852752972e-06,
      "loss": 0.5407,
      "step": 734
    },
    {
      "epoch": 0.12,
      "grad_norm": 40.67656506510299,
      "learning_rate": 9.796862865866183e-06,
      "loss": 0.5123,
      "step": 735
    },
    {
      "epoch": 0.12,
      "grad_norm": 28.08456392008881,
      "learning_rate": 9.796125897404973e-06,
      "loss": 0.4395,
      "step": 736
    },
    {
      "epoch": 0.12,
      "grad_norm": 24.581896318986935,
      "learning_rate": 9.795387622346866e-06,
      "loss": 0.4684,
      "step": 737
    },
    {
      "epoch": 0.12,
      "grad_norm": 42.86869425333663,
      "learning_rate": 9.794648040892983e-06,
      "loss": 0.4985,
      "step": 738
    },
    {
      "epoch": 0.12,
      "grad_norm": 35.17206995762081,
      "learning_rate": 9.793907153244808e-06,
      "loss": 0.57,
      "step": 739
    },
    {
      "epoch": 0.12,
      "grad_norm": 31.35202313496837,
      "learning_rate": 9.79316495960418e-06,
      "loss": 0.4461,
      "step": 740
    },
    {
      "epoch": 0.12,
      "grad_norm": 22.032846135926587,
      "learning_rate": 9.792421460173294e-06,
      "loss": 0.5234,
      "step": 741
    },
    {
      "epoch": 0.12,
      "grad_norm": 21.94914471309661,
      "learning_rate": 9.791676655154696e-06,
      "loss": 0.5382,
      "step": 742
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.235393642987467,
      "learning_rate": 9.790930544751297e-06,
      "loss": 0.4793,
      "step": 743
    },
    {
      "epoch": 0.12,
      "grad_norm": 15.009076321870372,
      "learning_rate": 9.790183129166351e-06,
      "loss": 0.4514,
      "step": 744
    },
    {
      "epoch": 0.12,
      "grad_norm": 22.857535630894642,
      "learning_rate": 9.78943440860348e-06,
      "loss": 0.5513,
      "step": 745
    },
    {
      "epoch": 0.12,
      "grad_norm": 28.103565851655105,
      "learning_rate": 9.788684383266655e-06,
      "loss": 0.4908,
      "step": 746
    },
    {
      "epoch": 0.12,
      "grad_norm": 29.43437633491117,
      "learning_rate": 9.787933053360203e-06,
      "loss": 0.5203,
      "step": 747
    },
    {
      "epoch": 0.12,
      "grad_norm": 23.694193741663607,
      "learning_rate": 9.78718041908881e-06,
      "loss": 0.455,
      "step": 748
    },
    {
      "epoch": 0.12,
      "grad_norm": 20.296911835573596,
      "learning_rate": 9.786426480657507e-06,
      "loss": 0.5169,
      "step": 749
    },
    {
      "epoch": 0.12,
      "grad_norm": 27.717028874014833,
      "learning_rate": 9.785671238271694e-06,
      "loss": 0.4094,
      "step": 750
    },
    {
      "epoch": 0.12,
      "grad_norm": 26.48782492758789,
      "learning_rate": 9.784914692137118e-06,
      "loss": 0.524,
      "step": 751
    },
    {
      "epoch": 0.12,
      "grad_norm": 34.02215545449661,
      "learning_rate": 9.784156842459882e-06,
      "loss": 0.5327,
      "step": 752
    },
    {
      "epoch": 0.12,
      "grad_norm": 51.791240360548045,
      "learning_rate": 9.783397689446447e-06,
      "loss": 0.5072,
      "step": 753
    },
    {
      "epoch": 0.12,
      "grad_norm": 19.318952483530822,
      "learning_rate": 9.782637233303624e-06,
      "loss": 0.4843,
      "step": 754
    },
    {
      "epoch": 0.12,
      "grad_norm": 30.38603084942431,
      "learning_rate": 9.781875474238587e-06,
      "loss": 0.4982,
      "step": 755
    },
    {
      "epoch": 0.12,
      "grad_norm": 21.212825663607752,
      "learning_rate": 9.781112412458858e-06,
      "loss": 0.5192,
      "step": 756
    },
    {
      "epoch": 0.12,
      "grad_norm": 21.236279975655908,
      "learning_rate": 9.780348048172315e-06,
      "loss": 0.467,
      "step": 757
    },
    {
      "epoch": 0.12,
      "grad_norm": 25.763818650187154,
      "learning_rate": 9.779582381587192e-06,
      "loss": 0.5267,
      "step": 758
    },
    {
      "epoch": 0.12,
      "grad_norm": 42.01728460728236,
      "learning_rate": 9.778815412912078e-06,
      "loss": 0.4894,
      "step": 759
    },
    {
      "epoch": 0.12,
      "grad_norm": 22.665619984522813,
      "learning_rate": 9.778047142355917e-06,
      "loss": 0.5251,
      "step": 760
    },
    {
      "epoch": 0.12,
      "grad_norm": 46.90950551132639,
      "learning_rate": 9.777277570128008e-06,
      "loss": 0.4437,
      "step": 761
    },
    {
      "epoch": 0.12,
      "grad_norm": 24.245146390369378,
      "learning_rate": 9.776506696438002e-06,
      "loss": 0.5017,
      "step": 762
    },
    {
      "epoch": 0.12,
      "grad_norm": 33.75557381511454,
      "learning_rate": 9.775734521495905e-06,
      "loss": 0.5034,
      "step": 763
    },
    {
      "epoch": 0.12,
      "grad_norm": 249.8105860568305,
      "learning_rate": 9.774961045512082e-06,
      "loss": 0.5436,
      "step": 764
    },
    {
      "epoch": 0.12,
      "grad_norm": 25.11208467122166,
      "learning_rate": 9.774186268697247e-06,
      "loss": 0.4478,
      "step": 765
    },
    {
      "epoch": 0.12,
      "grad_norm": 43.728047738023335,
      "learning_rate": 9.773410191262471e-06,
      "loss": 0.5351,
      "step": 766
    },
    {
      "epoch": 0.12,
      "grad_norm": 36.982489729286506,
      "learning_rate": 9.772632813419181e-06,
      "loss": 0.4478,
      "step": 767
    },
    {
      "epoch": 0.12,
      "grad_norm": 28.577056074073887,
      "learning_rate": 9.771854135379153e-06,
      "loss": 0.5379,
      "step": 768
    },
    {
      "epoch": 0.12,
      "grad_norm": 21.66953740046426,
      "learning_rate": 9.771074157354521e-06,
      "loss": 0.5311,
      "step": 769
    },
    {
      "epoch": 0.12,
      "grad_norm": 22.018434431381095,
      "learning_rate": 9.770292879557774e-06,
      "loss": 0.4516,
      "step": 770
    },
    {
      "epoch": 0.12,
      "grad_norm": 28.9316001223727,
      "learning_rate": 9.769510302201751e-06,
      "loss": 0.4782,
      "step": 771
    },
    {
      "epoch": 0.12,
      "grad_norm": 38.465798868528,
      "learning_rate": 9.76872642549965e-06,
      "loss": 0.5506,
      "step": 772
    },
    {
      "epoch": 0.12,
      "grad_norm": 42.98895755490222,
      "learning_rate": 9.767941249665022e-06,
      "loss": 0.5215,
      "step": 773
    },
    {
      "epoch": 0.12,
      "grad_norm": 37.5542346787441,
      "learning_rate": 9.767154774911767e-06,
      "loss": 0.4245,
      "step": 774
    },
    {
      "epoch": 0.12,
      "grad_norm": 126.0337025925342,
      "learning_rate": 9.766367001454144e-06,
      "loss": 0.4668,
      "step": 775
    },
    {
      "epoch": 0.13,
      "grad_norm": 26.965340966727034,
      "learning_rate": 9.765577929506764e-06,
      "loss": 0.5684,
      "step": 776
    },
    {
      "epoch": 0.13,
      "grad_norm": 39.08373874742864,
      "learning_rate": 9.764787559284592e-06,
      "loss": 0.5215,
      "step": 777
    },
    {
      "epoch": 0.13,
      "grad_norm": 38.35500361272542,
      "learning_rate": 9.763995891002946e-06,
      "loss": 0.4699,
      "step": 778
    },
    {
      "epoch": 0.13,
      "grad_norm": 13.772844282819213,
      "learning_rate": 9.763202924877502e-06,
      "loss": 0.4488,
      "step": 779
    },
    {
      "epoch": 0.13,
      "grad_norm": 24.341551103401898,
      "learning_rate": 9.762408661124279e-06,
      "loss": 0.5404,
      "step": 780
    },
    {
      "epoch": 0.13,
      "grad_norm": 38.58160278798945,
      "learning_rate": 9.761613099959663e-06,
      "loss": 0.5149,
      "step": 781
    },
    {
      "epoch": 0.13,
      "grad_norm": 15.453459024065733,
      "learning_rate": 9.760816241600383e-06,
      "loss": 0.4269,
      "step": 782
    },
    {
      "epoch": 0.13,
      "grad_norm": 16.782477172057998,
      "learning_rate": 9.760018086263525e-06,
      "loss": 0.5336,
      "step": 783
    },
    {
      "epoch": 0.13,
      "grad_norm": 15.057640537762172,
      "learning_rate": 9.759218634166531e-06,
      "loss": 0.5252,
      "step": 784
    },
    {
      "epoch": 0.13,
      "grad_norm": 15.444438544851318,
      "learning_rate": 9.75841788552719e-06,
      "loss": 0.4455,
      "step": 785
    },
    {
      "epoch": 0.13,
      "grad_norm": 17.696856495068964,
      "learning_rate": 9.757615840563654e-06,
      "loss": 0.5463,
      "step": 786
    },
    {
      "epoch": 0.13,
      "grad_norm": 29.698541802400126,
      "learning_rate": 9.756812499494417e-06,
      "loss": 0.4737,
      "step": 787
    },
    {
      "epoch": 0.13,
      "grad_norm": 14.152192660611997,
      "learning_rate": 9.756007862538333e-06,
      "loss": 0.4847,
      "step": 788
    },
    {
      "epoch": 0.13,
      "grad_norm": 15.374085135518834,
      "learning_rate": 9.755201929914607e-06,
      "loss": 0.5425,
      "step": 789
    },
    {
      "epoch": 0.13,
      "grad_norm": 15.601690712889747,
      "learning_rate": 9.754394701842797e-06,
      "loss": 0.5351,
      "step": 790
    },
    {
      "epoch": 0.13,
      "grad_norm": 12.39724747079049,
      "learning_rate": 9.753586178542815e-06,
      "loss": 0.4753,
      "step": 791
    },
    {
      "epoch": 0.13,
      "grad_norm": 88.4459986252391,
      "learning_rate": 9.752776360234925e-06,
      "loss": 0.6004,
      "step": 792
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.642484366322941,
      "learning_rate": 9.751965247139743e-06,
      "loss": 0.5001,
      "step": 793
    },
    {
      "epoch": 0.13,
      "grad_norm": 20.049962860527497,
      "learning_rate": 9.751152839478238e-06,
      "loss": 0.487,
      "step": 794
    },
    {
      "epoch": 0.13,
      "grad_norm": 34.15693630886806,
      "learning_rate": 9.750339137471733e-06,
      "loss": 0.4758,
      "step": 795
    },
    {
      "epoch": 0.13,
      "grad_norm": 12.007307300973832,
      "learning_rate": 9.749524141341905e-06,
      "loss": 0.4805,
      "step": 796
    },
    {
      "epoch": 0.13,
      "grad_norm": 21.15144517678615,
      "learning_rate": 9.74870785131078e-06,
      "loss": 0.4894,
      "step": 797
    },
    {
      "epoch": 0.13,
      "grad_norm": 20.318874521302774,
      "learning_rate": 9.747890267600735e-06,
      "loss": 0.5165,
      "step": 798
    },
    {
      "epoch": 0.13,
      "grad_norm": 45.477751430851235,
      "learning_rate": 9.747071390434507e-06,
      "loss": 0.5599,
      "step": 799
    },
    {
      "epoch": 0.13,
      "grad_norm": 30.523846900686554,
      "learning_rate": 9.746251220035176e-06,
      "loss": 0.5623,
      "step": 800
    },
    {
      "epoch": 0.13,
      "grad_norm": 17.029920017968752,
      "learning_rate": 9.745429756626184e-06,
      "loss": 0.5061,
      "step": 801
    },
    {
      "epoch": 0.13,
      "grad_norm": 12.798832651167722,
      "learning_rate": 9.744607000431317e-06,
      "loss": 0.4786,
      "step": 802
    },
    {
      "epoch": 0.13,
      "grad_norm": 42.98569617285503,
      "learning_rate": 9.743782951674715e-06,
      "loss": 0.4658,
      "step": 803
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.936594426370887,
      "learning_rate": 9.742957610580876e-06,
      "loss": 0.4848,
      "step": 804
    },
    {
      "epoch": 0.13,
      "grad_norm": 41.80080208942185,
      "learning_rate": 9.742130977374643e-06,
      "loss": 0.5086,
      "step": 805
    },
    {
      "epoch": 0.13,
      "grad_norm": 13.581925144344769,
      "learning_rate": 9.741303052281214e-06,
      "loss": 0.434,
      "step": 806
    },
    {
      "epoch": 0.13,
      "grad_norm": 125.50288228823116,
      "learning_rate": 9.740473835526139e-06,
      "loss": 0.5607,
      "step": 807
    },
    {
      "epoch": 0.13,
      "grad_norm": 62.456442248613335,
      "learning_rate": 9.73964332733532e-06,
      "loss": 0.517,
      "step": 808
    },
    {
      "epoch": 0.13,
      "grad_norm": 12.800522872105265,
      "learning_rate": 9.738811527935008e-06,
      "loss": 0.4212,
      "step": 809
    },
    {
      "epoch": 0.13,
      "grad_norm": 14.914885095457565,
      "learning_rate": 9.737978437551812e-06,
      "loss": 0.4883,
      "step": 810
    },
    {
      "epoch": 0.13,
      "grad_norm": 14.822141050807216,
      "learning_rate": 9.737144056412685e-06,
      "loss": 0.4288,
      "step": 811
    },
    {
      "epoch": 0.13,
      "grad_norm": 29.969505488786965,
      "learning_rate": 9.736308384744935e-06,
      "loss": 0.4473,
      "step": 812
    },
    {
      "epoch": 0.13,
      "grad_norm": 18.497496614624033,
      "learning_rate": 9.735471422776225e-06,
      "loss": 0.474,
      "step": 813
    },
    {
      "epoch": 0.13,
      "grad_norm": 16.927159822840608,
      "learning_rate": 9.734633170734568e-06,
      "loss": 0.5147,
      "step": 814
    },
    {
      "epoch": 0.13,
      "grad_norm": 36.87017319293176,
      "learning_rate": 9.733793628848323e-06,
      "loss": 0.4876,
      "step": 815
    },
    {
      "epoch": 0.13,
      "grad_norm": 31.494967767495705,
      "learning_rate": 9.732952797346205e-06,
      "loss": 0.4122,
      "step": 816
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.752764823995623,
      "learning_rate": 9.73211067645728e-06,
      "loss": 0.4765,
      "step": 817
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.528607287800185,
      "learning_rate": 9.731267266410967e-06,
      "loss": 0.5333,
      "step": 818
    },
    {
      "epoch": 0.13,
      "grad_norm": 22.194691483947587,
      "learning_rate": 9.73042256743703e-06,
      "loss": 0.4357,
      "step": 819
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.869727929302632,
      "learning_rate": 9.729576579765594e-06,
      "loss": 0.4917,
      "step": 820
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.926837035753058,
      "learning_rate": 9.728729303627124e-06,
      "loss": 0.4927,
      "step": 821
    },
    {
      "epoch": 0.13,
      "grad_norm": 12.706235684641722,
      "learning_rate": 9.727880739252444e-06,
      "loss": 0.5219,
      "step": 822
    },
    {
      "epoch": 0.13,
      "grad_norm": 21.21492251551645,
      "learning_rate": 9.727030886872724e-06,
      "loss": 0.4769,
      "step": 823
    },
    {
      "epoch": 0.13,
      "grad_norm": 38.43607689859831,
      "learning_rate": 9.726179746719492e-06,
      "loss": 0.4821,
      "step": 824
    },
    {
      "epoch": 0.13,
      "grad_norm": 15.941592316429077,
      "learning_rate": 9.725327319024618e-06,
      "loss": 0.4896,
      "step": 825
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.47631918971038,
      "learning_rate": 9.724473604020327e-06,
      "loss": 0.4711,
      "step": 826
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.842969169683593,
      "learning_rate": 9.723618601939197e-06,
      "loss": 0.4248,
      "step": 827
    },
    {
      "epoch": 0.13,
      "grad_norm": 47.55235027940581,
      "learning_rate": 9.722762313014152e-06,
      "loss": 0.4227,
      "step": 828
    },
    {
      "epoch": 0.13,
      "grad_norm": 13.836324713863256,
      "learning_rate": 9.721904737478468e-06,
      "loss": 0.4579,
      "step": 829
    },
    {
      "epoch": 0.13,
      "grad_norm": 19.415280801615253,
      "learning_rate": 9.721045875565774e-06,
      "loss": 0.5543,
      "step": 830
    },
    {
      "epoch": 0.13,
      "grad_norm": 23.7578698124494,
      "learning_rate": 9.720185727510047e-06,
      "loss": 0.5065,
      "step": 831
    },
    {
      "epoch": 0.13,
      "grad_norm": 18.216402001014256,
      "learning_rate": 9.719324293545615e-06,
      "loss": 0.5267,
      "step": 832
    },
    {
      "epoch": 0.13,
      "grad_norm": 13.880159011854113,
      "learning_rate": 9.718461573907158e-06,
      "loss": 0.4669,
      "step": 833
    },
    {
      "epoch": 0.13,
      "grad_norm": 19.51078919073798,
      "learning_rate": 9.717597568829702e-06,
      "loss": 0.4535,
      "step": 834
    },
    {
      "epoch": 0.13,
      "grad_norm": 16.18987952909316,
      "learning_rate": 9.716732278548628e-06,
      "loss": 0.4872,
      "step": 835
    },
    {
      "epoch": 0.13,
      "grad_norm": 23.482734741641927,
      "learning_rate": 9.715865703299663e-06,
      "loss": 0.4869,
      "step": 836
    },
    {
      "epoch": 0.13,
      "grad_norm": 22.22853214736185,
      "learning_rate": 9.714997843318887e-06,
      "loss": 0.5543,
      "step": 837
    },
    {
      "epoch": 0.14,
      "grad_norm": 29.02386420869639,
      "learning_rate": 9.714128698842728e-06,
      "loss": 0.4664,
      "step": 838
    },
    {
      "epoch": 0.14,
      "grad_norm": 26.823224206670098,
      "learning_rate": 9.713258270107966e-06,
      "loss": 0.5409,
      "step": 839
    },
    {
      "epoch": 0.14,
      "grad_norm": 22.070761335419267,
      "learning_rate": 9.71238655735173e-06,
      "loss": 0.4323,
      "step": 840
    },
    {
      "epoch": 0.14,
      "grad_norm": 28.72337850581609,
      "learning_rate": 9.711513560811498e-06,
      "loss": 0.5368,
      "step": 841
    },
    {
      "epoch": 0.14,
      "grad_norm": 26.71920995935433,
      "learning_rate": 9.710639280725097e-06,
      "loss": 0.4952,
      "step": 842
    },
    {
      "epoch": 0.14,
      "grad_norm": 22.411715576414686,
      "learning_rate": 9.709763717330709e-06,
      "loss": 0.4597,
      "step": 843
    },
    {
      "epoch": 0.14,
      "grad_norm": 33.22750454262311,
      "learning_rate": 9.708886870866856e-06,
      "loss": 0.4676,
      "step": 844
    },
    {
      "epoch": 0.14,
      "grad_norm": 33.57076272100097,
      "learning_rate": 9.708008741572422e-06,
      "loss": 0.4582,
      "step": 845
    },
    {
      "epoch": 0.14,
      "grad_norm": 35.80275606572998,
      "learning_rate": 9.707129329686626e-06,
      "loss": 0.498,
      "step": 846
    },
    {
      "epoch": 0.14,
      "grad_norm": 167.16560738947774,
      "learning_rate": 9.706248635449048e-06,
      "loss": 0.504,
      "step": 847
    },
    {
      "epoch": 0.14,
      "grad_norm": 16.265516129201238,
      "learning_rate": 9.705366659099617e-06,
      "loss": 0.4637,
      "step": 848
    },
    {
      "epoch": 0.14,
      "grad_norm": 71.6112277441031,
      "learning_rate": 9.704483400878602e-06,
      "loss": 0.483,
      "step": 849
    },
    {
      "epoch": 0.14,
      "grad_norm": 80.8414976295329,
      "learning_rate": 9.703598861026627e-06,
      "loss": 0.5556,
      "step": 850
    },
    {
      "epoch": 0.14,
      "grad_norm": 26.965311598824623,
      "learning_rate": 9.702713039784668e-06,
      "loss": 0.5404,
      "step": 851
    },
    {
      "epoch": 0.14,
      "grad_norm": 58.389752719763734,
      "learning_rate": 9.701825937394045e-06,
      "loss": 0.5265,
      "step": 852
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.279052830232325,
      "learning_rate": 9.700937554096432e-06,
      "loss": 0.4486,
      "step": 853
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4043328176802556,
      "learning_rate": 9.700047890133845e-06,
      "loss": 0.4749,
      "step": 854
    },
    {
      "epoch": 0.14,
      "grad_norm": 19.533761976762747,
      "learning_rate": 9.699156945748657e-06,
      "loss": 0.5307,
      "step": 855
    },
    {
      "epoch": 0.14,
      "grad_norm": 59.65067546295453,
      "learning_rate": 9.698264721183584e-06,
      "loss": 0.4929,
      "step": 856
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2623776914107365,
      "learning_rate": 9.69737121668169e-06,
      "loss": 0.4804,
      "step": 857
    },
    {
      "epoch": 0.14,
      "grad_norm": 16.671407692878628,
      "learning_rate": 9.696476432486395e-06,
      "loss": 0.4658,
      "step": 858
    },
    {
      "epoch": 0.14,
      "grad_norm": 32.75790084169422,
      "learning_rate": 9.695580368841462e-06,
      "loss": 0.5437,
      "step": 859
    },
    {
      "epoch": 0.14,
      "grad_norm": 31.097498922921837,
      "learning_rate": 9.694683025991e-06,
      "loss": 0.4413,
      "step": 860
    },
    {
      "epoch": 0.14,
      "grad_norm": 28.834648231069195,
      "learning_rate": 9.693784404179472e-06,
      "loss": 0.5846,
      "step": 861
    },
    {
      "epoch": 0.14,
      "grad_norm": 38.580221165450446,
      "learning_rate": 9.692884503651687e-06,
      "loss": 0.4923,
      "step": 862
    },
    {
      "epoch": 0.14,
      "grad_norm": 24.460690915658258,
      "learning_rate": 9.691983324652804e-06,
      "loss": 0.3778,
      "step": 863
    },
    {
      "epoch": 0.14,
      "grad_norm": 16.431107726245582,
      "learning_rate": 9.691080867428328e-06,
      "loss": 0.4853,
      "step": 864
    },
    {
      "epoch": 0.14,
      "grad_norm": 86.35788132544359,
      "learning_rate": 9.690177132224113e-06,
      "loss": 0.5914,
      "step": 865
    },
    {
      "epoch": 0.14,
      "grad_norm": 64.46183182063074,
      "learning_rate": 9.689272119286361e-06,
      "loss": 0.4576,
      "step": 866
    },
    {
      "epoch": 0.14,
      "grad_norm": 17.320974530922435,
      "learning_rate": 9.688365828861625e-06,
      "loss": 0.509,
      "step": 867
    },
    {
      "epoch": 0.14,
      "grad_norm": 126.20985491765774,
      "learning_rate": 9.6874582611968e-06,
      "loss": 0.4724,
      "step": 868
    },
    {
      "epoch": 0.14,
      "grad_norm": 24.26624464865846,
      "learning_rate": 9.686549416539135e-06,
      "loss": 0.5037,
      "step": 869
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4698098921507048,
      "learning_rate": 9.685639295136224e-06,
      "loss": 0.453,
      "step": 870
    },
    {
      "epoch": 0.14,
      "grad_norm": 22.00530610603084,
      "learning_rate": 9.684727897236008e-06,
      "loss": 0.4667,
      "step": 871
    },
    {
      "epoch": 0.14,
      "grad_norm": 40.4987937675432,
      "learning_rate": 9.683815223086777e-06,
      "loss": 0.5267,
      "step": 872
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4037830578460488,
      "learning_rate": 9.68290127293717e-06,
      "loss": 0.5089,
      "step": 873
    },
    {
      "epoch": 0.14,
      "grad_norm": 24.533577589405628,
      "learning_rate": 9.68198604703617e-06,
      "loss": 0.5373,
      "step": 874
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1957333630601856,
      "learning_rate": 9.681069545633113e-06,
      "loss": 0.4467,
      "step": 875
    },
    {
      "epoch": 0.14,
      "grad_norm": 44.51402900276449,
      "learning_rate": 9.680151768977676e-06,
      "loss": 0.4955,
      "step": 876
    },
    {
      "epoch": 0.14,
      "grad_norm": 13.810054162013374,
      "learning_rate": 9.679232717319887e-06,
      "loss": 0.4937,
      "step": 877
    },
    {
      "epoch": 0.14,
      "grad_norm": 23.56182578277013,
      "learning_rate": 9.678312390910123e-06,
      "loss": 0.4777,
      "step": 878
    },
    {
      "epoch": 0.14,
      "grad_norm": 13.25269393349784,
      "learning_rate": 9.677390789999106e-06,
      "loss": 0.5698,
      "step": 879
    },
    {
      "epoch": 0.14,
      "grad_norm": 17.385739310511497,
      "learning_rate": 9.676467914837904e-06,
      "loss": 0.4292,
      "step": 880
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.418399895330307,
      "learning_rate": 9.675543765677935e-06,
      "loss": 0.482,
      "step": 881
    },
    {
      "epoch": 0.14,
      "grad_norm": 51.4925584104897,
      "learning_rate": 9.674618342770962e-06,
      "loss": 0.5084,
      "step": 882
    },
    {
      "epoch": 0.14,
      "grad_norm": 18.287028909763716,
      "learning_rate": 9.673691646369094e-06,
      "loss": 0.4899,
      "step": 883
    },
    {
      "epoch": 0.14,
      "grad_norm": 17.300724558715963,
      "learning_rate": 9.672763676724792e-06,
      "loss": 0.4877,
      "step": 884
    },
    {
      "epoch": 0.14,
      "grad_norm": 12.894477117895924,
      "learning_rate": 9.671834434090861e-06,
      "loss": 0.5086,
      "step": 885
    },
    {
      "epoch": 0.14,
      "grad_norm": 16.74662240257578,
      "learning_rate": 9.670903918720446e-06,
      "loss": 0.4973,
      "step": 886
    },
    {
      "epoch": 0.14,
      "grad_norm": 17.663540427089483,
      "learning_rate": 9.669972130867053e-06,
      "loss": 0.5045,
      "step": 887
    },
    {
      "epoch": 0.14,
      "grad_norm": 12.807825064701182,
      "learning_rate": 9.66903907078452e-06,
      "loss": 0.4121,
      "step": 888
    },
    {
      "epoch": 0.14,
      "grad_norm": 13.209317194839818,
      "learning_rate": 9.668104738727045e-06,
      "loss": 0.5457,
      "step": 889
    },
    {
      "epoch": 0.14,
      "grad_norm": 12.572594206483387,
      "learning_rate": 9.667169134949158e-06,
      "loss": 0.4775,
      "step": 890
    },
    {
      "epoch": 0.14,
      "grad_norm": 12.933546620391244,
      "learning_rate": 9.666232259705751e-06,
      "loss": 0.4305,
      "step": 891
    },
    {
      "epoch": 0.14,
      "grad_norm": 12.733915139145058,
      "learning_rate": 9.66529411325205e-06,
      "loss": 0.5353,
      "step": 892
    },
    {
      "epoch": 0.14,
      "grad_norm": 53.315614659048926,
      "learning_rate": 9.664354695843632e-06,
      "loss": 0.4776,
      "step": 893
    },
    {
      "epoch": 0.14,
      "grad_norm": 21.38241119271368,
      "learning_rate": 9.66341400773642e-06,
      "loss": 0.4456,
      "step": 894
    },
    {
      "epoch": 0.14,
      "grad_norm": 18.981549837730675,
      "learning_rate": 9.662472049186688e-06,
      "loss": 0.4922,
      "step": 895
    },
    {
      "epoch": 0.14,
      "grad_norm": 15.365771384488077,
      "learning_rate": 9.661528820451045e-06,
      "loss": 0.5116,
      "step": 896
    },
    {
      "epoch": 0.14,
      "grad_norm": 17.38066872462953,
      "learning_rate": 9.660584321786456e-06,
      "loss": 0.4082,
      "step": 897
    },
    {
      "epoch": 0.14,
      "grad_norm": 24.258314947389227,
      "learning_rate": 9.65963855345023e-06,
      "loss": 0.5222,
      "step": 898
    },
    {
      "epoch": 0.14,
      "grad_norm": 18.11395784802861,
      "learning_rate": 9.658691515700016e-06,
      "loss": 0.512,
      "step": 899
    },
    {
      "epoch": 0.15,
      "grad_norm": 34.84165914930186,
      "learning_rate": 9.657743208793818e-06,
      "loss": 0.4277,
      "step": 900
    },
    {
      "epoch": 0.15,
      "grad_norm": 26.304413274962798,
      "learning_rate": 9.656793632989976e-06,
      "loss": 0.5686,
      "step": 901
    },
    {
      "epoch": 0.15,
      "grad_norm": 21.38646161739341,
      "learning_rate": 9.655842788547184e-06,
      "loss": 0.4312,
      "step": 902
    },
    {
      "epoch": 0.15,
      "grad_norm": 14.560823648173352,
      "learning_rate": 9.654890675724478e-06,
      "loss": 0.4912,
      "step": 903
    },
    {
      "epoch": 0.15,
      "grad_norm": 19.80098513475386,
      "learning_rate": 9.653937294781237e-06,
      "loss": 0.4724,
      "step": 904
    },
    {
      "epoch": 0.15,
      "grad_norm": 22.69963368981203,
      "learning_rate": 9.652982645977193e-06,
      "loss": 0.4892,
      "step": 905
    },
    {
      "epoch": 0.15,
      "grad_norm": 24.81523226142492,
      "learning_rate": 9.652026729572415e-06,
      "loss": 0.4879,
      "step": 906
    },
    {
      "epoch": 0.15,
      "grad_norm": 25.74256499384213,
      "learning_rate": 9.651069545827321e-06,
      "loss": 0.5019,
      "step": 907
    },
    {
      "epoch": 0.15,
      "grad_norm": 19.296978907218705,
      "learning_rate": 9.650111095002676e-06,
      "loss": 0.4618,
      "step": 908
    },
    {
      "epoch": 0.15,
      "grad_norm": 26.53211183276057,
      "learning_rate": 9.649151377359588e-06,
      "loss": 0.5478,
      "step": 909
    },
    {
      "epoch": 0.15,
      "grad_norm": 18.359990850346854,
      "learning_rate": 9.64819039315951e-06,
      "loss": 0.5059,
      "step": 910
    },
    {
      "epoch": 0.15,
      "grad_norm": 27.146350026207703,
      "learning_rate": 9.647228142664241e-06,
      "loss": 0.5308,
      "step": 911
    },
    {
      "epoch": 0.15,
      "grad_norm": 14.601144986459667,
      "learning_rate": 9.646264626135924e-06,
      "loss": 0.4795,
      "step": 912
    },
    {
      "epoch": 0.15,
      "grad_norm": 19.379563200214534,
      "learning_rate": 9.645299843837047e-06,
      "loss": 0.5059,
      "step": 913
    },
    {
      "epoch": 0.15,
      "grad_norm": 22.483035712335138,
      "learning_rate": 9.644333796030444e-06,
      "loss": 0.4789,
      "step": 914
    },
    {
      "epoch": 0.15,
      "grad_norm": 18.399389340651457,
      "learning_rate": 9.643366482979296e-06,
      "loss": 0.5264,
      "step": 915
    },
    {
      "epoch": 0.15,
      "grad_norm": 33.64923947907195,
      "learning_rate": 9.64239790494712e-06,
      "loss": 0.4322,
      "step": 916
    },
    {
      "epoch": 0.15,
      "grad_norm": 16.24641618243107,
      "learning_rate": 9.641428062197789e-06,
      "loss": 0.5115,
      "step": 917
    },
    {
      "epoch": 0.15,
      "grad_norm": 21.46289283536512,
      "learning_rate": 9.640456954995509e-06,
      "loss": 0.4851,
      "step": 918
    },
    {
      "epoch": 0.15,
      "grad_norm": 21.980764490788626,
      "learning_rate": 9.639484583604841e-06,
      "loss": 0.55,
      "step": 919
    },
    {
      "epoch": 0.15,
      "grad_norm": 29.88506289570659,
      "learning_rate": 9.638510948290684e-06,
      "loss": 0.4992,
      "step": 920
    },
    {
      "epoch": 0.15,
      "grad_norm": 15.293853172724683,
      "learning_rate": 9.637536049318284e-06,
      "loss": 0.5399,
      "step": 921
    },
    {
      "epoch": 0.15,
      "grad_norm": 26.935886223016613,
      "learning_rate": 9.63655988695323e-06,
      "loss": 0.5603,
      "step": 922
    },
    {
      "epoch": 0.15,
      "grad_norm": 57.60764415721829,
      "learning_rate": 9.635582461461455e-06,
      "loss": 0.5477,
      "step": 923
    },
    {
      "epoch": 0.15,
      "grad_norm": 13.905764808920038,
      "learning_rate": 9.634603773109235e-06,
      "loss": 0.4264,
      "step": 924
    },
    {
      "epoch": 0.15,
      "grad_norm": 12.149995219723257,
      "learning_rate": 9.633623822163196e-06,
      "loss": 0.5126,
      "step": 925
    },
    {
      "epoch": 0.15,
      "grad_norm": 24.346304474653277,
      "learning_rate": 9.6326426088903e-06,
      "loss": 0.5666,
      "step": 926
    },
    {
      "epoch": 0.15,
      "grad_norm": 63.50948535867169,
      "learning_rate": 9.631660133557858e-06,
      "loss": 0.4929,
      "step": 927
    },
    {
      "epoch": 0.15,
      "grad_norm": 234.82827664074006,
      "learning_rate": 9.630676396433524e-06,
      "loss": 0.4471,
      "step": 928
    },
    {
      "epoch": 0.15,
      "grad_norm": 17.407425843058338,
      "learning_rate": 9.629691397785294e-06,
      "loss": 0.5085,
      "step": 929
    },
    {
      "epoch": 0.15,
      "grad_norm": 41.441168088491075,
      "learning_rate": 9.628705137881509e-06,
      "loss": 0.445,
      "step": 930
    },
    {
      "epoch": 0.15,
      "grad_norm": 67.10488979067183,
      "learning_rate": 9.627717616990853e-06,
      "loss": 0.451,
      "step": 931
    },
    {
      "epoch": 0.15,
      "grad_norm": 18.811327948317675,
      "learning_rate": 9.626728835382353e-06,
      "loss": 0.5455,
      "step": 932
    },
    {
      "epoch": 0.15,
      "grad_norm": 16.04187581422627,
      "learning_rate": 9.625738793325384e-06,
      "loss": 0.481,
      "step": 933
    },
    {
      "epoch": 0.15,
      "grad_norm": 31.82019051690305,
      "learning_rate": 9.624747491089657e-06,
      "loss": 0.557,
      "step": 934
    },
    {
      "epoch": 0.15,
      "grad_norm": 13.427645062972102,
      "learning_rate": 9.623754928945233e-06,
      "loss": 0.4604,
      "step": 935
    },
    {
      "epoch": 0.15,
      "grad_norm": 12.145064333624996,
      "learning_rate": 9.622761107162511e-06,
      "loss": 0.3854,
      "step": 936
    },
    {
      "epoch": 0.15,
      "grad_norm": 24.39856168953827,
      "learning_rate": 9.621766026012236e-06,
      "loss": 0.4084,
      "step": 937
    },
    {
      "epoch": 0.15,
      "grad_norm": 17.287502710482805,
      "learning_rate": 9.620769685765497e-06,
      "loss": 0.5109,
      "step": 938
    },
    {
      "epoch": 0.15,
      "grad_norm": 15.869683750374959,
      "learning_rate": 9.619772086693721e-06,
      "loss": 0.4883,
      "step": 939
    },
    {
      "epoch": 0.15,
      "grad_norm": 11.026502508984617,
      "learning_rate": 9.618773229068685e-06,
      "loss": 0.5207,
      "step": 940
    },
    {
      "epoch": 0.15,
      "grad_norm": 19.94769006365636,
      "learning_rate": 9.617773113162505e-06,
      "loss": 0.4478,
      "step": 941
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6611989853004538,
      "learning_rate": 9.616771739247639e-06,
      "loss": 0.4926,
      "step": 942
    },
    {
      "epoch": 0.15,
      "grad_norm": 29.964661697397037,
      "learning_rate": 9.615769107596888e-06,
      "loss": 0.4905,
      "step": 943
    },
    {
      "epoch": 0.15,
      "grad_norm": 16.413503397210974,
      "learning_rate": 9.614765218483398e-06,
      "loss": 0.4928,
      "step": 944
    },
    {
      "epoch": 0.15,
      "grad_norm": 20.424786699059403,
      "learning_rate": 9.613760072180656e-06,
      "loss": 0.5097,
      "step": 945
    },
    {
      "epoch": 0.15,
      "grad_norm": 14.706280523892099,
      "learning_rate": 9.612753668962492e-06,
      "loss": 0.4844,
      "step": 946
    },
    {
      "epoch": 0.15,
      "grad_norm": 11.79609289935516,
      "learning_rate": 9.611746009103077e-06,
      "loss": 0.4802,
      "step": 947
    },
    {
      "epoch": 0.15,
      "grad_norm": 11.701576662868932,
      "learning_rate": 9.610737092876924e-06,
      "loss": 0.5289,
      "step": 948
    },
    {
      "epoch": 0.15,
      "grad_norm": 11.092264536235662,
      "learning_rate": 9.609726920558893e-06,
      "loss": 0.4706,
      "step": 949
    },
    {
      "epoch": 0.15,
      "grad_norm": 29.435131489183174,
      "learning_rate": 9.60871549242418e-06,
      "loss": 0.4624,
      "step": 950
    },
    {
      "epoch": 0.15,
      "grad_norm": 10.296228534653492,
      "learning_rate": 9.607702808748329e-06,
      "loss": 0.4705,
      "step": 951
    },
    {
      "epoch": 0.15,
      "grad_norm": 29.626415975793552,
      "learning_rate": 9.60668886980722e-06,
      "loss": 0.4618,
      "step": 952
    },
    {
      "epoch": 0.15,
      "grad_norm": 10.609088950874494,
      "learning_rate": 9.605673675877077e-06,
      "loss": 0.4183,
      "step": 953
    },
    {
      "epoch": 0.15,
      "grad_norm": 23.558458313023788,
      "learning_rate": 9.604657227234468e-06,
      "loss": 0.5323,
      "step": 954
    },
    {
      "epoch": 0.15,
      "grad_norm": 11.798142775792458,
      "learning_rate": 9.603639524156306e-06,
      "loss": 0.5021,
      "step": 955
    },
    {
      "epoch": 0.15,
      "grad_norm": 12.267768452020524,
      "learning_rate": 9.602620566919834e-06,
      "loss": 0.4778,
      "step": 956
    },
    {
      "epoch": 0.15,
      "grad_norm": 44.57079061569597,
      "learning_rate": 9.601600355802649e-06,
      "loss": 0.4635,
      "step": 957
    },
    {
      "epoch": 0.15,
      "grad_norm": 16.502880927289095,
      "learning_rate": 9.600578891082683e-06,
      "loss": 0.5217,
      "step": 958
    },
    {
      "epoch": 0.15,
      "grad_norm": 21.689018152019912,
      "learning_rate": 9.599556173038211e-06,
      "loss": 0.4269,
      "step": 959
    },
    {
      "epoch": 0.15,
      "grad_norm": 14.531986554194695,
      "learning_rate": 9.59853220194785e-06,
      "loss": 0.4162,
      "step": 960
    },
    {
      "epoch": 0.15,
      "grad_norm": 32.57856162932613,
      "learning_rate": 9.597506978090558e-06,
      "loss": 0.4277,
      "step": 961
    },
    {
      "epoch": 0.15,
      "grad_norm": 19.578718641830992,
      "learning_rate": 9.596480501745635e-06,
      "loss": 0.5532,
      "step": 962
    },
    {
      "epoch": 0.16,
      "grad_norm": 20.304897535333694,
      "learning_rate": 9.59545277319272e-06,
      "loss": 0.4184,
      "step": 963
    },
    {
      "epoch": 0.16,
      "grad_norm": 16.85257954647598,
      "learning_rate": 9.594423792711796e-06,
      "loss": 0.5085,
      "step": 964
    },
    {
      "epoch": 0.16,
      "grad_norm": 13.092244342434318,
      "learning_rate": 9.593393560583184e-06,
      "loss": 0.5103,
      "step": 965
    },
    {
      "epoch": 0.16,
      "grad_norm": 40.08298882340636,
      "learning_rate": 9.592362077087548e-06,
      "loss": 0.5485,
      "step": 966
    },
    {
      "epoch": 0.16,
      "grad_norm": 20.483878434898017,
      "learning_rate": 9.591329342505894e-06,
      "loss": 0.4658,
      "step": 967
    },
    {
      "epoch": 0.16,
      "grad_norm": 17.234616929066743,
      "learning_rate": 9.590295357119565e-06,
      "loss": 0.4832,
      "step": 968
    },
    {
      "epoch": 0.16,
      "grad_norm": 11.947411677599721,
      "learning_rate": 9.589260121210249e-06,
      "loss": 0.5231,
      "step": 969
    },
    {
      "epoch": 0.16,
      "grad_norm": 17.75116681043414,
      "learning_rate": 9.588223635059973e-06,
      "loss": 0.455,
      "step": 970
    },
    {
      "epoch": 0.16,
      "grad_norm": 16.700109917417805,
      "learning_rate": 9.587185898951104e-06,
      "loss": 0.4448,
      "step": 971
    },
    {
      "epoch": 0.16,
      "grad_norm": 14.172756837957563,
      "learning_rate": 9.58614691316635e-06,
      "loss": 0.4755,
      "step": 972
    },
    {
      "epoch": 0.16,
      "grad_norm": 13.130934611983747,
      "learning_rate": 9.585106677988756e-06,
      "loss": 0.4213,
      "step": 973
    },
    {
      "epoch": 0.16,
      "grad_norm": 13.93130557338124,
      "learning_rate": 9.584065193701717e-06,
      "loss": 0.3945,
      "step": 974
    },
    {
      "epoch": 0.16,
      "grad_norm": 14.306927039039563,
      "learning_rate": 9.583022460588956e-06,
      "loss": 0.484,
      "step": 975
    },
    {
      "epoch": 0.16,
      "grad_norm": 19.085808779093785,
      "learning_rate": 9.581978478934547e-06,
      "loss": 0.4268,
      "step": 976
    },
    {
      "epoch": 0.16,
      "grad_norm": 12.889267451609474,
      "learning_rate": 9.580933249022896e-06,
      "loss": 0.4535,
      "step": 977
    },
    {
      "epoch": 0.16,
      "grad_norm": 20.593346115144758,
      "learning_rate": 9.579886771138754e-06,
      "loss": 0.5139,
      "step": 978
    },
    {
      "epoch": 0.16,
      "grad_norm": 13.172950097501262,
      "learning_rate": 9.578839045567212e-06,
      "loss": 0.5372,
      "step": 979
    },
    {
      "epoch": 0.16,
      "grad_norm": 11.895476707079597,
      "learning_rate": 9.577790072593696e-06,
      "loss": 0.4928,
      "step": 980
    },
    {
      "epoch": 0.16,
      "grad_norm": 12.14817049309049,
      "learning_rate": 9.576739852503976e-06,
      "loss": 0.4348,
      "step": 981
    },
    {
      "epoch": 0.16,
      "grad_norm": 18.441425954475015,
      "learning_rate": 9.575688385584162e-06,
      "loss": 0.4806,
      "step": 982
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.128038132817784,
      "learning_rate": 9.574635672120705e-06,
      "loss": 0.4751,
      "step": 983
    },
    {
      "epoch": 0.16,
      "grad_norm": 18.973098666003366,
      "learning_rate": 9.573581712400386e-06,
      "loss": 0.4333,
      "step": 984
    },
    {
      "epoch": 0.16,
      "grad_norm": 16.796792041474703,
      "learning_rate": 9.57252650671034e-06,
      "loss": 0.4671,
      "step": 985
    },
    {
      "epoch": 0.16,
      "grad_norm": 22.827199612453057,
      "learning_rate": 9.57147005533803e-06,
      "loss": 0.4823,
      "step": 986
    },
    {
      "epoch": 0.16,
      "grad_norm": 17.57179750769814,
      "learning_rate": 9.570412358571264e-06,
      "loss": 0.5187,
      "step": 987
    },
    {
      "epoch": 0.16,
      "grad_norm": 13.410651668499781,
      "learning_rate": 9.569353416698187e-06,
      "loss": 0.5501,
      "step": 988
    },
    {
      "epoch": 0.16,
      "grad_norm": 19.774738320693395,
      "learning_rate": 9.568293230007284e-06,
      "loss": 0.563,
      "step": 989
    },
    {
      "epoch": 0.16,
      "grad_norm": 12.193087345047092,
      "learning_rate": 9.56723179878738e-06,
      "loss": 0.5148,
      "step": 990
    },
    {
      "epoch": 0.16,
      "grad_norm": 13.06236893677764,
      "learning_rate": 9.566169123327638e-06,
      "loss": 0.504,
      "step": 991
    },
    {
      "epoch": 0.16,
      "grad_norm": 15.382162286616634,
      "learning_rate": 9.565105203917559e-06,
      "loss": 0.4874,
      "step": 992
    },
    {
      "epoch": 0.16,
      "grad_norm": 12.493745017555534,
      "learning_rate": 9.564040040846984e-06,
      "loss": 0.4602,
      "step": 993
    },
    {
      "epoch": 0.16,
      "grad_norm": 13.95916096139023,
      "learning_rate": 9.562973634406096e-06,
      "loss": 0.4502,
      "step": 994
    },
    {
      "epoch": 0.16,
      "grad_norm": 14.877131856974902,
      "learning_rate": 9.561905984885407e-06,
      "loss": 0.4852,
      "step": 995
    },
    {
      "epoch": 0.16,
      "grad_norm": 17.394463385448358,
      "learning_rate": 9.560837092575781e-06,
      "loss": 0.4095,
      "step": 996
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.687329726858136,
      "learning_rate": 9.55976695776841e-06,
      "loss": 0.5273,
      "step": 997
    },
    {
      "epoch": 0.16,
      "grad_norm": 14.113261260705103,
      "learning_rate": 9.558695580754828e-06,
      "loss": 0.5204,
      "step": 998
    },
    {
      "epoch": 0.16,
      "grad_norm": 12.484344485846112,
      "learning_rate": 9.55762296182691e-06,
      "loss": 0.4668,
      "step": 999
    },
    {
      "epoch": 0.16,
      "grad_norm": 35.16102900153671,
      "learning_rate": 9.556549101276865e-06,
      "loss": 0.49,
      "step": 1000
    },
    {
      "epoch": 0.16,
      "grad_norm": 13.601122733500624,
      "learning_rate": 9.555473999397242e-06,
      "loss": 0.4518,
      "step": 1001
    },
    {
      "epoch": 0.16,
      "grad_norm": 19.973026517191816,
      "learning_rate": 9.55439765648093e-06,
      "loss": 0.5142,
      "step": 1002
    },
    {
      "epoch": 0.16,
      "grad_norm": 12.74323779918366,
      "learning_rate": 9.553320072821154e-06,
      "loss": 0.5706,
      "step": 1003
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.528235325119805,
      "learning_rate": 9.552241248711478e-06,
      "loss": 0.4327,
      "step": 1004
    },
    {
      "epoch": 0.16,
      "grad_norm": 13.804643031667975,
      "learning_rate": 9.551161184445801e-06,
      "loss": 0.4451,
      "step": 1005
    },
    {
      "epoch": 0.16,
      "grad_norm": 21.665667807762887,
      "learning_rate": 9.550079880318363e-06,
      "loss": 0.4484,
      "step": 1006
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3581474010418773,
      "learning_rate": 9.548997336623744e-06,
      "loss": 0.4666,
      "step": 1007
    },
    {
      "epoch": 0.16,
      "grad_norm": 15.79477809188708,
      "learning_rate": 9.547913553656856e-06,
      "loss": 0.4617,
      "step": 1008
    },
    {
      "epoch": 0.16,
      "grad_norm": 14.946613656419078,
      "learning_rate": 9.54682853171295e-06,
      "loss": 0.459,
      "step": 1009
    },
    {
      "epoch": 0.16,
      "grad_norm": 31.964391597437995,
      "learning_rate": 9.545742271087622e-06,
      "loss": 0.562,
      "step": 1010
    },
    {
      "epoch": 0.16,
      "grad_norm": 15.483170294856006,
      "learning_rate": 9.54465477207679e-06,
      "loss": 0.5087,
      "step": 1011
    },
    {
      "epoch": 0.16,
      "grad_norm": 22.400086590843667,
      "learning_rate": 9.543566034976728e-06,
      "loss": 0.4621,
      "step": 1012
    },
    {
      "epoch": 0.16,
      "grad_norm": 14.139788394314213,
      "learning_rate": 9.542476060084035e-06,
      "loss": 0.498,
      "step": 1013
    },
    {
      "epoch": 0.16,
      "grad_norm": 12.555769415941954,
      "learning_rate": 9.541384847695645e-06,
      "loss": 0.4352,
      "step": 1014
    },
    {
      "epoch": 0.16,
      "grad_norm": 16.57216535857035,
      "learning_rate": 9.540292398108842e-06,
      "loss": 0.4626,
      "step": 1015
    },
    {
      "epoch": 0.16,
      "grad_norm": 13.499571704566081,
      "learning_rate": 9.539198711621234e-06,
      "loss": 0.4518,
      "step": 1016
    },
    {
      "epoch": 0.16,
      "grad_norm": 13.234664693544344,
      "learning_rate": 9.538103788530773e-06,
      "loss": 0.4402,
      "step": 1017
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.545299194691657,
      "learning_rate": 9.537007629135745e-06,
      "loss": 0.4387,
      "step": 1018
    },
    {
      "epoch": 0.16,
      "grad_norm": 14.18574041740857,
      "learning_rate": 9.53591023373478e-06,
      "loss": 0.4308,
      "step": 1019
    },
    {
      "epoch": 0.16,
      "grad_norm": 14.5217921791035,
      "learning_rate": 9.53481160262683e-06,
      "loss": 0.4181,
      "step": 1020
    },
    {
      "epoch": 0.16,
      "grad_norm": 79.58655211806712,
      "learning_rate": 9.5337117361112e-06,
      "loss": 0.4888,
      "step": 1021
    },
    {
      "epoch": 0.16,
      "grad_norm": 11.888995146115528,
      "learning_rate": 9.532610634487519e-06,
      "loss": 0.4874,
      "step": 1022
    },
    {
      "epoch": 0.16,
      "grad_norm": 19.08221064076107,
      "learning_rate": 9.531508298055758e-06,
      "loss": 0.5789,
      "step": 1023
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.97861608552254,
      "learning_rate": 9.530404727116225e-06,
      "loss": 0.454,
      "step": 1024
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2010297306300344,
      "learning_rate": 9.529299921969563e-06,
      "loss": 0.4417,
      "step": 1025
    },
    {
      "epoch": 0.17,
      "grad_norm": 12.80708567197285,
      "learning_rate": 9.528193882916753e-06,
      "loss": 0.4841,
      "step": 1026
    },
    {
      "epoch": 0.17,
      "grad_norm": 22.150938767211656,
      "learning_rate": 9.52708661025911e-06,
      "loss": 0.464,
      "step": 1027
    },
    {
      "epoch": 0.17,
      "grad_norm": 13.515352821430874,
      "learning_rate": 9.525978104298282e-06,
      "loss": 0.4845,
      "step": 1028
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.1434025319412555,
      "learning_rate": 9.524868365336259e-06,
      "loss": 0.5679,
      "step": 1029
    },
    {
      "epoch": 0.17,
      "grad_norm": 10.126513833511396,
      "learning_rate": 9.523757393675365e-06,
      "loss": 0.4565,
      "step": 1030
    },
    {
      "epoch": 0.17,
      "grad_norm": 18.142325378882944,
      "learning_rate": 9.522645189618258e-06,
      "loss": 0.4757,
      "step": 1031
    },
    {
      "epoch": 0.17,
      "grad_norm": 17.91740615647938,
      "learning_rate": 9.521531753467935e-06,
      "loss": 0.4313,
      "step": 1032
    },
    {
      "epoch": 0.17,
      "grad_norm": 17.608969009291357,
      "learning_rate": 9.520417085527725e-06,
      "loss": 0.5528,
      "step": 1033
    },
    {
      "epoch": 0.17,
      "grad_norm": 36.056968469451256,
      "learning_rate": 9.519301186101297e-06,
      "loss": 0.4697,
      "step": 1034
    },
    {
      "epoch": 0.17,
      "grad_norm": 11.163717110117974,
      "learning_rate": 9.518184055492648e-06,
      "loss": 0.5005,
      "step": 1035
    },
    {
      "epoch": 0.17,
      "grad_norm": 20.275226388453046,
      "learning_rate": 9.517065694006118e-06,
      "loss": 0.4848,
      "step": 1036
    },
    {
      "epoch": 0.17,
      "grad_norm": 13.672904576323003,
      "learning_rate": 9.51594610194638e-06,
      "loss": 0.4894,
      "step": 1037
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4247903404095903,
      "learning_rate": 9.514825279618443e-06,
      "loss": 0.4839,
      "step": 1038
    },
    {
      "epoch": 0.17,
      "grad_norm": 21.880846935059704,
      "learning_rate": 9.513703227327646e-06,
      "loss": 0.4823,
      "step": 1039
    },
    {
      "epoch": 0.17,
      "grad_norm": 16.33271826037092,
      "learning_rate": 9.512579945379668e-06,
      "loss": 0.4896,
      "step": 1040
    },
    {
      "epoch": 0.17,
      "grad_norm": 34.12762389932343,
      "learning_rate": 9.511455434080525e-06,
      "loss": 0.4468,
      "step": 1041
    },
    {
      "epoch": 0.17,
      "grad_norm": 47.95397372075502,
      "learning_rate": 9.51032969373656e-06,
      "loss": 0.5101,
      "step": 1042
    },
    {
      "epoch": 0.17,
      "grad_norm": 13.865484661601567,
      "learning_rate": 9.50920272465446e-06,
      "loss": 0.5096,
      "step": 1043
    },
    {
      "epoch": 0.17,
      "grad_norm": 19.991221702462603,
      "learning_rate": 9.508074527141243e-06,
      "loss": 0.4924,
      "step": 1044
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.593405773468348,
      "learning_rate": 9.506945101504256e-06,
      "loss": 0.4791,
      "step": 1045
    },
    {
      "epoch": 0.17,
      "grad_norm": 12.228471517393855,
      "learning_rate": 9.50581444805119e-06,
      "loss": 0.5499,
      "step": 1046
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2120997499365314,
      "learning_rate": 9.504682567090064e-06,
      "loss": 0.4672,
      "step": 1047
    },
    {
      "epoch": 0.17,
      "grad_norm": 12.897666136524794,
      "learning_rate": 9.503549458929236e-06,
      "loss": 0.4759,
      "step": 1048
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.044085415950523,
      "learning_rate": 9.502415123877393e-06,
      "loss": 0.4373,
      "step": 1049
    },
    {
      "epoch": 0.17,
      "grad_norm": 16.48255701514367,
      "learning_rate": 9.501279562243561e-06,
      "loss": 0.5126,
      "step": 1050
    },
    {
      "epoch": 0.17,
      "grad_norm": 10.895304992621552,
      "learning_rate": 9.500142774337097e-06,
      "loss": 0.5163,
      "step": 1051
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.148211940230544,
      "learning_rate": 9.499004760467694e-06,
      "loss": 0.4558,
      "step": 1052
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.859836359554315,
      "learning_rate": 9.49786552094538e-06,
      "loss": 0.4317,
      "step": 1053
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.776650772075477,
      "learning_rate": 9.496725056080512e-06,
      "loss": 0.5136,
      "step": 1054
    },
    {
      "epoch": 0.17,
      "grad_norm": 20.957341788516025,
      "learning_rate": 9.495583366183788e-06,
      "loss": 0.4864,
      "step": 1055
    },
    {
      "epoch": 0.17,
      "grad_norm": 13.940883647140035,
      "learning_rate": 9.494440451566232e-06,
      "loss": 0.4566,
      "step": 1056
    },
    {
      "epoch": 0.17,
      "grad_norm": 18.690551730173826,
      "learning_rate": 9.493296312539206e-06,
      "loss": 0.5571,
      "step": 1057
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.209947090213658,
      "learning_rate": 9.492150949414408e-06,
      "loss": 0.5108,
      "step": 1058
    },
    {
      "epoch": 0.17,
      "grad_norm": 11.507618148830549,
      "learning_rate": 9.491004362503862e-06,
      "loss": 0.4612,
      "step": 1059
    },
    {
      "epoch": 0.17,
      "grad_norm": 42.213896385302455,
      "learning_rate": 9.489856552119934e-06,
      "loss": 0.5285,
      "step": 1060
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.12317444130324,
      "learning_rate": 9.488707518575319e-06,
      "loss": 0.401,
      "step": 1061
    },
    {
      "epoch": 0.17,
      "grad_norm": 21.878887963752344,
      "learning_rate": 9.487557262183043e-06,
      "loss": 0.4288,
      "step": 1062
    },
    {
      "epoch": 0.17,
      "grad_norm": 17.748352108962035,
      "learning_rate": 9.486405783256467e-06,
      "loss": 0.4927,
      "step": 1063
    },
    {
      "epoch": 0.17,
      "grad_norm": 27.55719417486496,
      "learning_rate": 9.485253082109289e-06,
      "loss": 0.4222,
      "step": 1064
    },
    {
      "epoch": 0.17,
      "grad_norm": 11.435198637025392,
      "learning_rate": 9.484099159055534e-06,
      "loss": 0.5437,
      "step": 1065
    },
    {
      "epoch": 0.17,
      "grad_norm": 17.12416456475259,
      "learning_rate": 9.482944014409563e-06,
      "loss": 0.4986,
      "step": 1066
    },
    {
      "epoch": 0.17,
      "grad_norm": 16.95971556895982,
      "learning_rate": 9.481787648486069e-06,
      "loss": 0.4986,
      "step": 1067
    },
    {
      "epoch": 0.17,
      "grad_norm": 10.982524039969313,
      "learning_rate": 9.480630061600079e-06,
      "loss": 0.4951,
      "step": 1068
    },
    {
      "epoch": 0.17,
      "grad_norm": 35.488620954847754,
      "learning_rate": 9.47947125406695e-06,
      "loss": 0.4603,
      "step": 1069
    },
    {
      "epoch": 0.17,
      "grad_norm": 11.756908885412711,
      "learning_rate": 9.478311226202375e-06,
      "loss": 0.5257,
      "step": 1070
    },
    {
      "epoch": 0.17,
      "grad_norm": 15.251827708371087,
      "learning_rate": 9.477149978322378e-06,
      "loss": 0.4619,
      "step": 1071
    },
    {
      "epoch": 0.17,
      "grad_norm": 13.01838895936322,
      "learning_rate": 9.475987510743311e-06,
      "loss": 0.5302,
      "step": 1072
    },
    {
      "epoch": 0.17,
      "grad_norm": 17.798763378924722,
      "learning_rate": 9.474823823781866e-06,
      "loss": 0.5385,
      "step": 1073
    },
    {
      "epoch": 0.17,
      "grad_norm": 14.464951985135798,
      "learning_rate": 9.473658917755063e-06,
      "loss": 0.4508,
      "step": 1074
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.965477806289659,
      "learning_rate": 9.472492792980252e-06,
      "loss": 0.4697,
      "step": 1075
    },
    {
      "epoch": 0.17,
      "grad_norm": 11.487399091539281,
      "learning_rate": 9.471325449775119e-06,
      "loss": 0.4747,
      "step": 1076
    },
    {
      "epoch": 0.17,
      "grad_norm": 11.878887877521795,
      "learning_rate": 9.470156888457682e-06,
      "loss": 0.4411,
      "step": 1077
    },
    {
      "epoch": 0.17,
      "grad_norm": 11.2156365414979,
      "learning_rate": 9.468987109346288e-06,
      "loss": 0.4737,
      "step": 1078
    },
    {
      "epoch": 0.17,
      "grad_norm": 10.988792573984448,
      "learning_rate": 9.467816112759616e-06,
      "loss": 0.5397,
      "step": 1079
    },
    {
      "epoch": 0.17,
      "grad_norm": 16.641094947310126,
      "learning_rate": 9.46664389901668e-06,
      "loss": 0.4609,
      "step": 1080
    },
    {
      "epoch": 0.17,
      "grad_norm": 23.815410443281895,
      "learning_rate": 9.465470468436822e-06,
      "loss": 0.4861,
      "step": 1081
    },
    {
      "epoch": 0.17,
      "grad_norm": 15.637081850511155,
      "learning_rate": 9.464295821339715e-06,
      "loss": 0.5018,
      "step": 1082
    },
    {
      "epoch": 0.17,
      "grad_norm": 12.22993165834608,
      "learning_rate": 9.46311995804537e-06,
      "loss": 0.4241,
      "step": 1083
    },
    {
      "epoch": 0.17,
      "grad_norm": 15.578914863482261,
      "learning_rate": 9.46194287887412e-06,
      "loss": 0.4695,
      "step": 1084
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.672083217835114,
      "learning_rate": 9.460764584146635e-06,
      "loss": 0.4722,
      "step": 1085
    },
    {
      "epoch": 0.17,
      "grad_norm": 15.475546803637759,
      "learning_rate": 9.459585074183919e-06,
      "loss": 0.4923,
      "step": 1086
    },
    {
      "epoch": 0.18,
      "grad_norm": 23.204796868705444,
      "learning_rate": 9.458404349307295e-06,
      "loss": 0.491,
      "step": 1087
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4269469763334337,
      "learning_rate": 9.457222409838433e-06,
      "loss": 0.499,
      "step": 1088
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.623904209983921,
      "learning_rate": 9.456039256099321e-06,
      "loss": 0.4336,
      "step": 1089
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.876087409419991,
      "learning_rate": 9.454854888412285e-06,
      "loss": 0.507,
      "step": 1090
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.398020604301625,
      "learning_rate": 9.453669307099978e-06,
      "loss": 0.5097,
      "step": 1091
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.654028564640374,
      "learning_rate": 9.452482512485386e-06,
      "loss": 0.4948,
      "step": 1092
    },
    {
      "epoch": 0.18,
      "grad_norm": 16.63963642095185,
      "learning_rate": 9.451294504891824e-06,
      "loss": 0.4649,
      "step": 1093
    },
    {
      "epoch": 0.18,
      "grad_norm": 16.693549034747875,
      "learning_rate": 9.450105284642938e-06,
      "loss": 0.5117,
      "step": 1094
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1640578879525318,
      "learning_rate": 9.448914852062705e-06,
      "loss": 0.4771,
      "step": 1095
    },
    {
      "epoch": 0.18,
      "grad_norm": 12.351754912346847,
      "learning_rate": 9.447723207475432e-06,
      "loss": 0.4351,
      "step": 1096
    },
    {
      "epoch": 0.18,
      "grad_norm": 18.223535066038824,
      "learning_rate": 9.446530351205754e-06,
      "loss": 0.5054,
      "step": 1097
    },
    {
      "epoch": 0.18,
      "grad_norm": 18.638419168632442,
      "learning_rate": 9.445336283578639e-06,
      "loss": 0.5412,
      "step": 1098
    },
    {
      "epoch": 0.18,
      "grad_norm": 23.48201034905293,
      "learning_rate": 9.444141004919385e-06,
      "loss": 0.5002,
      "step": 1099
    },
    {
      "epoch": 0.18,
      "grad_norm": 24.948653576937204,
      "learning_rate": 9.442944515553616e-06,
      "loss": 0.4789,
      "step": 1100
    },
    {
      "epoch": 0.18,
      "grad_norm": 13.252203421307753,
      "learning_rate": 9.441746815807292e-06,
      "loss": 0.4985,
      "step": 1101
    },
    {
      "epoch": 0.18,
      "grad_norm": 22.332822954586746,
      "learning_rate": 9.440547906006697e-06,
      "loss": 0.5795,
      "step": 1102
    },
    {
      "epoch": 0.18,
      "grad_norm": 12.880891534138579,
      "learning_rate": 9.439347786478448e-06,
      "loss": 0.5138,
      "step": 1103
    },
    {
      "epoch": 0.18,
      "grad_norm": 84.51759881771268,
      "learning_rate": 9.438146457549491e-06,
      "loss": 0.4963,
      "step": 1104
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.341451261441499,
      "learning_rate": 9.436943919547101e-06,
      "loss": 0.4957,
      "step": 1105
    },
    {
      "epoch": 0.18,
      "grad_norm": 12.041994111552437,
      "learning_rate": 9.435740172798881e-06,
      "loss": 0.4649,
      "step": 1106
    },
    {
      "epoch": 0.18,
      "grad_norm": 17.791394882417716,
      "learning_rate": 9.434535217632768e-06,
      "loss": 0.4981,
      "step": 1107
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.120870935680138,
      "learning_rate": 9.433329054377023e-06,
      "loss": 0.4958,
      "step": 1108
    },
    {
      "epoch": 0.18,
      "grad_norm": 15.132556810973604,
      "learning_rate": 9.432121683360235e-06,
      "loss": 0.5198,
      "step": 1109
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.80646202797608,
      "learning_rate": 9.430913104911331e-06,
      "loss": 0.4607,
      "step": 1110
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.030925059144677,
      "learning_rate": 9.429703319359558e-06,
      "loss": 0.4939,
      "step": 1111
    },
    {
      "epoch": 0.18,
      "grad_norm": 50.65573650694543,
      "learning_rate": 9.428492327034496e-06,
      "loss": 0.4355,
      "step": 1112
    },
    {
      "epoch": 0.18,
      "grad_norm": 12.188372492125053,
      "learning_rate": 9.427280128266049e-06,
      "loss": 0.4479,
      "step": 1113
    },
    {
      "epoch": 0.18,
      "grad_norm": 18.90073627757158,
      "learning_rate": 9.42606672338446e-06,
      "loss": 0.5341,
      "step": 1114
    },
    {
      "epoch": 0.18,
      "grad_norm": 18.311445664301328,
      "learning_rate": 9.424852112720289e-06,
      "loss": 0.4275,
      "step": 1115
    },
    {
      "epoch": 0.18,
      "grad_norm": 23.94973155526467,
      "learning_rate": 9.42363629660443e-06,
      "loss": 0.4815,
      "step": 1116
    },
    {
      "epoch": 0.18,
      "grad_norm": 45.12101300161726,
      "learning_rate": 9.422419275368107e-06,
      "loss": 0.4646,
      "step": 1117
    },
    {
      "epoch": 0.18,
      "grad_norm": 33.981706193205305,
      "learning_rate": 9.421201049342867e-06,
      "loss": 0.4381,
      "step": 1118
    },
    {
      "epoch": 0.18,
      "grad_norm": 29.862699745094982,
      "learning_rate": 9.41998161886059e-06,
      "loss": 0.4864,
      "step": 1119
    },
    {
      "epoch": 0.18,
      "grad_norm": 31.87468476084896,
      "learning_rate": 9.418760984253485e-06,
      "loss": 0.4884,
      "step": 1120
    },
    {
      "epoch": 0.18,
      "grad_norm": 18.145408793170006,
      "learning_rate": 9.417539145854082e-06,
      "loss": 0.497,
      "step": 1121
    },
    {
      "epoch": 0.18,
      "grad_norm": 33.32486798994945,
      "learning_rate": 9.416316103995246e-06,
      "loss": 0.4717,
      "step": 1122
    },
    {
      "epoch": 0.18,
      "grad_norm": 26.18179504580407,
      "learning_rate": 9.415091859010165e-06,
      "loss": 0.4646,
      "step": 1123
    },
    {
      "epoch": 0.18,
      "grad_norm": 21.537429592767765,
      "learning_rate": 9.413866411232361e-06,
      "loss": 0.4961,
      "step": 1124
    },
    {
      "epoch": 0.18,
      "grad_norm": 14.387111792037299,
      "learning_rate": 9.41263976099568e-06,
      "loss": 0.4726,
      "step": 1125
    },
    {
      "epoch": 0.18,
      "grad_norm": 22.69719255173724,
      "learning_rate": 9.411411908634289e-06,
      "loss": 0.4469,
      "step": 1126
    },
    {
      "epoch": 0.18,
      "grad_norm": 17.25897190677966,
      "learning_rate": 9.410182854482693e-06,
      "loss": 0.4663,
      "step": 1127
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3087366906080624,
      "learning_rate": 9.40895259887572e-06,
      "loss": 0.4837,
      "step": 1128
    },
    {
      "epoch": 0.18,
      "grad_norm": 32.081016827106815,
      "learning_rate": 9.407721142148528e-06,
      "loss": 0.4668,
      "step": 1129
    },
    {
      "epoch": 0.18,
      "grad_norm": 14.396009357773716,
      "learning_rate": 9.406488484636595e-06,
      "loss": 0.5448,
      "step": 1130
    },
    {
      "epoch": 0.18,
      "grad_norm": 30.392799335214516,
      "learning_rate": 9.405254626675736e-06,
      "loss": 0.5459,
      "step": 1131
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.52677066210452,
      "learning_rate": 9.404019568602081e-06,
      "loss": 0.4165,
      "step": 1132
    },
    {
      "epoch": 0.18,
      "grad_norm": 17.485077966435156,
      "learning_rate": 9.402783310752102e-06,
      "loss": 0.4647,
      "step": 1133
    },
    {
      "epoch": 0.18,
      "grad_norm": 26.14712993730686,
      "learning_rate": 9.401545853462584e-06,
      "loss": 0.4324,
      "step": 1134
    },
    {
      "epoch": 0.18,
      "grad_norm": 206.99213813279073,
      "learning_rate": 9.400307197070646e-06,
      "loss": 0.4845,
      "step": 1135
    },
    {
      "epoch": 0.18,
      "grad_norm": 16.172639217316522,
      "learning_rate": 9.399067341913732e-06,
      "loss": 0.4469,
      "step": 1136
    },
    {
      "epoch": 0.18,
      "grad_norm": 20.885640090923513,
      "learning_rate": 9.397826288329615e-06,
      "loss": 0.4737,
      "step": 1137
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.240214926222737,
      "learning_rate": 9.396584036656388e-06,
      "loss": 0.4828,
      "step": 1138
    },
    {
      "epoch": 0.18,
      "grad_norm": 19.212690911137276,
      "learning_rate": 9.395340587232478e-06,
      "loss": 0.5325,
      "step": 1139
    },
    {
      "epoch": 0.18,
      "grad_norm": 19.765604811065216,
      "learning_rate": 9.394095940396632e-06,
      "loss": 0.5173,
      "step": 1140
    },
    {
      "epoch": 0.18,
      "grad_norm": 88.99170683412197,
      "learning_rate": 9.39285009648793e-06,
      "loss": 0.3996,
      "step": 1141
    },
    {
      "epoch": 0.18,
      "grad_norm": 17.9497957789715,
      "learning_rate": 9.391603055845771e-06,
      "loss": 0.5323,
      "step": 1142
    },
    {
      "epoch": 0.18,
      "grad_norm": 10.257355218001061,
      "learning_rate": 9.390354818809886e-06,
      "loss": 0.5556,
      "step": 1143
    },
    {
      "epoch": 0.18,
      "grad_norm": 12.07665114170986,
      "learning_rate": 9.389105385720325e-06,
      "loss": 0.4289,
      "step": 1144
    },
    {
      "epoch": 0.18,
      "grad_norm": 21.817882344400445,
      "learning_rate": 9.387854756917472e-06,
      "loss": 0.4549,
      "step": 1145
    },
    {
      "epoch": 0.18,
      "grad_norm": 16.62139903255941,
      "learning_rate": 9.386602932742032e-06,
      "loss": 0.4868,
      "step": 1146
    },
    {
      "epoch": 0.18,
      "grad_norm": 10.06137492986953,
      "learning_rate": 9.385349913535034e-06,
      "loss": 0.5043,
      "step": 1147
    },
    {
      "epoch": 0.18,
      "grad_norm": 13.864672919800334,
      "learning_rate": 9.384095699637839e-06,
      "loss": 0.4445,
      "step": 1148
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.253327053047788,
      "learning_rate": 9.382840291392124e-06,
      "loss": 0.4224,
      "step": 1149
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.046173474935681,
      "learning_rate": 9.3815836891399e-06,
      "loss": 0.4754,
      "step": 1150
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.489242303268826,
      "learning_rate": 9.380325893223503e-06,
      "loss": 0.5442,
      "step": 1151
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0869260441775574,
      "learning_rate": 9.379066903985588e-06,
      "loss": 0.4446,
      "step": 1152
    },
    {
      "epoch": 0.19,
      "grad_norm": 19.592258237974765,
      "learning_rate": 9.377806721769138e-06,
      "loss": 0.4478,
      "step": 1153
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.748546834920383,
      "learning_rate": 9.37654534691746e-06,
      "loss": 0.4895,
      "step": 1154
    },
    {
      "epoch": 0.19,
      "grad_norm": 11.62312580828369,
      "learning_rate": 9.37528277977419e-06,
      "loss": 0.4243,
      "step": 1155
    },
    {
      "epoch": 0.19,
      "grad_norm": 35.12877145727759,
      "learning_rate": 9.374019020683287e-06,
      "loss": 0.4845,
      "step": 1156
    },
    {
      "epoch": 0.19,
      "grad_norm": 22.879201605154623,
      "learning_rate": 9.372754069989032e-06,
      "loss": 0.4797,
      "step": 1157
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.476809555230036,
      "learning_rate": 9.371487928036032e-06,
      "loss": 0.5033,
      "step": 1158
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.057681349307448,
      "learning_rate": 9.37022059516922e-06,
      "loss": 0.4588,
      "step": 1159
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.16209405167654,
      "learning_rate": 9.368952071733851e-06,
      "loss": 0.4869,
      "step": 1160
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.408254828463715,
      "learning_rate": 9.367682358075509e-06,
      "loss": 0.5148,
      "step": 1161
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.635243574782355,
      "learning_rate": 9.366411454540095e-06,
      "loss": 0.5032,
      "step": 1162
    },
    {
      "epoch": 0.19,
      "grad_norm": 11.406930155882574,
      "learning_rate": 9.36513936147384e-06,
      "loss": 0.4921,
      "step": 1163
    },
    {
      "epoch": 0.19,
      "grad_norm": 17.416469116732234,
      "learning_rate": 9.363866079223299e-06,
      "loss": 0.4587,
      "step": 1164
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.425267658806005,
      "learning_rate": 9.362591608135346e-06,
      "loss": 0.5475,
      "step": 1165
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.431501851374636,
      "learning_rate": 9.361315948557182e-06,
      "loss": 0.4846,
      "step": 1166
    },
    {
      "epoch": 0.19,
      "grad_norm": 17.832089371763782,
      "learning_rate": 9.360039100836335e-06,
      "loss": 0.508,
      "step": 1167
    },
    {
      "epoch": 0.19,
      "grad_norm": 13.133379324501101,
      "learning_rate": 9.358761065320653e-06,
      "loss": 0.4656,
      "step": 1168
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.798702732647197,
      "learning_rate": 9.357481842358302e-06,
      "loss": 0.566,
      "step": 1169
    },
    {
      "epoch": 0.19,
      "grad_norm": 14.117209200933333,
      "learning_rate": 9.356201432297788e-06,
      "loss": 0.4778,
      "step": 1170
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.311854056583549,
      "learning_rate": 9.354919835487922e-06,
      "loss": 0.4823,
      "step": 1171
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.52169299827595,
      "learning_rate": 9.353637052277848e-06,
      "loss": 0.4855,
      "step": 1172
    },
    {
      "epoch": 0.19,
      "grad_norm": 14.440455087182773,
      "learning_rate": 9.352353083017034e-06,
      "loss": 0.4369,
      "step": 1173
    },
    {
      "epoch": 0.19,
      "grad_norm": 17.625658108640863,
      "learning_rate": 9.351067928055265e-06,
      "loss": 0.5355,
      "step": 1174
    },
    {
      "epoch": 0.19,
      "grad_norm": 19.054361595410825,
      "learning_rate": 9.349781587742655e-06,
      "loss": 0.4444,
      "step": 1175
    },
    {
      "epoch": 0.19,
      "grad_norm": 50.298468256397996,
      "learning_rate": 9.348494062429639e-06,
      "loss": 0.493,
      "step": 1176
    },
    {
      "epoch": 0.19,
      "grad_norm": 22.228128249307108,
      "learning_rate": 9.347205352466972e-06,
      "loss": 0.4421,
      "step": 1177
    },
    {
      "epoch": 0.19,
      "grad_norm": 23.71583495647995,
      "learning_rate": 9.345915458205736e-06,
      "loss": 0.4444,
      "step": 1178
    },
    {
      "epoch": 0.19,
      "grad_norm": 19.1439750637546,
      "learning_rate": 9.344624379997335e-06,
      "loss": 0.4557,
      "step": 1179
    },
    {
      "epoch": 0.19,
      "grad_norm": 13.819056776149889,
      "learning_rate": 9.343332118193492e-06,
      "loss": 0.4726,
      "step": 1180
    },
    {
      "epoch": 0.19,
      "grad_norm": 17.508459236707278,
      "learning_rate": 9.342038673146253e-06,
      "loss": 0.5048,
      "step": 1181
    },
    {
      "epoch": 0.19,
      "grad_norm": 119.09385352197333,
      "learning_rate": 9.340744045207992e-06,
      "loss": 0.4849,
      "step": 1182
    },
    {
      "epoch": 0.19,
      "grad_norm": 16.957858104714944,
      "learning_rate": 9.3394482347314e-06,
      "loss": 0.4315,
      "step": 1183
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.342867913210528,
      "learning_rate": 9.338151242069491e-06,
      "loss": 0.4139,
      "step": 1184
    },
    {
      "epoch": 0.19,
      "grad_norm": 16.215044926373327,
      "learning_rate": 9.336853067575603e-06,
      "loss": 0.4756,
      "step": 1185
    },
    {
      "epoch": 0.19,
      "grad_norm": 53.467968004359186,
      "learning_rate": 9.335553711603393e-06,
      "loss": 0.4518,
      "step": 1186
    },
    {
      "epoch": 0.19,
      "grad_norm": 18.85378554402728,
      "learning_rate": 9.33425317450684e-06,
      "loss": 0.4421,
      "step": 1187
    },
    {
      "epoch": 0.19,
      "grad_norm": 35.407259253720966,
      "learning_rate": 9.332951456640249e-06,
      "loss": 0.4914,
      "step": 1188
    },
    {
      "epoch": 0.19,
      "grad_norm": 19.981261755815293,
      "learning_rate": 9.331648558358244e-06,
      "loss": 0.4497,
      "step": 1189
    },
    {
      "epoch": 0.19,
      "grad_norm": 20.609169845081027,
      "learning_rate": 9.330344480015767e-06,
      "loss": 0.4358,
      "step": 1190
    },
    {
      "epoch": 0.19,
      "grad_norm": 99.9583634238747,
      "learning_rate": 9.329039221968087e-06,
      "loss": 0.4899,
      "step": 1191
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.451622322687537,
      "learning_rate": 9.327732784570794e-06,
      "loss": 0.4941,
      "step": 1192
    },
    {
      "epoch": 0.19,
      "grad_norm": 21.35559268182892,
      "learning_rate": 9.326425168179795e-06,
      "loss": 0.4657,
      "step": 1193
    },
    {
      "epoch": 0.19,
      "grad_norm": 13.504441820715579,
      "learning_rate": 9.325116373151322e-06,
      "loss": 0.5195,
      "step": 1194
    },
    {
      "epoch": 0.19,
      "grad_norm": 20.2715934180167,
      "learning_rate": 9.323806399841927e-06,
      "loss": 0.4898,
      "step": 1195
    },
    {
      "epoch": 0.19,
      "grad_norm": 15.877236035520705,
      "learning_rate": 9.32249524860848e-06,
      "loss": 0.4756,
      "step": 1196
    },
    {
      "epoch": 0.19,
      "grad_norm": 13.665288639285857,
      "learning_rate": 9.321182919808179e-06,
      "loss": 0.4715,
      "step": 1197
    },
    {
      "epoch": 0.19,
      "grad_norm": 22.36898029781077,
      "learning_rate": 9.319869413798535e-06,
      "loss": 0.4858,
      "step": 1198
    },
    {
      "epoch": 0.19,
      "grad_norm": 22.996504095711675,
      "learning_rate": 9.318554730937385e-06,
      "loss": 0.4565,
      "step": 1199
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3377875501110854,
      "learning_rate": 9.317238871582886e-06,
      "loss": 0.4319,
      "step": 1200
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.255664931655975,
      "learning_rate": 9.31592183609351e-06,
      "loss": 0.41,
      "step": 1201
    },
    {
      "epoch": 0.19,
      "grad_norm": 13.201775552399635,
      "learning_rate": 9.31460362482806e-06,
      "loss": 0.4497,
      "step": 1202
    },
    {
      "epoch": 0.19,
      "grad_norm": 23.992040097790134,
      "learning_rate": 9.313284238145648e-06,
      "loss": 0.4768,
      "step": 1203
    },
    {
      "epoch": 0.19,
      "grad_norm": 38.7723169133457,
      "learning_rate": 9.311963676405716e-06,
      "loss": 0.4836,
      "step": 1204
    },
    {
      "epoch": 0.19,
      "grad_norm": 19.343054276542055,
      "learning_rate": 9.310641939968016e-06,
      "loss": 0.5546,
      "step": 1205
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.365010498680524,
      "learning_rate": 9.309319029192627e-06,
      "loss": 0.5018,
      "step": 1206
    },
    {
      "epoch": 0.19,
      "grad_norm": 13.967270033279258,
      "learning_rate": 9.307994944439949e-06,
      "loss": 0.5239,
      "step": 1207
    },
    {
      "epoch": 0.19,
      "grad_norm": 11.218295828753742,
      "learning_rate": 9.306669686070697e-06,
      "loss": 0.4324,
      "step": 1208
    },
    {
      "epoch": 0.19,
      "grad_norm": 13.54751828727378,
      "learning_rate": 9.30534325444591e-06,
      "loss": 0.5218,
      "step": 1209
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1739378918121595,
      "learning_rate": 9.304015649926941e-06,
      "loss": 0.4509,
      "step": 1210
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.13142482689222,
      "learning_rate": 9.30268687287547e-06,
      "loss": 0.467,
      "step": 1211
    },
    {
      "epoch": 0.2,
      "grad_norm": 11.397876469421153,
      "learning_rate": 9.30135692365349e-06,
      "loss": 0.4784,
      "step": 1212
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.158325384787432,
      "learning_rate": 9.300025802623316e-06,
      "loss": 0.481,
      "step": 1213
    },
    {
      "epoch": 0.2,
      "grad_norm": 29.734777233129346,
      "learning_rate": 9.29869351014758e-06,
      "loss": 0.4222,
      "step": 1214
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2468332697212356,
      "learning_rate": 9.29736004658924e-06,
      "loss": 0.4808,
      "step": 1215
    },
    {
      "epoch": 0.2,
      "grad_norm": 15.88352860449723,
      "learning_rate": 9.296025412311567e-06,
      "loss": 0.4934,
      "step": 1216
    },
    {
      "epoch": 0.2,
      "grad_norm": 16.768370728471233,
      "learning_rate": 9.294689607678151e-06,
      "loss": 0.5011,
      "step": 1217
    },
    {
      "epoch": 0.2,
      "grad_norm": 12.063144961541461,
      "learning_rate": 9.293352633052901e-06,
      "loss": 0.4005,
      "step": 1218
    },
    {
      "epoch": 0.2,
      "grad_norm": 10.141293802230301,
      "learning_rate": 9.292014488800046e-06,
      "loss": 0.4002,
      "step": 1219
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3431301998195286,
      "learning_rate": 9.290675175284135e-06,
      "loss": 0.4932,
      "step": 1220
    },
    {
      "epoch": 0.2,
      "grad_norm": 20.47786188148704,
      "learning_rate": 9.289334692870033e-06,
      "loss": 0.5413,
      "step": 1221
    },
    {
      "epoch": 0.2,
      "grad_norm": 14.956801794450936,
      "learning_rate": 9.287993041922924e-06,
      "loss": 0.5173,
      "step": 1222
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.611344499766316,
      "learning_rate": 9.286650222808314e-06,
      "loss": 0.5498,
      "step": 1223
    },
    {
      "epoch": 0.2,
      "grad_norm": 21.48874694353932,
      "learning_rate": 9.285306235892017e-06,
      "loss": 0.4429,
      "step": 1224
    },
    {
      "epoch": 0.2,
      "grad_norm": 14.468222040691122,
      "learning_rate": 9.283961081540178e-06,
      "loss": 0.4165,
      "step": 1225
    },
    {
      "epoch": 0.2,
      "grad_norm": 11.28054113136257,
      "learning_rate": 9.282614760119252e-06,
      "loss": 0.4845,
      "step": 1226
    },
    {
      "epoch": 0.2,
      "grad_norm": 25.150239852820814,
      "learning_rate": 9.281267271996015e-06,
      "loss": 0.553,
      "step": 1227
    },
    {
      "epoch": 0.2,
      "grad_norm": 13.20899492909649,
      "learning_rate": 9.27991861753756e-06,
      "loss": 0.4772,
      "step": 1228
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.79608666825319,
      "learning_rate": 9.278568797111294e-06,
      "loss": 0.472,
      "step": 1229
    },
    {
      "epoch": 0.2,
      "grad_norm": 11.215769122133224,
      "learning_rate": 9.277217811084952e-06,
      "loss": 0.4343,
      "step": 1230
    },
    {
      "epoch": 0.2,
      "grad_norm": 20.16144853859444,
      "learning_rate": 9.275865659826573e-06,
      "loss": 0.5042,
      "step": 1231
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.989537269658579,
      "learning_rate": 9.274512343704525e-06,
      "loss": 0.4415,
      "step": 1232
    },
    {
      "epoch": 0.2,
      "grad_norm": 15.480275734066542,
      "learning_rate": 9.273157863087486e-06,
      "loss": 0.4042,
      "step": 1233
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.801990898470137,
      "learning_rate": 9.271802218344455e-06,
      "loss": 0.4928,
      "step": 1234
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.223933352370969,
      "learning_rate": 9.270445409844749e-06,
      "loss": 0.4635,
      "step": 1235
    },
    {
      "epoch": 0.2,
      "grad_norm": 11.055781175774458,
      "learning_rate": 9.269087437957996e-06,
      "loss": 0.4356,
      "step": 1236
    },
    {
      "epoch": 0.2,
      "grad_norm": 26.43649069801985,
      "learning_rate": 9.267728303054146e-06,
      "loss": 0.462,
      "step": 1237
    },
    {
      "epoch": 0.2,
      "grad_norm": 20.609353355665974,
      "learning_rate": 9.266368005503465e-06,
      "loss": 0.4623,
      "step": 1238
    },
    {
      "epoch": 0.2,
      "grad_norm": 12.13612454486031,
      "learning_rate": 9.265006545676538e-06,
      "loss": 0.4657,
      "step": 1239
    },
    {
      "epoch": 0.2,
      "grad_norm": 10.603635138028705,
      "learning_rate": 9.263643923944262e-06,
      "loss": 0.4725,
      "step": 1240
    },
    {
      "epoch": 0.2,
      "grad_norm": 13.04866956416687,
      "learning_rate": 9.262280140677852e-06,
      "loss": 0.548,
      "step": 1241
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.216422309171557,
      "learning_rate": 9.260915196248842e-06,
      "loss": 0.4126,
      "step": 1242
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2481489264021828,
      "learning_rate": 9.259549091029082e-06,
      "loss": 0.4892,
      "step": 1243
    },
    {
      "epoch": 0.2,
      "grad_norm": 52.20123113474192,
      "learning_rate": 9.258181825390732e-06,
      "loss": 0.4764,
      "step": 1244
    },
    {
      "epoch": 0.2,
      "grad_norm": 11.05519660226427,
      "learning_rate": 9.256813399706276e-06,
      "loss": 0.4485,
      "step": 1245
    },
    {
      "epoch": 0.2,
      "grad_norm": 29.124364956214794,
      "learning_rate": 9.25544381434851e-06,
      "loss": 0.4824,
      "step": 1246
    },
    {
      "epoch": 0.2,
      "grad_norm": 11.517095297925277,
      "learning_rate": 9.254073069690545e-06,
      "loss": 0.4474,
      "step": 1247
    },
    {
      "epoch": 0.2,
      "grad_norm": 10.852439500211396,
      "learning_rate": 9.252701166105815e-06,
      "loss": 0.4288,
      "step": 1248
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.538429558245012,
      "learning_rate": 9.251328103968059e-06,
      "loss": 0.5233,
      "step": 1249
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.702156735349676,
      "learning_rate": 9.24995388365134e-06,
      "loss": 0.4236,
      "step": 1250
    },
    {
      "epoch": 0.2,
      "grad_norm": 11.099613402994994,
      "learning_rate": 9.24857850553003e-06,
      "loss": 0.497,
      "step": 1251
    },
    {
      "epoch": 0.2,
      "grad_norm": 13.707701686477801,
      "learning_rate": 9.247201969978825e-06,
      "loss": 0.4985,
      "step": 1252
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.403183216867559,
      "learning_rate": 9.245824277372729e-06,
      "loss": 0.4521,
      "step": 1253
    },
    {
      "epoch": 0.2,
      "grad_norm": 11.367205305396112,
      "learning_rate": 9.24444542808706e-06,
      "loss": 0.5147,
      "step": 1254
    },
    {
      "epoch": 0.2,
      "grad_norm": 10.27569694009789,
      "learning_rate": 9.24306542249746e-06,
      "loss": 0.5031,
      "step": 1255
    },
    {
      "epoch": 0.2,
      "grad_norm": 16.95109154151481,
      "learning_rate": 9.241684260979877e-06,
      "loss": 0.5319,
      "step": 1256
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.806869077294007,
      "learning_rate": 9.240301943910578e-06,
      "loss": 0.4816,
      "step": 1257
    },
    {
      "epoch": 0.2,
      "grad_norm": 79.63590440279026,
      "learning_rate": 9.238918471666148e-06,
      "loss": 0.4712,
      "step": 1258
    },
    {
      "epoch": 0.2,
      "grad_norm": 14.741811734696071,
      "learning_rate": 9.237533844623477e-06,
      "loss": 0.4267,
      "step": 1259
    },
    {
      "epoch": 0.2,
      "grad_norm": 20.51911950479914,
      "learning_rate": 9.236148063159778e-06,
      "loss": 0.5143,
      "step": 1260
    },
    {
      "epoch": 0.2,
      "grad_norm": 18.09495880257626,
      "learning_rate": 9.234761127652578e-06,
      "loss": 0.3864,
      "step": 1261
    },
    {
      "epoch": 0.2,
      "grad_norm": 18.274083317285942,
      "learning_rate": 9.233373038479716e-06,
      "loss": 0.5526,
      "step": 1262
    },
    {
      "epoch": 0.2,
      "grad_norm": 17.36187993715572,
      "learning_rate": 9.231983796019342e-06,
      "loss": 0.4587,
      "step": 1263
    },
    {
      "epoch": 0.2,
      "grad_norm": 14.530279273544366,
      "learning_rate": 9.230593400649928e-06,
      "loss": 0.463,
      "step": 1264
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.589281296572754,
      "learning_rate": 9.229201852750254e-06,
      "loss": 0.4684,
      "step": 1265
    },
    {
      "epoch": 0.2,
      "grad_norm": 15.277295313947509,
      "learning_rate": 9.227809152699418e-06,
      "loss": 0.5006,
      "step": 1266
    },
    {
      "epoch": 0.2,
      "grad_norm": 23.395107243578348,
      "learning_rate": 9.226415300876828e-06,
      "loss": 0.5185,
      "step": 1267
    },
    {
      "epoch": 0.2,
      "grad_norm": 12.391051482791196,
      "learning_rate": 9.225020297662208e-06,
      "loss": 0.5072,
      "step": 1268
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.149735535469054,
      "learning_rate": 9.223624143435595e-06,
      "loss": 0.4678,
      "step": 1269
    },
    {
      "epoch": 0.2,
      "grad_norm": 22.73384019925979,
      "learning_rate": 9.222226838577342e-06,
      "loss": 0.4558,
      "step": 1270
    },
    {
      "epoch": 0.2,
      "grad_norm": 11.62899862665488,
      "learning_rate": 9.22082838346811e-06,
      "loss": 0.5048,
      "step": 1271
    },
    {
      "epoch": 0.2,
      "grad_norm": 57.15693829281563,
      "learning_rate": 9.21942877848888e-06,
      "loss": 0.46,
      "step": 1272
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.628456781608687,
      "learning_rate": 9.21802802402094e-06,
      "loss": 0.4475,
      "step": 1273
    },
    {
      "epoch": 0.21,
      "grad_norm": 9.087844092420713,
      "learning_rate": 9.216626120445897e-06,
      "loss": 0.4458,
      "step": 1274
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.491669651180336,
      "learning_rate": 9.215223068145666e-06,
      "loss": 0.5246,
      "step": 1275
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.977727515705651,
      "learning_rate": 9.213818867502478e-06,
      "loss": 0.4452,
      "step": 1276
    },
    {
      "epoch": 0.21,
      "grad_norm": 14.24986174431144,
      "learning_rate": 9.212413518898878e-06,
      "loss": 0.4236,
      "step": 1277
    },
    {
      "epoch": 0.21,
      "grad_norm": 11.413776483419992,
      "learning_rate": 9.211007022717716e-06,
      "loss": 0.4034,
      "step": 1278
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.385466089490842,
      "learning_rate": 9.209599379342165e-06,
      "loss": 0.4427,
      "step": 1279
    },
    {
      "epoch": 0.21,
      "grad_norm": 10.940816814118238,
      "learning_rate": 9.208190589155705e-06,
      "loss": 0.5313,
      "step": 1280
    },
    {
      "epoch": 0.21,
      "grad_norm": 11.958495319276546,
      "learning_rate": 9.206780652542131e-06,
      "loss": 0.4437,
      "step": 1281
    },
    {
      "epoch": 0.21,
      "grad_norm": 26.791168871675765,
      "learning_rate": 9.205369569885544e-06,
      "loss": 0.5029,
      "step": 1282
    },
    {
      "epoch": 0.21,
      "grad_norm": 14.324589467082395,
      "learning_rate": 9.203957341570367e-06,
      "loss": 0.4548,
      "step": 1283
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.410091724695577,
      "learning_rate": 9.202543967981325e-06,
      "loss": 0.4778,
      "step": 1284
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.702845859851132,
      "learning_rate": 9.201129449503463e-06,
      "loss": 0.4368,
      "step": 1285
    },
    {
      "epoch": 0.21,
      "grad_norm": 15.623418610245421,
      "learning_rate": 9.199713786522135e-06,
      "loss": 0.4786,
      "step": 1286
    },
    {
      "epoch": 0.21,
      "grad_norm": 11.056804684065105,
      "learning_rate": 9.198296979423006e-06,
      "loss": 0.4497,
      "step": 1287
    },
    {
      "epoch": 0.21,
      "grad_norm": 24.75961125509385,
      "learning_rate": 9.196879028592052e-06,
      "loss": 0.3898,
      "step": 1288
    },
    {
      "epoch": 0.21,
      "grad_norm": 26.85212124355255,
      "learning_rate": 9.195459934415563e-06,
      "loss": 0.4554,
      "step": 1289
    },
    {
      "epoch": 0.21,
      "grad_norm": 17.310065012772057,
      "learning_rate": 9.19403969728014e-06,
      "loss": 0.4364,
      "step": 1290
    },
    {
      "epoch": 0.21,
      "grad_norm": 18.515392506319202,
      "learning_rate": 9.192618317572693e-06,
      "loss": 0.4527,
      "step": 1291
    },
    {
      "epoch": 0.21,
      "grad_norm": 17.17954595614451,
      "learning_rate": 9.191195795680447e-06,
      "loss": 0.4027,
      "step": 1292
    },
    {
      "epoch": 0.21,
      "grad_norm": 10.205084116468981,
      "learning_rate": 9.189772131990935e-06,
      "loss": 0.4488,
      "step": 1293
    },
    {
      "epoch": 0.21,
      "grad_norm": 16.050933430684896,
      "learning_rate": 9.188347326892002e-06,
      "loss": 0.4471,
      "step": 1294
    },
    {
      "epoch": 0.21,
      "grad_norm": 10.003518400123445,
      "learning_rate": 9.186921380771806e-06,
      "loss": 0.5134,
      "step": 1295
    },
    {
      "epoch": 0.21,
      "grad_norm": 27.695627595422607,
      "learning_rate": 9.185494294018811e-06,
      "loss": 0.4264,
      "step": 1296
    },
    {
      "epoch": 0.21,
      "grad_norm": 20.00949196301132,
      "learning_rate": 9.184066067021798e-06,
      "loss": 0.434,
      "step": 1297
    },
    {
      "epoch": 0.21,
      "grad_norm": 11.777361036554433,
      "learning_rate": 9.182636700169854e-06,
      "loss": 0.4422,
      "step": 1298
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3876563356128973,
      "learning_rate": 9.181206193852376e-06,
      "loss": 0.4462,
      "step": 1299
    },
    {
      "epoch": 0.21,
      "grad_norm": 13.688711584071397,
      "learning_rate": 9.179774548459077e-06,
      "loss": 0.4714,
      "step": 1300
    },
    {
      "epoch": 0.21,
      "grad_norm": 18.579101840787082,
      "learning_rate": 9.178341764379974e-06,
      "loss": 0.4762,
      "step": 1301
    },
    {
      "epoch": 0.21,
      "grad_norm": 23.57544635182901,
      "learning_rate": 9.176907842005398e-06,
      "loss": 0.5258,
      "step": 1302
    },
    {
      "epoch": 0.21,
      "grad_norm": 17.226077759206042,
      "learning_rate": 9.17547278172599e-06,
      "loss": 0.4234,
      "step": 1303
    },
    {
      "epoch": 0.21,
      "grad_norm": 10.54545814839091,
      "learning_rate": 9.174036583932695e-06,
      "loss": 0.5277,
      "step": 1304
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4144205359549633,
      "learning_rate": 9.17259924901678e-06,
      "loss": 0.5127,
      "step": 1305
    },
    {
      "epoch": 0.21,
      "grad_norm": 11.255516150321295,
      "learning_rate": 9.17116077736981e-06,
      "loss": 0.5251,
      "step": 1306
    },
    {
      "epoch": 0.21,
      "grad_norm": 13.807324052286138,
      "learning_rate": 9.169721169383666e-06,
      "loss": 0.4744,
      "step": 1307
    },
    {
      "epoch": 0.21,
      "grad_norm": 21.28383155767525,
      "learning_rate": 9.168280425450535e-06,
      "loss": 0.4663,
      "step": 1308
    },
    {
      "epoch": 0.21,
      "grad_norm": 19.963916602666423,
      "learning_rate": 9.166838545962917e-06,
      "loss": 0.4622,
      "step": 1309
    },
    {
      "epoch": 0.21,
      "grad_norm": 55.23616177983737,
      "learning_rate": 9.165395531313622e-06,
      "loss": 0.4964,
      "step": 1310
    },
    {
      "epoch": 0.21,
      "grad_norm": 16.042122077655637,
      "learning_rate": 9.163951381895761e-06,
      "loss": 0.5034,
      "step": 1311
    },
    {
      "epoch": 0.21,
      "grad_norm": 11.422662953418456,
      "learning_rate": 9.162506098102765e-06,
      "loss": 0.5072,
      "step": 1312
    },
    {
      "epoch": 0.21,
      "grad_norm": 16.476629498388053,
      "learning_rate": 9.161059680328368e-06,
      "loss": 0.4352,
      "step": 1313
    },
    {
      "epoch": 0.21,
      "grad_norm": 12.834599456728844,
      "learning_rate": 9.159612128966615e-06,
      "loss": 0.4169,
      "step": 1314
    },
    {
      "epoch": 0.21,
      "grad_norm": 15.989437922598462,
      "learning_rate": 9.158163444411857e-06,
      "loss": 0.5346,
      "step": 1315
    },
    {
      "epoch": 0.21,
      "grad_norm": 13.174938125846978,
      "learning_rate": 9.156713627058754e-06,
      "loss": 0.4222,
      "step": 1316
    },
    {
      "epoch": 0.21,
      "grad_norm": 18.44157670851661,
      "learning_rate": 9.155262677302281e-06,
      "loss": 0.4736,
      "step": 1317
    },
    {
      "epoch": 0.21,
      "grad_norm": 14.525230537507207,
      "learning_rate": 9.153810595537713e-06,
      "loss": 0.3699,
      "step": 1318
    },
    {
      "epoch": 0.21,
      "grad_norm": 16.931319891885362,
      "learning_rate": 9.152357382160639e-06,
      "loss": 0.4742,
      "step": 1319
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4073910594513346,
      "learning_rate": 9.150903037566954e-06,
      "loss": 0.4362,
      "step": 1320
    },
    {
      "epoch": 0.21,
      "grad_norm": 16.41880097822788,
      "learning_rate": 9.14944756215286e-06,
      "loss": 0.4477,
      "step": 1321
    },
    {
      "epoch": 0.21,
      "grad_norm": 31.637023932907365,
      "learning_rate": 9.14799095631487e-06,
      "loss": 0.4892,
      "step": 1322
    },
    {
      "epoch": 0.21,
      "grad_norm": 14.99614971919869,
      "learning_rate": 9.146533220449804e-06,
      "loss": 0.4099,
      "step": 1323
    },
    {
      "epoch": 0.21,
      "grad_norm": 23.106372062139346,
      "learning_rate": 9.14507435495479e-06,
      "loss": 0.489,
      "step": 1324
    },
    {
      "epoch": 0.21,
      "grad_norm": 25.684235961142882,
      "learning_rate": 9.143614360227261e-06,
      "loss": 0.4874,
      "step": 1325
    },
    {
      "epoch": 0.21,
      "grad_norm": 11.872847415985758,
      "learning_rate": 9.142153236664961e-06,
      "loss": 0.4499,
      "step": 1326
    },
    {
      "epoch": 0.21,
      "grad_norm": 19.62887129872119,
      "learning_rate": 9.14069098466594e-06,
      "loss": 0.418,
      "step": 1327
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4187584199052041,
      "learning_rate": 9.139227604628556e-06,
      "loss": 0.518,
      "step": 1328
    },
    {
      "epoch": 0.21,
      "grad_norm": 18.748798834806667,
      "learning_rate": 9.137763096951474e-06,
      "loss": 0.4456,
      "step": 1329
    },
    {
      "epoch": 0.21,
      "grad_norm": 17.390382067510025,
      "learning_rate": 9.136297462033667e-06,
      "loss": 0.4928,
      "step": 1330
    },
    {
      "epoch": 0.21,
      "grad_norm": 22.52121710577523,
      "learning_rate": 9.134830700274413e-06,
      "loss": 0.4352,
      "step": 1331
    },
    {
      "epoch": 0.21,
      "grad_norm": 30.130918738808376,
      "learning_rate": 9.133362812073298e-06,
      "loss": 0.4262,
      "step": 1332
    },
    {
      "epoch": 0.21,
      "grad_norm": 51.20820463640828,
      "learning_rate": 9.131893797830219e-06,
      "loss": 0.51,
      "step": 1333
    },
    {
      "epoch": 0.21,
      "grad_norm": 51.896539210940006,
      "learning_rate": 9.130423657945369e-06,
      "loss": 0.3873,
      "step": 1334
    },
    {
      "epoch": 0.22,
      "grad_norm": 38.01746440850158,
      "learning_rate": 9.128952392819262e-06,
      "loss": 0.488,
      "step": 1335
    },
    {
      "epoch": 0.22,
      "grad_norm": 29.387651645730315,
      "learning_rate": 9.127480002852706e-06,
      "loss": 0.5168,
      "step": 1336
    },
    {
      "epoch": 0.22,
      "grad_norm": 28.395164882192013,
      "learning_rate": 9.126006488446824e-06,
      "loss": 0.5219,
      "step": 1337
    },
    {
      "epoch": 0.22,
      "grad_norm": 46.6306114392875,
      "learning_rate": 9.12453185000304e-06,
      "loss": 0.5417,
      "step": 1338
    },
    {
      "epoch": 0.22,
      "grad_norm": 23.70115531328106,
      "learning_rate": 9.123056087923088e-06,
      "loss": 0.5601,
      "step": 1339
    },
    {
      "epoch": 0.22,
      "grad_norm": 14.480551692948808,
      "learning_rate": 9.121579202609004e-06,
      "loss": 0.5108,
      "step": 1340
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4914957166525074,
      "learning_rate": 9.12010119446313e-06,
      "loss": 0.4777,
      "step": 1341
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4780775033050049,
      "learning_rate": 9.118622063888124e-06,
      "loss": 0.4867,
      "step": 1342
    },
    {
      "epoch": 0.22,
      "grad_norm": 24.729274280863823,
      "learning_rate": 9.117141811286935e-06,
      "loss": 0.4694,
      "step": 1343
    },
    {
      "epoch": 0.22,
      "grad_norm": 17.06349807428754,
      "learning_rate": 9.115660437062828e-06,
      "loss": 0.4714,
      "step": 1344
    },
    {
      "epoch": 0.22,
      "grad_norm": 26.024820950268502,
      "learning_rate": 9.114177941619369e-06,
      "loss": 0.5125,
      "step": 1345
    },
    {
      "epoch": 0.22,
      "grad_norm": 15.231664802558168,
      "learning_rate": 9.11269432536043e-06,
      "loss": 0.4924,
      "step": 1346
    },
    {
      "epoch": 0.22,
      "grad_norm": 20.323088561043797,
      "learning_rate": 9.11120958869019e-06,
      "loss": 0.4952,
      "step": 1347
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3646948365850502,
      "learning_rate": 9.109723732013132e-06,
      "loss": 0.4997,
      "step": 1348
    },
    {
      "epoch": 0.22,
      "grad_norm": 22.515671785607097,
      "learning_rate": 9.108236755734045e-06,
      "loss": 0.4593,
      "step": 1349
    },
    {
      "epoch": 0.22,
      "grad_norm": 14.87568346975854,
      "learning_rate": 9.10674866025802e-06,
      "loss": 0.4786,
      "step": 1350
    },
    {
      "epoch": 0.22,
      "grad_norm": 32.87193528521304,
      "learning_rate": 9.105259445990457e-06,
      "loss": 0.5309,
      "step": 1351
    },
    {
      "epoch": 0.22,
      "grad_norm": 52.92662729038108,
      "learning_rate": 9.10376911333706e-06,
      "loss": 0.4188,
      "step": 1352
    },
    {
      "epoch": 0.22,
      "grad_norm": 27.640186962170592,
      "learning_rate": 9.102277662703834e-06,
      "loss": 0.4655,
      "step": 1353
    },
    {
      "epoch": 0.22,
      "grad_norm": 13.667125251100636,
      "learning_rate": 9.100785094497093e-06,
      "loss": 0.4639,
      "step": 1354
    },
    {
      "epoch": 0.22,
      "grad_norm": 14.083472155379368,
      "learning_rate": 9.099291409123454e-06,
      "loss": 0.5523,
      "step": 1355
    },
    {
      "epoch": 0.22,
      "grad_norm": 20.55259754733359,
      "learning_rate": 9.097796606989838e-06,
      "loss": 0.4389,
      "step": 1356
    },
    {
      "epoch": 0.22,
      "grad_norm": 20.704468162449896,
      "learning_rate": 9.09630068850347e-06,
      "loss": 0.4698,
      "step": 1357
    },
    {
      "epoch": 0.22,
      "grad_norm": 30.26170179842706,
      "learning_rate": 9.094803654071877e-06,
      "loss": 0.4954,
      "step": 1358
    },
    {
      "epoch": 0.22,
      "grad_norm": 15.73055675354421,
      "learning_rate": 9.093305504102897e-06,
      "loss": 0.358,
      "step": 1359
    },
    {
      "epoch": 0.22,
      "grad_norm": 21.843516125015547,
      "learning_rate": 9.091806239004664e-06,
      "loss": 0.4993,
      "step": 1360
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.426705648488882,
      "learning_rate": 9.090305859185619e-06,
      "loss": 0.476,
      "step": 1361
    },
    {
      "epoch": 0.22,
      "grad_norm": 32.36476620726132,
      "learning_rate": 9.088804365054511e-06,
      "loss": 0.4966,
      "step": 1362
    },
    {
      "epoch": 0.22,
      "grad_norm": 16.53031496507569,
      "learning_rate": 9.087301757020384e-06,
      "loss": 0.4351,
      "step": 1363
    },
    {
      "epoch": 0.22,
      "grad_norm": 19.424222770714906,
      "learning_rate": 9.08579803549259e-06,
      "loss": 0.4574,
      "step": 1364
    },
    {
      "epoch": 0.22,
      "grad_norm": 22.71946720673779,
      "learning_rate": 9.084293200880787e-06,
      "loss": 0.4796,
      "step": 1365
    },
    {
      "epoch": 0.22,
      "grad_norm": 15.686661171616713,
      "learning_rate": 9.082787253594932e-06,
      "loss": 0.4305,
      "step": 1366
    },
    {
      "epoch": 0.22,
      "grad_norm": 18.501625328224105,
      "learning_rate": 9.081280194045286e-06,
      "loss": 0.3989,
      "step": 1367
    },
    {
      "epoch": 0.22,
      "grad_norm": 19.83478580526505,
      "learning_rate": 9.079772022642413e-06,
      "loss": 0.5081,
      "step": 1368
    },
    {
      "epoch": 0.22,
      "grad_norm": 24.332286338091723,
      "learning_rate": 9.078262739797184e-06,
      "loss": 0.4863,
      "step": 1369
    },
    {
      "epoch": 0.22,
      "grad_norm": 13.818230009541045,
      "learning_rate": 9.076752345920764e-06,
      "loss": 0.4919,
      "step": 1370
    },
    {
      "epoch": 0.22,
      "grad_norm": 19.801251483646638,
      "learning_rate": 9.075240841424629e-06,
      "loss": 0.4621,
      "step": 1371
    },
    {
      "epoch": 0.22,
      "grad_norm": 18.71701743471966,
      "learning_rate": 9.073728226720555e-06,
      "loss": 0.4796,
      "step": 1372
    },
    {
      "epoch": 0.22,
      "grad_norm": 13.81863544332554,
      "learning_rate": 9.07221450222062e-06,
      "loss": 0.5006,
      "step": 1373
    },
    {
      "epoch": 0.22,
      "grad_norm": 11.449025877618004,
      "learning_rate": 9.070699668337202e-06,
      "loss": 0.4857,
      "step": 1374
    },
    {
      "epoch": 0.22,
      "grad_norm": 21.725905979424546,
      "learning_rate": 9.069183725482984e-06,
      "loss": 0.4309,
      "step": 1375
    },
    {
      "epoch": 0.22,
      "grad_norm": 14.0013196207162,
      "learning_rate": 9.067666674070952e-06,
      "loss": 0.4357,
      "step": 1376
    },
    {
      "epoch": 0.22,
      "grad_norm": 15.803010596255133,
      "learning_rate": 9.066148514514395e-06,
      "loss": 0.4687,
      "step": 1377
    },
    {
      "epoch": 0.22,
      "grad_norm": 18.275906721830868,
      "learning_rate": 9.064629247226895e-06,
      "loss": 0.4528,
      "step": 1378
    },
    {
      "epoch": 0.22,
      "grad_norm": 13.973481170144813,
      "learning_rate": 9.063108872622348e-06,
      "loss": 0.4766,
      "step": 1379
    },
    {
      "epoch": 0.22,
      "grad_norm": 21.71026300091794,
      "learning_rate": 9.061587391114942e-06,
      "loss": 0.5199,
      "step": 1380
    },
    {
      "epoch": 0.22,
      "grad_norm": 44.02811839197106,
      "learning_rate": 9.060064803119173e-06,
      "loss": 0.5202,
      "step": 1381
    },
    {
      "epoch": 0.22,
      "grad_norm": 74.85422997274796,
      "learning_rate": 9.058541109049836e-06,
      "loss": 0.4556,
      "step": 1382
    },
    {
      "epoch": 0.22,
      "grad_norm": 18.579628251511966,
      "learning_rate": 9.057016309322026e-06,
      "loss": 0.5966,
      "step": 1383
    },
    {
      "epoch": 0.22,
      "grad_norm": 26.67411198452988,
      "learning_rate": 9.055490404351141e-06,
      "loss": 0.4774,
      "step": 1384
    },
    {
      "epoch": 0.22,
      "grad_norm": 16.633871110327483,
      "learning_rate": 9.05396339455288e-06,
      "loss": 0.5119,
      "step": 1385
    },
    {
      "epoch": 0.22,
      "grad_norm": 15.382117263458648,
      "learning_rate": 9.05243528034324e-06,
      "loss": 0.5104,
      "step": 1386
    },
    {
      "epoch": 0.22,
      "grad_norm": 18.573334075367782,
      "learning_rate": 9.050906062138527e-06,
      "loss": 0.5515,
      "step": 1387
    },
    {
      "epoch": 0.22,
      "grad_norm": 18.358890556324596,
      "learning_rate": 9.049375740355338e-06,
      "loss": 0.5047,
      "step": 1388
    },
    {
      "epoch": 0.22,
      "grad_norm": 25.633611702870787,
      "learning_rate": 9.047844315410574e-06,
      "loss": 0.4835,
      "step": 1389
    },
    {
      "epoch": 0.22,
      "grad_norm": 63.715255472140306,
      "learning_rate": 9.04631178772144e-06,
      "loss": 0.4322,
      "step": 1390
    },
    {
      "epoch": 0.22,
      "grad_norm": 22.65127175358151,
      "learning_rate": 9.044778157705437e-06,
      "loss": 0.4812,
      "step": 1391
    },
    {
      "epoch": 0.22,
      "grad_norm": 28.245544505170365,
      "learning_rate": 9.04324342578037e-06,
      "loss": 0.4849,
      "step": 1392
    },
    {
      "epoch": 0.22,
      "grad_norm": 21.291562515955235,
      "learning_rate": 9.04170759236434e-06,
      "loss": 0.4698,
      "step": 1393
    },
    {
      "epoch": 0.22,
      "grad_norm": 11.663448905980461,
      "learning_rate": 9.040170657875753e-06,
      "loss": 0.433,
      "step": 1394
    },
    {
      "epoch": 0.22,
      "grad_norm": 16.078475537493784,
      "learning_rate": 9.038632622733311e-06,
      "loss": 0.4937,
      "step": 1395
    },
    {
      "epoch": 0.22,
      "grad_norm": 19.058760656396302,
      "learning_rate": 9.037093487356016e-06,
      "loss": 0.5158,
      "step": 1396
    },
    {
      "epoch": 0.23,
      "grad_norm": 18.42271472261091,
      "learning_rate": 9.035553252163172e-06,
      "loss": 0.5191,
      "step": 1397
    },
    {
      "epoch": 0.23,
      "grad_norm": 11.883669861386426,
      "learning_rate": 9.034011917574382e-06,
      "loss": 0.5046,
      "step": 1398
    },
    {
      "epoch": 0.23,
      "grad_norm": 20.729675986626503,
      "learning_rate": 9.032469484009548e-06,
      "loss": 0.4744,
      "step": 1399
    },
    {
      "epoch": 0.23,
      "grad_norm": 17.518435951467687,
      "learning_rate": 9.030925951888869e-06,
      "loss": 0.5217,
      "step": 1400
    },
    {
      "epoch": 0.23,
      "grad_norm": 11.350922828878298,
      "learning_rate": 9.029381321632849e-06,
      "loss": 0.4789,
      "step": 1401
    },
    {
      "epoch": 0.23,
      "grad_norm": 13.321753440462835,
      "learning_rate": 9.027835593662285e-06,
      "loss": 0.4575,
      "step": 1402
    },
    {
      "epoch": 0.23,
      "grad_norm": 15.102292166377094,
      "learning_rate": 9.026288768398278e-06,
      "loss": 0.4668,
      "step": 1403
    },
    {
      "epoch": 0.23,
      "grad_norm": 15.530500443242154,
      "learning_rate": 9.024740846262225e-06,
      "loss": 0.468,
      "step": 1404
    },
    {
      "epoch": 0.23,
      "grad_norm": 18.54755125323169,
      "learning_rate": 9.023191827675824e-06,
      "loss": 0.4875,
      "step": 1405
    },
    {
      "epoch": 0.23,
      "grad_norm": 13.903256952424416,
      "learning_rate": 9.021641713061069e-06,
      "loss": 0.4899,
      "step": 1406
    },
    {
      "epoch": 0.23,
      "grad_norm": 12.819147567488159,
      "learning_rate": 9.020090502840253e-06,
      "loss": 0.4806,
      "step": 1407
    },
    {
      "epoch": 0.23,
      "grad_norm": 17.95248407686516,
      "learning_rate": 9.01853819743597e-06,
      "loss": 0.5195,
      "step": 1408
    },
    {
      "epoch": 0.23,
      "grad_norm": 22.108535873567885,
      "learning_rate": 9.016984797271112e-06,
      "loss": 0.4763,
      "step": 1409
    },
    {
      "epoch": 0.23,
      "grad_norm": 19.011571501621837,
      "learning_rate": 9.015430302768865e-06,
      "loss": 0.4444,
      "step": 1410
    },
    {
      "epoch": 0.23,
      "grad_norm": 15.251096900371449,
      "learning_rate": 9.013874714352716e-06,
      "loss": 0.5195,
      "step": 1411
    },
    {
      "epoch": 0.23,
      "grad_norm": 12.918221136442906,
      "learning_rate": 9.012318032446454e-06,
      "loss": 0.533,
      "step": 1412
    },
    {
      "epoch": 0.23,
      "grad_norm": 9.912184588390328,
      "learning_rate": 9.010760257474158e-06,
      "loss": 0.474,
      "step": 1413
    },
    {
      "epoch": 0.23,
      "grad_norm": 13.116186421493087,
      "learning_rate": 9.009201389860212e-06,
      "loss": 0.5148,
      "step": 1414
    },
    {
      "epoch": 0.23,
      "grad_norm": 24.100969375026544,
      "learning_rate": 9.007641430029292e-06,
      "loss": 0.4445,
      "step": 1415
    },
    {
      "epoch": 0.23,
      "grad_norm": 9.298245587728536,
      "learning_rate": 9.006080378406377e-06,
      "loss": 0.4616,
      "step": 1416
    },
    {
      "epoch": 0.23,
      "grad_norm": 44.95262205780494,
      "learning_rate": 9.004518235416737e-06,
      "loss": 0.5426,
      "step": 1417
    },
    {
      "epoch": 0.23,
      "grad_norm": 16.77950724248937,
      "learning_rate": 9.002955001485945e-06,
      "loss": 0.4738,
      "step": 1418
    },
    {
      "epoch": 0.23,
      "grad_norm": 11.743634809295369,
      "learning_rate": 9.001390677039868e-06,
      "loss": 0.4751,
      "step": 1419
    },
    {
      "epoch": 0.23,
      "grad_norm": 40.90380610995379,
      "learning_rate": 8.999825262504672e-06,
      "loss": 0.4761,
      "step": 1420
    },
    {
      "epoch": 0.23,
      "grad_norm": 12.379649063995942,
      "learning_rate": 8.998258758306819e-06,
      "loss": 0.4371,
      "step": 1421
    },
    {
      "epoch": 0.23,
      "grad_norm": 12.479741586427682,
      "learning_rate": 8.996691164873068e-06,
      "loss": 0.4947,
      "step": 1422
    },
    {
      "epoch": 0.23,
      "grad_norm": 32.0125924555458,
      "learning_rate": 8.995122482630473e-06,
      "loss": 0.4498,
      "step": 1423
    },
    {
      "epoch": 0.23,
      "grad_norm": 16.72696221970109,
      "learning_rate": 8.993552712006388e-06,
      "loss": 0.4878,
      "step": 1424
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.738150180837083,
      "learning_rate": 8.99198185342846e-06,
      "loss": 0.4538,
      "step": 1425
    },
    {
      "epoch": 0.23,
      "grad_norm": 14.171858095111103,
      "learning_rate": 8.990409907324638e-06,
      "loss": 0.4423,
      "step": 1426
    },
    {
      "epoch": 0.23,
      "grad_norm": 37.14674702156252,
      "learning_rate": 8.98883687412316e-06,
      "loss": 0.5038,
      "step": 1427
    },
    {
      "epoch": 0.23,
      "grad_norm": 12.514200655491752,
      "learning_rate": 8.987262754252565e-06,
      "loss": 0.4195,
      "step": 1428
    },
    {
      "epoch": 0.23,
      "grad_norm": 18.73595518963675,
      "learning_rate": 8.985687548141685e-06,
      "loss": 0.4918,
      "step": 1429
    },
    {
      "epoch": 0.23,
      "grad_norm": 13.393744028649486,
      "learning_rate": 8.984111256219651e-06,
      "loss": 0.4598,
      "step": 1430
    },
    {
      "epoch": 0.23,
      "grad_norm": 33.74959136417218,
      "learning_rate": 8.982533878915889e-06,
      "loss": 0.4762,
      "step": 1431
    },
    {
      "epoch": 0.23,
      "grad_norm": 14.250598375787344,
      "learning_rate": 8.98095541666012e-06,
      "loss": 0.4775,
      "step": 1432
    },
    {
      "epoch": 0.23,
      "grad_norm": 12.825538490570228,
      "learning_rate": 8.979375869882358e-06,
      "loss": 0.4703,
      "step": 1433
    },
    {
      "epoch": 0.23,
      "grad_norm": 10.506107271922906,
      "learning_rate": 8.977795239012916e-06,
      "loss": 0.4326,
      "step": 1434
    },
    {
      "epoch": 0.23,
      "grad_norm": 14.52408901744981,
      "learning_rate": 8.976213524482404e-06,
      "loss": 0.4851,
      "step": 1435
    },
    {
      "epoch": 0.23,
      "grad_norm": 16.00035703593011,
      "learning_rate": 8.974630726721723e-06,
      "loss": 0.5184,
      "step": 1436
    },
    {
      "epoch": 0.23,
      "grad_norm": 33.75190294547055,
      "learning_rate": 8.97304684616207e-06,
      "loss": 0.5033,
      "step": 1437
    },
    {
      "epoch": 0.23,
      "grad_norm": 23.38672065857748,
      "learning_rate": 8.97146188323494e-06,
      "loss": 0.4523,
      "step": 1438
    },
    {
      "epoch": 0.23,
      "grad_norm": 28.17702616900368,
      "learning_rate": 8.969875838372117e-06,
      "loss": 0.4683,
      "step": 1439
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.73695502523755,
      "learning_rate": 8.968288712005688e-06,
      "loss": 0.4534,
      "step": 1440
    },
    {
      "epoch": 0.23,
      "grad_norm": 9.039933951501318,
      "learning_rate": 8.966700504568025e-06,
      "loss": 0.4471,
      "step": 1441
    },
    {
      "epoch": 0.23,
      "grad_norm": 15.24768091596694,
      "learning_rate": 8.965111216491803e-06,
      "loss": 0.4524,
      "step": 1442
    },
    {
      "epoch": 0.23,
      "grad_norm": 11.855969482707387,
      "learning_rate": 8.963520848209985e-06,
      "loss": 0.4514,
      "step": 1443
    },
    {
      "epoch": 0.23,
      "grad_norm": 12.785279525959147,
      "learning_rate": 8.961929400155833e-06,
      "loss": 0.497,
      "step": 1444
    },
    {
      "epoch": 0.23,
      "grad_norm": 11.922075292698853,
      "learning_rate": 8.960336872762903e-06,
      "loss": 0.4884,
      "step": 1445
    },
    {
      "epoch": 0.23,
      "grad_norm": 12.026701372383078,
      "learning_rate": 8.958743266465041e-06,
      "loss": 0.4873,
      "step": 1446
    },
    {
      "epoch": 0.23,
      "grad_norm": 24.539774701349426,
      "learning_rate": 8.957148581696389e-06,
      "loss": 0.4496,
      "step": 1447
    },
    {
      "epoch": 0.23,
      "grad_norm": 12.641384795833185,
      "learning_rate": 8.955552818891384e-06,
      "loss": 0.4594,
      "step": 1448
    },
    {
      "epoch": 0.23,
      "grad_norm": 14.37102050702069,
      "learning_rate": 8.953955978484756e-06,
      "loss": 0.4741,
      "step": 1449
    },
    {
      "epoch": 0.23,
      "grad_norm": 9.790732735897041,
      "learning_rate": 8.95235806091153e-06,
      "loss": 0.4133,
      "step": 1450
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.507999678566856,
      "learning_rate": 8.950759066607017e-06,
      "loss": 0.4779,
      "step": 1451
    },
    {
      "epoch": 0.23,
      "grad_norm": 13.959000156122002,
      "learning_rate": 8.949158996006834e-06,
      "loss": 0.5271,
      "step": 1452
    },
    {
      "epoch": 0.23,
      "grad_norm": 18.363854419623745,
      "learning_rate": 8.94755784954688e-06,
      "loss": 0.4959,
      "step": 1453
    },
    {
      "epoch": 0.23,
      "grad_norm": 10.21605100554187,
      "learning_rate": 8.945955627663353e-06,
      "loss": 0.4159,
      "step": 1454
    },
    {
      "epoch": 0.23,
      "grad_norm": 37.323527678676356,
      "learning_rate": 8.944352330792741e-06,
      "loss": 0.4545,
      "step": 1455
    },
    {
      "epoch": 0.23,
      "grad_norm": 10.789811674804827,
      "learning_rate": 8.942747959371829e-06,
      "loss": 0.4231,
      "step": 1456
    },
    {
      "epoch": 0.23,
      "grad_norm": 14.194037810394756,
      "learning_rate": 8.941142513837689e-06,
      "loss": 0.4463,
      "step": 1457
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.884912495630356,
      "learning_rate": 8.939535994627692e-06,
      "loss": 0.4535,
      "step": 1458
    },
    {
      "epoch": 0.24,
      "grad_norm": 11.895915153642727,
      "learning_rate": 8.937928402179495e-06,
      "loss": 0.4292,
      "step": 1459
    },
    {
      "epoch": 0.24,
      "grad_norm": 12.67063274596574,
      "learning_rate": 8.936319736931051e-06,
      "loss": 0.443,
      "step": 1460
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.584497654189247,
      "learning_rate": 8.934709999320605e-06,
      "loss": 0.5142,
      "step": 1461
    },
    {
      "epoch": 0.24,
      "grad_norm": 16.904185061229505,
      "learning_rate": 8.933099189786697e-06,
      "loss": 0.5184,
      "step": 1462
    },
    {
      "epoch": 0.24,
      "grad_norm": 18.79326128728486,
      "learning_rate": 8.93148730876815e-06,
      "loss": 0.4394,
      "step": 1463
    },
    {
      "epoch": 0.24,
      "grad_norm": 25.77048213505705,
      "learning_rate": 8.92987435670409e-06,
      "loss": 0.4672,
      "step": 1464
    },
    {
      "epoch": 0.24,
      "grad_norm": 14.731230767942538,
      "learning_rate": 8.928260334033927e-06,
      "loss": 0.4991,
      "step": 1465
    },
    {
      "epoch": 0.24,
      "grad_norm": 18.317646303564654,
      "learning_rate": 8.926645241197365e-06,
      "loss": 0.4601,
      "step": 1466
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.913102317586345,
      "learning_rate": 8.925029078634401e-06,
      "loss": 0.4413,
      "step": 1467
    },
    {
      "epoch": 0.24,
      "grad_norm": 13.502409598859076,
      "learning_rate": 8.923411846785322e-06,
      "loss": 0.424,
      "step": 1468
    },
    {
      "epoch": 0.24,
      "grad_norm": 13.153480065630106,
      "learning_rate": 8.921793546090709e-06,
      "loss": 0.4972,
      "step": 1469
    },
    {
      "epoch": 0.24,
      "grad_norm": 17.658353229158543,
      "learning_rate": 8.920174176991426e-06,
      "loss": 0.3933,
      "step": 1470
    },
    {
      "epoch": 0.24,
      "grad_norm": 11.598314892462291,
      "learning_rate": 8.918553739928637e-06,
      "loss": 0.4715,
      "step": 1471
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.894333455457994,
      "learning_rate": 8.916932235343797e-06,
      "loss": 0.4046,
      "step": 1472
    },
    {
      "epoch": 0.24,
      "grad_norm": 11.202730447168385,
      "learning_rate": 8.915309663678641e-06,
      "loss": 0.3356,
      "step": 1473
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.222870010651823,
      "learning_rate": 8.913686025375207e-06,
      "loss": 0.4715,
      "step": 1474
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.921090167960903,
      "learning_rate": 8.91206132087582e-06,
      "loss": 0.4393,
      "step": 1475
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.665442244238676,
      "learning_rate": 8.910435550623093e-06,
      "loss": 0.5106,
      "step": 1476
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.385219651649345,
      "learning_rate": 8.908808715059929e-06,
      "loss": 0.4484,
      "step": 1477
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.989471358964002,
      "learning_rate": 8.907180814629526e-06,
      "loss": 0.3998,
      "step": 1478
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.953810607363012,
      "learning_rate": 8.905551849775365e-06,
      "loss": 0.4225,
      "step": 1479
    },
    {
      "epoch": 0.24,
      "grad_norm": 19.387315516413985,
      "learning_rate": 8.903921820941224e-06,
      "loss": 0.4943,
      "step": 1480
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.360662032342251,
      "learning_rate": 8.902290728571165e-06,
      "loss": 0.4566,
      "step": 1481
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.828819831140725,
      "learning_rate": 8.900658573109546e-06,
      "loss": 0.4469,
      "step": 1482
    },
    {
      "epoch": 0.24,
      "grad_norm": 16.486301244107388,
      "learning_rate": 8.89902535500101e-06,
      "loss": 0.4308,
      "step": 1483
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.946320977649018,
      "learning_rate": 8.897391074690489e-06,
      "loss": 0.4986,
      "step": 1484
    },
    {
      "epoch": 0.24,
      "grad_norm": 24.17010586858406,
      "learning_rate": 8.895755732623207e-06,
      "loss": 0.4612,
      "step": 1485
    },
    {
      "epoch": 0.24,
      "grad_norm": 11.499776398578296,
      "learning_rate": 8.894119329244675e-06,
      "loss": 0.5218,
      "step": 1486
    },
    {
      "epoch": 0.24,
      "grad_norm": 16.767031555023646,
      "learning_rate": 8.892481865000698e-06,
      "loss": 0.48,
      "step": 1487
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.26755103261537,
      "learning_rate": 8.890843340337363e-06,
      "loss": 0.4486,
      "step": 1488
    },
    {
      "epoch": 0.24,
      "grad_norm": 19.02391252980494,
      "learning_rate": 8.88920375570105e-06,
      "loss": 0.3607,
      "step": 1489
    },
    {
      "epoch": 0.24,
      "grad_norm": 18.768405413748543,
      "learning_rate": 8.887563111538428e-06,
      "loss": 0.4413,
      "step": 1490
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.266156968094913,
      "learning_rate": 8.885921408296454e-06,
      "loss": 0.4467,
      "step": 1491
    },
    {
      "epoch": 0.24,
      "grad_norm": 12.162884144522227,
      "learning_rate": 8.88427864642237e-06,
      "loss": 0.4993,
      "step": 1492
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.316698072328538,
      "learning_rate": 8.882634826363714e-06,
      "loss": 0.3914,
      "step": 1493
    },
    {
      "epoch": 0.24,
      "grad_norm": 21.78965111471558,
      "learning_rate": 8.880989948568303e-06,
      "loss": 0.4835,
      "step": 1494
    },
    {
      "epoch": 0.24,
      "grad_norm": 12.386533619233864,
      "learning_rate": 8.879344013484253e-06,
      "loss": 0.456,
      "step": 1495
    },
    {
      "epoch": 0.24,
      "grad_norm": 18.561662849828668,
      "learning_rate": 8.877697021559958e-06,
      "loss": 0.5363,
      "step": 1496
    },
    {
      "epoch": 0.24,
      "grad_norm": 15.611816621633134,
      "learning_rate": 8.876048973244105e-06,
      "loss": 0.4692,
      "step": 1497
    },
    {
      "epoch": 0.24,
      "grad_norm": 15.247695530444396,
      "learning_rate": 8.874399868985668e-06,
      "loss": 0.4311,
      "step": 1498
    },
    {
      "epoch": 0.24,
      "grad_norm": 13.373173976261493,
      "learning_rate": 8.872749709233907e-06,
      "loss": 0.5271,
      "step": 1499
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.322046954447156,
      "learning_rate": 8.871098494438375e-06,
      "loss": 0.4505,
      "step": 1500
    },
    {
      "epoch": 0.24,
      "grad_norm": 19.172725477371777,
      "learning_rate": 8.869446225048903e-06,
      "loss": 0.4634,
      "step": 1501
    },
    {
      "epoch": 0.24,
      "grad_norm": 28.065674841026837,
      "learning_rate": 8.867792901515617e-06,
      "loss": 0.4358,
      "step": 1502
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.915751787905425,
      "learning_rate": 8.866138524288929e-06,
      "loss": 0.4824,
      "step": 1503
    },
    {
      "epoch": 0.24,
      "grad_norm": 21.636668840373233,
      "learning_rate": 8.864483093819537e-06,
      "loss": 0.5179,
      "step": 1504
    },
    {
      "epoch": 0.24,
      "grad_norm": 11.284744019902902,
      "learning_rate": 8.862826610558427e-06,
      "loss": 0.4411,
      "step": 1505
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.870551271729145,
      "learning_rate": 8.861169074956865e-06,
      "loss": 0.3877,
      "step": 1506
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.510066653139628,
      "learning_rate": 8.859510487466415e-06,
      "loss": 0.5143,
      "step": 1507
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.854886437852786,
      "learning_rate": 8.85785084853892e-06,
      "loss": 0.4592,
      "step": 1508
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.567428624446746,
      "learning_rate": 8.856190158626512e-06,
      "loss": 0.5008,
      "step": 1509
    },
    {
      "epoch": 0.24,
      "grad_norm": 11.22432118493056,
      "learning_rate": 8.854528418181609e-06,
      "loss": 0.4681,
      "step": 1510
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.210935296363807,
      "learning_rate": 8.852865627656912e-06,
      "loss": 0.4547,
      "step": 1511
    },
    {
      "epoch": 0.24,
      "grad_norm": 17.77219464157935,
      "learning_rate": 8.851201787505415e-06,
      "loss": 0.4548,
      "step": 1512
    },
    {
      "epoch": 0.24,
      "grad_norm": 12.259867303667486,
      "learning_rate": 8.84953689818039e-06,
      "loss": 0.4927,
      "step": 1513
    },
    {
      "epoch": 0.24,
      "grad_norm": 13.915689227371422,
      "learning_rate": 8.847870960135403e-06,
      "loss": 0.379,
      "step": 1514
    },
    {
      "epoch": 0.24,
      "grad_norm": 15.157403765320169,
      "learning_rate": 8.8462039738243e-06,
      "loss": 0.4818,
      "step": 1515
    },
    {
      "epoch": 0.24,
      "grad_norm": 13.476337575029335,
      "learning_rate": 8.84453593970121e-06,
      "loss": 0.4334,
      "step": 1516
    },
    {
      "epoch": 0.24,
      "grad_norm": 12.843487372142736,
      "learning_rate": 8.842866858220558e-06,
      "loss": 0.4789,
      "step": 1517
    },
    {
      "epoch": 0.24,
      "grad_norm": 12.881006581618307,
      "learning_rate": 8.841196729837044e-06,
      "loss": 0.4581,
      "step": 1518
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.77858180743534,
      "learning_rate": 8.839525555005656e-06,
      "loss": 0.4953,
      "step": 1519
    },
    {
      "epoch": 0.24,
      "grad_norm": 12.042844525487867,
      "learning_rate": 8.837853334181669e-06,
      "loss": 0.5429,
      "step": 1520
    },
    {
      "epoch": 0.25,
      "grad_norm": 12.585240439976088,
      "learning_rate": 8.836180067820646e-06,
      "loss": 0.4218,
      "step": 1521
    },
    {
      "epoch": 0.25,
      "grad_norm": 15.120633185594732,
      "learning_rate": 8.834505756378425e-06,
      "loss": 0.4639,
      "step": 1522
    },
    {
      "epoch": 0.25,
      "grad_norm": 9.74554076952203,
      "learning_rate": 8.832830400311137e-06,
      "loss": 0.4783,
      "step": 1523
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.379328626539223,
      "learning_rate": 8.831154000075196e-06,
      "loss": 0.4685,
      "step": 1524
    },
    {
      "epoch": 0.25,
      "grad_norm": 13.914355972277242,
      "learning_rate": 8.829476556127301e-06,
      "loss": 0.4383,
      "step": 1525
    },
    {
      "epoch": 0.25,
      "grad_norm": 24.298526008086267,
      "learning_rate": 8.82779806892443e-06,
      "loss": 0.4679,
      "step": 1526
    },
    {
      "epoch": 0.25,
      "grad_norm": 13.703473101307404,
      "learning_rate": 8.826118538923851e-06,
      "loss": 0.4968,
      "step": 1527
    },
    {
      "epoch": 0.25,
      "grad_norm": 14.10033117811136,
      "learning_rate": 8.824437966583114e-06,
      "loss": 0.512,
      "step": 1528
    },
    {
      "epoch": 0.25,
      "grad_norm": 9.869979959375344,
      "learning_rate": 8.822756352360056e-06,
      "loss": 0.4619,
      "step": 1529
    },
    {
      "epoch": 0.25,
      "grad_norm": 18.959301298861334,
      "learning_rate": 8.82107369671279e-06,
      "loss": 0.4479,
      "step": 1530
    },
    {
      "epoch": 0.25,
      "grad_norm": 24.512693186427196,
      "learning_rate": 8.819390000099723e-06,
      "loss": 0.4853,
      "step": 1531
    },
    {
      "epoch": 0.25,
      "grad_norm": 11.015412494470718,
      "learning_rate": 8.817705262979536e-06,
      "loss": 0.4275,
      "step": 1532
    },
    {
      "epoch": 0.25,
      "grad_norm": 19.078467763508947,
      "learning_rate": 8.8160194858112e-06,
      "loss": 0.5253,
      "step": 1533
    },
    {
      "epoch": 0.25,
      "grad_norm": 40.32751858306284,
      "learning_rate": 8.814332669053968e-06,
      "loss": 0.4961,
      "step": 1534
    },
    {
      "epoch": 0.25,
      "grad_norm": 12.806281402721412,
      "learning_rate": 8.812644813167372e-06,
      "loss": 0.4689,
      "step": 1535
    },
    {
      "epoch": 0.25,
      "grad_norm": 10.506546895101797,
      "learning_rate": 8.810955918611235e-06,
      "loss": 0.3931,
      "step": 1536
    },
    {
      "epoch": 0.25,
      "grad_norm": 12.33836872639905,
      "learning_rate": 8.809265985845655e-06,
      "loss": 0.461,
      "step": 1537
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.458692726322872,
      "learning_rate": 8.807575015331019e-06,
      "loss": 0.4868,
      "step": 1538
    },
    {
      "epoch": 0.25,
      "grad_norm": 27.8547530549637,
      "learning_rate": 8.805883007527992e-06,
      "loss": 0.471,
      "step": 1539
    },
    {
      "epoch": 0.25,
      "grad_norm": 54.47938342993424,
      "learning_rate": 8.804189962897521e-06,
      "loss": 0.5103,
      "step": 1540
    },
    {
      "epoch": 0.25,
      "grad_norm": 11.672482910770182,
      "learning_rate": 8.802495881900844e-06,
      "loss": 0.4434,
      "step": 1541
    },
    {
      "epoch": 0.25,
      "grad_norm": 14.586492126303982,
      "learning_rate": 8.80080076499947e-06,
      "loss": 0.504,
      "step": 1542
    },
    {
      "epoch": 0.25,
      "grad_norm": 16.949378974758428,
      "learning_rate": 8.7991046126552e-06,
      "loss": 0.5287,
      "step": 1543
    },
    {
      "epoch": 0.25,
      "grad_norm": 18.159470018985314,
      "learning_rate": 8.79740742533011e-06,
      "loss": 0.4861,
      "step": 1544
    },
    {
      "epoch": 0.25,
      "grad_norm": 17.607804835672827,
      "learning_rate": 8.795709203486563e-06,
      "loss": 0.4885,
      "step": 1545
    },
    {
      "epoch": 0.25,
      "grad_norm": 12.159775804109794,
      "learning_rate": 8.794009947587197e-06,
      "loss": 0.4585,
      "step": 1546
    },
    {
      "epoch": 0.25,
      "grad_norm": 26.62923218034464,
      "learning_rate": 8.79230965809494e-06,
      "loss": 0.4859,
      "step": 1547
    },
    {
      "epoch": 0.25,
      "grad_norm": 14.250902243362471,
      "learning_rate": 8.790608335472995e-06,
      "loss": 0.5175,
      "step": 1548
    },
    {
      "epoch": 0.25,
      "grad_norm": 11.675925681964848,
      "learning_rate": 8.788905980184851e-06,
      "loss": 0.4865,
      "step": 1549
    },
    {
      "epoch": 0.25,
      "grad_norm": 14.087372519844426,
      "learning_rate": 8.787202592694278e-06,
      "loss": 0.4431,
      "step": 1550
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5630420020553932,
      "learning_rate": 8.785498173465323e-06,
      "loss": 0.4954,
      "step": 1551
    },
    {
      "epoch": 0.25,
      "grad_norm": 14.049407890795996,
      "learning_rate": 8.783792722962316e-06,
      "loss": 0.5069,
      "step": 1552
    },
    {
      "epoch": 0.25,
      "grad_norm": 14.50248217894901,
      "learning_rate": 8.782086241649874e-06,
      "loss": 0.4288,
      "step": 1553
    },
    {
      "epoch": 0.25,
      "grad_norm": 85.24993064665662,
      "learning_rate": 8.780378729992884e-06,
      "loss": 0.4363,
      "step": 1554
    },
    {
      "epoch": 0.25,
      "grad_norm": 20.94501475124581,
      "learning_rate": 8.778670188456519e-06,
      "loss": 0.489,
      "step": 1555
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2893407157129522,
      "learning_rate": 8.776960617506237e-06,
      "loss": 0.4891,
      "step": 1556
    },
    {
      "epoch": 0.25,
      "grad_norm": 11.287533596726686,
      "learning_rate": 8.77525001760777e-06,
      "loss": 0.4055,
      "step": 1557
    },
    {
      "epoch": 0.25,
      "grad_norm": 13.671952396695993,
      "learning_rate": 8.773538389227134e-06,
      "loss": 0.5449,
      "step": 1558
    },
    {
      "epoch": 0.25,
      "grad_norm": 13.004745757255726,
      "learning_rate": 8.771825732830622e-06,
      "loss": 0.4723,
      "step": 1559
    },
    {
      "epoch": 0.25,
      "grad_norm": 25.78654896117246,
      "learning_rate": 8.77011204888481e-06,
      "loss": 0.4569,
      "step": 1560
    },
    {
      "epoch": 0.25,
      "grad_norm": 13.476302103720993,
      "learning_rate": 8.76839733785655e-06,
      "loss": 0.4803,
      "step": 1561
    },
    {
      "epoch": 0.25,
      "grad_norm": 15.924910595720602,
      "learning_rate": 8.766681600212981e-06,
      "loss": 0.4544,
      "step": 1562
    },
    {
      "epoch": 0.25,
      "grad_norm": 11.899806530334011,
      "learning_rate": 8.764964836421515e-06,
      "loss": 0.4417,
      "step": 1563
    },
    {
      "epoch": 0.25,
      "grad_norm": 11.213178257009929,
      "learning_rate": 8.763247046949843e-06,
      "loss": 0.4845,
      "step": 1564
    },
    {
      "epoch": 0.25,
      "grad_norm": 15.640810050489614,
      "learning_rate": 8.761528232265944e-06,
      "loss": 0.4844,
      "step": 1565
    },
    {
      "epoch": 0.25,
      "grad_norm": 14.732590348529142,
      "learning_rate": 8.759808392838066e-06,
      "loss": 0.4117,
      "step": 1566
    },
    {
      "epoch": 0.25,
      "grad_norm": 11.777350749626775,
      "learning_rate": 8.75808752913474e-06,
      "loss": 0.4687,
      "step": 1567
    },
    {
      "epoch": 0.25,
      "grad_norm": 10.857837687193514,
      "learning_rate": 8.756365641624782e-06,
      "loss": 0.4967,
      "step": 1568
    },
    {
      "epoch": 0.25,
      "grad_norm": 14.225323037480715,
      "learning_rate": 8.754642730777276e-06,
      "loss": 0.5513,
      "step": 1569
    },
    {
      "epoch": 0.25,
      "grad_norm": 19.721041422185902,
      "learning_rate": 8.752918797061593e-06,
      "loss": 0.4795,
      "step": 1570
    },
    {
      "epoch": 0.25,
      "grad_norm": 20.624679107638986,
      "learning_rate": 8.751193840947382e-06,
      "loss": 0.4806,
      "step": 1571
    },
    {
      "epoch": 0.25,
      "grad_norm": 12.15213570097301,
      "learning_rate": 8.749467862904565e-06,
      "loss": 0.4434,
      "step": 1572
    },
    {
      "epoch": 0.25,
      "grad_norm": 12.669716573387246,
      "learning_rate": 8.747740863403348e-06,
      "loss": 0.4534,
      "step": 1573
    },
    {
      "epoch": 0.25,
      "grad_norm": 10.42812835328904,
      "learning_rate": 8.746012842914214e-06,
      "loss": 0.3991,
      "step": 1574
    },
    {
      "epoch": 0.25,
      "grad_norm": 10.628703469906624,
      "learning_rate": 8.74428380190792e-06,
      "loss": 0.4875,
      "step": 1575
    },
    {
      "epoch": 0.25,
      "grad_norm": 11.53381864997919,
      "learning_rate": 8.742553740855507e-06,
      "loss": 0.4444,
      "step": 1576
    },
    {
      "epoch": 0.25,
      "grad_norm": 11.361858392222972,
      "learning_rate": 8.74082266022829e-06,
      "loss": 0.5031,
      "step": 1577
    },
    {
      "epoch": 0.25,
      "grad_norm": 19.741825356892356,
      "learning_rate": 8.739090560497864e-06,
      "loss": 0.4969,
      "step": 1578
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.369046679916586,
      "learning_rate": 8.737357442136104e-06,
      "loss": 0.5179,
      "step": 1579
    },
    {
      "epoch": 0.25,
      "grad_norm": 9.38618474385405,
      "learning_rate": 8.735623305615153e-06,
      "loss": 0.4716,
      "step": 1580
    },
    {
      "epoch": 0.25,
      "grad_norm": 14.374576774025085,
      "learning_rate": 8.733888151407441e-06,
      "loss": 0.4917,
      "step": 1581
    },
    {
      "epoch": 0.25,
      "grad_norm": 11.434308699435563,
      "learning_rate": 8.732151979985671e-06,
      "loss": 0.5607,
      "step": 1582
    },
    {
      "epoch": 0.26,
      "grad_norm": 10.999567354293504,
      "learning_rate": 8.730414791822825e-06,
      "loss": 0.4855,
      "step": 1583
    },
    {
      "epoch": 0.26,
      "grad_norm": 16.43151531480138,
      "learning_rate": 8.72867658739216e-06,
      "loss": 0.3989,
      "step": 1584
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.507633457444578,
      "learning_rate": 8.726937367167211e-06,
      "loss": 0.5235,
      "step": 1585
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.395906798739117,
      "learning_rate": 8.72519713162179e-06,
      "loss": 0.4209,
      "step": 1586
    },
    {
      "epoch": 0.26,
      "grad_norm": 10.075748424826264,
      "learning_rate": 8.723455881229984e-06,
      "loss": 0.4558,
      "step": 1587
    },
    {
      "epoch": 0.26,
      "grad_norm": 74.68050799151031,
      "learning_rate": 8.721713616466158e-06,
      "loss": 0.4714,
      "step": 1588
    },
    {
      "epoch": 0.26,
      "grad_norm": 11.17493845346469,
      "learning_rate": 8.719970337804953e-06,
      "loss": 0.4418,
      "step": 1589
    },
    {
      "epoch": 0.26,
      "grad_norm": 15.108421406802913,
      "learning_rate": 8.718226045721287e-06,
      "loss": 0.499,
      "step": 1590
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.386312616900841,
      "learning_rate": 8.716480740690353e-06,
      "loss": 0.5241,
      "step": 1591
    },
    {
      "epoch": 0.26,
      "grad_norm": 11.152886129264175,
      "learning_rate": 8.71473442318762e-06,
      "loss": 0.4956,
      "step": 1592
    },
    {
      "epoch": 0.26,
      "grad_norm": 9.824383606819817,
      "learning_rate": 8.712987093688833e-06,
      "loss": 0.5248,
      "step": 1593
    },
    {
      "epoch": 0.26,
      "grad_norm": 10.840339097815546,
      "learning_rate": 8.711238752670012e-06,
      "loss": 0.4019,
      "step": 1594
    },
    {
      "epoch": 0.26,
      "grad_norm": 14.305540928187915,
      "learning_rate": 8.709489400607453e-06,
      "loss": 0.5416,
      "step": 1595
    },
    {
      "epoch": 0.26,
      "grad_norm": 14.227153851722683,
      "learning_rate": 8.70773903797773e-06,
      "loss": 0.5591,
      "step": 1596
    },
    {
      "epoch": 0.26,
      "grad_norm": 11.108728143104305,
      "learning_rate": 8.705987665257688e-06,
      "loss": 0.5017,
      "step": 1597
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.939312849928044,
      "learning_rate": 8.704235282924449e-06,
      "loss": 0.4614,
      "step": 1598
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.875892599537317,
      "learning_rate": 8.702481891455414e-06,
      "loss": 0.4704,
      "step": 1599
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.207126504701245,
      "learning_rate": 8.70072749132825e-06,
      "loss": 0.4264,
      "step": 1600
    },
    {
      "epoch": 0.26,
      "grad_norm": 10.784673634393716,
      "learning_rate": 8.698972083020905e-06,
      "loss": 0.4546,
      "step": 1601
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3230511787201216,
      "learning_rate": 8.697215667011605e-06,
      "loss": 0.4705,
      "step": 1602
    },
    {
      "epoch": 0.26,
      "grad_norm": 11.98221144030863,
      "learning_rate": 8.69545824377884e-06,
      "loss": 0.4449,
      "step": 1603
    },
    {
      "epoch": 0.26,
      "grad_norm": 23.821188858311075,
      "learning_rate": 8.693699813801387e-06,
      "loss": 0.5079,
      "step": 1604
    },
    {
      "epoch": 0.26,
      "grad_norm": 15.651753885235879,
      "learning_rate": 8.691940377558284e-06,
      "loss": 0.4985,
      "step": 1605
    },
    {
      "epoch": 0.26,
      "grad_norm": 10.218542923354468,
      "learning_rate": 8.690179935528853e-06,
      "loss": 0.4156,
      "step": 1606
    },
    {
      "epoch": 0.26,
      "grad_norm": 14.92000654913018,
      "learning_rate": 8.68841848819269e-06,
      "loss": 0.4425,
      "step": 1607
    },
    {
      "epoch": 0.26,
      "grad_norm": 11.039339432108823,
      "learning_rate": 8.686656036029657e-06,
      "loss": 0.4664,
      "step": 1608
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.023226576557596,
      "learning_rate": 8.684892579519897e-06,
      "loss": 0.447,
      "step": 1609
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.386391500978715,
      "learning_rate": 8.683128119143824e-06,
      "loss": 0.4835,
      "step": 1610
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.272351642503492,
      "learning_rate": 8.681362655382125e-06,
      "loss": 0.4901,
      "step": 1611
    },
    {
      "epoch": 0.26,
      "grad_norm": 11.562683062850953,
      "learning_rate": 8.67959618871576e-06,
      "loss": 0.5142,
      "step": 1612
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.337599537261909,
      "learning_rate": 8.677828719625967e-06,
      "loss": 0.4586,
      "step": 1613
    },
    {
      "epoch": 0.26,
      "grad_norm": 12.549774493591219,
      "learning_rate": 8.676060248594248e-06,
      "loss": 0.4516,
      "step": 1614
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.91220969728968,
      "learning_rate": 8.674290776102388e-06,
      "loss": 0.48,
      "step": 1615
    },
    {
      "epoch": 0.26,
      "grad_norm": 31.60787626908987,
      "learning_rate": 8.67252030263244e-06,
      "loss": 0.4157,
      "step": 1616
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.012919896435109,
      "learning_rate": 8.670748828666725e-06,
      "loss": 0.501,
      "step": 1617
    },
    {
      "epoch": 0.26,
      "grad_norm": 9.562450922303592,
      "learning_rate": 8.668976354687844e-06,
      "loss": 0.4023,
      "step": 1618
    },
    {
      "epoch": 0.26,
      "grad_norm": 9.126666057795351,
      "learning_rate": 8.66720288117867e-06,
      "loss": 0.5263,
      "step": 1619
    },
    {
      "epoch": 0.26,
      "grad_norm": 19.048680437263343,
      "learning_rate": 8.665428408622343e-06,
      "loss": 0.4949,
      "step": 1620
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.304286820862748,
      "learning_rate": 8.66365293750228e-06,
      "loss": 0.4747,
      "step": 1621
    },
    {
      "epoch": 0.26,
      "grad_norm": 9.330854118739088,
      "learning_rate": 8.661876468302167e-06,
      "loss": 0.3907,
      "step": 1622
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.773192018598685,
      "learning_rate": 8.660099001505965e-06,
      "loss": 0.462,
      "step": 1623
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.423545288887948,
      "learning_rate": 8.658320537597901e-06,
      "loss": 0.4659,
      "step": 1624
    },
    {
      "epoch": 0.26,
      "grad_norm": 13.619672130449494,
      "learning_rate": 8.656541077062483e-06,
      "loss": 0.4232,
      "step": 1625
    },
    {
      "epoch": 0.26,
      "grad_norm": 12.155801981142217,
      "learning_rate": 8.654760620384482e-06,
      "loss": 0.4828,
      "step": 1626
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.853544841782938,
      "learning_rate": 8.652979168048944e-06,
      "loss": 0.4484,
      "step": 1627
    },
    {
      "epoch": 0.26,
      "grad_norm": 14.509172060832219,
      "learning_rate": 8.651196720541186e-06,
      "loss": 0.5376,
      "step": 1628
    },
    {
      "epoch": 0.26,
      "grad_norm": 14.780221815046163,
      "learning_rate": 8.649413278346795e-06,
      "loss": 0.5588,
      "step": 1629
    },
    {
      "epoch": 0.26,
      "grad_norm": 10.350896913841805,
      "learning_rate": 8.64762884195163e-06,
      "loss": 0.4948,
      "step": 1630
    },
    {
      "epoch": 0.26,
      "grad_norm": 10.515572713753683,
      "learning_rate": 8.64584341184182e-06,
      "loss": 0.4799,
      "step": 1631
    },
    {
      "epoch": 0.26,
      "grad_norm": 11.242723253405977,
      "learning_rate": 8.644056988503769e-06,
      "loss": 0.4776,
      "step": 1632
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.682261032499714,
      "learning_rate": 8.642269572424143e-06,
      "loss": 0.5032,
      "step": 1633
    },
    {
      "epoch": 0.26,
      "grad_norm": 13.137230347762788,
      "learning_rate": 8.640481164089887e-06,
      "loss": 0.4574,
      "step": 1634
    },
    {
      "epoch": 0.26,
      "grad_norm": 13.13361040266308,
      "learning_rate": 8.63869176398821e-06,
      "loss": 0.4564,
      "step": 1635
    },
    {
      "epoch": 0.26,
      "grad_norm": 32.3337013144351,
      "learning_rate": 8.636901372606596e-06,
      "loss": 0.4893,
      "step": 1636
    },
    {
      "epoch": 0.26,
      "grad_norm": 10.334518888758286,
      "learning_rate": 8.635109990432797e-06,
      "loss": 0.4794,
      "step": 1637
    },
    {
      "epoch": 0.26,
      "grad_norm": 16.42994113303752,
      "learning_rate": 8.633317617954832e-06,
      "loss": 0.4015,
      "step": 1638
    },
    {
      "epoch": 0.26,
      "grad_norm": 15.055502561901461,
      "learning_rate": 8.631524255660997e-06,
      "loss": 0.4392,
      "step": 1639
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.92981169540569,
      "learning_rate": 8.629729904039853e-06,
      "loss": 0.441,
      "step": 1640
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3895174555023133,
      "learning_rate": 8.627934563580226e-06,
      "loss": 0.4982,
      "step": 1641
    },
    {
      "epoch": 0.26,
      "grad_norm": 15.557621464092596,
      "learning_rate": 8.62613823477122e-06,
      "loss": 0.4587,
      "step": 1642
    },
    {
      "epoch": 0.26,
      "grad_norm": 10.518765854296685,
      "learning_rate": 8.624340918102206e-06,
      "loss": 0.4543,
      "step": 1643
    },
    {
      "epoch": 0.26,
      "grad_norm": 12.869296176655247,
      "learning_rate": 8.622542614062816e-06,
      "loss": 0.4249,
      "step": 1644
    },
    {
      "epoch": 0.27,
      "grad_norm": 17.10485778520921,
      "learning_rate": 8.620743323142966e-06,
      "loss": 0.4686,
      "step": 1645
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3853653315378351,
      "learning_rate": 8.618943045832826e-06,
      "loss": 0.4945,
      "step": 1646
    },
    {
      "epoch": 0.27,
      "grad_norm": 9.63686716824368,
      "learning_rate": 8.617141782622844e-06,
      "loss": 0.5502,
      "step": 1647
    },
    {
      "epoch": 0.27,
      "grad_norm": 11.779907949552356,
      "learning_rate": 8.615339534003735e-06,
      "loss": 0.4504,
      "step": 1648
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.028652664341822,
      "learning_rate": 8.613536300466476e-06,
      "loss": 0.4973,
      "step": 1649
    },
    {
      "epoch": 0.27,
      "grad_norm": 10.808902802279686,
      "learning_rate": 8.611732082502324e-06,
      "loss": 0.4591,
      "step": 1650
    },
    {
      "epoch": 0.27,
      "grad_norm": 10.749152932092182,
      "learning_rate": 8.609926880602794e-06,
      "loss": 0.4115,
      "step": 1651
    },
    {
      "epoch": 0.27,
      "grad_norm": 10.026370896564668,
      "learning_rate": 8.608120695259674e-06,
      "loss": 0.4868,
      "step": 1652
    },
    {
      "epoch": 0.27,
      "grad_norm": 11.50421339016215,
      "learning_rate": 8.606313526965017e-06,
      "loss": 0.5115,
      "step": 1653
    },
    {
      "epoch": 0.27,
      "grad_norm": 18.41372557705101,
      "learning_rate": 8.604505376211148e-06,
      "loss": 0.5554,
      "step": 1654
    },
    {
      "epoch": 0.27,
      "grad_norm": 10.236731374543647,
      "learning_rate": 8.602696243490653e-06,
      "loss": 0.4143,
      "step": 1655
    },
    {
      "epoch": 0.27,
      "grad_norm": 10.620595490367121,
      "learning_rate": 8.600886129296396e-06,
      "loss": 0.4151,
      "step": 1656
    },
    {
      "epoch": 0.27,
      "grad_norm": 33.25172395196379,
      "learning_rate": 8.599075034121496e-06,
      "loss": 0.4621,
      "step": 1657
    },
    {
      "epoch": 0.27,
      "grad_norm": 12.978323308186885,
      "learning_rate": 8.59726295845935e-06,
      "loss": 0.5235,
      "step": 1658
    },
    {
      "epoch": 0.27,
      "grad_norm": 13.602390675921882,
      "learning_rate": 8.595449902803612e-06,
      "loss": 0.4089,
      "step": 1659
    },
    {
      "epoch": 0.27,
      "grad_norm": 13.696820801308398,
      "learning_rate": 8.593635867648214e-06,
      "loss": 0.4428,
      "step": 1660
    },
    {
      "epoch": 0.27,
      "grad_norm": 11.41039944464447,
      "learning_rate": 8.591820853487344e-06,
      "loss": 0.5256,
      "step": 1661
    },
    {
      "epoch": 0.27,
      "grad_norm": 9.054526887437678,
      "learning_rate": 8.590004860815466e-06,
      "loss": 0.4178,
      "step": 1662
    },
    {
      "epoch": 0.27,
      "grad_norm": 12.486451819130309,
      "learning_rate": 8.588187890127305e-06,
      "loss": 0.4175,
      "step": 1663
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.706921743690426,
      "learning_rate": 8.586369941917852e-06,
      "loss": 0.5075,
      "step": 1664
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.05660143987441,
      "learning_rate": 8.58455101668237e-06,
      "loss": 0.4665,
      "step": 1665
    },
    {
      "epoch": 0.27,
      "grad_norm": 14.011998156857956,
      "learning_rate": 8.58273111491638e-06,
      "loss": 0.4425,
      "step": 1666
    },
    {
      "epoch": 0.27,
      "grad_norm": 9.781817735925221,
      "learning_rate": 8.580910237115678e-06,
      "loss": 0.5101,
      "step": 1667
    },
    {
      "epoch": 0.27,
      "grad_norm": 20.66535014417799,
      "learning_rate": 8.579088383776318e-06,
      "loss": 0.5304,
      "step": 1668
    },
    {
      "epoch": 0.27,
      "grad_norm": 16.90363131393331,
      "learning_rate": 8.577265555394626e-06,
      "loss": 0.4364,
      "step": 1669
    },
    {
      "epoch": 0.27,
      "grad_norm": 11.00326759488563,
      "learning_rate": 8.575441752467185e-06,
      "loss": 0.4369,
      "step": 1670
    },
    {
      "epoch": 0.27,
      "grad_norm": 11.101639684614764,
      "learning_rate": 8.573616975490855e-06,
      "loss": 0.4911,
      "step": 1671
    },
    {
      "epoch": 0.27,
      "grad_norm": 14.278542819965034,
      "learning_rate": 8.571791224962754e-06,
      "loss": 0.4537,
      "step": 1672
    },
    {
      "epoch": 0.27,
      "grad_norm": 9.27675994511827,
      "learning_rate": 8.569964501380266e-06,
      "loss": 0.4728,
      "step": 1673
    },
    {
      "epoch": 0.27,
      "grad_norm": 34.81570469805829,
      "learning_rate": 8.56813680524104e-06,
      "loss": 0.4279,
      "step": 1674
    },
    {
      "epoch": 0.27,
      "grad_norm": 18.298980035789068,
      "learning_rate": 8.566308137042995e-06,
      "loss": 0.4922,
      "step": 1675
    },
    {
      "epoch": 0.27,
      "grad_norm": 13.033749923288475,
      "learning_rate": 8.564478497284306e-06,
      "loss": 0.4888,
      "step": 1676
    },
    {
      "epoch": 0.27,
      "grad_norm": 12.503427217368314,
      "learning_rate": 8.562647886463417e-06,
      "loss": 0.4862,
      "step": 1677
    },
    {
      "epoch": 0.27,
      "grad_norm": 31.240571004033875,
      "learning_rate": 8.560816305079041e-06,
      "loss": 0.3986,
      "step": 1678
    },
    {
      "epoch": 0.27,
      "grad_norm": 12.130460752457049,
      "learning_rate": 8.558983753630149e-06,
      "loss": 0.4324,
      "step": 1679
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4086478235800388,
      "learning_rate": 8.557150232615977e-06,
      "loss": 0.4886,
      "step": 1680
    },
    {
      "epoch": 0.27,
      "grad_norm": 14.851887094705658,
      "learning_rate": 8.55531574253603e-06,
      "loss": 0.4839,
      "step": 1681
    },
    {
      "epoch": 0.27,
      "grad_norm": 13.72788461712924,
      "learning_rate": 8.55348028389007e-06,
      "loss": 0.4828,
      "step": 1682
    },
    {
      "epoch": 0.27,
      "grad_norm": 16.252121533670486,
      "learning_rate": 8.55164385717813e-06,
      "loss": 0.4857,
      "step": 1683
    },
    {
      "epoch": 0.27,
      "grad_norm": 11.574199275771631,
      "learning_rate": 8.549806462900503e-06,
      "loss": 0.5071,
      "step": 1684
    },
    {
      "epoch": 0.27,
      "grad_norm": 17.054549576769286,
      "learning_rate": 8.547968101557742e-06,
      "loss": 0.4587,
      "step": 1685
    },
    {
      "epoch": 0.27,
      "grad_norm": 10.51320614334958,
      "learning_rate": 8.54612877365067e-06,
      "loss": 0.4483,
      "step": 1686
    },
    {
      "epoch": 0.27,
      "grad_norm": 12.978725054756705,
      "learning_rate": 8.544288479680371e-06,
      "loss": 0.3968,
      "step": 1687
    },
    {
      "epoch": 0.27,
      "grad_norm": 12.153282940935025,
      "learning_rate": 8.542447220148191e-06,
      "loss": 0.5234,
      "step": 1688
    },
    {
      "epoch": 0.27,
      "grad_norm": 19.47603342477601,
      "learning_rate": 8.540604995555741e-06,
      "loss": 0.4826,
      "step": 1689
    },
    {
      "epoch": 0.27,
      "grad_norm": 14.18731041762881,
      "learning_rate": 8.538761806404892e-06,
      "loss": 0.4605,
      "step": 1690
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3946656817674892,
      "learning_rate": 8.53691765319778e-06,
      "loss": 0.4517,
      "step": 1691
    },
    {
      "epoch": 0.27,
      "grad_norm": 15.467965220831184,
      "learning_rate": 8.535072536436805e-06,
      "loss": 0.4843,
      "step": 1692
    },
    {
      "epoch": 0.27,
      "grad_norm": 15.594412590202573,
      "learning_rate": 8.533226456624624e-06,
      "loss": 0.4398,
      "step": 1693
    },
    {
      "epoch": 0.27,
      "grad_norm": 14.116100439626473,
      "learning_rate": 8.531379414264165e-06,
      "loss": 0.4495,
      "step": 1694
    },
    {
      "epoch": 0.27,
      "grad_norm": 21.007705304321693,
      "learning_rate": 8.52953140985861e-06,
      "loss": 0.5625,
      "step": 1695
    },
    {
      "epoch": 0.27,
      "grad_norm": 19.74921427902309,
      "learning_rate": 8.527682443911405e-06,
      "loss": 0.4496,
      "step": 1696
    },
    {
      "epoch": 0.27,
      "grad_norm": 15.638990948192355,
      "learning_rate": 8.525832516926262e-06,
      "loss": 0.5645,
      "step": 1697
    },
    {
      "epoch": 0.27,
      "grad_norm": 20.959114363780845,
      "learning_rate": 8.52398162940715e-06,
      "loss": 0.3857,
      "step": 1698
    },
    {
      "epoch": 0.27,
      "grad_norm": 16.137100407316332,
      "learning_rate": 8.522129781858306e-06,
      "loss": 0.5022,
      "step": 1699
    },
    {
      "epoch": 0.27,
      "grad_norm": 20.951808536602996,
      "learning_rate": 8.52027697478422e-06,
      "loss": 0.3644,
      "step": 1700
    },
    {
      "epoch": 0.27,
      "grad_norm": 26.158019756074676,
      "learning_rate": 8.518423208689647e-06,
      "loss": 0.4882,
      "step": 1701
    },
    {
      "epoch": 0.27,
      "grad_norm": 24.297157246455917,
      "learning_rate": 8.516568484079609e-06,
      "loss": 0.493,
      "step": 1702
    },
    {
      "epoch": 0.27,
      "grad_norm": 31.579394701798257,
      "learning_rate": 8.514712801459379e-06,
      "loss": 0.5093,
      "step": 1703
    },
    {
      "epoch": 0.27,
      "grad_norm": 25.97274536984953,
      "learning_rate": 8.5128561613345e-06,
      "loss": 0.4537,
      "step": 1704
    },
    {
      "epoch": 0.27,
      "grad_norm": 66.11590215744103,
      "learning_rate": 8.510998564210769e-06,
      "loss": 0.4128,
      "step": 1705
    },
    {
      "epoch": 0.27,
      "grad_norm": 29.710324056001934,
      "learning_rate": 8.509140010594248e-06,
      "loss": 0.4696,
      "step": 1706
    },
    {
      "epoch": 0.28,
      "grad_norm": 26.61163396301926,
      "learning_rate": 8.50728050099126e-06,
      "loss": 0.4059,
      "step": 1707
    },
    {
      "epoch": 0.28,
      "grad_norm": 15.268046430898593,
      "learning_rate": 8.505420035908383e-06,
      "loss": 0.4423,
      "step": 1708
    },
    {
      "epoch": 0.28,
      "grad_norm": 17.46164155677649,
      "learning_rate": 8.503558615852461e-06,
      "loss": 0.5137,
      "step": 1709
    },
    {
      "epoch": 0.28,
      "grad_norm": 19.987528095451438,
      "learning_rate": 8.501696241330594e-06,
      "loss": 0.4872,
      "step": 1710
    },
    {
      "epoch": 0.28,
      "grad_norm": 17.94169289381315,
      "learning_rate": 8.49983291285015e-06,
      "loss": 0.477,
      "step": 1711
    },
    {
      "epoch": 0.28,
      "grad_norm": 12.380447024361388,
      "learning_rate": 8.497968630918743e-06,
      "loss": 0.4179,
      "step": 1712
    },
    {
      "epoch": 0.28,
      "grad_norm": 18.153599009873076,
      "learning_rate": 8.496103396044262e-06,
      "loss": 0.4607,
      "step": 1713
    },
    {
      "epoch": 0.28,
      "grad_norm": 10.624039119459969,
      "learning_rate": 8.494237208734843e-06,
      "loss": 0.4405,
      "step": 1714
    },
    {
      "epoch": 0.28,
      "grad_norm": 13.64180077056077,
      "learning_rate": 8.492370069498892e-06,
      "loss": 0.4563,
      "step": 1715
    },
    {
      "epoch": 0.28,
      "grad_norm": 11.591772801802806,
      "learning_rate": 8.490501978845064e-06,
      "loss": 0.4074,
      "step": 1716
    },
    {
      "epoch": 0.28,
      "grad_norm": 13.188904265649606,
      "learning_rate": 8.488632937282281e-06,
      "loss": 0.428,
      "step": 1717
    },
    {
      "epoch": 0.28,
      "grad_norm": 14.284839185824236,
      "learning_rate": 8.486762945319722e-06,
      "loss": 0.4865,
      "step": 1718
    },
    {
      "epoch": 0.28,
      "grad_norm": 12.929936700489185,
      "learning_rate": 8.484892003466823e-06,
      "loss": 0.4191,
      "step": 1719
    },
    {
      "epoch": 0.28,
      "grad_norm": 10.59284587057886,
      "learning_rate": 8.48302011223328e-06,
      "loss": 0.3476,
      "step": 1720
    },
    {
      "epoch": 0.28,
      "grad_norm": 11.791662393615871,
      "learning_rate": 8.48114727212905e-06,
      "loss": 0.4482,
      "step": 1721
    },
    {
      "epoch": 0.28,
      "grad_norm": 15.446081038534517,
      "learning_rate": 8.479273483664344e-06,
      "loss": 0.4697,
      "step": 1722
    },
    {
      "epoch": 0.28,
      "grad_norm": 12.360144621178796,
      "learning_rate": 8.477398747349632e-06,
      "loss": 0.4608,
      "step": 1723
    },
    {
      "epoch": 0.28,
      "grad_norm": 10.344554653061945,
      "learning_rate": 8.47552306369565e-06,
      "loss": 0.4937,
      "step": 1724
    },
    {
      "epoch": 0.28,
      "grad_norm": 13.93541684719766,
      "learning_rate": 8.473646433213378e-06,
      "loss": 0.4346,
      "step": 1725
    },
    {
      "epoch": 0.28,
      "grad_norm": 14.310477905520795,
      "learning_rate": 8.471768856414069e-06,
      "loss": 0.4748,
      "step": 1726
    },
    {
      "epoch": 0.28,
      "grad_norm": 17.036615066349903,
      "learning_rate": 8.469890333809223e-06,
      "loss": 0.4512,
      "step": 1727
    },
    {
      "epoch": 0.28,
      "grad_norm": 20.868647610127773,
      "learning_rate": 8.468010865910601e-06,
      "loss": 0.4477,
      "step": 1728
    },
    {
      "epoch": 0.28,
      "grad_norm": 15.253249853719144,
      "learning_rate": 8.466130453230224e-06,
      "loss": 0.4814,
      "step": 1729
    },
    {
      "epoch": 0.28,
      "grad_norm": 14.94229099565896,
      "learning_rate": 8.464249096280368e-06,
      "loss": 0.4414,
      "step": 1730
    },
    {
      "epoch": 0.28,
      "grad_norm": 14.263295107453342,
      "learning_rate": 8.462366795573564e-06,
      "loss": 0.4329,
      "step": 1731
    },
    {
      "epoch": 0.28,
      "grad_norm": 10.049319993606705,
      "learning_rate": 8.460483551622606e-06,
      "loss": 0.4954,
      "step": 1732
    },
    {
      "epoch": 0.28,
      "grad_norm": 22.40481352155851,
      "learning_rate": 8.458599364940537e-06,
      "loss": 0.498,
      "step": 1733
    },
    {
      "epoch": 0.28,
      "grad_norm": 18.061119651418057,
      "learning_rate": 8.456714236040664e-06,
      "loss": 0.4554,
      "step": 1734
    },
    {
      "epoch": 0.28,
      "grad_norm": 11.911538626267598,
      "learning_rate": 8.45482816543655e-06,
      "loss": 0.5636,
      "step": 1735
    },
    {
      "epoch": 0.28,
      "grad_norm": 10.988129127948227,
      "learning_rate": 8.45294115364201e-06,
      "loss": 0.4689,
      "step": 1736
    },
    {
      "epoch": 0.28,
      "grad_norm": 73.01351768083636,
      "learning_rate": 8.45105320117112e-06,
      "loss": 0.5221,
      "step": 1737
    },
    {
      "epoch": 0.28,
      "grad_norm": 21.848488524356906,
      "learning_rate": 8.449164308538209e-06,
      "loss": 0.5006,
      "step": 1738
    },
    {
      "epoch": 0.28,
      "grad_norm": 13.244241721761913,
      "learning_rate": 8.447274476257863e-06,
      "loss": 0.5025,
      "step": 1739
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3658830912486526,
      "learning_rate": 8.445383704844925e-06,
      "loss": 0.4787,
      "step": 1740
    },
    {
      "epoch": 0.28,
      "grad_norm": 14.451635095346411,
      "learning_rate": 8.443491994814493e-06,
      "loss": 0.5581,
      "step": 1741
    },
    {
      "epoch": 0.28,
      "grad_norm": 13.183600117805668,
      "learning_rate": 8.441599346681921e-06,
      "loss": 0.5061,
      "step": 1742
    },
    {
      "epoch": 0.28,
      "grad_norm": 13.941269381075315,
      "learning_rate": 8.43970576096282e-06,
      "loss": 0.399,
      "step": 1743
    },
    {
      "epoch": 0.28,
      "grad_norm": 16.947461660431472,
      "learning_rate": 8.437811238173053e-06,
      "loss": 0.5306,
      "step": 1744
    },
    {
      "epoch": 0.28,
      "grad_norm": 15.532524262980653,
      "learning_rate": 8.43591577882874e-06,
      "loss": 0.3515,
      "step": 1745
    },
    {
      "epoch": 0.28,
      "grad_norm": 20.360542633207107,
      "learning_rate": 8.43401938344626e-06,
      "loss": 0.426,
      "step": 1746
    },
    {
      "epoch": 0.28,
      "grad_norm": 28.687001438847798,
      "learning_rate": 8.432122052542238e-06,
      "loss": 0.3623,
      "step": 1747
    },
    {
      "epoch": 0.28,
      "grad_norm": 18.50405036340982,
      "learning_rate": 8.430223786633563e-06,
      "loss": 0.4165,
      "step": 1748
    },
    {
      "epoch": 0.28,
      "grad_norm": 16.497973916622897,
      "learning_rate": 8.428324586237374e-06,
      "loss": 0.4901,
      "step": 1749
    },
    {
      "epoch": 0.28,
      "grad_norm": 27.731133264989,
      "learning_rate": 8.426424451871063e-06,
      "loss": 0.5145,
      "step": 1750
    },
    {
      "epoch": 0.28,
      "grad_norm": 17.873825700678175,
      "learning_rate": 8.424523384052284e-06,
      "loss": 0.4882,
      "step": 1751
    },
    {
      "epoch": 0.28,
      "grad_norm": 16.363451631667527,
      "learning_rate": 8.422621383298936e-06,
      "loss": 0.488,
      "step": 1752
    },
    {
      "epoch": 0.28,
      "grad_norm": 14.406654371583244,
      "learning_rate": 8.42071845012918e-06,
      "loss": 0.4437,
      "step": 1753
    },
    {
      "epoch": 0.28,
      "grad_norm": 17.08646492843234,
      "learning_rate": 8.418814585061423e-06,
      "loss": 0.3818,
      "step": 1754
    },
    {
      "epoch": 0.28,
      "grad_norm": 21.484578175918124,
      "learning_rate": 8.416909788614335e-06,
      "loss": 0.4347,
      "step": 1755
    },
    {
      "epoch": 0.28,
      "grad_norm": 24.31309885712686,
      "learning_rate": 8.415004061306833e-06,
      "loss": 0.4899,
      "step": 1756
    },
    {
      "epoch": 0.28,
      "grad_norm": 21.308017888189397,
      "learning_rate": 8.413097403658089e-06,
      "loss": 0.4511,
      "step": 1757
    },
    {
      "epoch": 0.28,
      "grad_norm": 16.101152504316246,
      "learning_rate": 8.411189816187528e-06,
      "loss": 0.4726,
      "step": 1758
    },
    {
      "epoch": 0.28,
      "grad_norm": 12.517548247178718,
      "learning_rate": 8.409281299414833e-06,
      "loss": 0.3902,
      "step": 1759
    },
    {
      "epoch": 0.28,
      "grad_norm": 11.62411927986583,
      "learning_rate": 8.407371853859935e-06,
      "loss": 0.4779,
      "step": 1760
    },
    {
      "epoch": 0.28,
      "grad_norm": 11.733966305725422,
      "learning_rate": 8.405461480043019e-06,
      "loss": 0.4881,
      "step": 1761
    },
    {
      "epoch": 0.28,
      "grad_norm": 25.606392439355545,
      "learning_rate": 8.403550178484521e-06,
      "loss": 0.4397,
      "step": 1762
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2400329454028334,
      "learning_rate": 8.401637949705138e-06,
      "loss": 0.5138,
      "step": 1763
    },
    {
      "epoch": 0.28,
      "grad_norm": 10.53117447597358,
      "learning_rate": 8.399724794225809e-06,
      "loss": 0.4055,
      "step": 1764
    },
    {
      "epoch": 0.28,
      "grad_norm": 12.439454360388517,
      "learning_rate": 8.397810712567732e-06,
      "loss": 0.4623,
      "step": 1765
    },
    {
      "epoch": 0.28,
      "grad_norm": 14.644494712122718,
      "learning_rate": 8.39589570525236e-06,
      "loss": 0.4962,
      "step": 1766
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.883141468823347,
      "learning_rate": 8.393979772801386e-06,
      "loss": 0.491,
      "step": 1767
    },
    {
      "epoch": 0.28,
      "grad_norm": 12.825058122805691,
      "learning_rate": 8.392062915736765e-06,
      "loss": 0.4009,
      "step": 1768
    },
    {
      "epoch": 0.29,
      "grad_norm": 28.902302990397818,
      "learning_rate": 8.390145134580705e-06,
      "loss": 0.4433,
      "step": 1769
    },
    {
      "epoch": 0.29,
      "grad_norm": 17.084891083493,
      "learning_rate": 8.38822642985566e-06,
      "loss": 0.443,
      "step": 1770
    },
    {
      "epoch": 0.29,
      "grad_norm": 11.494499528812158,
      "learning_rate": 8.386306802084339e-06,
      "loss": 0.462,
      "step": 1771
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.853799513547169,
      "learning_rate": 8.3843862517897e-06,
      "loss": 0.4567,
      "step": 1772
    },
    {
      "epoch": 0.29,
      "grad_norm": 15.655371380456858,
      "learning_rate": 8.382464779494954e-06,
      "loss": 0.3956,
      "step": 1773
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.221529865290183,
      "learning_rate": 8.380542385723566e-06,
      "loss": 0.4512,
      "step": 1774
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.463946808499726,
      "learning_rate": 8.378619070999245e-06,
      "loss": 0.4624,
      "step": 1775
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.43068870456697,
      "learning_rate": 8.37669483584596e-06,
      "loss": 0.4958,
      "step": 1776
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.864957123568374,
      "learning_rate": 8.37476968078792e-06,
      "loss": 0.5371,
      "step": 1777
    },
    {
      "epoch": 0.29,
      "grad_norm": 11.297187493578013,
      "learning_rate": 8.372843606349594e-06,
      "loss": 0.4424,
      "step": 1778
    },
    {
      "epoch": 0.29,
      "grad_norm": 9.646721072413927,
      "learning_rate": 8.370916613055695e-06,
      "loss": 0.468,
      "step": 1779
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.959320936924168,
      "learning_rate": 8.368988701431192e-06,
      "loss": 0.4723,
      "step": 1780
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.986117639114127,
      "learning_rate": 8.3670598720013e-06,
      "loss": 0.4734,
      "step": 1781
    },
    {
      "epoch": 0.29,
      "grad_norm": 11.176154087809401,
      "learning_rate": 8.365130125291485e-06,
      "loss": 0.5121,
      "step": 1782
    },
    {
      "epoch": 0.29,
      "grad_norm": 14.449742613048508,
      "learning_rate": 8.363199461827464e-06,
      "loss": 0.5307,
      "step": 1783
    },
    {
      "epoch": 0.29,
      "grad_norm": 13.136152815654695,
      "learning_rate": 8.361267882135203e-06,
      "loss": 0.4419,
      "step": 1784
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.453501766924287,
      "learning_rate": 8.359335386740916e-06,
      "loss": 0.4166,
      "step": 1785
    },
    {
      "epoch": 0.29,
      "grad_norm": 15.46461678716022,
      "learning_rate": 8.357401976171069e-06,
      "loss": 0.4481,
      "step": 1786
    },
    {
      "epoch": 0.29,
      "grad_norm": 17.689025478266142,
      "learning_rate": 8.355467650952375e-06,
      "loss": 0.4319,
      "step": 1787
    },
    {
      "epoch": 0.29,
      "grad_norm": 32.92352334013537,
      "learning_rate": 8.353532411611801e-06,
      "loss": 0.5377,
      "step": 1788
    },
    {
      "epoch": 0.29,
      "grad_norm": 10.656649854795825,
      "learning_rate": 8.351596258676558e-06,
      "loss": 0.4499,
      "step": 1789
    },
    {
      "epoch": 0.29,
      "grad_norm": 14.819779590057792,
      "learning_rate": 8.349659192674104e-06,
      "loss": 0.4946,
      "step": 1790
    },
    {
      "epoch": 0.29,
      "grad_norm": 16.66328151920925,
      "learning_rate": 8.347721214132154e-06,
      "loss": 0.4765,
      "step": 1791
    },
    {
      "epoch": 0.29,
      "grad_norm": 10.049763762552724,
      "learning_rate": 8.345782323578664e-06,
      "loss": 0.4039,
      "step": 1792
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.481263059860153,
      "learning_rate": 8.343842521541844e-06,
      "loss": 0.5103,
      "step": 1793
    },
    {
      "epoch": 0.29,
      "grad_norm": 10.685545764508802,
      "learning_rate": 8.341901808550147e-06,
      "loss": 0.4901,
      "step": 1794
    },
    {
      "epoch": 0.29,
      "grad_norm": 14.482133335822734,
      "learning_rate": 8.339960185132275e-06,
      "loss": 0.5208,
      "step": 1795
    },
    {
      "epoch": 0.29,
      "grad_norm": 14.870071432656541,
      "learning_rate": 8.338017651817183e-06,
      "loss": 0.5241,
      "step": 1796
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2869410834011459,
      "learning_rate": 8.336074209134071e-06,
      "loss": 0.446,
      "step": 1797
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.778060979023511,
      "learning_rate": 8.334129857612383e-06,
      "loss": 0.5232,
      "step": 1798
    },
    {
      "epoch": 0.29,
      "grad_norm": 13.569415755455735,
      "learning_rate": 8.332184597781818e-06,
      "loss": 0.4619,
      "step": 1799
    },
    {
      "epoch": 0.29,
      "grad_norm": 15.430439255771944,
      "learning_rate": 8.330238430172315e-06,
      "loss": 0.434,
      "step": 1800
    },
    {
      "epoch": 0.29,
      "grad_norm": 16.751167726668026,
      "learning_rate": 8.328291355314067e-06,
      "loss": 0.4808,
      "step": 1801
    },
    {
      "epoch": 0.29,
      "grad_norm": 32.192879792975624,
      "learning_rate": 8.326343373737506e-06,
      "loss": 0.4244,
      "step": 1802
    },
    {
      "epoch": 0.29,
      "grad_norm": 13.470468212317533,
      "learning_rate": 8.32439448597332e-06,
      "loss": 0.4402,
      "step": 1803
    },
    {
      "epoch": 0.29,
      "grad_norm": 13.624004760927283,
      "learning_rate": 8.322444692552437e-06,
      "loss": 0.466,
      "step": 1804
    },
    {
      "epoch": 0.29,
      "grad_norm": 9.130367188054828,
      "learning_rate": 8.320493994006039e-06,
      "loss": 0.5162,
      "step": 1805
    },
    {
      "epoch": 0.29,
      "grad_norm": 11.621331870317807,
      "learning_rate": 8.318542390865546e-06,
      "loss": 0.4334,
      "step": 1806
    },
    {
      "epoch": 0.29,
      "grad_norm": 13.091968878269356,
      "learning_rate": 8.316589883662629e-06,
      "loss": 0.4858,
      "step": 1807
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.098732968845795,
      "learning_rate": 8.314636472929206e-06,
      "loss": 0.3972,
      "step": 1808
    },
    {
      "epoch": 0.29,
      "grad_norm": 10.672514691461252,
      "learning_rate": 8.31268215919744e-06,
      "loss": 0.4977,
      "step": 1809
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.796987146402977,
      "learning_rate": 8.310726942999736e-06,
      "loss": 0.5081,
      "step": 1810
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.209953574810269,
      "learning_rate": 8.308770824868757e-06,
      "loss": 0.4743,
      "step": 1811
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.60332827446013,
      "learning_rate": 8.306813805337395e-06,
      "loss": 0.5323,
      "step": 1812
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.5702925008449204,
      "learning_rate": 8.3048558849388e-06,
      "loss": 0.4797,
      "step": 1813
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.547047920633775,
      "learning_rate": 8.302897064206363e-06,
      "loss": 0.4777,
      "step": 1814
    },
    {
      "epoch": 0.29,
      "grad_norm": 29.593462606059088,
      "learning_rate": 8.300937343673722e-06,
      "loss": 0.5231,
      "step": 1815
    },
    {
      "epoch": 0.29,
      "grad_norm": 15.393494232963628,
      "learning_rate": 8.298976723874757e-06,
      "loss": 0.4042,
      "step": 1816
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.934206236025013,
      "learning_rate": 8.297015205343595e-06,
      "loss": 0.396,
      "step": 1817
    },
    {
      "epoch": 0.29,
      "grad_norm": 10.092173847840467,
      "learning_rate": 8.295052788614608e-06,
      "loss": 0.4053,
      "step": 1818
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.395020247515282,
      "learning_rate": 8.293089474222414e-06,
      "loss": 0.4308,
      "step": 1819
    },
    {
      "epoch": 0.29,
      "grad_norm": 15.60235545023061,
      "learning_rate": 8.291125262701874e-06,
      "loss": 0.5248,
      "step": 1820
    },
    {
      "epoch": 0.29,
      "grad_norm": 10.22186120958288,
      "learning_rate": 8.289160154588088e-06,
      "loss": 0.5464,
      "step": 1821
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.876645115159355,
      "learning_rate": 8.287194150416413e-06,
      "loss": 0.4784,
      "step": 1822
    },
    {
      "epoch": 0.29,
      "grad_norm": 10.272816749331893,
      "learning_rate": 8.285227250722439e-06,
      "loss": 0.4482,
      "step": 1823
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.691659178315458,
      "learning_rate": 8.283259456042e-06,
      "loss": 0.4232,
      "step": 1824
    },
    {
      "epoch": 0.29,
      "grad_norm": 9.952105268912957,
      "learning_rate": 8.281290766911187e-06,
      "loss": 0.3881,
      "step": 1825
    },
    {
      "epoch": 0.29,
      "grad_norm": 12.686149505205082,
      "learning_rate": 8.279321183866317e-06,
      "loss": 0.4323,
      "step": 1826
    },
    {
      "epoch": 0.29,
      "grad_norm": 9.753069243138839,
      "learning_rate": 8.277350707443964e-06,
      "loss": 0.4607,
      "step": 1827
    },
    {
      "epoch": 0.29,
      "grad_norm": 9.998994473107198,
      "learning_rate": 8.275379338180936e-06,
      "loss": 0.4328,
      "step": 1828
    },
    {
      "epoch": 0.29,
      "grad_norm": 9.150496834566596,
      "learning_rate": 8.273407076614288e-06,
      "loss": 0.4904,
      "step": 1829
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.63007979752757,
      "learning_rate": 8.271433923281322e-06,
      "loss": 0.5007,
      "step": 1830
    },
    {
      "epoch": 0.3,
      "grad_norm": 9.24104133253299,
      "learning_rate": 8.26945987871958e-06,
      "loss": 0.4586,
      "step": 1831
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.338792035956157,
      "learning_rate": 8.26748494346684e-06,
      "loss": 0.4354,
      "step": 1832
    },
    {
      "epoch": 0.3,
      "grad_norm": 24.69294378934424,
      "learning_rate": 8.265509118061135e-06,
      "loss": 0.4355,
      "step": 1833
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.42627842149161,
      "learning_rate": 8.26353240304073e-06,
      "loss": 0.4569,
      "step": 1834
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.299007469167014,
      "learning_rate": 8.26155479894414e-06,
      "loss": 0.4631,
      "step": 1835
    },
    {
      "epoch": 0.3,
      "grad_norm": 10.704611672321414,
      "learning_rate": 8.259576306310118e-06,
      "loss": 0.4947,
      "step": 1836
    },
    {
      "epoch": 0.3,
      "grad_norm": 11.265032240080957,
      "learning_rate": 8.257596925677657e-06,
      "loss": 0.4217,
      "step": 1837
    },
    {
      "epoch": 0.3,
      "grad_norm": 92.58870948430163,
      "learning_rate": 8.255616657586e-06,
      "loss": 0.4556,
      "step": 1838
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.5883870427664615,
      "learning_rate": 8.253635502574623e-06,
      "loss": 0.4548,
      "step": 1839
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.650623163245917,
      "learning_rate": 8.251653461183249e-06,
      "loss": 0.4217,
      "step": 1840
    },
    {
      "epoch": 0.3,
      "grad_norm": 12.186534988639481,
      "learning_rate": 8.249670533951838e-06,
      "loss": 0.506,
      "step": 1841
    },
    {
      "epoch": 0.3,
      "grad_norm": 12.829229322790615,
      "learning_rate": 8.247686721420596e-06,
      "loss": 0.5017,
      "step": 1842
    },
    {
      "epoch": 0.3,
      "grad_norm": 15.44682078639622,
      "learning_rate": 8.24570202412997e-06,
      "loss": 0.4305,
      "step": 1843
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2939011149731274,
      "learning_rate": 8.243716442620644e-06,
      "loss": 0.5034,
      "step": 1844
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.00205765835229,
      "learning_rate": 8.241729977433543e-06,
      "loss": 0.4363,
      "step": 1845
    },
    {
      "epoch": 0.3,
      "grad_norm": 13.265499855379051,
      "learning_rate": 8.239742629109839e-06,
      "loss": 0.4812,
      "step": 1846
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.657523160013294,
      "learning_rate": 8.237754398190937e-06,
      "loss": 0.4435,
      "step": 1847
    },
    {
      "epoch": 0.3,
      "grad_norm": 11.544997580430103,
      "learning_rate": 8.235765285218491e-06,
      "loss": 0.4829,
      "step": 1848
    },
    {
      "epoch": 0.3,
      "grad_norm": 10.082268117093488,
      "learning_rate": 8.233775290734385e-06,
      "loss": 0.4429,
      "step": 1849
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.050997447869134,
      "learning_rate": 8.23178441528075e-06,
      "loss": 0.4625,
      "step": 1850
    },
    {
      "epoch": 0.3,
      "grad_norm": 9.01415487882834,
      "learning_rate": 8.229792659399957e-06,
      "loss": 0.4355,
      "step": 1851
    },
    {
      "epoch": 0.3,
      "grad_norm": 12.185236088347908,
      "learning_rate": 8.227800023634614e-06,
      "loss": 0.4719,
      "step": 1852
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1899598950708967,
      "learning_rate": 8.225806508527568e-06,
      "loss": 0.5037,
      "step": 1853
    },
    {
      "epoch": 0.3,
      "grad_norm": 11.544543319070447,
      "learning_rate": 8.22381211462191e-06,
      "loss": 0.538,
      "step": 1854
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2869544339427652,
      "learning_rate": 8.221816842460967e-06,
      "loss": 0.4602,
      "step": 1855
    },
    {
      "epoch": 0.3,
      "grad_norm": 18.29256238011236,
      "learning_rate": 8.219820692588306e-06,
      "loss": 0.5011,
      "step": 1856
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.201743014993786,
      "learning_rate": 8.217823665547733e-06,
      "loss": 0.448,
      "step": 1857
    },
    {
      "epoch": 0.3,
      "grad_norm": 9.624443087298454,
      "learning_rate": 8.215825761883295e-06,
      "loss": 0.487,
      "step": 1858
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.31640888031695,
      "learning_rate": 8.213826982139273e-06,
      "loss": 0.4834,
      "step": 1859
    },
    {
      "epoch": 0.3,
      "grad_norm": 15.376094189156644,
      "learning_rate": 8.211827326860193e-06,
      "loss": 0.4967,
      "step": 1860
    },
    {
      "epoch": 0.3,
      "grad_norm": 31.569835376328527,
      "learning_rate": 8.209826796590814e-06,
      "loss": 0.3853,
      "step": 1861
    },
    {
      "epoch": 0.3,
      "grad_norm": 17.178171678263944,
      "learning_rate": 8.207825391876137e-06,
      "loss": 0.5247,
      "step": 1862
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.806153492820137,
      "learning_rate": 8.205823113261397e-06,
      "loss": 0.5024,
      "step": 1863
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.56398242710549,
      "learning_rate": 8.203819961292074e-06,
      "loss": 0.4615,
      "step": 1864
    },
    {
      "epoch": 0.3,
      "grad_norm": 11.222232088655792,
      "learning_rate": 8.20181593651388e-06,
      "loss": 0.4978,
      "step": 1865
    },
    {
      "epoch": 0.3,
      "grad_norm": 12.826614289597524,
      "learning_rate": 8.199811039472764e-06,
      "loss": 0.5161,
      "step": 1866
    },
    {
      "epoch": 0.3,
      "grad_norm": 10.938278612428531,
      "learning_rate": 8.197805270714918e-06,
      "loss": 0.4294,
      "step": 1867
    },
    {
      "epoch": 0.3,
      "grad_norm": 9.912293474116812,
      "learning_rate": 8.19579863078677e-06,
      "loss": 0.3732,
      "step": 1868
    },
    {
      "epoch": 0.3,
      "grad_norm": 10.496731848682673,
      "learning_rate": 8.193791120234982e-06,
      "loss": 0.4378,
      "step": 1869
    },
    {
      "epoch": 0.3,
      "grad_norm": 34.9320832228751,
      "learning_rate": 8.191782739606455e-06,
      "loss": 0.4813,
      "step": 1870
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4325991702648226,
      "learning_rate": 8.189773489448328e-06,
      "loss": 0.4877,
      "step": 1871
    },
    {
      "epoch": 0.3,
      "grad_norm": 13.798553487796895,
      "learning_rate": 8.187763370307975e-06,
      "loss": 0.4346,
      "step": 1872
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3825097704921212,
      "learning_rate": 8.185752382733007e-06,
      "loss": 0.4209,
      "step": 1873
    },
    {
      "epoch": 0.3,
      "grad_norm": 21.821756747862477,
      "learning_rate": 8.183740527271278e-06,
      "loss": 0.4819,
      "step": 1874
    },
    {
      "epoch": 0.3,
      "grad_norm": 9.180679109090125,
      "learning_rate": 8.181727804470867e-06,
      "loss": 0.481,
      "step": 1875
    },
    {
      "epoch": 0.3,
      "grad_norm": 19.508129795862565,
      "learning_rate": 8.179714214880096e-06,
      "loss": 0.4727,
      "step": 1876
    },
    {
      "epoch": 0.3,
      "grad_norm": 25.065243455455967,
      "learning_rate": 8.177699759047525e-06,
      "loss": 0.5055,
      "step": 1877
    },
    {
      "epoch": 0.3,
      "grad_norm": 10.185311759250396,
      "learning_rate": 8.175684437521946e-06,
      "loss": 0.4373,
      "step": 1878
    },
    {
      "epoch": 0.3,
      "grad_norm": 22.61750829268189,
      "learning_rate": 8.173668250852389e-06,
      "loss": 0.4964,
      "step": 1879
    },
    {
      "epoch": 0.3,
      "grad_norm": 11.757430876743639,
      "learning_rate": 8.171651199588118e-06,
      "loss": 0.4633,
      "step": 1880
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2860555364682593,
      "learning_rate": 8.16963328427863e-06,
      "loss": 0.4721,
      "step": 1881
    },
    {
      "epoch": 0.3,
      "grad_norm": 32.43435771204616,
      "learning_rate": 8.167614505473667e-06,
      "loss": 0.5167,
      "step": 1882
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.451686489945402,
      "learning_rate": 8.165594863723197e-06,
      "loss": 0.4788,
      "step": 1883
    },
    {
      "epoch": 0.3,
      "grad_norm": 11.28991473592401,
      "learning_rate": 8.163574359577422e-06,
      "loss": 0.5085,
      "step": 1884
    },
    {
      "epoch": 0.3,
      "grad_norm": 13.531241944417877,
      "learning_rate": 8.16155299358679e-06,
      "loss": 0.4545,
      "step": 1885
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.499327205424262,
      "learning_rate": 8.159530766301974e-06,
      "loss": 0.3623,
      "step": 1886
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.023961630403385,
      "learning_rate": 8.157507678273884e-06,
      "loss": 0.4487,
      "step": 1887
    },
    {
      "epoch": 0.3,
      "grad_norm": 13.223444847237513,
      "learning_rate": 8.155483730053664e-06,
      "loss": 0.5283,
      "step": 1888
    },
    {
      "epoch": 0.3,
      "grad_norm": 15.480474154831146,
      "learning_rate": 8.153458922192696e-06,
      "loss": 0.4348,
      "step": 1889
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.396814694578618,
      "learning_rate": 8.15143325524259e-06,
      "loss": 0.4423,
      "step": 1890
    },
    {
      "epoch": 0.3,
      "grad_norm": 13.468279093750061,
      "learning_rate": 8.149406729755198e-06,
      "loss": 0.5127,
      "step": 1891
    },
    {
      "epoch": 0.3,
      "grad_norm": 68.76089004738085,
      "learning_rate": 8.147379346282599e-06,
      "loss": 0.5119,
      "step": 1892
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.262141138936306,
      "learning_rate": 8.145351105377107e-06,
      "loss": 0.4351,
      "step": 1893
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.069347919688143,
      "learning_rate": 8.14332200759127e-06,
      "loss": 0.4801,
      "step": 1894
    },
    {
      "epoch": 0.31,
      "grad_norm": 14.72688403141868,
      "learning_rate": 8.141292053477873e-06,
      "loss": 0.4662,
      "step": 1895
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.901064449722384,
      "learning_rate": 8.139261243589933e-06,
      "loss": 0.499,
      "step": 1896
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.230140264100207,
      "learning_rate": 8.137229578480694e-06,
      "loss": 0.4091,
      "step": 1897
    },
    {
      "epoch": 0.31,
      "grad_norm": 16.255200287087952,
      "learning_rate": 8.135197058703638e-06,
      "loss": 0.4332,
      "step": 1898
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.125695915327382,
      "learning_rate": 8.133163684812484e-06,
      "loss": 0.4844,
      "step": 1899
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.838622263720925,
      "learning_rate": 8.131129457361176e-06,
      "loss": 0.4612,
      "step": 1900
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.1326989493283195,
      "learning_rate": 8.129094376903891e-06,
      "loss": 0.4366,
      "step": 1901
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.168084706186646,
      "learning_rate": 8.127058443995046e-06,
      "loss": 0.4836,
      "step": 1902
    },
    {
      "epoch": 0.31,
      "grad_norm": 10.455059345979217,
      "learning_rate": 8.125021659189281e-06,
      "loss": 0.4367,
      "step": 1903
    },
    {
      "epoch": 0.31,
      "grad_norm": 13.884392601375222,
      "learning_rate": 8.122984023041476e-06,
      "loss": 0.4665,
      "step": 1904
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3049386307029434,
      "learning_rate": 8.120945536106738e-06,
      "loss": 0.4846,
      "step": 1905
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.702064323236561,
      "learning_rate": 8.118906198940403e-06,
      "loss": 0.4946,
      "step": 1906
    },
    {
      "epoch": 0.31,
      "grad_norm": 32.61278915327996,
      "learning_rate": 8.11686601209805e-06,
      "loss": 0.3984,
      "step": 1907
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2554731141994322,
      "learning_rate": 8.114824976135478e-06,
      "loss": 0.5,
      "step": 1908
    },
    {
      "epoch": 0.31,
      "grad_norm": 10.745495582843695,
      "learning_rate": 8.11278309160872e-06,
      "loss": 0.4234,
      "step": 1909
    },
    {
      "epoch": 0.31,
      "grad_norm": 10.66981560973259,
      "learning_rate": 8.110740359074046e-06,
      "loss": 0.5033,
      "step": 1910
    },
    {
      "epoch": 0.31,
      "grad_norm": 10.578922072691991,
      "learning_rate": 8.108696779087949e-06,
      "loss": 0.5003,
      "step": 1911
    },
    {
      "epoch": 0.31,
      "grad_norm": 9.596061895270065,
      "learning_rate": 8.106652352207157e-06,
      "loss": 0.4477,
      "step": 1912
    },
    {
      "epoch": 0.31,
      "grad_norm": 9.589909903234725,
      "learning_rate": 8.10460707898863e-06,
      "loss": 0.4094,
      "step": 1913
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.899410763774922,
      "learning_rate": 8.102560959989554e-06,
      "loss": 0.4883,
      "step": 1914
    },
    {
      "epoch": 0.31,
      "grad_norm": 13.645961532350412,
      "learning_rate": 8.100513995767352e-06,
      "loss": 0.4235,
      "step": 1915
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.938508570814165,
      "learning_rate": 8.098466186879669e-06,
      "loss": 0.5199,
      "step": 1916
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.950029226032058,
      "learning_rate": 8.096417533884388e-06,
      "loss": 0.4538,
      "step": 1917
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.646892522456374,
      "learning_rate": 8.094368037339619e-06,
      "loss": 0.4831,
      "step": 1918
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3334363217075338,
      "learning_rate": 8.092317697803698e-06,
      "loss": 0.4679,
      "step": 1919
    },
    {
      "epoch": 0.31,
      "grad_norm": 11.715286427949287,
      "learning_rate": 8.090266515835197e-06,
      "loss": 0.4546,
      "step": 1920
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.314716226954758,
      "learning_rate": 8.088214491992912e-06,
      "loss": 0.4988,
      "step": 1921
    },
    {
      "epoch": 0.31,
      "grad_norm": 10.860102370839819,
      "learning_rate": 8.086161626835873e-06,
      "loss": 0.4248,
      "step": 1922
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.83879659651721,
      "learning_rate": 8.084107920923336e-06,
      "loss": 0.4431,
      "step": 1923
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.975574715341837,
      "learning_rate": 8.08205337481479e-06,
      "loss": 0.4992,
      "step": 1924
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.29652862993854,
      "learning_rate": 8.079997989069945e-06,
      "loss": 0.535,
      "step": 1925
    },
    {
      "epoch": 0.31,
      "grad_norm": 9.782181581164688,
      "learning_rate": 8.077941764248746e-06,
      "loss": 0.4299,
      "step": 1926
    },
    {
      "epoch": 0.31,
      "grad_norm": 10.712994439521564,
      "learning_rate": 8.075884700911368e-06,
      "loss": 0.436,
      "step": 1927
    },
    {
      "epoch": 0.31,
      "grad_norm": 70.16961276790295,
      "learning_rate": 8.07382679961821e-06,
      "loss": 0.4369,
      "step": 1928
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.9019188456405205,
      "learning_rate": 8.071768060929903e-06,
      "loss": 0.4507,
      "step": 1929
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.738944432790222,
      "learning_rate": 8.0697084854073e-06,
      "loss": 0.428,
      "step": 1930
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.755314278216263,
      "learning_rate": 8.06764807361149e-06,
      "loss": 0.4552,
      "step": 1931
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.552347306803121,
      "learning_rate": 8.065586826103782e-06,
      "loss": 0.4201,
      "step": 1932
    },
    {
      "epoch": 0.31,
      "grad_norm": 12.204452645409837,
      "learning_rate": 8.063524743445721e-06,
      "loss": 0.4231,
      "step": 1933
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.463657697420989,
      "learning_rate": 8.061461826199075e-06,
      "loss": 0.4653,
      "step": 1934
    },
    {
      "epoch": 0.31,
      "grad_norm": 15.841048325652578,
      "learning_rate": 8.059398074925835e-06,
      "loss": 0.4855,
      "step": 1935
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.851660165817965,
      "learning_rate": 8.057333490188228e-06,
      "loss": 0.4298,
      "step": 1936
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.213266969864146,
      "learning_rate": 8.055268072548704e-06,
      "loss": 0.4506,
      "step": 1937
    },
    {
      "epoch": 0.31,
      "grad_norm": 9.568082338514536,
      "learning_rate": 8.053201822569933e-06,
      "loss": 0.4869,
      "step": 1938
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.310918595384004,
      "learning_rate": 8.051134740814827e-06,
      "loss": 0.4371,
      "step": 1939
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.870477648185713,
      "learning_rate": 8.049066827846513e-06,
      "loss": 0.419,
      "step": 1940
    },
    {
      "epoch": 0.31,
      "grad_norm": 12.027804923363858,
      "learning_rate": 8.046998084228347e-06,
      "loss": 0.4988,
      "step": 1941
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.648922814481721,
      "learning_rate": 8.044928510523911e-06,
      "loss": 0.444,
      "step": 1942
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.50971594506674,
      "learning_rate": 8.042858107297015e-06,
      "loss": 0.4588,
      "step": 1943
    },
    {
      "epoch": 0.31,
      "grad_norm": 12.976089748059135,
      "learning_rate": 8.040786875111694e-06,
      "loss": 0.4834,
      "step": 1944
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.845809531793806,
      "learning_rate": 8.03871481453221e-06,
      "loss": 0.4694,
      "step": 1945
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.41022219356413,
      "learning_rate": 8.036641926123043e-06,
      "loss": 0.502,
      "step": 1946
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.180883378926618,
      "learning_rate": 8.034568210448914e-06,
      "loss": 0.4382,
      "step": 1947
    },
    {
      "epoch": 0.31,
      "grad_norm": 10.401262917198043,
      "learning_rate": 8.032493668074756e-06,
      "loss": 0.5562,
      "step": 1948
    },
    {
      "epoch": 0.31,
      "grad_norm": 9.167436909072268,
      "learning_rate": 8.03041829956573e-06,
      "loss": 0.5043,
      "step": 1949
    },
    {
      "epoch": 0.31,
      "grad_norm": 20.834765823767345,
      "learning_rate": 8.028342105487226e-06,
      "loss": 0.5405,
      "step": 1950
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.377730462124356,
      "learning_rate": 8.026265086404856e-06,
      "loss": 0.4395,
      "step": 1951
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.364589313361443,
      "learning_rate": 8.024187242884456e-06,
      "loss": 0.4614,
      "step": 1952
    },
    {
      "epoch": 0.31,
      "grad_norm": 10.056923211944085,
      "learning_rate": 8.02210857549209e-06,
      "loss": 0.427,
      "step": 1953
    },
    {
      "epoch": 0.31,
      "grad_norm": 13.082564912931845,
      "learning_rate": 8.020029084794043e-06,
      "loss": 0.5227,
      "step": 1954
    },
    {
      "epoch": 0.31,
      "grad_norm": 13.815038797792258,
      "learning_rate": 8.017948771356824e-06,
      "loss": 0.4767,
      "step": 1955
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.602087374786478,
      "learning_rate": 8.01586763574717e-06,
      "loss": 0.4167,
      "step": 1956
    },
    {
      "epoch": 0.32,
      "grad_norm": 11.522124543890541,
      "learning_rate": 8.01378567853204e-06,
      "loss": 0.4911,
      "step": 1957
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.672087614491452,
      "learning_rate": 8.011702900278614e-06,
      "loss": 0.4614,
      "step": 1958
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.544293939827445,
      "learning_rate": 8.009619301554303e-06,
      "loss": 0.3818,
      "step": 1959
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.22851174037217,
      "learning_rate": 8.007534882926731e-06,
      "loss": 0.3736,
      "step": 1960
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.551389094659921,
      "learning_rate": 8.005449644963756e-06,
      "loss": 0.4239,
      "step": 1961
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.327251297403608,
      "learning_rate": 8.00336358823345e-06,
      "loss": 0.4564,
      "step": 1962
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.488387428536163,
      "learning_rate": 8.001276713304116e-06,
      "loss": 0.4512,
      "step": 1963
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.095152155382439,
      "learning_rate": 7.999189020744273e-06,
      "loss": 0.4354,
      "step": 1964
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.848915129494917,
      "learning_rate": 7.997100511122669e-06,
      "loss": 0.4663,
      "step": 1965
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.688257044754639,
      "learning_rate": 7.99501118500827e-06,
      "loss": 0.4697,
      "step": 1966
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.9624149026775415,
      "learning_rate": 7.992921042970264e-06,
      "loss": 0.4553,
      "step": 1967
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.079864951095308,
      "learning_rate": 7.990830085578068e-06,
      "loss": 0.501,
      "step": 1968
    },
    {
      "epoch": 0.32,
      "grad_norm": 10.962010976951062,
      "learning_rate": 7.988738313401312e-06,
      "loss": 0.4525,
      "step": 1969
    },
    {
      "epoch": 0.32,
      "grad_norm": 17.514958168458968,
      "learning_rate": 7.986645727009856e-06,
      "loss": 0.4788,
      "step": 1970
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.327362671635237,
      "learning_rate": 7.984552326973776e-06,
      "loss": 0.4974,
      "step": 1971
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.209274732033065,
      "learning_rate": 7.982458113863373e-06,
      "loss": 0.4522,
      "step": 1972
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.846222876316418,
      "learning_rate": 7.980363088249167e-06,
      "loss": 0.4566,
      "step": 1973
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.629175978444325,
      "learning_rate": 7.978267250701904e-06,
      "loss": 0.4105,
      "step": 1974
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.878768115665074,
      "learning_rate": 7.976170601792543e-06,
      "loss": 0.5323,
      "step": 1975
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.855274719217585,
      "learning_rate": 7.97407314209227e-06,
      "loss": 0.4617,
      "step": 1976
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.564497052781214,
      "learning_rate": 7.971974872172497e-06,
      "loss": 0.3646,
      "step": 1977
    },
    {
      "epoch": 0.32,
      "grad_norm": 11.833836147658264,
      "learning_rate": 7.969875792604842e-06,
      "loss": 0.4901,
      "step": 1978
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.56754924312359,
      "learning_rate": 7.967775903961158e-06,
      "loss": 0.4463,
      "step": 1979
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.47823879417704,
      "learning_rate": 7.96567520681351e-06,
      "loss": 0.4937,
      "step": 1980
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.731319187584853,
      "learning_rate": 7.963573701734185e-06,
      "loss": 0.4191,
      "step": 1981
    },
    {
      "epoch": 0.32,
      "grad_norm": 13.508514477577917,
      "learning_rate": 7.961471389295694e-06,
      "loss": 0.4543,
      "step": 1982
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.257723933190785,
      "learning_rate": 7.959368270070763e-06,
      "loss": 0.4433,
      "step": 1983
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.145727241741314,
      "learning_rate": 7.957264344632338e-06,
      "loss": 0.4946,
      "step": 1984
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.868389453337813,
      "learning_rate": 7.95515961355359e-06,
      "loss": 0.5473,
      "step": 1985
    },
    {
      "epoch": 0.32,
      "grad_norm": 13.454982620634105,
      "learning_rate": 7.953054077407903e-06,
      "loss": 0.4962,
      "step": 1986
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.891587913483195,
      "learning_rate": 7.950947736768884e-06,
      "loss": 0.4443,
      "step": 1987
    },
    {
      "epoch": 0.32,
      "grad_norm": 38.44425384819595,
      "learning_rate": 7.948840592210358e-06,
      "loss": 0.5078,
      "step": 1988
    },
    {
      "epoch": 0.32,
      "grad_norm": 12.573910690161226,
      "learning_rate": 7.94673264430637e-06,
      "loss": 0.4534,
      "step": 1989
    },
    {
      "epoch": 0.32,
      "grad_norm": 35.55337855559341,
      "learning_rate": 7.94462389363118e-06,
      "loss": 0.4442,
      "step": 1990
    },
    {
      "epoch": 0.32,
      "grad_norm": 10.219659486498314,
      "learning_rate": 7.942514340759275e-06,
      "loss": 0.5177,
      "step": 1991
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.255376693527574,
      "learning_rate": 7.940403986265353e-06,
      "loss": 0.4671,
      "step": 1992
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.222703127820901,
      "learning_rate": 7.93829283072433e-06,
      "loss": 0.4613,
      "step": 1993
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.09425685687342,
      "learning_rate": 7.936180874711347e-06,
      "loss": 0.3954,
      "step": 1994
    },
    {
      "epoch": 0.32,
      "grad_norm": 54.9231081497017,
      "learning_rate": 7.934068118801758e-06,
      "loss": 0.432,
      "step": 1995
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.428750671611828,
      "learning_rate": 7.931954563571134e-06,
      "loss": 0.4382,
      "step": 1996
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.2183006470045346,
      "learning_rate": 7.92984020959527e-06,
      "loss": 0.3958,
      "step": 1997
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.335164577925307,
      "learning_rate": 7.92772505745017e-06,
      "loss": 0.4372,
      "step": 1998
    },
    {
      "epoch": 0.32,
      "grad_norm": 24.078572772560086,
      "learning_rate": 7.925609107712057e-06,
      "loss": 0.4183,
      "step": 1999
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.2906227673759405,
      "learning_rate": 7.923492360957383e-06,
      "loss": 0.4798,
      "step": 2000
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.142958731667773,
      "learning_rate": 7.921374817762801e-06,
      "loss": 0.5039,
      "step": 2001
    },
    {
      "epoch": 0.32,
      "grad_norm": 19.0528308052335,
      "learning_rate": 7.919256478705192e-06,
      "loss": 0.5041,
      "step": 2002
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.837959980382905,
      "learning_rate": 7.917137344361647e-06,
      "loss": 0.3984,
      "step": 2003
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.670517563732714,
      "learning_rate": 7.91501741530948e-06,
      "loss": 0.5087,
      "step": 2004
    },
    {
      "epoch": 0.32,
      "grad_norm": 10.43309152337332,
      "learning_rate": 7.912896692126216e-06,
      "loss": 0.476,
      "step": 2005
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.4988024680866046,
      "learning_rate": 7.910775175389595e-06,
      "loss": 0.4517,
      "step": 2006
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.35127470440275,
      "learning_rate": 7.908652865677584e-06,
      "loss": 0.472,
      "step": 2007
    },
    {
      "epoch": 0.32,
      "grad_norm": 12.317547405148845,
      "learning_rate": 7.906529763568354e-06,
      "loss": 0.4157,
      "step": 2008
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.015749196698476,
      "learning_rate": 7.904405869640296e-06,
      "loss": 0.4391,
      "step": 2009
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.343033581989047,
      "learning_rate": 7.902281184472021e-06,
      "loss": 0.4766,
      "step": 2010
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.933565716348321,
      "learning_rate": 7.900155708642347e-06,
      "loss": 0.465,
      "step": 2011
    },
    {
      "epoch": 0.32,
      "grad_norm": 11.706468200632765,
      "learning_rate": 7.898029442730316e-06,
      "loss": 0.4902,
      "step": 2012
    },
    {
      "epoch": 0.32,
      "grad_norm": 10.987783285993412,
      "learning_rate": 7.89590238731518e-06,
      "loss": 0.4418,
      "step": 2013
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.162176854245928,
      "learning_rate": 7.893774542976408e-06,
      "loss": 0.4849,
      "step": 2014
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.223371427753035,
      "learning_rate": 7.891645910293683e-06,
      "loss": 0.5018,
      "step": 2015
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.09631578016545,
      "learning_rate": 7.889516489846904e-06,
      "loss": 0.4901,
      "step": 2016
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.316351170743074,
      "learning_rate": 7.887386282216182e-06,
      "loss": 0.4034,
      "step": 2017
    },
    {
      "epoch": 0.33,
      "grad_norm": 16.097680569091715,
      "learning_rate": 7.885255287981845e-06,
      "loss": 0.3753,
      "step": 2018
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.612760669605551,
      "learning_rate": 7.883123507724437e-06,
      "loss": 0.5049,
      "step": 2019
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5947704432095369,
      "learning_rate": 7.88099094202471e-06,
      "loss": 0.5394,
      "step": 2020
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.808357102121382,
      "learning_rate": 7.878857591463636e-06,
      "loss": 0.5089,
      "step": 2021
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.28224412938231,
      "learning_rate": 7.876723456622399e-06,
      "loss": 0.4979,
      "step": 2022
    },
    {
      "epoch": 0.33,
      "grad_norm": 31.551334357942554,
      "learning_rate": 7.874588538082394e-06,
      "loss": 0.455,
      "step": 2023
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.498760988145656,
      "learning_rate": 7.872452836425233e-06,
      "loss": 0.4674,
      "step": 2024
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.094283612058878,
      "learning_rate": 7.870316352232739e-06,
      "loss": 0.4831,
      "step": 2025
    },
    {
      "epoch": 0.33,
      "grad_norm": 16.183512331735344,
      "learning_rate": 7.868179086086951e-06,
      "loss": 0.5141,
      "step": 2026
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.35041804261931,
      "learning_rate": 7.866041038570117e-06,
      "loss": 0.4515,
      "step": 2027
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.301796465197682,
      "learning_rate": 7.863902210264702e-06,
      "loss": 0.4529,
      "step": 2028
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.120195296631943,
      "learning_rate": 7.861762601753379e-06,
      "loss": 0.4221,
      "step": 2029
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.395519338347155,
      "learning_rate": 7.859622213619043e-06,
      "loss": 0.4484,
      "step": 2030
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.729944583169548,
      "learning_rate": 7.857481046444785e-06,
      "loss": 0.4282,
      "step": 2031
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.234628197285998,
      "learning_rate": 7.855339100813925e-06,
      "loss": 0.4497,
      "step": 2032
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.165959591914228,
      "learning_rate": 7.853196377309986e-06,
      "loss": 0.4526,
      "step": 2033
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.66118104021425,
      "learning_rate": 7.851052876516708e-06,
      "loss": 0.3996,
      "step": 2034
    },
    {
      "epoch": 0.33,
      "grad_norm": 11.064428195705318,
      "learning_rate": 7.848908599018033e-06,
      "loss": 0.4387,
      "step": 2035
    },
    {
      "epoch": 0.33,
      "grad_norm": 21.674799254722483,
      "learning_rate": 7.846763545398124e-06,
      "loss": 0.4562,
      "step": 2036
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.665159723126349,
      "learning_rate": 7.844617716241358e-06,
      "loss": 0.4895,
      "step": 2037
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.828176669011881,
      "learning_rate": 7.842471112132311e-06,
      "loss": 0.4237,
      "step": 2038
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.771015524828297,
      "learning_rate": 7.84032373365578e-06,
      "loss": 0.4572,
      "step": 2039
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.470157843075591,
      "learning_rate": 7.83817558139677e-06,
      "loss": 0.4257,
      "step": 2040
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.816088864620952,
      "learning_rate": 7.836026655940497e-06,
      "loss": 0.4593,
      "step": 2041
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.926826724743756,
      "learning_rate": 7.833876957872388e-06,
      "loss": 0.4447,
      "step": 2042
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.733389651568566,
      "learning_rate": 7.83172648777808e-06,
      "loss": 0.3852,
      "step": 2043
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.178703399655718,
      "learning_rate": 7.829575246243417e-06,
      "loss": 0.4546,
      "step": 2044
    },
    {
      "epoch": 0.33,
      "grad_norm": 12.673246330340408,
      "learning_rate": 7.82742323385446e-06,
      "loss": 0.3604,
      "step": 2045
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.202751873354109,
      "learning_rate": 7.825270451197477e-06,
      "loss": 0.5041,
      "step": 2046
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.536499853476276,
      "learning_rate": 7.823116898858945e-06,
      "loss": 0.4271,
      "step": 2047
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.7787794304338105,
      "learning_rate": 7.820962577425548e-06,
      "loss": 0.3711,
      "step": 2048
    },
    {
      "epoch": 0.33,
      "grad_norm": 10.228048575577825,
      "learning_rate": 7.818807487484186e-06,
      "loss": 0.5705,
      "step": 2049
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.546504991684403,
      "learning_rate": 7.816651629621963e-06,
      "loss": 0.4827,
      "step": 2050
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.813017879852273,
      "learning_rate": 7.814495004426195e-06,
      "loss": 0.4707,
      "step": 2051
    },
    {
      "epoch": 0.33,
      "grad_norm": 15.28523074975952,
      "learning_rate": 7.812337612484404e-06,
      "loss": 0.4166,
      "step": 2052
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.386043289355734,
      "learning_rate": 7.810179454384326e-06,
      "loss": 0.4567,
      "step": 2053
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.4206538396143165,
      "learning_rate": 7.808020530713902e-06,
      "loss": 0.4662,
      "step": 2054
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.491156846605655,
      "learning_rate": 7.805860842061282e-06,
      "loss": 0.4322,
      "step": 2055
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.654295994570372,
      "learning_rate": 7.803700389014824e-06,
      "loss": 0.4602,
      "step": 2056
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.005011798801162,
      "learning_rate": 7.801539172163097e-06,
      "loss": 0.5186,
      "step": 2057
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.305515339484451,
      "learning_rate": 7.799377192094872e-06,
      "loss": 0.4235,
      "step": 2058
    },
    {
      "epoch": 0.33,
      "grad_norm": 12.36815159657513,
      "learning_rate": 7.797214449399136e-06,
      "loss": 0.5129,
      "step": 2059
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.659070385983556,
      "learning_rate": 7.79505094466508e-06,
      "loss": 0.4414,
      "step": 2060
    },
    {
      "epoch": 0.33,
      "grad_norm": 11.807310655809136,
      "learning_rate": 7.792886678482096e-06,
      "loss": 0.4407,
      "step": 2061
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.892352247191939,
      "learning_rate": 7.7907216514398e-06,
      "loss": 0.4644,
      "step": 2062
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.633186861359742,
      "learning_rate": 7.788555864127995e-06,
      "loss": 0.4407,
      "step": 2063
    },
    {
      "epoch": 0.33,
      "grad_norm": 14.73566911462073,
      "learning_rate": 7.786389317136708e-06,
      "loss": 0.4191,
      "step": 2064
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.0006853378797445,
      "learning_rate": 7.784222011056164e-06,
      "loss": 0.4985,
      "step": 2065
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.547553933894957,
      "learning_rate": 7.782053946476795e-06,
      "loss": 0.4996,
      "step": 2066
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.568642721501863,
      "learning_rate": 7.779885123989244e-06,
      "loss": 0.4591,
      "step": 2067
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.148444163935931,
      "learning_rate": 7.777715544184358e-06,
      "loss": 0.4206,
      "step": 2068
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.522686804883667,
      "learning_rate": 7.775545207653188e-06,
      "loss": 0.4912,
      "step": 2069
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.12409953213317,
      "learning_rate": 7.773374114986996e-06,
      "loss": 0.4576,
      "step": 2070
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.5868985531589,
      "learning_rate": 7.771202266777247e-06,
      "loss": 0.4994,
      "step": 2071
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.250759787321442,
      "learning_rate": 7.76902966361561e-06,
      "loss": 0.4187,
      "step": 2072
    },
    {
      "epoch": 0.33,
      "grad_norm": 11.71473578028989,
      "learning_rate": 7.766856306093966e-06,
      "loss": 0.5041,
      "step": 2073
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.351637843124184,
      "learning_rate": 7.764682194804394e-06,
      "loss": 0.4178,
      "step": 2074
    },
    {
      "epoch": 0.33,
      "grad_norm": 13.798845304222317,
      "learning_rate": 7.762507330339185e-06,
      "loss": 0.4457,
      "step": 2075
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.495819809272955,
      "learning_rate": 7.76033171329083e-06,
      "loss": 0.4479,
      "step": 2076
    },
    {
      "epoch": 0.33,
      "grad_norm": 12.345739655749375,
      "learning_rate": 7.758155344252025e-06,
      "loss": 0.4024,
      "step": 2077
    },
    {
      "epoch": 0.33,
      "grad_norm": 16.992975632417618,
      "learning_rate": 7.755978223815678e-06,
      "loss": 0.4583,
      "step": 2078
    },
    {
      "epoch": 0.33,
      "grad_norm": 22.440057369445576,
      "learning_rate": 7.753800352574891e-06,
      "loss": 0.4172,
      "step": 2079
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.775185137519819,
      "learning_rate": 7.751621731122981e-06,
      "loss": 0.4132,
      "step": 2080
    },
    {
      "epoch": 0.34,
      "grad_norm": 10.170048435338325,
      "learning_rate": 7.749442360053462e-06,
      "loss": 0.4111,
      "step": 2081
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.793958641544817,
      "learning_rate": 7.747262239960055e-06,
      "loss": 0.5114,
      "step": 2082
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.693722456524318,
      "learning_rate": 7.745081371436686e-06,
      "loss": 0.4481,
      "step": 2083
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.121932971643975,
      "learning_rate": 7.742899755077482e-06,
      "loss": 0.4307,
      "step": 2084
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.818245063688625,
      "learning_rate": 7.740717391476778e-06,
      "loss": 0.3956,
      "step": 2085
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1355429817902725,
      "learning_rate": 7.738534281229106e-06,
      "loss": 0.4366,
      "step": 2086
    },
    {
      "epoch": 0.34,
      "grad_norm": 39.04255247917219,
      "learning_rate": 7.736350424929209e-06,
      "loss": 0.4057,
      "step": 2087
    },
    {
      "epoch": 0.34,
      "grad_norm": 17.076772033874832,
      "learning_rate": 7.734165823172028e-06,
      "loss": 0.5252,
      "step": 2088
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.699921254458682,
      "learning_rate": 7.731980476552708e-06,
      "loss": 0.4246,
      "step": 2089
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.962129819741002,
      "learning_rate": 7.7297943856666e-06,
      "loss": 0.4588,
      "step": 2090
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1635822861437144,
      "learning_rate": 7.727607551109252e-06,
      "loss": 0.4531,
      "step": 2091
    },
    {
      "epoch": 0.34,
      "grad_norm": 12.569732998456995,
      "learning_rate": 7.725419973476422e-06,
      "loss": 0.5591,
      "step": 2092
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.502310798658302,
      "learning_rate": 7.723231653364065e-06,
      "loss": 0.4124,
      "step": 2093
    },
    {
      "epoch": 0.34,
      "grad_norm": 9.27403344407039,
      "learning_rate": 7.72104259136834e-06,
      "loss": 0.462,
      "step": 2094
    },
    {
      "epoch": 0.34,
      "grad_norm": 19.366640599819426,
      "learning_rate": 7.718852788085604e-06,
      "loss": 0.5064,
      "step": 2095
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.891207342638202,
      "learning_rate": 7.716662244112425e-06,
      "loss": 0.4654,
      "step": 2096
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.626515686155651,
      "learning_rate": 7.714470960045565e-06,
      "loss": 0.4983,
      "step": 2097
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2476040495688359,
      "learning_rate": 7.712278936481992e-06,
      "loss": 0.4259,
      "step": 2098
    },
    {
      "epoch": 0.34,
      "grad_norm": 10.962365975774091,
      "learning_rate": 7.710086174018871e-06,
      "loss": 0.4554,
      "step": 2099
    },
    {
      "epoch": 0.34,
      "grad_norm": 9.099182736273677,
      "learning_rate": 7.707892673253572e-06,
      "loss": 0.4748,
      "step": 2100
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.8360204500473705,
      "learning_rate": 7.705698434783666e-06,
      "loss": 0.4618,
      "step": 2101
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.4648276335413675,
      "learning_rate": 7.703503459206922e-06,
      "loss": 0.5301,
      "step": 2102
    },
    {
      "epoch": 0.34,
      "grad_norm": 21.087349952832422,
      "learning_rate": 7.701307747121314e-06,
      "loss": 0.4925,
      "step": 2103
    },
    {
      "epoch": 0.34,
      "grad_norm": 12.295310339288026,
      "learning_rate": 7.699111299125015e-06,
      "loss": 0.4571,
      "step": 2104
    },
    {
      "epoch": 0.34,
      "grad_norm": 40.4547538075996,
      "learning_rate": 7.696914115816395e-06,
      "loss": 0.5021,
      "step": 2105
    },
    {
      "epoch": 0.34,
      "grad_norm": 11.517189106915135,
      "learning_rate": 7.69471619779403e-06,
      "loss": 0.4373,
      "step": 2106
    },
    {
      "epoch": 0.34,
      "grad_norm": 24.37179142332863,
      "learning_rate": 7.692517545656691e-06,
      "loss": 0.4854,
      "step": 2107
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.491746271381572,
      "learning_rate": 7.690318160003356e-06,
      "loss": 0.4795,
      "step": 2108
    },
    {
      "epoch": 0.34,
      "grad_norm": 15.018576914890293,
      "learning_rate": 7.688118041433192e-06,
      "loss": 0.4134,
      "step": 2109
    },
    {
      "epoch": 0.34,
      "grad_norm": 12.373447919772515,
      "learning_rate": 7.685917190545576e-06,
      "loss": 0.3929,
      "step": 2110
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.635106381013662,
      "learning_rate": 7.683715607940078e-06,
      "loss": 0.4987,
      "step": 2111
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.420356528520459,
      "learning_rate": 7.681513294216476e-06,
      "loss": 0.4002,
      "step": 2112
    },
    {
      "epoch": 0.34,
      "grad_norm": 12.314805447930347,
      "learning_rate": 7.679310249974732e-06,
      "loss": 0.4496,
      "step": 2113
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.152977500534956,
      "learning_rate": 7.677106475815021e-06,
      "loss": 0.4743,
      "step": 2114
    },
    {
      "epoch": 0.34,
      "grad_norm": 10.265360543514397,
      "learning_rate": 7.674901972337712e-06,
      "loss": 0.479,
      "step": 2115
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.892626919665304,
      "learning_rate": 7.672696740143372e-06,
      "loss": 0.486,
      "step": 2116
    },
    {
      "epoch": 0.34,
      "grad_norm": 16.447314171336238,
      "learning_rate": 7.670490779832767e-06,
      "loss": 0.4447,
      "step": 2117
    },
    {
      "epoch": 0.34,
      "grad_norm": 9.010660934520317,
      "learning_rate": 7.668284092006859e-06,
      "loss": 0.4467,
      "step": 2118
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.582599163752811,
      "learning_rate": 7.666076677266813e-06,
      "loss": 0.4534,
      "step": 2119
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.439480988458671,
      "learning_rate": 7.66386853621399e-06,
      "loss": 0.468,
      "step": 2120
    },
    {
      "epoch": 0.34,
      "grad_norm": 9.709460865730035,
      "learning_rate": 7.661659669449948e-06,
      "loss": 0.5387,
      "step": 2121
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.938826734532013,
      "learning_rate": 7.659450077576444e-06,
      "loss": 0.4394,
      "step": 2122
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.444920566375533,
      "learning_rate": 7.657239761195428e-06,
      "loss": 0.4769,
      "step": 2123
    },
    {
      "epoch": 0.34,
      "grad_norm": 18.326600827972936,
      "learning_rate": 7.655028720909057e-06,
      "loss": 0.4388,
      "step": 2124
    },
    {
      "epoch": 0.34,
      "grad_norm": 13.71086697772542,
      "learning_rate": 7.652816957319674e-06,
      "loss": 0.4927,
      "step": 2125
    },
    {
      "epoch": 0.34,
      "grad_norm": 10.03794445672883,
      "learning_rate": 7.650604471029825e-06,
      "loss": 0.4382,
      "step": 2126
    },
    {
      "epoch": 0.34,
      "grad_norm": 9.903624924246099,
      "learning_rate": 7.648391262642257e-06,
      "loss": 0.4745,
      "step": 2127
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.909242764401975,
      "learning_rate": 7.646177332759906e-06,
      "loss": 0.4062,
      "step": 2128
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.002591919166264,
      "learning_rate": 7.643962681985904e-06,
      "loss": 0.4751,
      "step": 2129
    },
    {
      "epoch": 0.34,
      "grad_norm": 38.99043966300326,
      "learning_rate": 7.641747310923588e-06,
      "loss": 0.4337,
      "step": 2130
    },
    {
      "epoch": 0.34,
      "grad_norm": 53.35011615674261,
      "learning_rate": 7.639531220176484e-06,
      "loss": 0.484,
      "step": 2131
    },
    {
      "epoch": 0.34,
      "grad_norm": 12.135960179233045,
      "learning_rate": 7.637314410348315e-06,
      "loss": 0.4266,
      "step": 2132
    },
    {
      "epoch": 0.34,
      "grad_norm": 18.839165241310706,
      "learning_rate": 7.635096882043006e-06,
      "loss": 0.482,
      "step": 2133
    },
    {
      "epoch": 0.34,
      "grad_norm": 11.104523626688536,
      "learning_rate": 7.632878635864666e-06,
      "loss": 0.4261,
      "step": 2134
    },
    {
      "epoch": 0.34,
      "grad_norm": 16.020175023305185,
      "learning_rate": 7.630659672417613e-06,
      "loss": 0.4656,
      "step": 2135
    },
    {
      "epoch": 0.34,
      "grad_norm": 10.038651041019715,
      "learning_rate": 7.628439992306349e-06,
      "loss": 0.4451,
      "step": 2136
    },
    {
      "epoch": 0.34,
      "grad_norm": 14.149303386908928,
      "learning_rate": 7.626219596135578e-06,
      "loss": 0.3957,
      "step": 2137
    },
    {
      "epoch": 0.34,
      "grad_norm": 13.526555384856678,
      "learning_rate": 7.623998484510197e-06,
      "loss": 0.4776,
      "step": 2138
    },
    {
      "epoch": 0.34,
      "grad_norm": 9.55388190204747,
      "learning_rate": 7.621776658035298e-06,
      "loss": 0.4811,
      "step": 2139
    },
    {
      "epoch": 0.34,
      "grad_norm": 9.65707369913588,
      "learning_rate": 7.619554117316165e-06,
      "loss": 0.5148,
      "step": 2140
    },
    {
      "epoch": 0.34,
      "grad_norm": 13.143875365853082,
      "learning_rate": 7.617330862958287e-06,
      "loss": 0.4574,
      "step": 2141
    },
    {
      "epoch": 0.35,
      "grad_norm": 103.27202474706992,
      "learning_rate": 7.615106895567331e-06,
      "loss": 0.4774,
      "step": 2142
    },
    {
      "epoch": 0.35,
      "grad_norm": 32.31772357178092,
      "learning_rate": 7.612882215749172e-06,
      "loss": 0.444,
      "step": 2143
    },
    {
      "epoch": 0.35,
      "grad_norm": 33.44351750884095,
      "learning_rate": 7.6106568241098745e-06,
      "loss": 0.4923,
      "step": 2144
    },
    {
      "epoch": 0.35,
      "grad_norm": 22.014947822189317,
      "learning_rate": 7.608430721255691e-06,
      "loss": 0.435,
      "step": 2145
    },
    {
      "epoch": 0.35,
      "grad_norm": 12.548360259447163,
      "learning_rate": 7.606203907793081e-06,
      "loss": 0.4872,
      "step": 2146
    },
    {
      "epoch": 0.35,
      "grad_norm": 16.240883693894045,
      "learning_rate": 7.603976384328684e-06,
      "loss": 0.4828,
      "step": 2147
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.36210250058226,
      "learning_rate": 7.601748151469341e-06,
      "loss": 0.4951,
      "step": 2148
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.30723751481526,
      "learning_rate": 7.599519209822085e-06,
      "loss": 0.4352,
      "step": 2149
    },
    {
      "epoch": 0.35,
      "grad_norm": 12.951984227901336,
      "learning_rate": 7.59728955999414e-06,
      "loss": 0.4375,
      "step": 2150
    },
    {
      "epoch": 0.35,
      "grad_norm": 27.459722761193934,
      "learning_rate": 7.595059202592923e-06,
      "loss": 0.4875,
      "step": 2151
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.207672555675573,
      "learning_rate": 7.5928281382260474e-06,
      "loss": 0.4358,
      "step": 2152
    },
    {
      "epoch": 0.35,
      "grad_norm": 22.450925334354615,
      "learning_rate": 7.590596367501314e-06,
      "loss": 0.4765,
      "step": 2153
    },
    {
      "epoch": 0.35,
      "grad_norm": 13.963648249263192,
      "learning_rate": 7.58836389102672e-06,
      "loss": 0.4238,
      "step": 2154
    },
    {
      "epoch": 0.35,
      "grad_norm": 10.411454674732726,
      "learning_rate": 7.586130709410454e-06,
      "loss": 0.5141,
      "step": 2155
    },
    {
      "epoch": 0.35,
      "grad_norm": 13.266203565270903,
      "learning_rate": 7.583896823260894e-06,
      "loss": 0.3885,
      "step": 2156
    },
    {
      "epoch": 0.35,
      "grad_norm": 12.371355393039966,
      "learning_rate": 7.581662233186618e-06,
      "loss": 0.4413,
      "step": 2157
    },
    {
      "epoch": 0.35,
      "grad_norm": 16.207917705661327,
      "learning_rate": 7.5794269397963814e-06,
      "loss": 0.4993,
      "step": 2158
    },
    {
      "epoch": 0.35,
      "grad_norm": 17.414883714143716,
      "learning_rate": 7.577190943699145e-06,
      "loss": 0.4873,
      "step": 2159
    },
    {
      "epoch": 0.35,
      "grad_norm": 14.78337604780799,
      "learning_rate": 7.574954245504056e-06,
      "loss": 0.4246,
      "step": 2160
    },
    {
      "epoch": 0.35,
      "grad_norm": 19.97413900319821,
      "learning_rate": 7.572716845820452e-06,
      "loss": 0.4673,
      "step": 2161
    },
    {
      "epoch": 0.35,
      "grad_norm": 16.428325361306584,
      "learning_rate": 7.57047874525786e-06,
      "loss": 0.4394,
      "step": 2162
    },
    {
      "epoch": 0.35,
      "grad_norm": 13.397722783432853,
      "learning_rate": 7.568239944426003e-06,
      "loss": 0.4537,
      "step": 2163
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.452390771139909,
      "learning_rate": 7.5660004439347916e-06,
      "loss": 0.433,
      "step": 2164
    },
    {
      "epoch": 0.35,
      "grad_norm": 19.412005650509236,
      "learning_rate": 7.563760244394325e-06,
      "loss": 0.4888,
      "step": 2165
    },
    {
      "epoch": 0.35,
      "grad_norm": 13.566416846853913,
      "learning_rate": 7.5615193464149005e-06,
      "loss": 0.4329,
      "step": 2166
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.602081209949365,
      "learning_rate": 7.5592777506069946e-06,
      "loss": 0.4011,
      "step": 2167
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.390458987787747,
      "learning_rate": 7.557035457581284e-06,
      "loss": 0.5084,
      "step": 2168
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.851761248183015,
      "learning_rate": 7.5547924679486294e-06,
      "loss": 0.4371,
      "step": 2169
    },
    {
      "epoch": 0.35,
      "grad_norm": 20.248081265506137,
      "learning_rate": 7.552548782320084e-06,
      "loss": 0.4674,
      "step": 2170
    },
    {
      "epoch": 0.35,
      "grad_norm": 15.864396244782142,
      "learning_rate": 7.550304401306887e-06,
      "loss": 0.4746,
      "step": 2171
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.837406519165391,
      "learning_rate": 7.5480593255204725e-06,
      "loss": 0.4208,
      "step": 2172
    },
    {
      "epoch": 0.35,
      "grad_norm": 37.15316655511652,
      "learning_rate": 7.545813555572461e-06,
      "loss": 0.4565,
      "step": 2173
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.960090103980175,
      "learning_rate": 7.54356709207466e-06,
      "loss": 0.4533,
      "step": 2174
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.78600158204266,
      "learning_rate": 7.5413199356390695e-06,
      "loss": 0.4274,
      "step": 2175
    },
    {
      "epoch": 0.35,
      "grad_norm": 18.535708229303637,
      "learning_rate": 7.539072086877877e-06,
      "loss": 0.405,
      "step": 2176
    },
    {
      "epoch": 0.35,
      "grad_norm": 12.412936408471017,
      "learning_rate": 7.536823546403458e-06,
      "loss": 0.4504,
      "step": 2177
    },
    {
      "epoch": 0.35,
      "grad_norm": 10.598808594918669,
      "learning_rate": 7.534574314828376e-06,
      "loss": 0.452,
      "step": 2178
    },
    {
      "epoch": 0.35,
      "grad_norm": 10.089986698972194,
      "learning_rate": 7.532324392765387e-06,
      "loss": 0.4743,
      "step": 2179
    },
    {
      "epoch": 0.35,
      "grad_norm": 18.410059972717555,
      "learning_rate": 7.530073780827427e-06,
      "loss": 0.4482,
      "step": 2180
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.594075646979574,
      "learning_rate": 7.527822479627629e-06,
      "loss": 0.4127,
      "step": 2181
    },
    {
      "epoch": 0.35,
      "grad_norm": 10.23008352433601,
      "learning_rate": 7.525570489779307e-06,
      "loss": 0.4261,
      "step": 2182
    },
    {
      "epoch": 0.35,
      "grad_norm": 14.269110040307998,
      "learning_rate": 7.523317811895965e-06,
      "loss": 0.4565,
      "step": 2183
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.724414686476505,
      "learning_rate": 7.5210644465912975e-06,
      "loss": 0.466,
      "step": 2184
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.093711851655809,
      "learning_rate": 7.518810394479179e-06,
      "loss": 0.4684,
      "step": 2185
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.692682829884761,
      "learning_rate": 7.516555656173678e-06,
      "loss": 0.4543,
      "step": 2186
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5719634520222734,
      "learning_rate": 7.514300232289049e-06,
      "loss": 0.523,
      "step": 2187
    },
    {
      "epoch": 0.35,
      "grad_norm": 12.966202937254597,
      "learning_rate": 7.512044123439728e-06,
      "loss": 0.4497,
      "step": 2188
    },
    {
      "epoch": 0.35,
      "grad_norm": 10.014092401338424,
      "learning_rate": 7.509787330240342e-06,
      "loss": 0.4055,
      "step": 2189
    },
    {
      "epoch": 0.35,
      "grad_norm": 17.747031475174325,
      "learning_rate": 7.507529853305706e-06,
      "loss": 0.389,
      "step": 2190
    },
    {
      "epoch": 0.35,
      "grad_norm": 9.510371366013034,
      "learning_rate": 7.505271693250817e-06,
      "loss": 0.4492,
      "step": 2191
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.601907829154497,
      "learning_rate": 7.50301285069086e-06,
      "loss": 0.5155,
      "step": 2192
    },
    {
      "epoch": 0.35,
      "grad_norm": 10.2548322800129,
      "learning_rate": 7.500753326241208e-06,
      "loss": 0.4372,
      "step": 2193
    },
    {
      "epoch": 0.35,
      "grad_norm": 17.409018910625846,
      "learning_rate": 7.498493120517415e-06,
      "loss": 0.4328,
      "step": 2194
    },
    {
      "epoch": 0.35,
      "grad_norm": 10.098137247245152,
      "learning_rate": 7.4962322341352256e-06,
      "loss": 0.4674,
      "step": 2195
    },
    {
      "epoch": 0.35,
      "grad_norm": 12.858562459473326,
      "learning_rate": 7.493970667710566e-06,
      "loss": 0.4751,
      "step": 2196
    },
    {
      "epoch": 0.35,
      "grad_norm": 14.568628082616911,
      "learning_rate": 7.491708421859549e-06,
      "loss": 0.4678,
      "step": 2197
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.636427350868132,
      "learning_rate": 7.489445497198475e-06,
      "loss": 0.4706,
      "step": 2198
    },
    {
      "epoch": 0.35,
      "grad_norm": 12.952326217702499,
      "learning_rate": 7.487181894343826e-06,
      "loss": 0.4606,
      "step": 2199
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.995260582823821,
      "learning_rate": 7.484917613912267e-06,
      "loss": 0.4686,
      "step": 2200
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.623753843407551,
      "learning_rate": 7.482652656520655e-06,
      "loss": 0.5072,
      "step": 2201
    },
    {
      "epoch": 0.35,
      "grad_norm": 17.111091450134992,
      "learning_rate": 7.480387022786023e-06,
      "loss": 0.4933,
      "step": 2202
    },
    {
      "epoch": 0.35,
      "grad_norm": 10.941808314099825,
      "learning_rate": 7.478120713325595e-06,
      "loss": 0.4933,
      "step": 2203
    },
    {
      "epoch": 0.36,
      "grad_norm": 10.55069129000571,
      "learning_rate": 7.4758537287567745e-06,
      "loss": 0.4529,
      "step": 2204
    },
    {
      "epoch": 0.36,
      "grad_norm": 11.07230794015543,
      "learning_rate": 7.4735860696971505e-06,
      "loss": 0.4514,
      "step": 2205
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.06690721760916,
      "learning_rate": 7.471317736764497e-06,
      "loss": 0.4681,
      "step": 2206
    },
    {
      "epoch": 0.36,
      "grad_norm": 14.06831495086204,
      "learning_rate": 7.46904873057677e-06,
      "loss": 0.458,
      "step": 2207
    },
    {
      "epoch": 0.36,
      "grad_norm": 12.915016507904822,
      "learning_rate": 7.466779051752107e-06,
      "loss": 0.4153,
      "step": 2208
    },
    {
      "epoch": 0.36,
      "grad_norm": 637.8867150776215,
      "learning_rate": 7.464508700908836e-06,
      "loss": 0.3987,
      "step": 2209
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.982455642267267,
      "learning_rate": 7.4622376786654596e-06,
      "loss": 0.4262,
      "step": 2210
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.845330031960573,
      "learning_rate": 7.459965985640665e-06,
      "loss": 0.4387,
      "step": 2211
    },
    {
      "epoch": 0.36,
      "grad_norm": 30.448791623191244,
      "learning_rate": 7.457693622453329e-06,
      "loss": 0.4366,
      "step": 2212
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.235333536107866,
      "learning_rate": 7.455420589722504e-06,
      "loss": 0.4419,
      "step": 2213
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2981043499163316,
      "learning_rate": 7.453146888067424e-06,
      "loss": 0.4386,
      "step": 2214
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.072360465820033,
      "learning_rate": 7.450872518107511e-06,
      "loss": 0.4866,
      "step": 2215
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.288121813588003,
      "learning_rate": 7.448597480462366e-06,
      "loss": 0.4251,
      "step": 2216
    },
    {
      "epoch": 0.36,
      "grad_norm": 12.120802218964878,
      "learning_rate": 7.446321775751772e-06,
      "loss": 0.4685,
      "step": 2217
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.596918772388062,
      "learning_rate": 7.444045404595692e-06,
      "loss": 0.4356,
      "step": 2218
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.5552650636579175,
      "learning_rate": 7.441768367614274e-06,
      "loss": 0.4194,
      "step": 2219
    },
    {
      "epoch": 0.36,
      "grad_norm": 10.001765814701153,
      "learning_rate": 7.439490665427844e-06,
      "loss": 0.4019,
      "step": 2220
    },
    {
      "epoch": 0.36,
      "grad_norm": 15.02221793958069,
      "learning_rate": 7.437212298656914e-06,
      "loss": 0.4609,
      "step": 2221
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.439664391213184,
      "learning_rate": 7.434933267922168e-06,
      "loss": 0.4554,
      "step": 2222
    },
    {
      "epoch": 0.36,
      "grad_norm": 16.4021217996922,
      "learning_rate": 7.432653573844483e-06,
      "loss": 0.4885,
      "step": 2223
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.205231614628408,
      "learning_rate": 7.430373217044909e-06,
      "loss": 0.4373,
      "step": 2224
    },
    {
      "epoch": 0.36,
      "grad_norm": 11.680814244296599,
      "learning_rate": 7.428092198144674e-06,
      "loss": 0.4013,
      "step": 2225
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1737927882822101,
      "learning_rate": 7.425810517765196e-06,
      "loss": 0.4812,
      "step": 2226
    },
    {
      "epoch": 0.36,
      "grad_norm": 38.753182680092394,
      "learning_rate": 7.423528176528063e-06,
      "loss": 0.497,
      "step": 2227
    },
    {
      "epoch": 0.36,
      "grad_norm": 13.986620768775001,
      "learning_rate": 7.4212451750550515e-06,
      "loss": 0.4887,
      "step": 2228
    },
    {
      "epoch": 0.36,
      "grad_norm": 37.92779065413966,
      "learning_rate": 7.418961513968111e-06,
      "loss": 0.5363,
      "step": 2229
    },
    {
      "epoch": 0.36,
      "grad_norm": 12.144955078335013,
      "learning_rate": 7.416677193889376e-06,
      "loss": 0.4719,
      "step": 2230
    },
    {
      "epoch": 0.36,
      "grad_norm": 12.797129593581774,
      "learning_rate": 7.4143922154411576e-06,
      "loss": 0.4709,
      "step": 2231
    },
    {
      "epoch": 0.36,
      "grad_norm": 12.313986572615347,
      "learning_rate": 7.412106579245945e-06,
      "loss": 0.4663,
      "step": 2232
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.426212812312611,
      "learning_rate": 7.409820285926411e-06,
      "loss": 0.5158,
      "step": 2233
    },
    {
      "epoch": 0.36,
      "grad_norm": 15.979716443754684,
      "learning_rate": 7.407533336105404e-06,
      "loss": 0.4804,
      "step": 2234
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.706177401035058,
      "learning_rate": 7.405245730405954e-06,
      "loss": 0.4257,
      "step": 2235
    },
    {
      "epoch": 0.36,
      "grad_norm": 16.37471380908726,
      "learning_rate": 7.402957469451263e-06,
      "loss": 0.4244,
      "step": 2236
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.157896888151061,
      "learning_rate": 7.4006685538647214e-06,
      "loss": 0.473,
      "step": 2237
    },
    {
      "epoch": 0.36,
      "grad_norm": 11.80268924173756,
      "learning_rate": 7.3983789842698894e-06,
      "loss": 0.4003,
      "step": 2238
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.955791515873411,
      "learning_rate": 7.3960887612905116e-06,
      "loss": 0.528,
      "step": 2239
    },
    {
      "epoch": 0.36,
      "grad_norm": 12.133625113152975,
      "learning_rate": 7.393797885550506e-06,
      "loss": 0.4865,
      "step": 2240
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.25714468987419,
      "learning_rate": 7.391506357673972e-06,
      "loss": 0.465,
      "step": 2241
    },
    {
      "epoch": 0.36,
      "grad_norm": 20.153509807388133,
      "learning_rate": 7.389214178285182e-06,
      "loss": 0.4241,
      "step": 2242
    },
    {
      "epoch": 0.36,
      "grad_norm": 11.188337004989473,
      "learning_rate": 7.386921348008592e-06,
      "loss": 0.4042,
      "step": 2243
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.77823647067846,
      "learning_rate": 7.384627867468831e-06,
      "loss": 0.4905,
      "step": 2244
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.724889248888449,
      "learning_rate": 7.382333737290708e-06,
      "loss": 0.4393,
      "step": 2245
    },
    {
      "epoch": 0.36,
      "grad_norm": 17.867966553701493,
      "learning_rate": 7.3800389580992025e-06,
      "loss": 0.4639,
      "step": 2246
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.434624281875605,
      "learning_rate": 7.37774353051948e-06,
      "loss": 0.4374,
      "step": 2247
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.950368977540833,
      "learning_rate": 7.375447455176877e-06,
      "loss": 0.4682,
      "step": 2248
    },
    {
      "epoch": 0.36,
      "grad_norm": 11.332165281344027,
      "learning_rate": 7.373150732696907e-06,
      "loss": 0.4589,
      "step": 2249
    },
    {
      "epoch": 0.36,
      "grad_norm": 10.247659934183176,
      "learning_rate": 7.370853363705261e-06,
      "loss": 0.4261,
      "step": 2250
    },
    {
      "epoch": 0.36,
      "grad_norm": 12.276095542889951,
      "learning_rate": 7.3685553488278064e-06,
      "loss": 0.4515,
      "step": 2251
    },
    {
      "epoch": 0.36,
      "grad_norm": 21.151116564101603,
      "learning_rate": 7.366256688690585e-06,
      "loss": 0.4158,
      "step": 2252
    },
    {
      "epoch": 0.36,
      "grad_norm": 10.02125046531735,
      "learning_rate": 7.363957383919815e-06,
      "loss": 0.4276,
      "step": 2253
    },
    {
      "epoch": 0.36,
      "grad_norm": 10.714814572295273,
      "learning_rate": 7.361657435141892e-06,
      "loss": 0.4743,
      "step": 2254
    },
    {
      "epoch": 0.36,
      "grad_norm": 14.540178462722155,
      "learning_rate": 7.3593568429833825e-06,
      "loss": 0.4944,
      "step": 2255
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.3959780615148425,
      "learning_rate": 7.357055608071034e-06,
      "loss": 0.4467,
      "step": 2256
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.326958353084601,
      "learning_rate": 7.354753731031765e-06,
      "loss": 0.4786,
      "step": 2257
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.1770041229199,
      "learning_rate": 7.352451212492671e-06,
      "loss": 0.5197,
      "step": 2258
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.30273349036512,
      "learning_rate": 7.350148053081021e-06,
      "loss": 0.4991,
      "step": 2259
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.378671762319642,
      "learning_rate": 7.3478442534242565e-06,
      "loss": 0.4633,
      "step": 2260
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.851291484288036,
      "learning_rate": 7.345539814150002e-06,
      "loss": 0.5002,
      "step": 2261
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.907208181633826,
      "learning_rate": 7.3432347358860445e-06,
      "loss": 0.5443,
      "step": 2262
    },
    {
      "epoch": 0.36,
      "grad_norm": 18.8556952246286,
      "learning_rate": 7.340929019260356e-06,
      "loss": 0.4805,
      "step": 2263
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.095614189610417,
      "learning_rate": 7.338622664901073e-06,
      "loss": 0.4518,
      "step": 2264
    },
    {
      "epoch": 0.36,
      "grad_norm": 10.839558533596918,
      "learning_rate": 7.336315673436512e-06,
      "loss": 0.4984,
      "step": 2265
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.086179670561378,
      "learning_rate": 7.33400804549516e-06,
      "loss": 0.4215,
      "step": 2266
    },
    {
      "epoch": 0.37,
      "grad_norm": 9.581925745922735,
      "learning_rate": 7.331699781705679e-06,
      "loss": 0.5005,
      "step": 2267
    },
    {
      "epoch": 0.37,
      "grad_norm": 10.225794330510054,
      "learning_rate": 7.329390882696904e-06,
      "loss": 0.5105,
      "step": 2268
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.997626559692268,
      "learning_rate": 7.327081349097842e-06,
      "loss": 0.4261,
      "step": 2269
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.620642326796976,
      "learning_rate": 7.324771181537676e-06,
      "loss": 0.4002,
      "step": 2270
    },
    {
      "epoch": 0.37,
      "grad_norm": 10.499137496831187,
      "learning_rate": 7.322460380645755e-06,
      "loss": 0.4664,
      "step": 2271
    },
    {
      "epoch": 0.37,
      "grad_norm": 11.33877314538186,
      "learning_rate": 7.32014894705161e-06,
      "loss": 0.3852,
      "step": 2272
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3855615616887829,
      "learning_rate": 7.317836881384934e-06,
      "loss": 0.5257,
      "step": 2273
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.439921804124257,
      "learning_rate": 7.315524184275602e-06,
      "loss": 0.4502,
      "step": 2274
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.8423278727698715,
      "learning_rate": 7.313210856353653e-06,
      "loss": 0.4849,
      "step": 2275
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.377224333850489,
      "learning_rate": 7.310896898249303e-06,
      "loss": 0.4851,
      "step": 2276
    },
    {
      "epoch": 0.37,
      "grad_norm": 11.695300727880074,
      "learning_rate": 7.3085823105929385e-06,
      "loss": 0.4897,
      "step": 2277
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.355416852444542,
      "learning_rate": 7.306267094015116e-06,
      "loss": 0.4391,
      "step": 2278
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.269960654369742,
      "learning_rate": 7.303951249146563e-06,
      "loss": 0.3958,
      "step": 2279
    },
    {
      "epoch": 0.37,
      "grad_norm": 10.942989723426711,
      "learning_rate": 7.301634776618183e-06,
      "loss": 0.5127,
      "step": 2280
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.990299487583623,
      "learning_rate": 7.299317677061043e-06,
      "loss": 0.3809,
      "step": 2281
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.428374292441154,
      "learning_rate": 7.2969999511063875e-06,
      "loss": 0.4492,
      "step": 2282
    },
    {
      "epoch": 0.37,
      "grad_norm": 9.68269509485545,
      "learning_rate": 7.294681599385629e-06,
      "loss": 0.4949,
      "step": 2283
    },
    {
      "epoch": 0.37,
      "grad_norm": 10.726385317223865,
      "learning_rate": 7.292362622530347e-06,
      "loss": 0.4387,
      "step": 2284
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.703692197288127,
      "learning_rate": 7.2900430211723e-06,
      "loss": 0.5122,
      "step": 2285
    },
    {
      "epoch": 0.37,
      "grad_norm": 11.41923267502462,
      "learning_rate": 7.287722795943407e-06,
      "loss": 0.477,
      "step": 2286
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.277230325607926,
      "learning_rate": 7.285401947475764e-06,
      "loss": 0.4471,
      "step": 2287
    },
    {
      "epoch": 0.37,
      "grad_norm": 10.295091473012938,
      "learning_rate": 7.283080476401634e-06,
      "loss": 0.4238,
      "step": 2288
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.223241963594739,
      "learning_rate": 7.280758383353447e-06,
      "loss": 0.5175,
      "step": 2289
    },
    {
      "epoch": 0.37,
      "grad_norm": 9.56715934664918,
      "learning_rate": 7.278435668963807e-06,
      "loss": 0.4083,
      "step": 2290
    },
    {
      "epoch": 0.37,
      "grad_norm": 60.505905171125264,
      "learning_rate": 7.276112333865485e-06,
      "loss": 0.5082,
      "step": 2291
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.0654693024042,
      "learning_rate": 7.273788378691422e-06,
      "loss": 0.4844,
      "step": 2292
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3780112708343384,
      "learning_rate": 7.271463804074728e-06,
      "loss": 0.466,
      "step": 2293
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.717702348084831,
      "learning_rate": 7.26913861064868e-06,
      "loss": 0.4415,
      "step": 2294
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.101783535247044,
      "learning_rate": 7.2668127990467266e-06,
      "loss": 0.4591,
      "step": 2295
    },
    {
      "epoch": 0.37,
      "grad_norm": 9.028238701024941,
      "learning_rate": 7.26448636990248e-06,
      "loss": 0.5111,
      "step": 2296
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.297986563283944,
      "learning_rate": 7.262159323849725e-06,
      "loss": 0.4944,
      "step": 2297
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.970588226967344,
      "learning_rate": 7.259831661522415e-06,
      "loss": 0.4636,
      "step": 2298
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3283109488606364,
      "learning_rate": 7.257503383554668e-06,
      "loss": 0.4609,
      "step": 2299
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.281383438785661,
      "learning_rate": 7.25517449058077e-06,
      "loss": 0.4399,
      "step": 2300
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.580084235730238,
      "learning_rate": 7.252844983235177e-06,
      "loss": 0.4196,
      "step": 2301
    },
    {
      "epoch": 0.37,
      "grad_norm": 20.15197987021541,
      "learning_rate": 7.250514862152509e-06,
      "loss": 0.4162,
      "step": 2302
    },
    {
      "epoch": 0.37,
      "grad_norm": 10.963679212689494,
      "learning_rate": 7.248184127967557e-06,
      "loss": 0.4434,
      "step": 2303
    },
    {
      "epoch": 0.37,
      "grad_norm": 12.34447337626002,
      "learning_rate": 7.245852781315278e-06,
      "loss": 0.4996,
      "step": 2304
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.822386625745935,
      "learning_rate": 7.243520822830794e-06,
      "loss": 0.4463,
      "step": 2305
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.376647823834363,
      "learning_rate": 7.241188253149395e-06,
      "loss": 0.412,
      "step": 2306
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.82513575055368,
      "learning_rate": 7.238855072906537e-06,
      "loss": 0.5113,
      "step": 2307
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.388090847297473,
      "learning_rate": 7.236521282737842e-06,
      "loss": 0.4658,
      "step": 2308
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.217588585050403,
      "learning_rate": 7.2341868832791e-06,
      "loss": 0.438,
      "step": 2309
    },
    {
      "epoch": 0.37,
      "grad_norm": 12.68103867624612,
      "learning_rate": 7.231851875166266e-06,
      "loss": 0.4049,
      "step": 2310
    },
    {
      "epoch": 0.37,
      "grad_norm": 9.662099721902655,
      "learning_rate": 7.229516259035458e-06,
      "loss": 0.3642,
      "step": 2311
    },
    {
      "epoch": 0.37,
      "grad_norm": 10.455954810168295,
      "learning_rate": 7.227180035522966e-06,
      "loss": 0.491,
      "step": 2312
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.941646924722033,
      "learning_rate": 7.224843205265239e-06,
      "loss": 0.4429,
      "step": 2313
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.373739081692871,
      "learning_rate": 7.222505768898894e-06,
      "loss": 0.4596,
      "step": 2314
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3125910203014157,
      "learning_rate": 7.220167727060714e-06,
      "loss": 0.5171,
      "step": 2315
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.691027964746766,
      "learning_rate": 7.217829080387648e-06,
      "loss": 0.461,
      "step": 2316
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.236353262272981,
      "learning_rate": 7.2154898295168045e-06,
      "loss": 0.5031,
      "step": 2317
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.06257014097761,
      "learning_rate": 7.21314997508546e-06,
      "loss": 0.4318,
      "step": 2318
    },
    {
      "epoch": 0.37,
      "grad_norm": 13.222896431949701,
      "learning_rate": 7.210809517731057e-06,
      "loss": 0.4274,
      "step": 2319
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.275052758516195,
      "learning_rate": 7.2084684580912e-06,
      "loss": 0.4484,
      "step": 2320
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.047955159935836,
      "learning_rate": 7.206126796803659e-06,
      "loss": 0.4387,
      "step": 2321
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.040850616521069,
      "learning_rate": 7.203784534506364e-06,
      "loss": 0.4846,
      "step": 2322
    },
    {
      "epoch": 0.37,
      "grad_norm": 11.99305418972947,
      "learning_rate": 7.201441671837417e-06,
      "loss": 0.4626,
      "step": 2323
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.570588079221245,
      "learning_rate": 7.199098209435073e-06,
      "loss": 0.4708,
      "step": 2324
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.232881339789681,
      "learning_rate": 7.19675414793776e-06,
      "loss": 0.4992,
      "step": 2325
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.622926516400964,
      "learning_rate": 7.194409487984063e-06,
      "loss": 0.4003,
      "step": 2326
    },
    {
      "epoch": 0.37,
      "grad_norm": 10.095281070298373,
      "learning_rate": 7.192064230212733e-06,
      "loss": 0.4063,
      "step": 2327
    },
    {
      "epoch": 0.38,
      "grad_norm": 9.798873220757585,
      "learning_rate": 7.189718375262681e-06,
      "loss": 0.5014,
      "step": 2328
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.798353927363444,
      "learning_rate": 7.1873719237729835e-06,
      "loss": 0.494,
      "step": 2329
    },
    {
      "epoch": 0.38,
      "grad_norm": 32.21359090778231,
      "learning_rate": 7.18502487638288e-06,
      "loss": 0.4877,
      "step": 2330
    },
    {
      "epoch": 0.38,
      "grad_norm": 76.69863828111654,
      "learning_rate": 7.182677233731768e-06,
      "loss": 0.4908,
      "step": 2331
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.306496828211518,
      "learning_rate": 7.1803289964592115e-06,
      "loss": 0.4528,
      "step": 2332
    },
    {
      "epoch": 0.38,
      "grad_norm": 12.959558298737406,
      "learning_rate": 7.177980165204935e-06,
      "loss": 0.4659,
      "step": 2333
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.383401295302071,
      "learning_rate": 7.1756307406088275e-06,
      "loss": 0.4712,
      "step": 2334
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.820948031824189,
      "learning_rate": 7.173280723310932e-06,
      "loss": 0.4685,
      "step": 2335
    },
    {
      "epoch": 0.38,
      "grad_norm": 9.370859669820504,
      "learning_rate": 7.170930113951462e-06,
      "loss": 0.4742,
      "step": 2336
    },
    {
      "epoch": 0.38,
      "grad_norm": 10.201344388532467,
      "learning_rate": 7.168578913170788e-06,
      "loss": 0.5038,
      "step": 2337
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.148699503788613,
      "learning_rate": 7.166227121609439e-06,
      "loss": 0.4438,
      "step": 2338
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.3972833657333625,
      "learning_rate": 7.16387473990811e-06,
      "loss": 0.3986,
      "step": 2339
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.6801814010577,
      "learning_rate": 7.1615217687076555e-06,
      "loss": 0.43,
      "step": 2340
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.677543195091026,
      "learning_rate": 7.159168208649086e-06,
      "loss": 0.4779,
      "step": 2341
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.197696416554334,
      "learning_rate": 7.15681406037358e-06,
      "loss": 0.389,
      "step": 2342
    },
    {
      "epoch": 0.38,
      "grad_norm": 13.278102046453421,
      "learning_rate": 7.154459324522474e-06,
      "loss": 0.452,
      "step": 2343
    },
    {
      "epoch": 0.38,
      "grad_norm": 9.484114423721536,
      "learning_rate": 7.152104001737254e-06,
      "loss": 0.453,
      "step": 2344
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.494850371531936,
      "learning_rate": 7.149748092659585e-06,
      "loss": 0.5003,
      "step": 2345
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.951922827002078,
      "learning_rate": 7.147391597931277e-06,
      "loss": 0.3957,
      "step": 2346
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2073194239313514,
      "learning_rate": 7.145034518194304e-06,
      "loss": 0.4873,
      "step": 2347
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2781448741845711,
      "learning_rate": 7.1426768540908e-06,
      "loss": 0.4385,
      "step": 2348
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.792667243618058,
      "learning_rate": 7.140318606263058e-06,
      "loss": 0.4329,
      "step": 2349
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.126932308540526,
      "learning_rate": 7.137959775353529e-06,
      "loss": 0.4599,
      "step": 2350
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.254358907380975,
      "learning_rate": 7.135600362004824e-06,
      "loss": 0.5906,
      "step": 2351
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.571658807909942,
      "learning_rate": 7.133240366859713e-06,
      "loss": 0.4699,
      "step": 2352
    },
    {
      "epoch": 0.38,
      "grad_norm": 11.56911111335395,
      "learning_rate": 7.130879790561122e-06,
      "loss": 0.488,
      "step": 2353
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3955096313395376,
      "learning_rate": 7.128518633752139e-06,
      "loss": 0.4683,
      "step": 2354
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.039838244323425,
      "learning_rate": 7.126156897076005e-06,
      "loss": 0.4212,
      "step": 2355
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.335872449497525,
      "learning_rate": 7.123794581176127e-06,
      "loss": 0.4778,
      "step": 2356
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4370740332278222,
      "learning_rate": 7.121431686696061e-06,
      "loss": 0.4848,
      "step": 2357
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.535473843128977,
      "learning_rate": 7.119068214279525e-06,
      "loss": 0.4218,
      "step": 2358
    },
    {
      "epoch": 0.38,
      "grad_norm": 10.693924844396516,
      "learning_rate": 7.116704164570398e-06,
      "loss": 0.4731,
      "step": 2359
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.212206817490371,
      "learning_rate": 7.114339538212707e-06,
      "loss": 0.4985,
      "step": 2360
    },
    {
      "epoch": 0.38,
      "grad_norm": 22.40544599207518,
      "learning_rate": 7.111974335850644e-06,
      "loss": 0.3914,
      "step": 2361
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2077120511604114,
      "learning_rate": 7.1096085581285555e-06,
      "loss": 0.4901,
      "step": 2362
    },
    {
      "epoch": 0.38,
      "grad_norm": 13.08913830297425,
      "learning_rate": 7.1072422056909426e-06,
      "loss": 0.4581,
      "step": 2363
    },
    {
      "epoch": 0.38,
      "grad_norm": 14.535403628425986,
      "learning_rate": 7.104875279182468e-06,
      "loss": 0.5037,
      "step": 2364
    },
    {
      "epoch": 0.38,
      "grad_norm": 11.29889237359968,
      "learning_rate": 7.102507779247947e-06,
      "loss": 0.4171,
      "step": 2365
    },
    {
      "epoch": 0.38,
      "grad_norm": 11.247829408752288,
      "learning_rate": 7.100139706532347e-06,
      "loss": 0.5091,
      "step": 2366
    },
    {
      "epoch": 0.38,
      "grad_norm": 10.342420806821993,
      "learning_rate": 7.097771061680805e-06,
      "loss": 0.4864,
      "step": 2367
    },
    {
      "epoch": 0.38,
      "grad_norm": 20.43916592740891,
      "learning_rate": 7.095401845338598e-06,
      "loss": 0.4294,
      "step": 2368
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.346033549745908,
      "learning_rate": 7.093032058151168e-06,
      "loss": 0.4017,
      "step": 2369
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4126253185001287,
      "learning_rate": 7.090661700764112e-06,
      "loss": 0.4767,
      "step": 2370
    },
    {
      "epoch": 0.38,
      "grad_norm": 9.442100200897753,
      "learning_rate": 7.088290773823177e-06,
      "loss": 0.4492,
      "step": 2371
    },
    {
      "epoch": 0.38,
      "grad_norm": 10.093730759580277,
      "learning_rate": 7.085919277974274e-06,
      "loss": 0.3833,
      "step": 2372
    },
    {
      "epoch": 0.38,
      "grad_norm": 14.209336948606353,
      "learning_rate": 7.083547213863458e-06,
      "loss": 0.4717,
      "step": 2373
    },
    {
      "epoch": 0.38,
      "grad_norm": 11.700448577681772,
      "learning_rate": 7.0811745821369495e-06,
      "loss": 0.4499,
      "step": 2374
    },
    {
      "epoch": 0.38,
      "grad_norm": 16.525786521393062,
      "learning_rate": 7.0788013834411165e-06,
      "loss": 0.3774,
      "step": 2375
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.971559601576116,
      "learning_rate": 7.0764276184224845e-06,
      "loss": 0.4552,
      "step": 2376
    },
    {
      "epoch": 0.38,
      "grad_norm": 10.053155738690688,
      "learning_rate": 7.07405328772773e-06,
      "loss": 0.4599,
      "step": 2377
    },
    {
      "epoch": 0.38,
      "grad_norm": 10.54922162754398,
      "learning_rate": 7.071678392003691e-06,
      "loss": 0.4537,
      "step": 2378
    },
    {
      "epoch": 0.38,
      "grad_norm": 9.390768274107499,
      "learning_rate": 7.069302931897352e-06,
      "loss": 0.4318,
      "step": 2379
    },
    {
      "epoch": 0.38,
      "grad_norm": 16.389897126320463,
      "learning_rate": 7.0669269080558515e-06,
      "loss": 0.4357,
      "step": 2380
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.359967822353355,
      "learning_rate": 7.064550321126489e-06,
      "loss": 0.4508,
      "step": 2381
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4346415714824337,
      "learning_rate": 7.0621731717567055e-06,
      "loss": 0.4711,
      "step": 2382
    },
    {
      "epoch": 0.38,
      "grad_norm": 10.367588822697945,
      "learning_rate": 7.059795460594109e-06,
      "loss": 0.4941,
      "step": 2383
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.644600413776572,
      "learning_rate": 7.057417188286449e-06,
      "loss": 0.4505,
      "step": 2384
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.441600520334161,
      "learning_rate": 7.0550383554816345e-06,
      "loss": 0.4506,
      "step": 2385
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.380972897998028,
      "learning_rate": 7.052658962827724e-06,
      "loss": 0.4344,
      "step": 2386
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.398920871556851,
      "learning_rate": 7.050279010972932e-06,
      "loss": 0.4521,
      "step": 2387
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.274252363518743,
      "learning_rate": 7.047898500565619e-06,
      "loss": 0.4474,
      "step": 2388
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0866906532737934,
      "learning_rate": 7.045517432254304e-06,
      "loss": 0.4404,
      "step": 2389
    },
    {
      "epoch": 0.39,
      "grad_norm": 19.589335321162775,
      "learning_rate": 7.043135806687655e-06,
      "loss": 0.5244,
      "step": 2390
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.120472291837094,
      "learning_rate": 7.040753624514494e-06,
      "loss": 0.4877,
      "step": 2391
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.983480443303751,
      "learning_rate": 7.038370886383793e-06,
      "loss": 0.4287,
      "step": 2392
    },
    {
      "epoch": 0.39,
      "grad_norm": 10.234894789703503,
      "learning_rate": 7.035987592944672e-06,
      "loss": 0.4438,
      "step": 2393
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.676336474260915,
      "learning_rate": 7.03360374484641e-06,
      "loss": 0.4813,
      "step": 2394
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.041342141168878,
      "learning_rate": 7.031219342738431e-06,
      "loss": 0.4503,
      "step": 2395
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.482946764121926,
      "learning_rate": 7.028834387270311e-06,
      "loss": 0.418,
      "step": 2396
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.444080281908883,
      "learning_rate": 7.02644887909178e-06,
      "loss": 0.442,
      "step": 2397
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.512584356362216,
      "learning_rate": 7.024062818852716e-06,
      "loss": 0.3888,
      "step": 2398
    },
    {
      "epoch": 0.39,
      "grad_norm": 9.405662255519774,
      "learning_rate": 7.021676207203145e-06,
      "loss": 0.4353,
      "step": 2399
    },
    {
      "epoch": 0.39,
      "grad_norm": 10.75244419428642,
      "learning_rate": 7.019289044793247e-06,
      "loss": 0.3938,
      "step": 2400
    },
    {
      "epoch": 0.39,
      "grad_norm": 11.346925884032942,
      "learning_rate": 7.016901332273352e-06,
      "loss": 0.5427,
      "step": 2401
    },
    {
      "epoch": 0.39,
      "grad_norm": 15.204562533791732,
      "learning_rate": 7.014513070293938e-06,
      "loss": 0.4767,
      "step": 2402
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.040572170452628,
      "learning_rate": 7.012124259505633e-06,
      "loss": 0.399,
      "step": 2403
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.442254116693553,
      "learning_rate": 7.0097349005592145e-06,
      "loss": 0.4117,
      "step": 2404
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.985072505865638,
      "learning_rate": 7.007344994105612e-06,
      "loss": 0.4158,
      "step": 2405
    },
    {
      "epoch": 0.39,
      "grad_norm": 12.478739903039797,
      "learning_rate": 7.004954540795899e-06,
      "loss": 0.518,
      "step": 2406
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.704643709428703,
      "learning_rate": 7.002563541281302e-06,
      "loss": 0.4486,
      "step": 2407
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.3844073556806995,
      "learning_rate": 7.000171996213196e-06,
      "loss": 0.3981,
      "step": 2408
    },
    {
      "epoch": 0.39,
      "grad_norm": 13.199684753424732,
      "learning_rate": 6.997779906243103e-06,
      "loss": 0.3728,
      "step": 2409
    },
    {
      "epoch": 0.39,
      "grad_norm": 16.461602843019307,
      "learning_rate": 6.995387272022695e-06,
      "loss": 0.3962,
      "step": 2410
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.678208730393507,
      "learning_rate": 6.99299409420379e-06,
      "loss": 0.5136,
      "step": 2411
    },
    {
      "epoch": 0.39,
      "grad_norm": 11.584951031330831,
      "learning_rate": 6.9906003734383565e-06,
      "loss": 0.4088,
      "step": 2412
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.206057432935938,
      "learning_rate": 6.98820611037851e-06,
      "loss": 0.4919,
      "step": 2413
    },
    {
      "epoch": 0.39,
      "grad_norm": 25.48820974719153,
      "learning_rate": 6.985811305676515e-06,
      "loss": 0.4919,
      "step": 2414
    },
    {
      "epoch": 0.39,
      "grad_norm": 23.947717235865277,
      "learning_rate": 6.98341595998478e-06,
      "loss": 0.4232,
      "step": 2415
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.898446328823951,
      "learning_rate": 6.981020073955866e-06,
      "loss": 0.3991,
      "step": 2416
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.457360816881248,
      "learning_rate": 6.978623648242474e-06,
      "loss": 0.4558,
      "step": 2417
    },
    {
      "epoch": 0.39,
      "grad_norm": 11.97962498058865,
      "learning_rate": 6.9762266834974605e-06,
      "loss": 0.4969,
      "step": 2418
    },
    {
      "epoch": 0.39,
      "grad_norm": 15.40563419821297,
      "learning_rate": 6.973829180373823e-06,
      "loss": 0.4339,
      "step": 2419
    },
    {
      "epoch": 0.39,
      "grad_norm": 19.828764356090712,
      "learning_rate": 6.971431139524709e-06,
      "loss": 0.4901,
      "step": 2420
    },
    {
      "epoch": 0.39,
      "grad_norm": 9.778713667308615,
      "learning_rate": 6.96903256160341e-06,
      "loss": 0.4309,
      "step": 2421
    },
    {
      "epoch": 0.39,
      "grad_norm": 9.36401245946902,
      "learning_rate": 6.966633447263362e-06,
      "loss": 0.4377,
      "step": 2422
    },
    {
      "epoch": 0.39,
      "grad_norm": 9.788069862285154,
      "learning_rate": 6.964233797158155e-06,
      "loss": 0.469,
      "step": 2423
    },
    {
      "epoch": 0.39,
      "grad_norm": 12.102995897781907,
      "learning_rate": 6.961833611941515e-06,
      "loss": 0.4568,
      "step": 2424
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.4561923233948,
      "learning_rate": 6.959432892267324e-06,
      "loss": 0.4617,
      "step": 2425
    },
    {
      "epoch": 0.39,
      "grad_norm": 9.187151635133098,
      "learning_rate": 6.957031638789598e-06,
      "loss": 0.4169,
      "step": 2426
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3874037784658793,
      "learning_rate": 6.954629852162509e-06,
      "loss": 0.4984,
      "step": 2427
    },
    {
      "epoch": 0.39,
      "grad_norm": 13.449698897659353,
      "learning_rate": 6.952227533040369e-06,
      "loss": 0.4636,
      "step": 2428
    },
    {
      "epoch": 0.39,
      "grad_norm": 22.568536652964223,
      "learning_rate": 6.949824682077635e-06,
      "loss": 0.4902,
      "step": 2429
    },
    {
      "epoch": 0.39,
      "grad_norm": 11.351741678071608,
      "learning_rate": 6.947421299928909e-06,
      "loss": 0.4346,
      "step": 2430
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.790244987952567,
      "learning_rate": 6.945017387248942e-06,
      "loss": 0.4104,
      "step": 2431
    },
    {
      "epoch": 0.39,
      "grad_norm": 13.487103643818642,
      "learning_rate": 6.942612944692624e-06,
      "loss": 0.3904,
      "step": 2432
    },
    {
      "epoch": 0.39,
      "grad_norm": 16.244672362302826,
      "learning_rate": 6.940207972914989e-06,
      "loss": 0.3976,
      "step": 2433
    },
    {
      "epoch": 0.39,
      "grad_norm": 10.504723629669721,
      "learning_rate": 6.9378024725712225e-06,
      "loss": 0.408,
      "step": 2434
    },
    {
      "epoch": 0.39,
      "grad_norm": 10.508221371748512,
      "learning_rate": 6.935396444316646e-06,
      "loss": 0.4445,
      "step": 2435
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0637657091295818,
      "learning_rate": 6.9329898888067295e-06,
      "loss": 0.4681,
      "step": 2436
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.746609200866317,
      "learning_rate": 6.930582806697082e-06,
      "loss": 0.4757,
      "step": 2437
    },
    {
      "epoch": 0.39,
      "grad_norm": 11.244148747738713,
      "learning_rate": 6.928175198643463e-06,
      "loss": 0.4374,
      "step": 2438
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.587036160693334,
      "learning_rate": 6.9257670653017674e-06,
      "loss": 0.4679,
      "step": 2439
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.316161717954362,
      "learning_rate": 6.92335840732804e-06,
      "loss": 0.387,
      "step": 2440
    },
    {
      "epoch": 0.39,
      "grad_norm": 10.624828080509273,
      "learning_rate": 6.9209492253784664e-06,
      "loss": 0.441,
      "step": 2441
    },
    {
      "epoch": 0.39,
      "grad_norm": 9.752619066860827,
      "learning_rate": 6.91853952010937e-06,
      "loss": 0.4813,
      "step": 2442
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.472394904483936,
      "learning_rate": 6.916129292177225e-06,
      "loss": 0.5247,
      "step": 2443
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.532384304401379,
      "learning_rate": 6.913718542238642e-06,
      "loss": 0.497,
      "step": 2444
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.684806398803388,
      "learning_rate": 6.911307270950376e-06,
      "loss": 0.4275,
      "step": 2445
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.8886421716911395,
      "learning_rate": 6.908895478969324e-06,
      "loss": 0.4319,
      "step": 2446
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.638392199346289,
      "learning_rate": 6.906483166952526e-06,
      "loss": 0.4775,
      "step": 2447
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.564972130318737,
      "learning_rate": 6.904070335557158e-06,
      "loss": 0.4477,
      "step": 2448
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.183747032671569,
      "learning_rate": 6.9016569854405466e-06,
      "loss": 0.4179,
      "step": 2449
    },
    {
      "epoch": 0.39,
      "grad_norm": 13.931457895049578,
      "learning_rate": 6.899243117260153e-06,
      "loss": 0.4345,
      "step": 2450
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.071694154845241,
      "learning_rate": 6.896828731673579e-06,
      "loss": 0.4728,
      "step": 2451
    },
    {
      "epoch": 0.4,
      "grad_norm": 9.848153462459301,
      "learning_rate": 6.894413829338576e-06,
      "loss": 0.4849,
      "step": 2452
    },
    {
      "epoch": 0.4,
      "grad_norm": 9.617163645216445,
      "learning_rate": 6.891998410913021e-06,
      "loss": 0.4632,
      "step": 2453
    },
    {
      "epoch": 0.4,
      "grad_norm": 12.02608942490304,
      "learning_rate": 6.88958247705495e-06,
      "loss": 0.5041,
      "step": 2454
    },
    {
      "epoch": 0.4,
      "grad_norm": 43.31490199784347,
      "learning_rate": 6.887166028422524e-06,
      "loss": 0.427,
      "step": 2455
    },
    {
      "epoch": 0.4,
      "grad_norm": 17.880979555375145,
      "learning_rate": 6.884749065674051e-06,
      "loss": 0.455,
      "step": 2456
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.553133422448308,
      "learning_rate": 6.88233158946798e-06,
      "loss": 0.4146,
      "step": 2457
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.792022571716306,
      "learning_rate": 6.879913600462898e-06,
      "loss": 0.4784,
      "step": 2458
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.120768171119673,
      "learning_rate": 6.87749509931753e-06,
      "loss": 0.4019,
      "step": 2459
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.628429299028117,
      "learning_rate": 6.875076086690744e-06,
      "loss": 0.4932,
      "step": 2460
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.923010388243224,
      "learning_rate": 6.8726565632415445e-06,
      "loss": 0.4385,
      "step": 2461
    },
    {
      "epoch": 0.4,
      "grad_norm": 9.313389831013437,
      "learning_rate": 6.8702365296290775e-06,
      "loss": 0.4669,
      "step": 2462
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.2075318171180305,
      "learning_rate": 6.867815986512627e-06,
      "loss": 0.3429,
      "step": 2463
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.542475948699439,
      "learning_rate": 6.865394934551613e-06,
      "loss": 0.4565,
      "step": 2464
    },
    {
      "epoch": 0.4,
      "grad_norm": 24.349009204481835,
      "learning_rate": 6.862973374405601e-06,
      "loss": 0.401,
      "step": 2465
    },
    {
      "epoch": 0.4,
      "grad_norm": 17.61356147439899,
      "learning_rate": 6.860551306734289e-06,
      "loss": 0.5061,
      "step": 2466
    },
    {
      "epoch": 0.4,
      "grad_norm": 10.353981452950123,
      "learning_rate": 6.858128732197513e-06,
      "loss": 0.5167,
      "step": 2467
    },
    {
      "epoch": 0.4,
      "grad_norm": 11.959934202270215,
      "learning_rate": 6.855705651455252e-06,
      "loss": 0.4518,
      "step": 2468
    },
    {
      "epoch": 0.4,
      "grad_norm": 12.390369952358583,
      "learning_rate": 6.853282065167618e-06,
      "loss": 0.4665,
      "step": 2469
    },
    {
      "epoch": 0.4,
      "grad_norm": 15.66352112030414,
      "learning_rate": 6.850857973994865e-06,
      "loss": 0.5094,
      "step": 2470
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.637520938717194,
      "learning_rate": 6.8484333785973786e-06,
      "loss": 0.4668,
      "step": 2471
    },
    {
      "epoch": 0.4,
      "grad_norm": 11.120490352544106,
      "learning_rate": 6.846008279635688e-06,
      "loss": 0.439,
      "step": 2472
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.498246428285075,
      "learning_rate": 6.8435826777704575e-06,
      "loss": 0.4388,
      "step": 2473
    },
    {
      "epoch": 0.4,
      "grad_norm": 9.319473845578853,
      "learning_rate": 6.841156573662486e-06,
      "loss": 0.4539,
      "step": 2474
    },
    {
      "epoch": 0.4,
      "grad_norm": 13.067272245181947,
      "learning_rate": 6.8387299679727125e-06,
      "loss": 0.4351,
      "step": 2475
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.951437447637335,
      "learning_rate": 6.836302861362211e-06,
      "loss": 0.5113,
      "step": 2476
    },
    {
      "epoch": 0.4,
      "grad_norm": 13.348753098698928,
      "learning_rate": 6.8338752544921915e-06,
      "loss": 0.4185,
      "step": 2477
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.626255046470796,
      "learning_rate": 6.831447148024002e-06,
      "loss": 0.4693,
      "step": 2478
    },
    {
      "epoch": 0.4,
      "grad_norm": 9.314454162590877,
      "learning_rate": 6.829018542619125e-06,
      "loss": 0.5144,
      "step": 2479
    },
    {
      "epoch": 0.4,
      "grad_norm": 25.22795601174396,
      "learning_rate": 6.82658943893918e-06,
      "loss": 0.3851,
      "step": 2480
    },
    {
      "epoch": 0.4,
      "grad_norm": 10.805834671419454,
      "learning_rate": 6.824159837645921e-06,
      "loss": 0.4493,
      "step": 2481
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.71174134334264,
      "learning_rate": 6.821729739401239e-06,
      "loss": 0.4594,
      "step": 2482
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.946191081683506,
      "learning_rate": 6.8192991448671605e-06,
      "loss": 0.3986,
      "step": 2483
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.371589467223759,
      "learning_rate": 6.8168680547058455e-06,
      "loss": 0.357,
      "step": 2484
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3604249155764994,
      "learning_rate": 6.81443646957959e-06,
      "loss": 0.4532,
      "step": 2485
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.487567277588702,
      "learning_rate": 6.812004390150825e-06,
      "loss": 0.469,
      "step": 2486
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.885175407638135,
      "learning_rate": 6.809571817082117e-06,
      "loss": 0.4407,
      "step": 2487
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.80959018378842,
      "learning_rate": 6.807138751036163e-06,
      "loss": 0.4614,
      "step": 2488
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.911564898686454,
      "learning_rate": 6.804705192675799e-06,
      "loss": 0.4443,
      "step": 2489
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.605052674841815,
      "learning_rate": 6.802271142663994e-06,
      "loss": 0.3692,
      "step": 2490
    },
    {
      "epoch": 0.4,
      "grad_norm": 10.279131357547781,
      "learning_rate": 6.799836601663851e-06,
      "loss": 0.4285,
      "step": 2491
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.78402630822119,
      "learning_rate": 6.797401570338604e-06,
      "loss": 0.4253,
      "step": 2492
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.243260681573895,
      "learning_rate": 6.794966049351625e-06,
      "loss": 0.4135,
      "step": 2493
    },
    {
      "epoch": 0.4,
      "grad_norm": 11.689243656869555,
      "learning_rate": 6.792530039366414e-06,
      "loss": 0.5574,
      "step": 2494
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.224508072563628,
      "learning_rate": 6.790093541046609e-06,
      "loss": 0.4303,
      "step": 2495
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.94453921064131,
      "learning_rate": 6.787656555055979e-06,
      "loss": 0.4364,
      "step": 2496
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.028222904594061,
      "learning_rate": 6.785219082058426e-06,
      "loss": 0.4769,
      "step": 2497
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.602281927823516,
      "learning_rate": 6.782781122717987e-06,
      "loss": 0.4231,
      "step": 2498
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.65427888621789,
      "learning_rate": 6.780342677698826e-06,
      "loss": 0.47,
      "step": 2499
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.401731655806332,
      "learning_rate": 6.777903747665245e-06,
      "loss": 0.4269,
      "step": 2500
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.54475510786984,
      "learning_rate": 6.775464333281674e-06,
      "loss": 0.5074,
      "step": 2501
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.795421086707475,
      "learning_rate": 6.773024435212678e-06,
      "loss": 0.5656,
      "step": 2502
    },
    {
      "epoch": 0.4,
      "grad_norm": 9.8236463999906,
      "learning_rate": 6.770584054122954e-06,
      "loss": 0.4163,
      "step": 2503
    },
    {
      "epoch": 0.4,
      "grad_norm": 19.534862380627295,
      "learning_rate": 6.7681431906773255e-06,
      "loss": 0.5112,
      "step": 2504
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.415365863433265,
      "learning_rate": 6.765701845540753e-06,
      "loss": 0.4402,
      "step": 2505
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.003525598342758,
      "learning_rate": 6.763260019378325e-06,
      "loss": 0.4953,
      "step": 2506
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.871693100856605,
      "learning_rate": 6.760817712855266e-06,
      "loss": 0.4365,
      "step": 2507
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.905281859691696,
      "learning_rate": 6.75837492663692e-06,
      "loss": 0.4187,
      "step": 2508
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.032936131322165,
      "learning_rate": 6.755931661388778e-06,
      "loss": 0.3569,
      "step": 2509
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.708119835835211,
      "learning_rate": 6.753487917776447e-06,
      "loss": 0.4506,
      "step": 2510
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.7092951775892935,
      "learning_rate": 6.751043696465674e-06,
      "loss": 0.4922,
      "step": 2511
    },
    {
      "epoch": 0.4,
      "grad_norm": 14.129649127754325,
      "learning_rate": 6.748598998122328e-06,
      "loss": 0.4639,
      "step": 2512
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.153527092736446,
      "learning_rate": 6.746153823412416e-06,
      "loss": 0.4482,
      "step": 2513
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.62264251203995,
      "learning_rate": 6.7437081730020695e-06,
      "loss": 0.4269,
      "step": 2514
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.32422041928711,
      "learning_rate": 6.7412620475575495e-06,
      "loss": 0.4972,
      "step": 2515
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.801432238383798,
      "learning_rate": 6.7388154477452505e-06,
      "loss": 0.4625,
      "step": 2516
    },
    {
      "epoch": 0.41,
      "grad_norm": 40.776797507388686,
      "learning_rate": 6.736368374231693e-06,
      "loss": 0.4438,
      "step": 2517
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.170686519146614,
      "learning_rate": 6.733920827683529e-06,
      "loss": 0.461,
      "step": 2518
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.751376694019045,
      "learning_rate": 6.731472808767532e-06,
      "loss": 0.4547,
      "step": 2519
    },
    {
      "epoch": 0.41,
      "grad_norm": 9.71397785458948,
      "learning_rate": 6.729024318150617e-06,
      "loss": 0.5061,
      "step": 2520
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.594982642960922,
      "learning_rate": 6.726575356499814e-06,
      "loss": 0.4371,
      "step": 2521
    },
    {
      "epoch": 0.41,
      "grad_norm": 9.250113129443198,
      "learning_rate": 6.724125924482292e-06,
      "loss": 0.3829,
      "step": 2522
    },
    {
      "epoch": 0.41,
      "grad_norm": 11.853882875881725,
      "learning_rate": 6.7216760227653426e-06,
      "loss": 0.3786,
      "step": 2523
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.861008928616773,
      "learning_rate": 6.7192256520163844e-06,
      "loss": 0.4565,
      "step": 2524
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.457622200387418,
      "learning_rate": 6.7167748129029705e-06,
      "loss": 0.3949,
      "step": 2525
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.6602605638169035,
      "learning_rate": 6.714323506092773e-06,
      "loss": 0.4781,
      "step": 2526
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.436911792902247,
      "learning_rate": 6.711871732253596e-06,
      "loss": 0.411,
      "step": 2527
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.101439601441904,
      "learning_rate": 6.709419492053373e-06,
      "loss": 0.4377,
      "step": 2528
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.153150686188736,
      "learning_rate": 6.706966786160159e-06,
      "loss": 0.464,
      "step": 2529
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.995667865135097,
      "learning_rate": 6.7045136152421395e-06,
      "loss": 0.4357,
      "step": 2530
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.872361788978383,
      "learning_rate": 6.702059979967627e-06,
      "loss": 0.413,
      "step": 2531
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.254810523532101,
      "learning_rate": 6.699605881005058e-06,
      "loss": 0.4111,
      "step": 2532
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.489493616878587,
      "learning_rate": 6.697151319022996e-06,
      "loss": 0.472,
      "step": 2533
    },
    {
      "epoch": 0.41,
      "grad_norm": 150.0070041924996,
      "learning_rate": 6.694696294690133e-06,
      "loss": 0.4326,
      "step": 2534
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.198379398120597,
      "learning_rate": 6.692240808675286e-06,
      "loss": 0.496,
      "step": 2535
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.548660764451137,
      "learning_rate": 6.689784861647395e-06,
      "loss": 0.4637,
      "step": 2536
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.23958418346717,
      "learning_rate": 6.6873284542755275e-06,
      "loss": 0.4693,
      "step": 2537
    },
    {
      "epoch": 0.41,
      "grad_norm": 35.062916405547824,
      "learning_rate": 6.684871587228878e-06,
      "loss": 0.4018,
      "step": 2538
    },
    {
      "epoch": 0.41,
      "grad_norm": 10.641131105627792,
      "learning_rate": 6.682414261176765e-06,
      "loss": 0.4659,
      "step": 2539
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.948874295505058,
      "learning_rate": 6.6799564767886305e-06,
      "loss": 0.4367,
      "step": 2540
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.4784267684022625,
      "learning_rate": 6.677498234734045e-06,
      "loss": 0.4525,
      "step": 2541
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.2501831605458245,
      "learning_rate": 6.675039535682699e-06,
      "loss": 0.4128,
      "step": 2542
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.473418512295367,
      "learning_rate": 6.67258038030441e-06,
      "loss": 0.4735,
      "step": 2543
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.2851383732486426,
      "learning_rate": 6.67012076926912e-06,
      "loss": 0.4395,
      "step": 2544
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.604074788966195,
      "learning_rate": 6.667660703246897e-06,
      "loss": 0.4685,
      "step": 2545
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.313606849498942,
      "learning_rate": 6.665200182907928e-06,
      "loss": 0.4202,
      "step": 2546
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.956636370713147,
      "learning_rate": 6.662739208922529e-06,
      "loss": 0.5373,
      "step": 2547
    },
    {
      "epoch": 0.41,
      "grad_norm": 12.055223231997422,
      "learning_rate": 6.660277781961135e-06,
      "loss": 0.4713,
      "step": 2548
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.286407851874011,
      "learning_rate": 6.6578159026943064e-06,
      "loss": 0.4367,
      "step": 2549
    },
    {
      "epoch": 0.41,
      "grad_norm": 10.644769568841399,
      "learning_rate": 6.655353571792729e-06,
      "loss": 0.4711,
      "step": 2550
    },
    {
      "epoch": 0.41,
      "grad_norm": 10.229302329082488,
      "learning_rate": 6.652890789927209e-06,
      "loss": 0.4848,
      "step": 2551
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.984718733276937,
      "learning_rate": 6.650427557768674e-06,
      "loss": 0.4504,
      "step": 2552
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.613730766824289,
      "learning_rate": 6.647963875988179e-06,
      "loss": 0.5137,
      "step": 2553
    },
    {
      "epoch": 0.41,
      "grad_norm": 74.0701663191425,
      "learning_rate": 6.645499745256898e-06,
      "loss": 0.5075,
      "step": 2554
    },
    {
      "epoch": 0.41,
      "grad_norm": 16.06784590189765,
      "learning_rate": 6.643035166246128e-06,
      "loss": 0.5358,
      "step": 2555
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.690869147987697,
      "learning_rate": 6.640570139627288e-06,
      "loss": 0.4751,
      "step": 2556
    },
    {
      "epoch": 0.41,
      "grad_norm": 9.133148844282418,
      "learning_rate": 6.638104666071918e-06,
      "loss": 0.405,
      "step": 2557
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.986444923080993,
      "learning_rate": 6.635638746251685e-06,
      "loss": 0.4346,
      "step": 2558
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.479772442887024,
      "learning_rate": 6.6331723808383674e-06,
      "loss": 0.4042,
      "step": 2559
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.2267677263907695,
      "learning_rate": 6.630705570503878e-06,
      "loss": 0.4256,
      "step": 2560
    },
    {
      "epoch": 0.41,
      "grad_norm": 10.699193198341884,
      "learning_rate": 6.628238315920239e-06,
      "loss": 0.4184,
      "step": 2561
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.317524374935241,
      "learning_rate": 6.6257706177595994e-06,
      "loss": 0.4855,
      "step": 2562
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2382407577660823,
      "learning_rate": 6.62330247669423e-06,
      "loss": 0.4169,
      "step": 2563
    },
    {
      "epoch": 0.41,
      "grad_norm": 9.899954610216126,
      "learning_rate": 6.62083389339652e-06,
      "loss": 0.459,
      "step": 2564
    },
    {
      "epoch": 0.41,
      "grad_norm": 10.494747103918552,
      "learning_rate": 6.618364868538978e-06,
      "loss": 0.4149,
      "step": 2565
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.027055467196949,
      "learning_rate": 6.6158954027942345e-06,
      "loss": 0.481,
      "step": 2566
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.969160671372257,
      "learning_rate": 6.6134254968350434e-06,
      "loss": 0.4484,
      "step": 2567
    },
    {
      "epoch": 0.41,
      "grad_norm": 13.324056038032303,
      "learning_rate": 6.610955151334269e-06,
      "loss": 0.4732,
      "step": 2568
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.95180886494875,
      "learning_rate": 6.608484366964908e-06,
      "loss": 0.3573,
      "step": 2569
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.450754776449508,
      "learning_rate": 6.606013144400065e-06,
      "loss": 0.3995,
      "step": 2570
    },
    {
      "epoch": 0.41,
      "grad_norm": 18.15265210573935,
      "learning_rate": 6.603541484312974e-06,
      "loss": 0.4307,
      "step": 2571
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.088011485070242,
      "learning_rate": 6.601069387376979e-06,
      "loss": 0.4299,
      "step": 2572
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.431935444216913,
      "learning_rate": 6.5985968542655495e-06,
      "loss": 0.3813,
      "step": 2573
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.621195191444515,
      "learning_rate": 6.596123885652272e-06,
      "loss": 0.4423,
      "step": 2574
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.477239791415972,
      "learning_rate": 6.593650482210851e-06,
      "loss": 0.4538,
      "step": 2575
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.6289922172591105,
      "learning_rate": 6.591176644615108e-06,
      "loss": 0.4273,
      "step": 2576
    },
    {
      "epoch": 0.42,
      "grad_norm": 10.908990019562408,
      "learning_rate": 6.588702373538987e-06,
      "loss": 0.3923,
      "step": 2577
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.5750552656757035,
      "learning_rate": 6.5862276696565454e-06,
      "loss": 0.4998,
      "step": 2578
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.361338125006084,
      "learning_rate": 6.583752533641963e-06,
      "loss": 0.507,
      "step": 2579
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.056490038779889,
      "learning_rate": 6.581276966169534e-06,
      "loss": 0.4807,
      "step": 2580
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.70429634791011,
      "learning_rate": 6.57880096791367e-06,
      "loss": 0.4081,
      "step": 2581
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.715377542307558,
      "learning_rate": 6.576324539548904e-06,
      "loss": 0.4719,
      "step": 2582
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.05825707480304,
      "learning_rate": 6.573847681749881e-06,
      "loss": 0.4021,
      "step": 2583
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.243919977210553,
      "learning_rate": 6.5713703951913665e-06,
      "loss": 0.4298,
      "step": 2584
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.998699956224827,
      "learning_rate": 6.5688926805482425e-06,
      "loss": 0.424,
      "step": 2585
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.238615815933871,
      "learning_rate": 6.566414538495504e-06,
      "loss": 0.4624,
      "step": 2586
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2656532034853407,
      "learning_rate": 6.563935969708266e-06,
      "loss": 0.4627,
      "step": 2587
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.543358463385838,
      "learning_rate": 6.561456974861761e-06,
      "loss": 0.5258,
      "step": 2588
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.347047652295883,
      "learning_rate": 6.558977554631334e-06,
      "loss": 0.4739,
      "step": 2589
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.814995499421517,
      "learning_rate": 6.556497709692447e-06,
      "loss": 0.4696,
      "step": 2590
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.558616012773064,
      "learning_rate": 6.554017440720679e-06,
      "loss": 0.4985,
      "step": 2591
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.622321412075058,
      "learning_rate": 6.551536748391724e-06,
      "loss": 0.421,
      "step": 2592
    },
    {
      "epoch": 0.42,
      "grad_norm": 12.876306529399512,
      "learning_rate": 6.54905563338139e-06,
      "loss": 0.4552,
      "step": 2593
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.392104080108402,
      "learning_rate": 6.546574096365601e-06,
      "loss": 0.4348,
      "step": 2594
    },
    {
      "epoch": 0.42,
      "grad_norm": 13.977814803803094,
      "learning_rate": 6.544092138020397e-06,
      "loss": 0.4589,
      "step": 2595
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.935803424687954,
      "learning_rate": 6.541609759021933e-06,
      "loss": 0.465,
      "step": 2596
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.633575173057532,
      "learning_rate": 6.539126960046474e-06,
      "loss": 0.4601,
      "step": 2597
    },
    {
      "epoch": 0.42,
      "grad_norm": 11.056524507265701,
      "learning_rate": 6.536643741770406e-06,
      "loss": 0.4567,
      "step": 2598
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.182785309469038,
      "learning_rate": 6.534160104870224e-06,
      "loss": 0.5201,
      "step": 2599
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.28855665371542,
      "learning_rate": 6.531676050022539e-06,
      "loss": 0.4843,
      "step": 2600
    },
    {
      "epoch": 0.42,
      "grad_norm": 9.904639065937364,
      "learning_rate": 6.529191577904079e-06,
      "loss": 0.5118,
      "step": 2601
    },
    {
      "epoch": 0.42,
      "grad_norm": 14.728811126547683,
      "learning_rate": 6.526706689191681e-06,
      "loss": 0.4585,
      "step": 2602
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.324117688241901,
      "learning_rate": 6.524221384562295e-06,
      "loss": 0.4603,
      "step": 2603
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.433413795561798,
      "learning_rate": 6.521735664692989e-06,
      "loss": 0.4852,
      "step": 2604
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.113528607261589,
      "learning_rate": 6.519249530260943e-06,
      "loss": 0.4166,
      "step": 2605
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.971934424921738,
      "learning_rate": 6.516762981943444e-06,
      "loss": 0.5678,
      "step": 2606
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.035636553942168,
      "learning_rate": 6.514276020417901e-06,
      "loss": 0.4264,
      "step": 2607
    },
    {
      "epoch": 0.42,
      "grad_norm": 10.311031102408492,
      "learning_rate": 6.511788646361828e-06,
      "loss": 0.4414,
      "step": 2608
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.555967568890328,
      "learning_rate": 6.509300860452854e-06,
      "loss": 0.5288,
      "step": 2609
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.213857548757916,
      "learning_rate": 6.506812663368722e-06,
      "loss": 0.4375,
      "step": 2610
    },
    {
      "epoch": 0.42,
      "grad_norm": 9.852319543974456,
      "learning_rate": 6.504324055787285e-06,
      "loss": 0.4732,
      "step": 2611
    },
    {
      "epoch": 0.42,
      "grad_norm": 9.401242355672261,
      "learning_rate": 6.501835038386509e-06,
      "loss": 0.4724,
      "step": 2612
    },
    {
      "epoch": 0.42,
      "grad_norm": 9.15327723603227,
      "learning_rate": 6.499345611844471e-06,
      "loss": 0.4765,
      "step": 2613
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.6293050332811,
      "learning_rate": 6.496855776839357e-06,
      "loss": 0.4075,
      "step": 2614
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.008175140096545,
      "learning_rate": 6.494365534049469e-06,
      "loss": 0.3909,
      "step": 2615
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.31794495599635,
      "learning_rate": 6.491874884153217e-06,
      "loss": 0.4094,
      "step": 2616
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.698969580833742,
      "learning_rate": 6.489383827829122e-06,
      "loss": 0.586,
      "step": 2617
    },
    {
      "epoch": 0.42,
      "grad_norm": 14.487316960761474,
      "learning_rate": 6.486892365755819e-06,
      "loss": 0.4841,
      "step": 2618
    },
    {
      "epoch": 0.42,
      "grad_norm": 10.588295451567266,
      "learning_rate": 6.4844004986120465e-06,
      "loss": 0.3707,
      "step": 2619
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.839548428583583,
      "learning_rate": 6.481908227076663e-06,
      "loss": 0.4402,
      "step": 2620
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.651427795130737,
      "learning_rate": 6.479415551828627e-06,
      "loss": 0.4055,
      "step": 2621
    },
    {
      "epoch": 0.42,
      "grad_norm": 12.094997417962459,
      "learning_rate": 6.476922473547016e-06,
      "loss": 0.3929,
      "step": 2622
    },
    {
      "epoch": 0.42,
      "grad_norm": 14.177669561568806,
      "learning_rate": 6.474428992911011e-06,
      "loss": 0.4072,
      "step": 2623
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.818888995850015,
      "learning_rate": 6.471935110599907e-06,
      "loss": 0.4229,
      "step": 2624
    },
    {
      "epoch": 0.42,
      "grad_norm": 13.74534096866166,
      "learning_rate": 6.469440827293103e-06,
      "loss": 0.4461,
      "step": 2625
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.590711105562193,
      "learning_rate": 6.466946143670113e-06,
      "loss": 0.4612,
      "step": 2626
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.866538942602566,
      "learning_rate": 6.464451060410556e-06,
      "loss": 0.4579,
      "step": 2627
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.839636044363688,
      "learning_rate": 6.461955578194163e-06,
      "loss": 0.4974,
      "step": 2628
    },
    {
      "epoch": 0.42,
      "grad_norm": 9.309756196488948,
      "learning_rate": 6.459459697700772e-06,
      "loss": 0.4103,
      "step": 2629
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.976798986316532,
      "learning_rate": 6.456963419610327e-06,
      "loss": 0.4768,
      "step": 2630
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.980065588092765,
      "learning_rate": 6.454466744602888e-06,
      "loss": 0.3736,
      "step": 2631
    },
    {
      "epoch": 0.42,
      "grad_norm": 14.256426159789708,
      "learning_rate": 6.451969673358613e-06,
      "loss": 0.4476,
      "step": 2632
    },
    {
      "epoch": 0.42,
      "grad_norm": 10.714377376956033,
      "learning_rate": 6.449472206557776e-06,
      "loss": 0.4111,
      "step": 2633
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.6693899040068505,
      "learning_rate": 6.4469743448807546e-06,
      "loss": 0.4755,
      "step": 2634
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.19485490173629,
      "learning_rate": 6.444476089008037e-06,
      "loss": 0.4269,
      "step": 2635
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.142633944501165,
      "learning_rate": 6.441977439620214e-06,
      "loss": 0.4413,
      "step": 2636
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.030040109818081,
      "learning_rate": 6.439478397397989e-06,
      "loss": 0.4921,
      "step": 2637
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.5924940255425195,
      "learning_rate": 6.436978963022168e-06,
      "loss": 0.4717,
      "step": 2638
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.234421506393416,
      "learning_rate": 6.434479137173667e-06,
      "loss": 0.5025,
      "step": 2639
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.166069220646565,
      "learning_rate": 6.431978920533509e-06,
      "loss": 0.4273,
      "step": 2640
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.235043799236979,
      "learning_rate": 6.429478313782817e-06,
      "loss": 0.5036,
      "step": 2641
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.492752842724464,
      "learning_rate": 6.426977317602833e-06,
      "loss": 0.3965,
      "step": 2642
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.576258178111879,
      "learning_rate": 6.424475932674889e-06,
      "loss": 0.4069,
      "step": 2643
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.655019012153899,
      "learning_rate": 6.421974159680437e-06,
      "loss": 0.4723,
      "step": 2644
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.068719422172653,
      "learning_rate": 6.419471999301027e-06,
      "loss": 0.4335,
      "step": 2645
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.869529164790956,
      "learning_rate": 6.416969452218317e-06,
      "loss": 0.4725,
      "step": 2646
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.730561542807674,
      "learning_rate": 6.414466519114068e-06,
      "loss": 0.4603,
      "step": 2647
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.507983135991724,
      "learning_rate": 6.411963200670148e-06,
      "loss": 0.4575,
      "step": 2648
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.51326306108928,
      "learning_rate": 6.409459497568533e-06,
      "loss": 0.5274,
      "step": 2649
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.064366392998285,
      "learning_rate": 6.406955410491298e-06,
      "loss": 0.4103,
      "step": 2650
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.377608754759293,
      "learning_rate": 6.404450940120628e-06,
      "loss": 0.2971,
      "step": 2651
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.057232338254683,
      "learning_rate": 6.4019460871388055e-06,
      "loss": 0.4087,
      "step": 2652
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.180230786056468,
      "learning_rate": 6.399440852228226e-06,
      "loss": 0.4278,
      "step": 2653
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.53148459455616,
      "learning_rate": 6.396935236071381e-06,
      "loss": 0.4284,
      "step": 2654
    },
    {
      "epoch": 0.43,
      "grad_norm": 12.063471195723691,
      "learning_rate": 6.394429239350872e-06,
      "loss": 0.4528,
      "step": 2655
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.682416310124413,
      "learning_rate": 6.3919228627494e-06,
      "loss": 0.3967,
      "step": 2656
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.271346342014903,
      "learning_rate": 6.38941610694977e-06,
      "loss": 0.4389,
      "step": 2657
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.10739246140022,
      "learning_rate": 6.386908972634897e-06,
      "loss": 0.4173,
      "step": 2658
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.187665782777974,
      "learning_rate": 6.384401460487787e-06,
      "loss": 0.4579,
      "step": 2659
    },
    {
      "epoch": 0.43,
      "grad_norm": 32.623136563088636,
      "learning_rate": 6.381893571191558e-06,
      "loss": 0.4661,
      "step": 2660
    },
    {
      "epoch": 0.43,
      "grad_norm": 10.435900444826837,
      "learning_rate": 6.379385305429428e-06,
      "loss": 0.4249,
      "step": 2661
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.100213755462196,
      "learning_rate": 6.376876663884719e-06,
      "loss": 0.5064,
      "step": 2662
    },
    {
      "epoch": 0.43,
      "grad_norm": 13.144890622794215,
      "learning_rate": 6.374367647240851e-06,
      "loss": 0.4276,
      "step": 2663
    },
    {
      "epoch": 0.43,
      "grad_norm": 27.716935599979614,
      "learning_rate": 6.371858256181352e-06,
      "loss": 0.534,
      "step": 2664
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.638379644538953,
      "learning_rate": 6.3693484913898494e-06,
      "loss": 0.4457,
      "step": 2665
    },
    {
      "epoch": 0.43,
      "grad_norm": 12.346828948137913,
      "learning_rate": 6.36683835355007e-06,
      "loss": 0.4207,
      "step": 2666
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.227501003418883,
      "learning_rate": 6.364327843345847e-06,
      "loss": 0.4385,
      "step": 2667
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.846696581638323,
      "learning_rate": 6.361816961461111e-06,
      "loss": 0.4229,
      "step": 2668
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.291758538528957,
      "learning_rate": 6.359305708579897e-06,
      "loss": 0.3588,
      "step": 2669
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.165142426568481,
      "learning_rate": 6.356794085386337e-06,
      "loss": 0.5033,
      "step": 2670
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.900133208792257,
      "learning_rate": 6.3542820925646696e-06,
      "loss": 0.3729,
      "step": 2671
    },
    {
      "epoch": 0.43,
      "grad_norm": 9.002292295083949,
      "learning_rate": 6.351769730799227e-06,
      "loss": 0.4564,
      "step": 2672
    },
    {
      "epoch": 0.43,
      "grad_norm": 11.306790240773466,
      "learning_rate": 6.349257000774452e-06,
      "loss": 0.4124,
      "step": 2673
    },
    {
      "epoch": 0.43,
      "grad_norm": 12.838228191538686,
      "learning_rate": 6.346743903174872e-06,
      "loss": 0.4436,
      "step": 2674
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.086842403856586,
      "learning_rate": 6.344230438685134e-06,
      "loss": 0.4187,
      "step": 2675
    },
    {
      "epoch": 0.43,
      "grad_norm": 33.80261339147973,
      "learning_rate": 6.3417166079899685e-06,
      "loss": 0.5015,
      "step": 2676
    },
    {
      "epoch": 0.43,
      "grad_norm": 9.243025034241391,
      "learning_rate": 6.339202411774215e-06,
      "loss": 0.4635,
      "step": 2677
    },
    {
      "epoch": 0.43,
      "grad_norm": 12.520925800838363,
      "learning_rate": 6.336687850722809e-06,
      "loss": 0.4702,
      "step": 2678
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.64152819211246,
      "learning_rate": 6.334172925520785e-06,
      "loss": 0.5261,
      "step": 2679
    },
    {
      "epoch": 0.43,
      "grad_norm": 15.415014487141626,
      "learning_rate": 6.3316576368532814e-06,
      "loss": 0.3912,
      "step": 2680
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.443083752000213,
      "learning_rate": 6.329141985405529e-06,
      "loss": 0.4338,
      "step": 2681
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.554378337447598,
      "learning_rate": 6.326625971862863e-06,
      "loss": 0.496,
      "step": 2682
    },
    {
      "epoch": 0.43,
      "grad_norm": 9.524150112189412,
      "learning_rate": 6.324109596910713e-06,
      "loss": 0.4937,
      "step": 2683
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.382812946871318,
      "learning_rate": 6.32159286123461e-06,
      "loss": 0.4313,
      "step": 2684
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.300908206372447,
      "learning_rate": 6.319075765520179e-06,
      "loss": 0.5239,
      "step": 2685
    },
    {
      "epoch": 0.43,
      "grad_norm": 11.120138264827535,
      "learning_rate": 6.316558310453153e-06,
      "loss": 0.4447,
      "step": 2686
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.554352031545959,
      "learning_rate": 6.314040496719349e-06,
      "loss": 0.3973,
      "step": 2687
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.3209089390675555,
      "learning_rate": 6.3115223250046934e-06,
      "loss": 0.4189,
      "step": 2688
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.634814001048279,
      "learning_rate": 6.309003795995205e-06,
      "loss": 0.4132,
      "step": 2689
    },
    {
      "epoch": 0.43,
      "grad_norm": 23.540451242394347,
      "learning_rate": 6.306484910376998e-06,
      "loss": 0.4192,
      "step": 2690
    },
    {
      "epoch": 0.43,
      "grad_norm": 75.74703327065372,
      "learning_rate": 6.303965668836288e-06,
      "loss": 0.3974,
      "step": 2691
    },
    {
      "epoch": 0.43,
      "grad_norm": 13.160979515850597,
      "learning_rate": 6.301446072059386e-06,
      "loss": 0.4044,
      "step": 2692
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.465364088756202,
      "learning_rate": 6.2989261207327e-06,
      "loss": 0.4867,
      "step": 2693
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.125097700183161,
      "learning_rate": 6.296405815542732e-06,
      "loss": 0.5052,
      "step": 2694
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.200506300280933,
      "learning_rate": 6.293885157176087e-06,
      "loss": 0.4071,
      "step": 2695
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.073838043357562,
      "learning_rate": 6.2913641463194566e-06,
      "loss": 0.456,
      "step": 2696
    },
    {
      "epoch": 0.43,
      "grad_norm": 9.10277824690059,
      "learning_rate": 6.288842783659638e-06,
      "loss": 0.5271,
      "step": 2697
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.737442402847915,
      "learning_rate": 6.286321069883517e-06,
      "loss": 0.488,
      "step": 2698
    },
    {
      "epoch": 0.43,
      "grad_norm": 9.246804857140315,
      "learning_rate": 6.283799005678077e-06,
      "loss": 0.4591,
      "step": 2699
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.925226873328308,
      "learning_rate": 6.281276591730403e-06,
      "loss": 0.5257,
      "step": 2700
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.64727114832731,
      "learning_rate": 6.278753828727664e-06,
      "loss": 0.4179,
      "step": 2701
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.680886340056413,
      "learning_rate": 6.276230717357136e-06,
      "loss": 0.4628,
      "step": 2702
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.50793460908045,
      "learning_rate": 6.273707258306178e-06,
      "loss": 0.4515,
      "step": 2703
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.796332425892704,
      "learning_rate": 6.271183452262255e-06,
      "loss": 0.5469,
      "step": 2704
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.328862873918428,
      "learning_rate": 6.268659299912918e-06,
      "loss": 0.439,
      "step": 2705
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.438906563087814,
      "learning_rate": 6.266134801945819e-06,
      "loss": 0.4875,
      "step": 2706
    },
    {
      "epoch": 0.44,
      "grad_norm": 20.978818076807062,
      "learning_rate": 6.263609959048696e-06,
      "loss": 0.4287,
      "step": 2707
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.6353419296450085,
      "learning_rate": 6.261084771909391e-06,
      "loss": 0.4785,
      "step": 2708
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.182548597994957,
      "learning_rate": 6.25855924121583e-06,
      "loss": 0.4581,
      "step": 2709
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.588111544800425,
      "learning_rate": 6.25603336765604e-06,
      "loss": 0.4956,
      "step": 2710
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.397898810042977,
      "learning_rate": 6.2535071519181385e-06,
      "loss": 0.4208,
      "step": 2711
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.482741719503859,
      "learning_rate": 6.250980594690335e-06,
      "loss": 0.4457,
      "step": 2712
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.50662115500216,
      "learning_rate": 6.248453696660934e-06,
      "loss": 0.4734,
      "step": 2713
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.04578343209705,
      "learning_rate": 6.245926458518333e-06,
      "loss": 0.5028,
      "step": 2714
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.830974507809927,
      "learning_rate": 6.243398880951019e-06,
      "loss": 0.4,
      "step": 2715
    },
    {
      "epoch": 0.44,
      "grad_norm": 10.052135634108433,
      "learning_rate": 6.240870964647577e-06,
      "loss": 0.4671,
      "step": 2716
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.808063612110583,
      "learning_rate": 6.2383427102966786e-06,
      "loss": 0.4894,
      "step": 2717
    },
    {
      "epoch": 0.44,
      "grad_norm": 15.91788038970479,
      "learning_rate": 6.2358141185870915e-06,
      "loss": 0.4024,
      "step": 2718
    },
    {
      "epoch": 0.44,
      "grad_norm": 13.861177998389012,
      "learning_rate": 6.233285190207673e-06,
      "loss": 0.3856,
      "step": 2719
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.701407921028038,
      "learning_rate": 6.2307559258473716e-06,
      "loss": 0.4649,
      "step": 2720
    },
    {
      "epoch": 0.44,
      "grad_norm": 10.36373193330169,
      "learning_rate": 6.228226326195232e-06,
      "loss": 0.51,
      "step": 2721
    },
    {
      "epoch": 0.44,
      "grad_norm": 9.405732683359846,
      "learning_rate": 6.225696391940383e-06,
      "loss": 0.4732,
      "step": 2722
    },
    {
      "epoch": 0.44,
      "grad_norm": 10.19915027692201,
      "learning_rate": 6.223166123772051e-06,
      "loss": 0.4777,
      "step": 2723
    },
    {
      "epoch": 0.44,
      "grad_norm": 20.878948203950827,
      "learning_rate": 6.220635522379551e-06,
      "loss": 0.5122,
      "step": 2724
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2779286199929991,
      "learning_rate": 6.2181045884522876e-06,
      "loss": 0.4779,
      "step": 2725
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.99472682464385,
      "learning_rate": 6.215573322679756e-06,
      "loss": 0.4154,
      "step": 2726
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.570595288035781,
      "learning_rate": 6.213041725751543e-06,
      "loss": 0.4424,
      "step": 2727
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.810825507530346,
      "learning_rate": 6.210509798357328e-06,
      "loss": 0.4294,
      "step": 2728
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.226577038235522,
      "learning_rate": 6.207977541186876e-06,
      "loss": 0.4348,
      "step": 2729
    },
    {
      "epoch": 0.44,
      "grad_norm": 21.599783684133605,
      "learning_rate": 6.205444954930043e-06,
      "loss": 0.4388,
      "step": 2730
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.0821169438375176,
      "learning_rate": 6.2029120402767765e-06,
      "loss": 0.3968,
      "step": 2731
    },
    {
      "epoch": 0.44,
      "grad_norm": 10.354288564491162,
      "learning_rate": 6.2003787979171105e-06,
      "loss": 0.4482,
      "step": 2732
    },
    {
      "epoch": 0.44,
      "grad_norm": 18.65268544986505,
      "learning_rate": 6.197845228541174e-06,
      "loss": 0.4291,
      "step": 2733
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.293287355011215,
      "learning_rate": 6.195311332839175e-06,
      "loss": 0.3855,
      "step": 2734
    },
    {
      "epoch": 0.44,
      "grad_norm": 9.126346477637044,
      "learning_rate": 6.192777111501422e-06,
      "loss": 0.464,
      "step": 2735
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.755672787511041,
      "learning_rate": 6.190242565218305e-06,
      "loss": 0.4329,
      "step": 2736
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.576648308887396,
      "learning_rate": 6.187707694680302e-06,
      "loss": 0.4338,
      "step": 2737
    },
    {
      "epoch": 0.44,
      "grad_norm": 9.046079700546041,
      "learning_rate": 6.185172500577986e-06,
      "loss": 0.4862,
      "step": 2738
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.162362047865712,
      "learning_rate": 6.182636983602009e-06,
      "loss": 0.4116,
      "step": 2739
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.153470463651396,
      "learning_rate": 6.180101144443121e-06,
      "loss": 0.4052,
      "step": 2740
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.072678562721228,
      "learning_rate": 6.17756498379215e-06,
      "loss": 0.4315,
      "step": 2741
    },
    {
      "epoch": 0.44,
      "grad_norm": 10.0204922821827,
      "learning_rate": 6.175028502340018e-06,
      "loss": 0.4718,
      "step": 2742
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.293669494645629,
      "learning_rate": 6.172491700777732e-06,
      "loss": 0.4177,
      "step": 2743
    },
    {
      "epoch": 0.44,
      "grad_norm": 12.653351949106673,
      "learning_rate": 6.169954579796387e-06,
      "loss": 0.453,
      "step": 2744
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.513040664021082,
      "learning_rate": 6.167417140087163e-06,
      "loss": 0.4009,
      "step": 2745
    },
    {
      "epoch": 0.44,
      "grad_norm": 11.099125822071976,
      "learning_rate": 6.164879382341331e-06,
      "loss": 0.4766,
      "step": 2746
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.429167475528589,
      "learning_rate": 6.162341307250246e-06,
      "loss": 0.4388,
      "step": 2747
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.987854435430918,
      "learning_rate": 6.159802915505347e-06,
      "loss": 0.4226,
      "step": 2748
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.607371448649292,
      "learning_rate": 6.157264207798165e-06,
      "loss": 0.4565,
      "step": 2749
    },
    {
      "epoch": 0.44,
      "grad_norm": 13.125636759146166,
      "learning_rate": 6.154725184820311e-06,
      "loss": 0.4975,
      "step": 2750
    },
    {
      "epoch": 0.44,
      "grad_norm": 17.391819986906896,
      "learning_rate": 6.15218584726349e-06,
      "loss": 0.5107,
      "step": 2751
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.271319401004447,
      "learning_rate": 6.149646195819481e-06,
      "loss": 0.3739,
      "step": 2752
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.362505493060931,
      "learning_rate": 6.147106231180159e-06,
      "loss": 0.4359,
      "step": 2753
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.547927449821129,
      "learning_rate": 6.144565954037479e-06,
      "loss": 0.4066,
      "step": 2754
    },
    {
      "epoch": 0.44,
      "grad_norm": 9.648426213645745,
      "learning_rate": 6.142025365083484e-06,
      "loss": 0.4349,
      "step": 2755
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.589219827220292,
      "learning_rate": 6.139484465010298e-06,
      "loss": 0.4896,
      "step": 2756
    },
    {
      "epoch": 0.44,
      "grad_norm": 9.825206933124118,
      "learning_rate": 6.136943254510135e-06,
      "loss": 0.4565,
      "step": 2757
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.640455220267073,
      "learning_rate": 6.134401734275288e-06,
      "loss": 0.4367,
      "step": 2758
    },
    {
      "epoch": 0.44,
      "grad_norm": 53.51769463223771,
      "learning_rate": 6.13185990499814e-06,
      "loss": 0.4198,
      "step": 2759
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.662277016620053,
      "learning_rate": 6.129317767371153e-06,
      "loss": 0.4572,
      "step": 2760
    },
    {
      "epoch": 0.44,
      "grad_norm": 13.15460624286685,
      "learning_rate": 6.126775322086876e-06,
      "loss": 0.4426,
      "step": 2761
    },
    {
      "epoch": 0.45,
      "grad_norm": 25.7354467811583,
      "learning_rate": 6.124232569837943e-06,
      "loss": 0.4108,
      "step": 2762
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.474621270538083,
      "learning_rate": 6.121689511317068e-06,
      "loss": 0.4513,
      "step": 2763
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.605291597863904,
      "learning_rate": 6.11914614721705e-06,
      "loss": 0.3929,
      "step": 2764
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.8261324061656605,
      "learning_rate": 6.116602478230772e-06,
      "loss": 0.372,
      "step": 2765
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.179267370371003,
      "learning_rate": 6.1140585050512e-06,
      "loss": 0.4222,
      "step": 2766
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.541338685622392,
      "learning_rate": 6.111514228371381e-06,
      "loss": 0.4386,
      "step": 2767
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2127581313492073,
      "learning_rate": 6.108969648884449e-06,
      "loss": 0.4427,
      "step": 2768
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.337868861712894,
      "learning_rate": 6.106424767283615e-06,
      "loss": 0.4009,
      "step": 2769
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2194106711466417,
      "learning_rate": 6.103879584262176e-06,
      "loss": 0.5144,
      "step": 2770
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.565767684952268,
      "learning_rate": 6.101334100513508e-06,
      "loss": 0.4337,
      "step": 2771
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.542000193867979,
      "learning_rate": 6.098788316731074e-06,
      "loss": 0.446,
      "step": 2772
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.130358310296823,
      "learning_rate": 6.096242233608414e-06,
      "loss": 0.4843,
      "step": 2773
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.2442525529707815,
      "learning_rate": 6.0936958518391505e-06,
      "loss": 0.4588,
      "step": 2774
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0900471694512461,
      "learning_rate": 6.0911491721169906e-06,
      "loss": 0.4631,
      "step": 2775
    },
    {
      "epoch": 0.45,
      "grad_norm": 13.842974242516187,
      "learning_rate": 6.088602195135717e-06,
      "loss": 0.3954,
      "step": 2776
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.958940559790359,
      "learning_rate": 6.086054921589198e-06,
      "loss": 0.435,
      "step": 2777
    },
    {
      "epoch": 0.45,
      "grad_norm": 10.32351654385405,
      "learning_rate": 6.083507352171382e-06,
      "loss": 0.4234,
      "step": 2778
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.094676318776658,
      "learning_rate": 6.080959487576297e-06,
      "loss": 0.3613,
      "step": 2779
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.954262361918845,
      "learning_rate": 6.078411328498049e-06,
      "loss": 0.4213,
      "step": 2780
    },
    {
      "epoch": 0.45,
      "grad_norm": 9.253297250451642,
      "learning_rate": 6.075862875630829e-06,
      "loss": 0.3958,
      "step": 2781
    },
    {
      "epoch": 0.45,
      "grad_norm": 16.426067194496017,
      "learning_rate": 6.073314129668907e-06,
      "loss": 0.4401,
      "step": 2782
    },
    {
      "epoch": 0.45,
      "grad_norm": 10.143026139236158,
      "learning_rate": 6.070765091306628e-06,
      "loss": 0.348,
      "step": 2783
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.466927515378565,
      "learning_rate": 6.068215761238423e-06,
      "loss": 0.4589,
      "step": 2784
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.634866082306,
      "learning_rate": 6.065666140158799e-06,
      "loss": 0.4574,
      "step": 2785
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.14251291095844,
      "learning_rate": 6.063116228762343e-06,
      "loss": 0.4704,
      "step": 2786
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2131468260960865,
      "learning_rate": 6.060566027743721e-06,
      "loss": 0.4954,
      "step": 2787
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.061025459156463,
      "learning_rate": 6.058015537797678e-06,
      "loss": 0.4638,
      "step": 2788
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.719036647159259,
      "learning_rate": 6.055464759619038e-06,
      "loss": 0.4568,
      "step": 2789
    },
    {
      "epoch": 0.45,
      "grad_norm": 28.169414343126586,
      "learning_rate": 6.052913693902706e-06,
      "loss": 0.4388,
      "step": 2790
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.163086036479778,
      "learning_rate": 6.050362341343658e-06,
      "loss": 0.4874,
      "step": 2791
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.635998107907762,
      "learning_rate": 6.047810702636956e-06,
      "loss": 0.4034,
      "step": 2792
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.52263108529722,
      "learning_rate": 6.045258778477735e-06,
      "loss": 0.4485,
      "step": 2793
    },
    {
      "epoch": 0.45,
      "grad_norm": 10.815233012690973,
      "learning_rate": 6.042706569561209e-06,
      "loss": 0.4747,
      "step": 2794
    },
    {
      "epoch": 0.45,
      "grad_norm": 12.806976379810907,
      "learning_rate": 6.040154076582672e-06,
      "loss": 0.4517,
      "step": 2795
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.8556525925025324,
      "learning_rate": 6.0376013002374924e-06,
      "loss": 0.4327,
      "step": 2796
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.872191256594274,
      "learning_rate": 6.035048241221118e-06,
      "loss": 0.4623,
      "step": 2797
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.3028402658989595,
      "learning_rate": 6.032494900229069e-06,
      "loss": 0.4775,
      "step": 2798
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.25367019782884,
      "learning_rate": 6.02994127795695e-06,
      "loss": 0.4862,
      "step": 2799
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2613711018501945,
      "learning_rate": 6.027387375100435e-06,
      "loss": 0.4681,
      "step": 2800
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.200438816117388,
      "learning_rate": 6.024833192355282e-06,
      "loss": 0.4748,
      "step": 2801
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.743007219615023,
      "learning_rate": 6.022278730417315e-06,
      "loss": 0.4841,
      "step": 2802
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.580469415826265,
      "learning_rate": 6.019723989982444e-06,
      "loss": 0.4707,
      "step": 2803
    },
    {
      "epoch": 0.45,
      "grad_norm": 12.027823925669162,
      "learning_rate": 6.0171689717466485e-06,
      "loss": 0.3815,
      "step": 2804
    },
    {
      "epoch": 0.45,
      "grad_norm": 23.82916723001742,
      "learning_rate": 6.014613676405985e-06,
      "loss": 0.46,
      "step": 2805
    },
    {
      "epoch": 0.45,
      "grad_norm": 17.540163612248104,
      "learning_rate": 6.01205810465659e-06,
      "loss": 0.4498,
      "step": 2806
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.800024694125482,
      "learning_rate": 6.009502257194669e-06,
      "loss": 0.536,
      "step": 2807
    },
    {
      "epoch": 0.45,
      "grad_norm": 13.892549418889722,
      "learning_rate": 6.0069461347165084e-06,
      "loss": 0.4169,
      "step": 2808
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.167414890031099,
      "learning_rate": 6.0043897379184615e-06,
      "loss": 0.4655,
      "step": 2809
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2945591617177967,
      "learning_rate": 6.001833067496964e-06,
      "loss": 0.5077,
      "step": 2810
    },
    {
      "epoch": 0.45,
      "grad_norm": 10.678785135980714,
      "learning_rate": 5.999276124148525e-06,
      "loss": 0.4045,
      "step": 2811
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.987631285032045,
      "learning_rate": 5.996718908569723e-06,
      "loss": 0.449,
      "step": 2812
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2610627119641213,
      "learning_rate": 5.9941614214572155e-06,
      "loss": 0.4673,
      "step": 2813
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.231435960396342,
      "learning_rate": 5.9916036635077325e-06,
      "loss": 0.4143,
      "step": 2814
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0977516409968608,
      "learning_rate": 5.98904563541808e-06,
      "loss": 0.4611,
      "step": 2815
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.81995309989696,
      "learning_rate": 5.986487337885129e-06,
      "loss": 0.4216,
      "step": 2816
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.894054853774277,
      "learning_rate": 5.983928771605839e-06,
      "loss": 0.4138,
      "step": 2817
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.9660314031712085,
      "learning_rate": 5.981369937277226e-06,
      "loss": 0.4342,
      "step": 2818
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.189830888470938,
      "learning_rate": 5.978810835596392e-06,
      "loss": 0.3284,
      "step": 2819
    },
    {
      "epoch": 0.45,
      "grad_norm": 9.087744432325385,
      "learning_rate": 5.976251467260505e-06,
      "loss": 0.4443,
      "step": 2820
    },
    {
      "epoch": 0.45,
      "grad_norm": 11.22671710824284,
      "learning_rate": 5.9736918329668094e-06,
      "loss": 0.4531,
      "step": 2821
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.402020343149909,
      "learning_rate": 5.971131933412617e-06,
      "loss": 0.4035,
      "step": 2822
    },
    {
      "epoch": 0.45,
      "grad_norm": 13.782679246204324,
      "learning_rate": 5.9685717692953175e-06,
      "loss": 0.5087,
      "step": 2823
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.430976506867861,
      "learning_rate": 5.966011341312368e-06,
      "loss": 0.3814,
      "step": 2824
    },
    {
      "epoch": 0.46,
      "grad_norm": 9.313739175264386,
      "learning_rate": 5.9634506501613006e-06,
      "loss": 0.4834,
      "step": 2825
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.9398412071528055,
      "learning_rate": 5.960889696539719e-06,
      "loss": 0.4025,
      "step": 2826
    },
    {
      "epoch": 0.46,
      "grad_norm": 12.281238599008825,
      "learning_rate": 5.958328481145297e-06,
      "loss": 0.4679,
      "step": 2827
    },
    {
      "epoch": 0.46,
      "grad_norm": 12.82645596281723,
      "learning_rate": 5.955767004675781e-06,
      "loss": 0.4318,
      "step": 2828
    },
    {
      "epoch": 0.46,
      "grad_norm": 10.184481941672383,
      "learning_rate": 5.953205267828985e-06,
      "loss": 0.4374,
      "step": 2829
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.812200149574807,
      "learning_rate": 5.950643271302802e-06,
      "loss": 0.4843,
      "step": 2830
    },
    {
      "epoch": 0.46,
      "grad_norm": 14.671128623119284,
      "learning_rate": 5.948081015795184e-06,
      "loss": 0.4648,
      "step": 2831
    },
    {
      "epoch": 0.46,
      "grad_norm": 10.950178348593365,
      "learning_rate": 5.945518502004165e-06,
      "loss": 0.4395,
      "step": 2832
    },
    {
      "epoch": 0.46,
      "grad_norm": 32.68427806120232,
      "learning_rate": 5.942955730627843e-06,
      "loss": 0.4276,
      "step": 2833
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4308996498046926,
      "learning_rate": 5.940392702364387e-06,
      "loss": 0.4826,
      "step": 2834
    },
    {
      "epoch": 0.46,
      "grad_norm": 10.46250085150172,
      "learning_rate": 5.9378294179120355e-06,
      "loss": 0.4748,
      "step": 2835
    },
    {
      "epoch": 0.46,
      "grad_norm": 10.569268625495289,
      "learning_rate": 5.935265877969101e-06,
      "loss": 0.4563,
      "step": 2836
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2777380071835278,
      "learning_rate": 5.932702083233959e-06,
      "loss": 0.4499,
      "step": 2837
    },
    {
      "epoch": 0.46,
      "grad_norm": 9.16965836799274,
      "learning_rate": 5.9301380344050595e-06,
      "loss": 0.392,
      "step": 2838
    },
    {
      "epoch": 0.46,
      "grad_norm": 12.919826187130859,
      "learning_rate": 5.927573732180923e-06,
      "loss": 0.4558,
      "step": 2839
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.805789016615026,
      "learning_rate": 5.925009177260128e-06,
      "loss": 0.4045,
      "step": 2840
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.169642680146085,
      "learning_rate": 5.922444370341339e-06,
      "loss": 0.466,
      "step": 2841
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.236952089752237,
      "learning_rate": 5.919879312123276e-06,
      "loss": 0.5086,
      "step": 2842
    },
    {
      "epoch": 0.46,
      "grad_norm": 15.8182504017186,
      "learning_rate": 5.91731400330473e-06,
      "loss": 0.4862,
      "step": 2843
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.729008949170497,
      "learning_rate": 5.914748444584565e-06,
      "loss": 0.5057,
      "step": 2844
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.149287078719682,
      "learning_rate": 5.912182636661707e-06,
      "loss": 0.49,
      "step": 2845
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.07467186332408,
      "learning_rate": 5.9096165802351555e-06,
      "loss": 0.5099,
      "step": 2846
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.040016582505954,
      "learning_rate": 5.907050276003974e-06,
      "loss": 0.4662,
      "step": 2847
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.104709325564337,
      "learning_rate": 5.904483724667294e-06,
      "loss": 0.4592,
      "step": 2848
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.122872684250046,
      "learning_rate": 5.901916926924314e-06,
      "loss": 0.4565,
      "step": 2849
    },
    {
      "epoch": 0.46,
      "grad_norm": 11.744291334932427,
      "learning_rate": 5.899349883474303e-06,
      "loss": 0.4251,
      "step": 2850
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.642401732751702,
      "learning_rate": 5.896782595016594e-06,
      "loss": 0.4696,
      "step": 2851
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.170065178320583,
      "learning_rate": 5.894215062250584e-06,
      "loss": 0.4063,
      "step": 2852
    },
    {
      "epoch": 0.46,
      "grad_norm": 9.699551418636885,
      "learning_rate": 5.891647285875742e-06,
      "loss": 0.4424,
      "step": 2853
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.752464275413466,
      "learning_rate": 5.889079266591602e-06,
      "loss": 0.4652,
      "step": 2854
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.291017194480416,
      "learning_rate": 5.886511005097762e-06,
      "loss": 0.5129,
      "step": 2855
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.123614591877015,
      "learning_rate": 5.883942502093887e-06,
      "loss": 0.5278,
      "step": 2856
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.322261647140844,
      "learning_rate": 5.881373758279709e-06,
      "loss": 0.4241,
      "step": 2857
    },
    {
      "epoch": 0.46,
      "grad_norm": 20.788485005371616,
      "learning_rate": 5.878804774355024e-06,
      "loss": 0.4148,
      "step": 2858
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.268956665265226,
      "learning_rate": 5.876235551019692e-06,
      "loss": 0.4414,
      "step": 2859
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.765342971635018,
      "learning_rate": 5.873666088973644e-06,
      "loss": 0.4584,
      "step": 2860
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.420105936739947,
      "learning_rate": 5.871096388916873e-06,
      "loss": 0.3963,
      "step": 2861
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.761903389367417,
      "learning_rate": 5.86852645154943e-06,
      "loss": 0.4693,
      "step": 2862
    },
    {
      "epoch": 0.46,
      "grad_norm": 12.312109800182375,
      "learning_rate": 5.865956277571441e-06,
      "loss": 0.4683,
      "step": 2863
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4145967125193315,
      "learning_rate": 5.863385867683093e-06,
      "loss": 0.5107,
      "step": 2864
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.776676672370537,
      "learning_rate": 5.860815222584635e-06,
      "loss": 0.4135,
      "step": 2865
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.701685739464139,
      "learning_rate": 5.858244342976383e-06,
      "loss": 0.3583,
      "step": 2866
    },
    {
      "epoch": 0.46,
      "grad_norm": 15.644562450203866,
      "learning_rate": 5.855673229558711e-06,
      "loss": 0.4483,
      "step": 2867
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.9802490518256715,
      "learning_rate": 5.853101883032069e-06,
      "loss": 0.416,
      "step": 2868
    },
    {
      "epoch": 0.46,
      "grad_norm": 14.950343144506773,
      "learning_rate": 5.850530304096956e-06,
      "loss": 0.386,
      "step": 2869
    },
    {
      "epoch": 0.46,
      "grad_norm": 31.015490967594783,
      "learning_rate": 5.847958493453941e-06,
      "loss": 0.4989,
      "step": 2870
    },
    {
      "epoch": 0.46,
      "grad_norm": 10.549421739557433,
      "learning_rate": 5.845386451803661e-06,
      "loss": 0.4447,
      "step": 2871
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.092159702157703,
      "learning_rate": 5.8428141798468086e-06,
      "loss": 0.4635,
      "step": 2872
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.091907129987086,
      "learning_rate": 5.84024167828414e-06,
      "loss": 0.3979,
      "step": 2873
    },
    {
      "epoch": 0.46,
      "grad_norm": 11.033593001540805,
      "learning_rate": 5.837668947816476e-06,
      "loss": 0.4813,
      "step": 2874
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.914908888124439,
      "learning_rate": 5.835095989144701e-06,
      "loss": 0.3649,
      "step": 2875
    },
    {
      "epoch": 0.46,
      "grad_norm": 12.274884053073727,
      "learning_rate": 5.83252280296976e-06,
      "loss": 0.4233,
      "step": 2876
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.681302157000443,
      "learning_rate": 5.8299493899926584e-06,
      "loss": 0.475,
      "step": 2877
    },
    {
      "epoch": 0.46,
      "grad_norm": 26.132700690007344,
      "learning_rate": 5.827375750914463e-06,
      "loss": 0.4379,
      "step": 2878
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.36417889758148,
      "learning_rate": 5.82480188643631e-06,
      "loss": 0.3889,
      "step": 2879
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.466491623862069,
      "learning_rate": 5.822227797259385e-06,
      "loss": 0.411,
      "step": 2880
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.4873387330737335,
      "learning_rate": 5.819653484084944e-06,
      "loss": 0.4027,
      "step": 2881
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.181501333812945,
      "learning_rate": 5.817078947614299e-06,
      "loss": 0.4879,
      "step": 2882
    },
    {
      "epoch": 0.46,
      "grad_norm": 10.290536697678343,
      "learning_rate": 5.814504188548828e-06,
      "loss": 0.4249,
      "step": 2883
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.627272995061737,
      "learning_rate": 5.811929207589963e-06,
      "loss": 0.4942,
      "step": 2884
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.988461582905527,
      "learning_rate": 5.8093540054392e-06,
      "loss": 0.447,
      "step": 2885
    },
    {
      "epoch": 0.46,
      "grad_norm": 10.953579135640085,
      "learning_rate": 5.806778582798096e-06,
      "loss": 0.5234,
      "step": 2886
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.88206805539415,
      "learning_rate": 5.804202940368267e-06,
      "loss": 0.4478,
      "step": 2887
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.429469932863361,
      "learning_rate": 5.801627078851391e-06,
      "loss": 0.3615,
      "step": 2888
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2323457238322424,
      "learning_rate": 5.799050998949199e-06,
      "loss": 0.478,
      "step": 2889
    },
    {
      "epoch": 0.47,
      "grad_norm": 14.648519009943133,
      "learning_rate": 5.796474701363491e-06,
      "loss": 0.466,
      "step": 2890
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.445335338769459,
      "learning_rate": 5.793898186796118e-06,
      "loss": 0.4188,
      "step": 2891
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.236850876240105,
      "learning_rate": 5.791321455948996e-06,
      "loss": 0.4102,
      "step": 2892
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.832761301808484,
      "learning_rate": 5.7887445095240954e-06,
      "loss": 0.4922,
      "step": 2893
    },
    {
      "epoch": 0.47,
      "grad_norm": 10.172347927307735,
      "learning_rate": 5.78616734822345e-06,
      "loss": 0.3856,
      "step": 2894
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.431278726392184,
      "learning_rate": 5.783589972749149e-06,
      "loss": 0.4382,
      "step": 2895
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.116597379648586,
      "learning_rate": 5.781012383803337e-06,
      "loss": 0.4563,
      "step": 2896
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.3252607913528935,
      "learning_rate": 5.7784345820882235e-06,
      "loss": 0.4752,
      "step": 2897
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.270174263282255,
      "learning_rate": 5.775856568306073e-06,
      "loss": 0.4731,
      "step": 2898
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.81204897196919,
      "learning_rate": 5.773278343159207e-06,
      "loss": 0.4886,
      "step": 2899
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.930999153154023,
      "learning_rate": 5.770699907350003e-06,
      "loss": 0.3932,
      "step": 2900
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.10985791126475,
      "learning_rate": 5.768121261580901e-06,
      "loss": 0.4893,
      "step": 2901
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.709401416530688,
      "learning_rate": 5.765542406554394e-06,
      "loss": 0.4504,
      "step": 2902
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3077089412384026,
      "learning_rate": 5.762963342973033e-06,
      "loss": 0.4347,
      "step": 2903
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.711515445818277,
      "learning_rate": 5.7603840715394264e-06,
      "loss": 0.4567,
      "step": 2904
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.151365844661109,
      "learning_rate": 5.757804592956237e-06,
      "loss": 0.404,
      "step": 2905
    },
    {
      "epoch": 0.47,
      "grad_norm": 12.131252411719633,
      "learning_rate": 5.75522490792619e-06,
      "loss": 0.4141,
      "step": 2906
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.658191696170597,
      "learning_rate": 5.752645017152058e-06,
      "loss": 0.4419,
      "step": 2907
    },
    {
      "epoch": 0.47,
      "grad_norm": 17.68514828944344,
      "learning_rate": 5.750064921336679e-06,
      "loss": 0.4387,
      "step": 2908
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.394743276584741,
      "learning_rate": 5.7474846211829376e-06,
      "loss": 0.4258,
      "step": 2909
    },
    {
      "epoch": 0.47,
      "grad_norm": 9.617387018194139,
      "learning_rate": 5.7449041173937825e-06,
      "loss": 0.405,
      "step": 2910
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.621199161538355,
      "learning_rate": 5.7423234106722105e-06,
      "loss": 0.4324,
      "step": 2911
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.34297665671962,
      "learning_rate": 5.739742501721282e-06,
      "loss": 0.4806,
      "step": 2912
    },
    {
      "epoch": 0.47,
      "grad_norm": 9.068450920415767,
      "learning_rate": 5.7371613912441015e-06,
      "loss": 0.5048,
      "step": 2913
    },
    {
      "epoch": 0.47,
      "grad_norm": 16.90882216362115,
      "learning_rate": 5.73458007994384e-06,
      "loss": 0.3789,
      "step": 2914
    },
    {
      "epoch": 0.47,
      "grad_norm": 9.358641242146204,
      "learning_rate": 5.731998568523716e-06,
      "loss": 0.3909,
      "step": 2915
    },
    {
      "epoch": 0.47,
      "grad_norm": 16.704391680023594,
      "learning_rate": 5.729416857687002e-06,
      "loss": 0.422,
      "step": 2916
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.963909238366637,
      "learning_rate": 5.726834948137033e-06,
      "loss": 0.4857,
      "step": 2917
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5659095219892367,
      "learning_rate": 5.7242528405771865e-06,
      "loss": 0.5525,
      "step": 2918
    },
    {
      "epoch": 0.47,
      "grad_norm": 14.18755182042403,
      "learning_rate": 5.721670535710902e-06,
      "loss": 0.4452,
      "step": 2919
    },
    {
      "epoch": 0.47,
      "grad_norm": 9.48614156534668,
      "learning_rate": 5.7190880342416725e-06,
      "loss": 0.3858,
      "step": 2920
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.49574028786523,
      "learning_rate": 5.7165053368730395e-06,
      "loss": 0.4062,
      "step": 2921
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.321588271493018,
      "learning_rate": 5.713922444308601e-06,
      "loss": 0.4526,
      "step": 2922
    },
    {
      "epoch": 0.47,
      "grad_norm": 9.10499448458488,
      "learning_rate": 5.711339357252011e-06,
      "loss": 0.4447,
      "step": 2923
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.363324139045963,
      "learning_rate": 5.708756076406972e-06,
      "loss": 0.4746,
      "step": 2924
    },
    {
      "epoch": 0.47,
      "grad_norm": 10.449096613407512,
      "learning_rate": 5.706172602477239e-06,
      "loss": 0.4404,
      "step": 2925
    },
    {
      "epoch": 0.47,
      "grad_norm": 13.987779361774265,
      "learning_rate": 5.703588936166625e-06,
      "loss": 0.3972,
      "step": 2926
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.740607792649446,
      "learning_rate": 5.701005078178989e-06,
      "loss": 0.4737,
      "step": 2927
    },
    {
      "epoch": 0.47,
      "grad_norm": 11.688351859744449,
      "learning_rate": 5.6984210292182485e-06,
      "loss": 0.4796,
      "step": 2928
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2708897545039788,
      "learning_rate": 5.695836789988365e-06,
      "loss": 0.4463,
      "step": 2929
    },
    {
      "epoch": 0.47,
      "grad_norm": 16.99989661488079,
      "learning_rate": 5.6932523611933595e-06,
      "loss": 0.3786,
      "step": 2930
    },
    {
      "epoch": 0.47,
      "grad_norm": 16.356109425026787,
      "learning_rate": 5.690667743537303e-06,
      "loss": 0.4863,
      "step": 2931
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.828661865708355,
      "learning_rate": 5.688082937724314e-06,
      "loss": 0.3906,
      "step": 2932
    },
    {
      "epoch": 0.47,
      "grad_norm": 11.780843911172074,
      "learning_rate": 5.685497944458564e-06,
      "loss": 0.4064,
      "step": 2933
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.758196365513205,
      "learning_rate": 5.68291276444428e-06,
      "loss": 0.4923,
      "step": 2934
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.375931704056862,
      "learning_rate": 5.680327398385735e-06,
      "loss": 0.4395,
      "step": 2935
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.5601526619090285,
      "learning_rate": 5.677741846987251e-06,
      "loss": 0.3962,
      "step": 2936
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.676956289824686,
      "learning_rate": 5.675156110953208e-06,
      "loss": 0.4603,
      "step": 2937
    },
    {
      "epoch": 0.47,
      "grad_norm": 17.553858492834614,
      "learning_rate": 5.672570190988027e-06,
      "loss": 0.4169,
      "step": 2938
    },
    {
      "epoch": 0.47,
      "grad_norm": 9.359757256831715,
      "learning_rate": 5.6699840877961884e-06,
      "loss": 0.4424,
      "step": 2939
    },
    {
      "epoch": 0.47,
      "grad_norm": 16.198840893683663,
      "learning_rate": 5.667397802082216e-06,
      "loss": 0.4074,
      "step": 2940
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1697897420996894,
      "learning_rate": 5.664811334550685e-06,
      "loss": 0.4499,
      "step": 2941
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.06793548040542,
      "learning_rate": 5.6622246859062205e-06,
      "loss": 0.4581,
      "step": 2942
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.485420283804375,
      "learning_rate": 5.6596378568534984e-06,
      "loss": 0.419,
      "step": 2943
    },
    {
      "epoch": 0.47,
      "grad_norm": 9.178239157599464,
      "learning_rate": 5.657050848097239e-06,
      "loss": 0.507,
      "step": 2944
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.32537508477043,
      "learning_rate": 5.654463660342219e-06,
      "loss": 0.3751,
      "step": 2945
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.248006033912738,
      "learning_rate": 5.651876294293256e-06,
      "loss": 0.4202,
      "step": 2946
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.670341710839232,
      "learning_rate": 5.649288750655222e-06,
      "loss": 0.4305,
      "step": 2947
    },
    {
      "epoch": 0.47,
      "grad_norm": 12.987442142332512,
      "learning_rate": 5.646701030133036e-06,
      "loss": 0.5025,
      "step": 2948
    },
    {
      "epoch": 0.48,
      "grad_norm": 9.008117523130684,
      "learning_rate": 5.64411313343166e-06,
      "loss": 0.4673,
      "step": 2949
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.79781854836096,
      "learning_rate": 5.641525061256113e-06,
      "loss": 0.4763,
      "step": 2950
    },
    {
      "epoch": 0.48,
      "grad_norm": 15.434043102403942,
      "learning_rate": 5.638936814311455e-06,
      "loss": 0.4265,
      "step": 2951
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.089252023063847,
      "learning_rate": 5.636348393302798e-06,
      "loss": 0.4643,
      "step": 2952
    },
    {
      "epoch": 0.48,
      "grad_norm": 12.361343945671232,
      "learning_rate": 5.633759798935296e-06,
      "loss": 0.466,
      "step": 2953
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.433692321283031,
      "learning_rate": 5.631171031914155e-06,
      "loss": 0.4898,
      "step": 2954
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.032368745184664,
      "learning_rate": 5.628582092944628e-06,
      "loss": 0.4258,
      "step": 2955
    },
    {
      "epoch": 0.48,
      "grad_norm": 10.37540584069632,
      "learning_rate": 5.625992982732011e-06,
      "loss": 0.4598,
      "step": 2956
    },
    {
      "epoch": 0.48,
      "grad_norm": 14.697101069490158,
      "learning_rate": 5.62340370198165e-06,
      "loss": 0.4282,
      "step": 2957
    },
    {
      "epoch": 0.48,
      "grad_norm": 17.498334520330957,
      "learning_rate": 5.620814251398936e-06,
      "loss": 0.4655,
      "step": 2958
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.65715098955851,
      "learning_rate": 5.6182246316893086e-06,
      "loss": 0.4558,
      "step": 2959
    },
    {
      "epoch": 0.48,
      "grad_norm": 11.129480613435163,
      "learning_rate": 5.615634843558248e-06,
      "loss": 0.4371,
      "step": 2960
    },
    {
      "epoch": 0.48,
      "grad_norm": 9.920528352568887,
      "learning_rate": 5.613044887711288e-06,
      "loss": 0.512,
      "step": 2961
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.168696657055204,
      "learning_rate": 5.6104547648539995e-06,
      "loss": 0.4567,
      "step": 2962
    },
    {
      "epoch": 0.48,
      "grad_norm": 10.557198106467714,
      "learning_rate": 5.6078644756920066e-06,
      "loss": 0.4361,
      "step": 2963
    },
    {
      "epoch": 0.48,
      "grad_norm": 9.018610417012438,
      "learning_rate": 5.605274020930975e-06,
      "loss": 0.4895,
      "step": 2964
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.826050775154176,
      "learning_rate": 5.6026834012766155e-06,
      "loss": 0.4424,
      "step": 2965
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.632067543121904,
      "learning_rate": 5.600092617434686e-06,
      "loss": 0.4667,
      "step": 2966
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.021073553700118,
      "learning_rate": 5.597501670110984e-06,
      "loss": 0.4327,
      "step": 2967
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.961601368770387,
      "learning_rate": 5.594910560011356e-06,
      "loss": 0.4471,
      "step": 2968
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.931766059110284,
      "learning_rate": 5.592319287841694e-06,
      "loss": 0.4603,
      "step": 2969
    },
    {
      "epoch": 0.48,
      "grad_norm": 35.18790916719359,
      "learning_rate": 5.589727854307932e-06,
      "loss": 0.3948,
      "step": 2970
    },
    {
      "epoch": 0.48,
      "grad_norm": 9.87055063415602,
      "learning_rate": 5.587136260116044e-06,
      "loss": 0.452,
      "step": 2971
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.274843826008723,
      "learning_rate": 5.584544505972056e-06,
      "loss": 0.4703,
      "step": 2972
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.381430190343625,
      "learning_rate": 5.58195259258203e-06,
      "loss": 0.4381,
      "step": 2973
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.8722634607789574,
      "learning_rate": 5.5793605206520776e-06,
      "loss": 0.45,
      "step": 2974
    },
    {
      "epoch": 0.48,
      "grad_norm": 15.477269450620478,
      "learning_rate": 5.576768290888348e-06,
      "loss": 0.4874,
      "step": 2975
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.040556301846228,
      "learning_rate": 5.574175903997038e-06,
      "loss": 0.427,
      "step": 2976
    },
    {
      "epoch": 0.48,
      "grad_norm": 9.656252760195777,
      "learning_rate": 5.571583360684385e-06,
      "loss": 0.4457,
      "step": 2977
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.557056097627112,
      "learning_rate": 5.568990661656668e-06,
      "loss": 0.4641,
      "step": 2978
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.912199265099999,
      "learning_rate": 5.566397807620209e-06,
      "loss": 0.4103,
      "step": 2979
    },
    {
      "epoch": 0.48,
      "grad_norm": 9.184608369367421,
      "learning_rate": 5.563804799281377e-06,
      "loss": 0.427,
      "step": 2980
    },
    {
      "epoch": 0.48,
      "grad_norm": 10.747081923284373,
      "learning_rate": 5.561211637346576e-06,
      "loss": 0.4327,
      "step": 2981
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.023265003576396,
      "learning_rate": 5.558618322522253e-06,
      "loss": 0.4619,
      "step": 2982
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.174136054754001,
      "learning_rate": 5.556024855514904e-06,
      "loss": 0.466,
      "step": 2983
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.940541998218045,
      "learning_rate": 5.553431237031054e-06,
      "loss": 0.4957,
      "step": 2984
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1117140702800752,
      "learning_rate": 5.5508374677772806e-06,
      "loss": 0.4374,
      "step": 2985
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.582501620033825,
      "learning_rate": 5.548243548460197e-06,
      "loss": 0.3632,
      "step": 2986
    },
    {
      "epoch": 0.48,
      "grad_norm": 9.799756014417358,
      "learning_rate": 5.545649479786459e-06,
      "loss": 0.3867,
      "step": 2987
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.731656731676633,
      "learning_rate": 5.543055262462762e-06,
      "loss": 0.4448,
      "step": 2988
    },
    {
      "epoch": 0.48,
      "grad_norm": 16.02364390092112,
      "learning_rate": 5.540460897195842e-06,
      "loss": 0.3909,
      "step": 2989
    },
    {
      "epoch": 0.48,
      "grad_norm": 16.91295147146881,
      "learning_rate": 5.537866384692474e-06,
      "loss": 0.4956,
      "step": 2990
    },
    {
      "epoch": 0.48,
      "grad_norm": 12.81289120047385,
      "learning_rate": 5.535271725659478e-06,
      "loss": 0.4124,
      "step": 2991
    },
    {
      "epoch": 0.48,
      "grad_norm": 23.592945826172226,
      "learning_rate": 5.53267692080371e-06,
      "loss": 0.3873,
      "step": 2992
    },
    {
      "epoch": 0.48,
      "grad_norm": 11.572809079223754,
      "learning_rate": 5.530081970832063e-06,
      "loss": 0.4434,
      "step": 2993
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0955485569805417,
      "learning_rate": 5.527486876451478e-06,
      "loss": 0.4443,
      "step": 2994
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.502408749778991,
      "learning_rate": 5.524891638368926e-06,
      "loss": 0.3865,
      "step": 2995
    },
    {
      "epoch": 0.48,
      "grad_norm": 12.115212538984887,
      "learning_rate": 5.522296257291422e-06,
      "loss": 0.4154,
      "step": 2996
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.34387453399514,
      "learning_rate": 5.519700733926022e-06,
      "loss": 0.4124,
      "step": 2997
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1951200354123026,
      "learning_rate": 5.517105068979816e-06,
      "loss": 0.4475,
      "step": 2998
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.482576751341709,
      "learning_rate": 5.514509263159935e-06,
      "loss": 0.4882,
      "step": 2999
    },
    {
      "epoch": 0.48,
      "grad_norm": 13.784224124370374,
      "learning_rate": 5.511913317173548e-06,
      "loss": 0.4766,
      "step": 3000
    },
    {
      "epoch": 0.48,
      "grad_norm": 14.866964138657936,
      "learning_rate": 5.509317231727863e-06,
      "loss": 0.3741,
      "step": 3001
    },
    {
      "epoch": 0.48,
      "grad_norm": 13.241596807622788,
      "learning_rate": 5.506721007530125e-06,
      "loss": 0.4146,
      "step": 3002
    },
    {
      "epoch": 0.48,
      "grad_norm": 9.675312569336509,
      "learning_rate": 5.504124645287616e-06,
      "loss": 0.4436,
      "step": 3003
    },
    {
      "epoch": 0.48,
      "grad_norm": 10.597914457616575,
      "learning_rate": 5.501528145707657e-06,
      "loss": 0.4038,
      "step": 3004
    },
    {
      "epoch": 0.48,
      "grad_norm": 13.941102971490304,
      "learning_rate": 5.498931509497607e-06,
      "loss": 0.4448,
      "step": 3005
    },
    {
      "epoch": 0.48,
      "grad_norm": 11.385167178869475,
      "learning_rate": 5.4963347373648615e-06,
      "loss": 0.3787,
      "step": 3006
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.840377027988868,
      "learning_rate": 5.49373783001685e-06,
      "loss": 0.4158,
      "step": 3007
    },
    {
      "epoch": 0.48,
      "grad_norm": 11.529640099088152,
      "learning_rate": 5.4911407881610455e-06,
      "loss": 0.4644,
      "step": 3008
    },
    {
      "epoch": 0.48,
      "grad_norm": 13.437315859413028,
      "learning_rate": 5.48854361250495e-06,
      "loss": 0.4401,
      "step": 3009
    },
    {
      "epoch": 0.48,
      "grad_norm": 11.128950454580467,
      "learning_rate": 5.48594630375611e-06,
      "loss": 0.4439,
      "step": 3010
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.270697033204776,
      "learning_rate": 5.4833488626221e-06,
      "loss": 0.3898,
      "step": 3011
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3686987977185772,
      "learning_rate": 5.480751289810537e-06,
      "loss": 0.5186,
      "step": 3012
    },
    {
      "epoch": 0.49,
      "grad_norm": 23.322942602698525,
      "learning_rate": 5.478153586029069e-06,
      "loss": 0.4117,
      "step": 3013
    },
    {
      "epoch": 0.49,
      "grad_norm": 14.058488251684079,
      "learning_rate": 5.4755557519853854e-06,
      "loss": 0.4722,
      "step": 3014
    },
    {
      "epoch": 0.49,
      "grad_norm": 17.811941792675054,
      "learning_rate": 5.472957788387206e-06,
      "loss": 0.5039,
      "step": 3015
    },
    {
      "epoch": 0.49,
      "grad_norm": 10.13398004252201,
      "learning_rate": 5.470359695942289e-06,
      "loss": 0.4678,
      "step": 3016
    },
    {
      "epoch": 0.49,
      "grad_norm": 23.56169653992629,
      "learning_rate": 5.467761475358423e-06,
      "loss": 0.5041,
      "step": 3017
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.61304770512132,
      "learning_rate": 5.465163127343438e-06,
      "loss": 0.4243,
      "step": 3018
    },
    {
      "epoch": 0.49,
      "grad_norm": 15.369763040543411,
      "learning_rate": 5.462564652605197e-06,
      "loss": 0.4737,
      "step": 3019
    },
    {
      "epoch": 0.49,
      "grad_norm": 20.573736604923365,
      "learning_rate": 5.459966051851589e-06,
      "loss": 0.4131,
      "step": 3020
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.053407611857278,
      "learning_rate": 5.457367325790555e-06,
      "loss": 0.4773,
      "step": 3021
    },
    {
      "epoch": 0.49,
      "grad_norm": 28.38364267826306,
      "learning_rate": 5.454768475130051e-06,
      "loss": 0.4635,
      "step": 3022
    },
    {
      "epoch": 0.49,
      "grad_norm": 12.223334601524416,
      "learning_rate": 5.452169500578079e-06,
      "loss": 0.5018,
      "step": 3023
    },
    {
      "epoch": 0.49,
      "grad_norm": 13.375112871644935,
      "learning_rate": 5.449570402842671e-06,
      "loss": 0.4219,
      "step": 3024
    },
    {
      "epoch": 0.49,
      "grad_norm": 10.124627722668103,
      "learning_rate": 5.446971182631893e-06,
      "loss": 0.4771,
      "step": 3025
    },
    {
      "epoch": 0.49,
      "grad_norm": 10.9666111989232,
      "learning_rate": 5.444371840653843e-06,
      "loss": 0.4355,
      "step": 3026
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.370208222662076,
      "learning_rate": 5.441772377616656e-06,
      "loss": 0.3884,
      "step": 3027
    },
    {
      "epoch": 0.49,
      "grad_norm": 12.281123527674065,
      "learning_rate": 5.439172794228495e-06,
      "loss": 0.4449,
      "step": 3028
    },
    {
      "epoch": 0.49,
      "grad_norm": 21.200330023634223,
      "learning_rate": 5.436573091197559e-06,
      "loss": 0.4655,
      "step": 3029
    },
    {
      "epoch": 0.49,
      "grad_norm": 9.250873829510336,
      "learning_rate": 5.43397326923208e-06,
      "loss": 0.3907,
      "step": 3030
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.509768267684336,
      "learning_rate": 5.431373329040317e-06,
      "loss": 0.3898,
      "step": 3031
    },
    {
      "epoch": 0.49,
      "grad_norm": 17.878157133365526,
      "learning_rate": 5.42877327133057e-06,
      "loss": 0.368,
      "step": 3032
    },
    {
      "epoch": 0.49,
      "grad_norm": 10.406487336045982,
      "learning_rate": 5.426173096811166e-06,
      "loss": 0.4079,
      "step": 3033
    },
    {
      "epoch": 0.49,
      "grad_norm": 19.36970195794106,
      "learning_rate": 5.423572806190461e-06,
      "loss": 0.4702,
      "step": 3034
    },
    {
      "epoch": 0.49,
      "grad_norm": 10.87141146465951,
      "learning_rate": 5.420972400176848e-06,
      "loss": 0.4941,
      "step": 3035
    },
    {
      "epoch": 0.49,
      "grad_norm": 13.345501179078067,
      "learning_rate": 5.41837187947875e-06,
      "loss": 0.3892,
      "step": 3036
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2426044466837947,
      "learning_rate": 5.41577124480462e-06,
      "loss": 0.4935,
      "step": 3037
    },
    {
      "epoch": 0.49,
      "grad_norm": 19.217449550179943,
      "learning_rate": 5.413170496862941e-06,
      "loss": 0.4399,
      "step": 3038
    },
    {
      "epoch": 0.49,
      "grad_norm": 11.832878542802371,
      "learning_rate": 5.4105696363622284e-06,
      "loss": 0.4419,
      "step": 3039
    },
    {
      "epoch": 0.49,
      "grad_norm": 11.958381639233135,
      "learning_rate": 5.4079686640110305e-06,
      "loss": 0.5091,
      "step": 3040
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.560510296147061,
      "learning_rate": 5.405367580517924e-06,
      "loss": 0.4742,
      "step": 3041
    },
    {
      "epoch": 0.49,
      "grad_norm": 9.395271131743407,
      "learning_rate": 5.402766386591511e-06,
      "loss": 0.4024,
      "step": 3042
    },
    {
      "epoch": 0.49,
      "grad_norm": 12.1148959973674,
      "learning_rate": 5.400165082940436e-06,
      "loss": 0.4839,
      "step": 3043
    },
    {
      "epoch": 0.49,
      "grad_norm": 10.142652226675239,
      "learning_rate": 5.397563670273359e-06,
      "loss": 0.4505,
      "step": 3044
    },
    {
      "epoch": 0.49,
      "grad_norm": 10.318194045151992,
      "learning_rate": 5.394962149298981e-06,
      "loss": 0.391,
      "step": 3045
    },
    {
      "epoch": 0.49,
      "grad_norm": 10.723830422267888,
      "learning_rate": 5.392360520726027e-06,
      "loss": 0.4882,
      "step": 3046
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.861229940220509,
      "learning_rate": 5.38975878526325e-06,
      "loss": 0.4333,
      "step": 3047
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.537563545070602,
      "learning_rate": 5.387156943619437e-06,
      "loss": 0.4207,
      "step": 3048
    },
    {
      "epoch": 0.49,
      "grad_norm": 21.353567527013375,
      "learning_rate": 5.384554996503401e-06,
      "loss": 0.3535,
      "step": 3049
    },
    {
      "epoch": 0.49,
      "grad_norm": 10.014262155330695,
      "learning_rate": 5.381952944623982e-06,
      "loss": 0.463,
      "step": 3050
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.7390009527428445,
      "learning_rate": 5.3793507886900535e-06,
      "loss": 0.4599,
      "step": 3051
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.64359595059921,
      "learning_rate": 5.3767485294105135e-06,
      "loss": 0.4664,
      "step": 3052
    },
    {
      "epoch": 0.49,
      "grad_norm": 9.1799534673305,
      "learning_rate": 5.374146167494286e-06,
      "loss": 0.4351,
      "step": 3053
    },
    {
      "epoch": 0.49,
      "grad_norm": 10.454901262697621,
      "learning_rate": 5.371543703650332e-06,
      "loss": 0.4987,
      "step": 3054
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.55000203393371,
      "learning_rate": 5.368941138587631e-06,
      "loss": 0.403,
      "step": 3055
    },
    {
      "epoch": 0.49,
      "grad_norm": 13.627411002834808,
      "learning_rate": 5.366338473015195e-06,
      "loss": 0.4829,
      "step": 3056
    },
    {
      "epoch": 0.49,
      "grad_norm": 18.606728169076742,
      "learning_rate": 5.3637357076420595e-06,
      "loss": 0.4886,
      "step": 3057
    },
    {
      "epoch": 0.49,
      "grad_norm": 38.03049332390407,
      "learning_rate": 5.36113284317729e-06,
      "loss": 0.4336,
      "step": 3058
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.497077389874061,
      "learning_rate": 5.358529880329982e-06,
      "loss": 0.4337,
      "step": 3059
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.918706290947075,
      "learning_rate": 5.355926819809248e-06,
      "loss": 0.4792,
      "step": 3060
    },
    {
      "epoch": 0.49,
      "grad_norm": 11.572269335513441,
      "learning_rate": 5.3533236623242405e-06,
      "loss": 0.4102,
      "step": 3061
    },
    {
      "epoch": 0.49,
      "grad_norm": 11.66655156277395,
      "learning_rate": 5.3507204085841256e-06,
      "loss": 0.4247,
      "step": 3062
    },
    {
      "epoch": 0.49,
      "grad_norm": 37.46023585109764,
      "learning_rate": 5.348117059298105e-06,
      "loss": 0.4745,
      "step": 3063
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.796431070246502,
      "learning_rate": 5.345513615175401e-06,
      "loss": 0.4963,
      "step": 3064
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.654833553240898,
      "learning_rate": 5.342910076925264e-06,
      "loss": 0.4775,
      "step": 3065
    },
    {
      "epoch": 0.49,
      "grad_norm": 9.49501314515276,
      "learning_rate": 5.3403064452569685e-06,
      "loss": 0.4459,
      "step": 3066
    },
    {
      "epoch": 0.49,
      "grad_norm": 13.945065133961046,
      "learning_rate": 5.337702720879819e-06,
      "loss": 0.4763,
      "step": 3067
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2143253464473334,
      "learning_rate": 5.335098904503139e-06,
      "loss": 0.4355,
      "step": 3068
    },
    {
      "epoch": 0.49,
      "grad_norm": 23.330296694736557,
      "learning_rate": 5.332494996836279e-06,
      "loss": 0.4484,
      "step": 3069
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.034016080741656,
      "learning_rate": 5.3298909985886195e-06,
      "loss": 0.4151,
      "step": 3070
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.3388509752263325,
      "learning_rate": 5.327286910469556e-06,
      "loss": 0.4361,
      "step": 3071
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.718726783764271,
      "learning_rate": 5.3246827331885165e-06,
      "loss": 0.4702,
      "step": 3072
    },
    {
      "epoch": 0.5,
      "grad_norm": 12.712503879988855,
      "learning_rate": 5.322078467454951e-06,
      "loss": 0.4829,
      "step": 3073
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.396557953699374,
      "learning_rate": 5.3194741139783324e-06,
      "loss": 0.4828,
      "step": 3074
    },
    {
      "epoch": 0.5,
      "grad_norm": 13.203373318257308,
      "learning_rate": 5.316869673468159e-06,
      "loss": 0.4,
      "step": 3075
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.396371056975073,
      "learning_rate": 5.314265146633954e-06,
      "loss": 0.4386,
      "step": 3076
    },
    {
      "epoch": 0.5,
      "grad_norm": 16.062307824542277,
      "learning_rate": 5.311660534185258e-06,
      "loss": 0.4475,
      "step": 3077
    },
    {
      "epoch": 0.5,
      "grad_norm": 11.43292352997585,
      "learning_rate": 5.3090558368316415e-06,
      "loss": 0.4478,
      "step": 3078
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.203531688523813,
      "learning_rate": 5.306451055282696e-06,
      "loss": 0.4492,
      "step": 3079
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.434731630693129,
      "learning_rate": 5.303846190248035e-06,
      "loss": 0.4465,
      "step": 3080
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.947973468452653,
      "learning_rate": 5.301241242437299e-06,
      "loss": 0.4533,
      "step": 3081
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.648690446391614,
      "learning_rate": 5.298636212560143e-06,
      "loss": 0.4875,
      "step": 3082
    },
    {
      "epoch": 0.5,
      "grad_norm": 16.854401199781517,
      "learning_rate": 5.29603110132625e-06,
      "loss": 0.5023,
      "step": 3083
    },
    {
      "epoch": 0.5,
      "grad_norm": 10.596059872730613,
      "learning_rate": 5.293425909445326e-06,
      "loss": 0.4721,
      "step": 3084
    },
    {
      "epoch": 0.5,
      "grad_norm": 10.37634818342419,
      "learning_rate": 5.290820637627095e-06,
      "loss": 0.4238,
      "step": 3085
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.161830551808453,
      "learning_rate": 5.288215286581306e-06,
      "loss": 0.4947,
      "step": 3086
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.0823576865966595,
      "learning_rate": 5.28560985701773e-06,
      "loss": 0.4233,
      "step": 3087
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.035264724788157,
      "learning_rate": 5.283004349646154e-06,
      "loss": 0.5188,
      "step": 3088
    },
    {
      "epoch": 0.5,
      "grad_norm": 10.55183737701892,
      "learning_rate": 5.280398765176392e-06,
      "loss": 0.4296,
      "step": 3089
    },
    {
      "epoch": 0.5,
      "grad_norm": 25.923769655534414,
      "learning_rate": 5.277793104318279e-06,
      "loss": 0.4785,
      "step": 3090
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.188738216200228,
      "learning_rate": 5.275187367781664e-06,
      "loss": 0.3882,
      "step": 3091
    },
    {
      "epoch": 0.5,
      "grad_norm": 13.504779817597388,
      "learning_rate": 5.272581556276428e-06,
      "loss": 0.4568,
      "step": 3092
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.599343150321767,
      "learning_rate": 5.26997567051246e-06,
      "loss": 0.3954,
      "step": 3093
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.049054956329651,
      "learning_rate": 5.267369711199678e-06,
      "loss": 0.4691,
      "step": 3094
    },
    {
      "epoch": 0.5,
      "grad_norm": 12.996168132368977,
      "learning_rate": 5.2647636790480175e-06,
      "loss": 0.388,
      "step": 3095
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.758409096157251,
      "learning_rate": 5.26215757476743e-06,
      "loss": 0.5424,
      "step": 3096
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.7150779883685106,
      "learning_rate": 5.259551399067895e-06,
      "loss": 0.4333,
      "step": 3097
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.93152449971319,
      "learning_rate": 5.256945152659404e-06,
      "loss": 0.4084,
      "step": 3098
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.568517183805675,
      "learning_rate": 5.254338836251971e-06,
      "loss": 0.4379,
      "step": 3099
    },
    {
      "epoch": 0.5,
      "grad_norm": 15.160476729858633,
      "learning_rate": 5.251732450555626e-06,
      "loss": 0.4607,
      "step": 3100
    },
    {
      "epoch": 0.5,
      "grad_norm": 10.439811695894882,
      "learning_rate": 5.249125996280426e-06,
      "loss": 0.3981,
      "step": 3101
    },
    {
      "epoch": 0.5,
      "grad_norm": 10.999900837084816,
      "learning_rate": 5.246519474136433e-06,
      "loss": 0.4628,
      "step": 3102
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.908807735109394,
      "learning_rate": 5.243912884833744e-06,
      "loss": 0.4227,
      "step": 3103
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.385455930999868,
      "learning_rate": 5.241306229082459e-06,
      "loss": 0.4072,
      "step": 3104
    },
    {
      "epoch": 0.5,
      "grad_norm": 11.539621899923734,
      "learning_rate": 5.2386995075927074e-06,
      "loss": 0.4327,
      "step": 3105
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3781149802070196,
      "learning_rate": 5.236092721074629e-06,
      "loss": 0.4656,
      "step": 3106
    },
    {
      "epoch": 0.5,
      "grad_norm": 16.24410489516164,
      "learning_rate": 5.233485870238385e-06,
      "loss": 0.4578,
      "step": 3107
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.272384842875574,
      "learning_rate": 5.2308789557941546e-06,
      "loss": 0.4584,
      "step": 3108
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.662367351895382,
      "learning_rate": 5.228271978452133e-06,
      "loss": 0.4544,
      "step": 3109
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.956113864429475,
      "learning_rate": 5.225664938922531e-06,
      "loss": 0.372,
      "step": 3110
    },
    {
      "epoch": 0.5,
      "grad_norm": 28.737245347872076,
      "learning_rate": 5.223057837915579e-06,
      "loss": 0.3723,
      "step": 3111
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.27407958667095,
      "learning_rate": 5.220450676141525e-06,
      "loss": 0.3873,
      "step": 3112
    },
    {
      "epoch": 0.5,
      "grad_norm": 12.691293350587973,
      "learning_rate": 5.217843454310628e-06,
      "loss": 0.4636,
      "step": 3113
    },
    {
      "epoch": 0.5,
      "grad_norm": 14.328866310325898,
      "learning_rate": 5.215236173133172e-06,
      "loss": 0.4295,
      "step": 3114
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.358182658239808,
      "learning_rate": 5.2126288333194455e-06,
      "loss": 0.3988,
      "step": 3115
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.642091199174931,
      "learning_rate": 5.210021435579766e-06,
      "loss": 0.3678,
      "step": 3116
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.63133591111443,
      "learning_rate": 5.207413980624459e-06,
      "loss": 0.4575,
      "step": 3117
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.137888328029778,
      "learning_rate": 5.204806469163865e-06,
      "loss": 0.4647,
      "step": 3118
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.4078111362129,
      "learning_rate": 5.202198901908343e-06,
      "loss": 0.4801,
      "step": 3119
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.27187302544425,
      "learning_rate": 5.199591279568268e-06,
      "loss": 0.4316,
      "step": 3120
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.509805384648743,
      "learning_rate": 5.196983602854024e-06,
      "loss": 0.3958,
      "step": 3121
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.5720357602583395,
      "learning_rate": 5.1943758724760185e-06,
      "loss": 0.434,
      "step": 3122
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.90536422673787,
      "learning_rate": 5.191768089144668e-06,
      "loss": 0.4171,
      "step": 3123
    },
    {
      "epoch": 0.5,
      "grad_norm": 12.997410464897989,
      "learning_rate": 5.189160253570402e-06,
      "loss": 0.3666,
      "step": 3124
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.246359227639362,
      "learning_rate": 5.186552366463671e-06,
      "loss": 0.4337,
      "step": 3125
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1508179337381892,
      "learning_rate": 5.18394442853493e-06,
      "loss": 0.4209,
      "step": 3126
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.517050473227762,
      "learning_rate": 5.181336440494659e-06,
      "loss": 0.3948,
      "step": 3127
    },
    {
      "epoch": 0.5,
      "grad_norm": 10.528817289920582,
      "learning_rate": 5.178728403053342e-06,
      "loss": 0.5046,
      "step": 3128
    },
    {
      "epoch": 0.5,
      "grad_norm": 13.429959161940932,
      "learning_rate": 5.176120316921481e-06,
      "loss": 0.4848,
      "step": 3129
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.854073549522074,
      "learning_rate": 5.173512182809592e-06,
      "loss": 0.46,
      "step": 3130
    },
    {
      "epoch": 0.5,
      "grad_norm": 19.92011495366333,
      "learning_rate": 5.170904001428201e-06,
      "loss": 0.4434,
      "step": 3131
    },
    {
      "epoch": 0.5,
      "grad_norm": 10.055177815685655,
      "learning_rate": 5.168295773487848e-06,
      "loss": 0.398,
      "step": 3132
    },
    {
      "epoch": 0.5,
      "grad_norm": 16.203570404030096,
      "learning_rate": 5.165687499699088e-06,
      "loss": 0.4296,
      "step": 3133
    },
    {
      "epoch": 0.5,
      "grad_norm": 12.984099710108099,
      "learning_rate": 5.163079180772486e-06,
      "loss": 0.4771,
      "step": 3134
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.192463042597208,
      "learning_rate": 5.160470817418619e-06,
      "loss": 0.4163,
      "step": 3135
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.565442429016356,
      "learning_rate": 5.15786241034808e-06,
      "loss": 0.5303,
      "step": 3136
    },
    {
      "epoch": 0.51,
      "grad_norm": 9.283318236375548,
      "learning_rate": 5.155253960271466e-06,
      "loss": 0.5087,
      "step": 3137
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.8695097196008525,
      "learning_rate": 5.152645467899397e-06,
      "loss": 0.4028,
      "step": 3138
    },
    {
      "epoch": 0.51,
      "grad_norm": 74.1354057670772,
      "learning_rate": 5.150036933942492e-06,
      "loss": 0.5656,
      "step": 3139
    },
    {
      "epoch": 0.51,
      "grad_norm": 21.05956954268088,
      "learning_rate": 5.147428359111391e-06,
      "loss": 0.4414,
      "step": 3140
    },
    {
      "epoch": 0.51,
      "grad_norm": 9.431954873443466,
      "learning_rate": 5.144819744116742e-06,
      "loss": 0.4265,
      "step": 3141
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.134236783618586,
      "learning_rate": 5.142211089669202e-06,
      "loss": 0.4002,
      "step": 3142
    },
    {
      "epoch": 0.51,
      "grad_norm": 11.640266005607186,
      "learning_rate": 5.1396023964794415e-06,
      "loss": 0.4356,
      "step": 3143
    },
    {
      "epoch": 0.51,
      "grad_norm": 9.804853887556776,
      "learning_rate": 5.1369936652581395e-06,
      "loss": 0.372,
      "step": 3144
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.88973125690465,
      "learning_rate": 5.134384896715987e-06,
      "loss": 0.5036,
      "step": 3145
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.043407790242097,
      "learning_rate": 5.131776091563685e-06,
      "loss": 0.3968,
      "step": 3146
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.726585578535852,
      "learning_rate": 5.129167250511943e-06,
      "loss": 0.4159,
      "step": 3147
    },
    {
      "epoch": 0.51,
      "grad_norm": 11.748898531703654,
      "learning_rate": 5.12655837427148e-06,
      "loss": 0.4366,
      "step": 3148
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.845501547923442,
      "learning_rate": 5.123949463553029e-06,
      "loss": 0.421,
      "step": 3149
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.116760495223623,
      "learning_rate": 5.121340519067327e-06,
      "loss": 0.4696,
      "step": 3150
    },
    {
      "epoch": 0.51,
      "grad_norm": 13.160256279405793,
      "learning_rate": 5.118731541525123e-06,
      "loss": 0.3845,
      "step": 3151
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.445196954418788,
      "learning_rate": 5.116122531637174e-06,
      "loss": 0.47,
      "step": 3152
    },
    {
      "epoch": 0.51,
      "grad_norm": 20.01810408514814,
      "learning_rate": 5.113513490114246e-06,
      "loss": 0.5091,
      "step": 3153
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.551508561900697,
      "learning_rate": 5.110904417667115e-06,
      "loss": 0.4925,
      "step": 3154
    },
    {
      "epoch": 0.51,
      "grad_norm": 9.167580798866489,
      "learning_rate": 5.108295315006563e-06,
      "loss": 0.4855,
      "step": 3155
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.875021697997089,
      "learning_rate": 5.1056861828433815e-06,
      "loss": 0.4352,
      "step": 3156
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.938787625207812,
      "learning_rate": 5.1030770218883706e-06,
      "loss": 0.457,
      "step": 3157
    },
    {
      "epoch": 0.51,
      "grad_norm": 9.064501761238272,
      "learning_rate": 5.1004678328523395e-06,
      "loss": 0.43,
      "step": 3158
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.12411076819494,
      "learning_rate": 5.097858616446099e-06,
      "loss": 0.4569,
      "step": 3159
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.906620388322995,
      "learning_rate": 5.0952493733804755e-06,
      "loss": 0.4269,
      "step": 3160
    },
    {
      "epoch": 0.51,
      "grad_norm": 12.391253158649452,
      "learning_rate": 5.092640104366297e-06,
      "loss": 0.4277,
      "step": 3161
    },
    {
      "epoch": 0.51,
      "grad_norm": 9.32471092918946,
      "learning_rate": 5.0900308101143996e-06,
      "loss": 0.4397,
      "step": 3162
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.951285899296328,
      "learning_rate": 5.087421491335629e-06,
      "loss": 0.4734,
      "step": 3163
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.255599789091882,
      "learning_rate": 5.0848121487408316e-06,
      "loss": 0.4302,
      "step": 3164
    },
    {
      "epoch": 0.51,
      "grad_norm": 15.442434553740744,
      "learning_rate": 5.08220278304087e-06,
      "loss": 0.428,
      "step": 3165
    },
    {
      "epoch": 0.51,
      "grad_norm": 18.82601737410301,
      "learning_rate": 5.079593394946603e-06,
      "loss": 0.4098,
      "step": 3166
    },
    {
      "epoch": 0.51,
      "grad_norm": 9.84379319270009,
      "learning_rate": 5.076983985168901e-06,
      "loss": 0.4539,
      "step": 3167
    },
    {
      "epoch": 0.51,
      "grad_norm": 9.90659556642468,
      "learning_rate": 5.074374554418641e-06,
      "loss": 0.4554,
      "step": 3168
    },
    {
      "epoch": 0.51,
      "grad_norm": 11.397066710868136,
      "learning_rate": 5.071765103406702e-06,
      "loss": 0.4496,
      "step": 3169
    },
    {
      "epoch": 0.51,
      "grad_norm": 11.263879305209539,
      "learning_rate": 5.06915563284397e-06,
      "loss": 0.4647,
      "step": 3170
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.528724922727628,
      "learning_rate": 5.066546143441336e-06,
      "loss": 0.4351,
      "step": 3171
    },
    {
      "epoch": 0.51,
      "grad_norm": 37.54356147082925,
      "learning_rate": 5.0639366359097e-06,
      "loss": 0.4124,
      "step": 3172
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.665071544218575,
      "learning_rate": 5.061327110959961e-06,
      "loss": 0.4289,
      "step": 3173
    },
    {
      "epoch": 0.51,
      "grad_norm": 12.385387920719054,
      "learning_rate": 5.058717569303027e-06,
      "loss": 0.4503,
      "step": 3174
    },
    {
      "epoch": 0.51,
      "grad_norm": 19.15063352982195,
      "learning_rate": 5.056108011649807e-06,
      "loss": 0.4769,
      "step": 3175
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.640173531036795,
      "learning_rate": 5.053498438711221e-06,
      "loss": 0.4878,
      "step": 3176
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.199516133663327,
      "learning_rate": 5.050888851198183e-06,
      "loss": 0.4238,
      "step": 3177
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.582901980695095,
      "learning_rate": 5.0482792498216194e-06,
      "loss": 0.4724,
      "step": 3178
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.17863000147381,
      "learning_rate": 5.045669635292458e-06,
      "loss": 0.4744,
      "step": 3179
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.154005527197196,
      "learning_rate": 5.04306000832163e-06,
      "loss": 0.4127,
      "step": 3180
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.614642539802718,
      "learning_rate": 5.040450369620068e-06,
      "loss": 0.4076,
      "step": 3181
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.974342715537309,
      "learning_rate": 5.03784071989871e-06,
      "loss": 0.4516,
      "step": 3182
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.174299976122674,
      "learning_rate": 5.0352310598684965e-06,
      "loss": 0.3798,
      "step": 3183
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.3646213021939735,
      "learning_rate": 5.032621390240371e-06,
      "loss": 0.4956,
      "step": 3184
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.55489457133509,
      "learning_rate": 5.030011711725281e-06,
      "loss": 0.4494,
      "step": 3185
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.31283624235548,
      "learning_rate": 5.027402025034171e-06,
      "loss": 0.4488,
      "step": 3186
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.56269849761301,
      "learning_rate": 5.024792330877997e-06,
      "loss": 0.4759,
      "step": 3187
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.464980892181738,
      "learning_rate": 5.022182629967707e-06,
      "loss": 0.4537,
      "step": 3188
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.208495684185179,
      "learning_rate": 5.019572923014258e-06,
      "loss": 0.4574,
      "step": 3189
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.479168826865888,
      "learning_rate": 5.016963210728604e-06,
      "loss": 0.386,
      "step": 3190
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.699488101399249,
      "learning_rate": 5.014353493821705e-06,
      "loss": 0.354,
      "step": 3191
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.722540149369177,
      "learning_rate": 5.011743773004518e-06,
      "loss": 0.413,
      "step": 3192
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.01603451813438,
      "learning_rate": 5.009134048988004e-06,
      "loss": 0.4014,
      "step": 3193
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.4619289800786675,
      "learning_rate": 5.006524322483123e-06,
      "loss": 0.4164,
      "step": 3194
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.421089848127978,
      "learning_rate": 5.0039145942008364e-06,
      "loss": 0.4155,
      "step": 3195
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.74082298964966,
      "learning_rate": 5.001304864852106e-06,
      "loss": 0.5006,
      "step": 3196
    },
    {
      "epoch": 0.52,
      "grad_norm": 9.56813355396322,
      "learning_rate": 4.998695135147896e-06,
      "loss": 0.4417,
      "step": 3197
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.218033955418096,
      "learning_rate": 4.996085405799166e-06,
      "loss": 0.3978,
      "step": 3198
    },
    {
      "epoch": 0.52,
      "grad_norm": 10.238680888843067,
      "learning_rate": 4.993475677516878e-06,
      "loss": 0.4642,
      "step": 3199
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.769383476423492,
      "learning_rate": 4.990865951011999e-06,
      "loss": 0.4639,
      "step": 3200
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.7344999244144175,
      "learning_rate": 4.9882562269954835e-06,
      "loss": 0.4304,
      "step": 3201
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.818351026349519,
      "learning_rate": 4.985646506178296e-06,
      "loss": 0.448,
      "step": 3202
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.396901218275492,
      "learning_rate": 4.983036789271398e-06,
      "loss": 0.4126,
      "step": 3203
    },
    {
      "epoch": 0.52,
      "grad_norm": 10.332381905235522,
      "learning_rate": 4.980427076985744e-06,
      "loss": 0.4669,
      "step": 3204
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.552738535529104,
      "learning_rate": 4.977817370032294e-06,
      "loss": 0.3936,
      "step": 3205
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.903559338216413,
      "learning_rate": 4.975207669122005e-06,
      "loss": 0.484,
      "step": 3206
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.578625600194476,
      "learning_rate": 4.97259797496583e-06,
      "loss": 0.4146,
      "step": 3207
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.693772792008131,
      "learning_rate": 4.9699882882747205e-06,
      "loss": 0.4651,
      "step": 3208
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.5363770574200775,
      "learning_rate": 4.967378609759629e-06,
      "loss": 0.4936,
      "step": 3209
    },
    {
      "epoch": 0.52,
      "grad_norm": 11.037016400219448,
      "learning_rate": 4.964768940131505e-06,
      "loss": 0.4497,
      "step": 3210
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4669779357922395,
      "learning_rate": 4.962159280101292e-06,
      "loss": 0.4945,
      "step": 3211
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.912619470396417,
      "learning_rate": 4.959549630379934e-06,
      "loss": 0.4214,
      "step": 3212
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.19765756419675,
      "learning_rate": 4.9569399916783724e-06,
      "loss": 0.4388,
      "step": 3213
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.304027798950356,
      "learning_rate": 4.954330364707543e-06,
      "loss": 0.4678,
      "step": 3214
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.225864610431844,
      "learning_rate": 4.9517207501783805e-06,
      "loss": 0.4518,
      "step": 3215
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.962754802835571,
      "learning_rate": 4.949111148801819e-06,
      "loss": 0.4363,
      "step": 3216
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.640241543859871,
      "learning_rate": 4.9465015612887815e-06,
      "loss": 0.4657,
      "step": 3217
    },
    {
      "epoch": 0.52,
      "grad_norm": 9.589996735112647,
      "learning_rate": 4.9438919883501934e-06,
      "loss": 0.5225,
      "step": 3218
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.680593210614576,
      "learning_rate": 4.941282430696976e-06,
      "loss": 0.4799,
      "step": 3219
    },
    {
      "epoch": 0.52,
      "grad_norm": 10.615049125574775,
      "learning_rate": 4.938672889040041e-06,
      "loss": 0.425,
      "step": 3220
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.338346451566445,
      "learning_rate": 4.936063364090302e-06,
      "loss": 0.4185,
      "step": 3221
    },
    {
      "epoch": 0.52,
      "grad_norm": 9.947471769384006,
      "learning_rate": 4.933453856558666e-06,
      "loss": 0.4888,
      "step": 3222
    },
    {
      "epoch": 0.52,
      "grad_norm": 9.263995730780012,
      "learning_rate": 4.930844367156032e-06,
      "loss": 0.4518,
      "step": 3223
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5495191993647182,
      "learning_rate": 4.9282348965933e-06,
      "loss": 0.5129,
      "step": 3224
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2757601222250934,
      "learning_rate": 4.925625445581361e-06,
      "loss": 0.4801,
      "step": 3225
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.251884362999813,
      "learning_rate": 4.923016014831099e-06,
      "loss": 0.3793,
      "step": 3226
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.7464754682644035,
      "learning_rate": 4.9204066050533975e-06,
      "loss": 0.3799,
      "step": 3227
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.4799308269106435,
      "learning_rate": 4.91779721695913e-06,
      "loss": 0.3796,
      "step": 3228
    },
    {
      "epoch": 0.52,
      "grad_norm": 9.847177443266865,
      "learning_rate": 4.915187851259169e-06,
      "loss": 0.3261,
      "step": 3229
    },
    {
      "epoch": 0.52,
      "grad_norm": 16.55318391205063,
      "learning_rate": 4.9125785086643726e-06,
      "loss": 0.3903,
      "step": 3230
    },
    {
      "epoch": 0.52,
      "grad_norm": 10.6058240412937,
      "learning_rate": 4.9099691898856e-06,
      "loss": 0.4462,
      "step": 3231
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0516805345376652,
      "learning_rate": 4.907359895633705e-06,
      "loss": 0.4591,
      "step": 3232
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.922762037333175,
      "learning_rate": 4.904750626619525e-06,
      "loss": 0.4699,
      "step": 3233
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.736964785025565,
      "learning_rate": 4.902141383553901e-06,
      "loss": 0.4723,
      "step": 3234
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.374462322562105,
      "learning_rate": 4.899532167147662e-06,
      "loss": 0.3979,
      "step": 3235
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.349469302762756,
      "learning_rate": 4.89692297811163e-06,
      "loss": 0.4259,
      "step": 3236
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.790210302078131,
      "learning_rate": 4.8943138171566184e-06,
      "loss": 0.4178,
      "step": 3237
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.993080409280082,
      "learning_rate": 4.89170468499344e-06,
      "loss": 0.4443,
      "step": 3238
    },
    {
      "epoch": 0.52,
      "grad_norm": 9.843376587122243,
      "learning_rate": 4.889095582332887e-06,
      "loss": 0.4894,
      "step": 3239
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.682398306055466,
      "learning_rate": 4.886486509885755e-06,
      "loss": 0.3884,
      "step": 3240
    },
    {
      "epoch": 0.52,
      "grad_norm": 10.548234451592345,
      "learning_rate": 4.883877468362828e-06,
      "loss": 0.4769,
      "step": 3241
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.027812752610894,
      "learning_rate": 4.88126845847488e-06,
      "loss": 0.4136,
      "step": 3242
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.656706476275129,
      "learning_rate": 4.878659480932674e-06,
      "loss": 0.3864,
      "step": 3243
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.234740228312786,
      "learning_rate": 4.876050536446973e-06,
      "loss": 0.4032,
      "step": 3244
    },
    {
      "epoch": 0.52,
      "grad_norm": 13.771710670820658,
      "learning_rate": 4.8734416257285215e-06,
      "loss": 0.4426,
      "step": 3245
    },
    {
      "epoch": 0.52,
      "grad_norm": 10.76068678183386,
      "learning_rate": 4.870832749488058e-06,
      "loss": 0.4358,
      "step": 3246
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.831224507993771,
      "learning_rate": 4.868223908436316e-06,
      "loss": 0.4289,
      "step": 3247
    },
    {
      "epoch": 0.52,
      "grad_norm": 10.620516785995491,
      "learning_rate": 4.865615103284014e-06,
      "loss": 0.4556,
      "step": 3248
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.642129847768723,
      "learning_rate": 4.863006334741861e-06,
      "loss": 0.4121,
      "step": 3249
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.997940337448587,
      "learning_rate": 4.860397603520559e-06,
      "loss": 0.5232,
      "step": 3250
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.6888717423334,
      "learning_rate": 4.8577889103308e-06,
      "loss": 0.4285,
      "step": 3251
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.064574250932244,
      "learning_rate": 4.855180255883259e-06,
      "loss": 0.4322,
      "step": 3252
    },
    {
      "epoch": 0.52,
      "grad_norm": 15.809160657905979,
      "learning_rate": 4.852571640888609e-06,
      "loss": 0.4233,
      "step": 3253
    },
    {
      "epoch": 0.52,
      "grad_norm": 10.771742827935897,
      "learning_rate": 4.84996306605751e-06,
      "loss": 0.5075,
      "step": 3254
    },
    {
      "epoch": 0.52,
      "grad_norm": 15.963208413064082,
      "learning_rate": 4.847354532100606e-06,
      "loss": 0.4981,
      "step": 3255
    },
    {
      "epoch": 0.52,
      "grad_norm": 12.312511769415389,
      "learning_rate": 4.844746039728535e-06,
      "loss": 0.4629,
      "step": 3256
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.7846026897490574,
      "learning_rate": 4.842137589651923e-06,
      "loss": 0.4286,
      "step": 3257
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.716225263554918,
      "learning_rate": 4.8395291825813824e-06,
      "loss": 0.36,
      "step": 3258
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.775068069989327,
      "learning_rate": 4.836920819227514e-06,
      "loss": 0.4722,
      "step": 3259
    },
    {
      "epoch": 0.53,
      "grad_norm": 16.483548625918782,
      "learning_rate": 4.834312500300915e-06,
      "loss": 0.4417,
      "step": 3260
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.066213906225504,
      "learning_rate": 4.831704226512153e-06,
      "loss": 0.4317,
      "step": 3261
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.026019939585913,
      "learning_rate": 4.829095998571801e-06,
      "loss": 0.453,
      "step": 3262
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.668837373565955,
      "learning_rate": 4.826487817190411e-06,
      "loss": 0.4663,
      "step": 3263
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.357696914469692,
      "learning_rate": 4.8238796830785206e-06,
      "loss": 0.4167,
      "step": 3264
    },
    {
      "epoch": 0.53,
      "grad_norm": 10.61704735251367,
      "learning_rate": 4.821271596946659e-06,
      "loss": 0.4522,
      "step": 3265
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.234798358369616,
      "learning_rate": 4.818663559505341e-06,
      "loss": 0.4076,
      "step": 3266
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.313799338989178,
      "learning_rate": 4.816055571465071e-06,
      "loss": 0.4501,
      "step": 3267
    },
    {
      "epoch": 0.53,
      "grad_norm": 22.07417178178273,
      "learning_rate": 4.813447633536331e-06,
      "loss": 0.4276,
      "step": 3268
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2356274572792492,
      "learning_rate": 4.810839746429598e-06,
      "loss": 0.4565,
      "step": 3269
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.413133051696658,
      "learning_rate": 4.808231910855335e-06,
      "loss": 0.4599,
      "step": 3270
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.870987924254158,
      "learning_rate": 4.805624127523982e-06,
      "loss": 0.4528,
      "step": 3271
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.387744826890694,
      "learning_rate": 4.8030163971459765e-06,
      "loss": 0.3826,
      "step": 3272
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.4389605678258945,
      "learning_rate": 4.800408720431735e-06,
      "loss": 0.3771,
      "step": 3273
    },
    {
      "epoch": 0.53,
      "grad_norm": 12.146302108638302,
      "learning_rate": 4.797801098091659e-06,
      "loss": 0.5332,
      "step": 3274
    },
    {
      "epoch": 0.53,
      "grad_norm": 14.623552266455311,
      "learning_rate": 4.795193530836136e-06,
      "loss": 0.5391,
      "step": 3275
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.646356063716102,
      "learning_rate": 4.792586019375543e-06,
      "loss": 0.4149,
      "step": 3276
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.043754738826692,
      "learning_rate": 4.789978564420235e-06,
      "loss": 0.4427,
      "step": 3277
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.173615627124765,
      "learning_rate": 4.787371166680555e-06,
      "loss": 0.3861,
      "step": 3278
    },
    {
      "epoch": 0.53,
      "grad_norm": 11.39499931391525,
      "learning_rate": 4.784763826866831e-06,
      "loss": 0.4053,
      "step": 3279
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.286508499635275,
      "learning_rate": 4.782156545689373e-06,
      "loss": 0.3881,
      "step": 3280
    },
    {
      "epoch": 0.53,
      "grad_norm": 9.275607912510944,
      "learning_rate": 4.779549323858476e-06,
      "loss": 0.4994,
      "step": 3281
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.138043520917561,
      "learning_rate": 4.776942162084423e-06,
      "loss": 0.3791,
      "step": 3282
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.301236075288075,
      "learning_rate": 4.77433506107747e-06,
      "loss": 0.4117,
      "step": 3283
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.71698639630238,
      "learning_rate": 4.771728021547868e-06,
      "loss": 0.3792,
      "step": 3284
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.168082779960844,
      "learning_rate": 4.769121044205847e-06,
      "loss": 0.4505,
      "step": 3285
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.017376726043553,
      "learning_rate": 4.766514129761616e-06,
      "loss": 0.4278,
      "step": 3286
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.521431120661022,
      "learning_rate": 4.763907278925372e-06,
      "loss": 0.4617,
      "step": 3287
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.124769879776053,
      "learning_rate": 4.761300492407293e-06,
      "loss": 0.4533,
      "step": 3288
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.606607693241983,
      "learning_rate": 4.758693770917543e-06,
      "loss": 0.4703,
      "step": 3289
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.516505479091492,
      "learning_rate": 4.756087115166257e-06,
      "loss": 0.4228,
      "step": 3290
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.388999064185464,
      "learning_rate": 4.753480525863566e-06,
      "loss": 0.3979,
      "step": 3291
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.437539219725819,
      "learning_rate": 4.750874003719577e-06,
      "loss": 0.481,
      "step": 3292
    },
    {
      "epoch": 0.53,
      "grad_norm": 14.92193121847193,
      "learning_rate": 4.7482675494443745e-06,
      "loss": 0.4296,
      "step": 3293
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.531106221388708,
      "learning_rate": 4.745661163748031e-06,
      "loss": 0.4271,
      "step": 3294
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.435712005507817,
      "learning_rate": 4.743054847340598e-06,
      "loss": 0.4044,
      "step": 3295
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.978765435992806,
      "learning_rate": 4.740448600932107e-06,
      "loss": 0.3826,
      "step": 3296
    },
    {
      "epoch": 0.53,
      "grad_norm": 20.565732379267395,
      "learning_rate": 4.73784242523257e-06,
      "loss": 0.4897,
      "step": 3297
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.300838360840096,
      "learning_rate": 4.735236320951986e-06,
      "loss": 0.4637,
      "step": 3298
    },
    {
      "epoch": 0.53,
      "grad_norm": 15.743621227201395,
      "learning_rate": 4.7326302888003224e-06,
      "loss": 0.5051,
      "step": 3299
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.099566468745147,
      "learning_rate": 4.730024329487541e-06,
      "loss": 0.4893,
      "step": 3300
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.294060126093305,
      "learning_rate": 4.7274184437235744e-06,
      "loss": 0.4727,
      "step": 3301
    },
    {
      "epoch": 0.53,
      "grad_norm": 11.08107344013705,
      "learning_rate": 4.7248126322183364e-06,
      "loss": 0.4077,
      "step": 3302
    },
    {
      "epoch": 0.53,
      "grad_norm": 9.235681587080398,
      "learning_rate": 4.722206895681723e-06,
      "loss": 0.3926,
      "step": 3303
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.5395705284714865,
      "learning_rate": 4.71960123482361e-06,
      "loss": 0.481,
      "step": 3304
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.045546721619823,
      "learning_rate": 4.716995650353847e-06,
      "loss": 0.4398,
      "step": 3305
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.780349409781917,
      "learning_rate": 4.714390142982272e-06,
      "loss": 0.4515,
      "step": 3306
    },
    {
      "epoch": 0.53,
      "grad_norm": 9.988003419463865,
      "learning_rate": 4.711784713418694e-06,
      "loss": 0.5617,
      "step": 3307
    },
    {
      "epoch": 0.53,
      "grad_norm": 10.121728223924425,
      "learning_rate": 4.709179362372906e-06,
      "loss": 0.4211,
      "step": 3308
    },
    {
      "epoch": 0.53,
      "grad_norm": 11.549337597089705,
      "learning_rate": 4.706574090554676e-06,
      "loss": 0.4652,
      "step": 3309
    },
    {
      "epoch": 0.53,
      "grad_norm": 9.490114941211655,
      "learning_rate": 4.70396889867375e-06,
      "loss": 0.4872,
      "step": 3310
    },
    {
      "epoch": 0.53,
      "grad_norm": 10.893928551943358,
      "learning_rate": 4.70136378743986e-06,
      "loss": 0.4701,
      "step": 3311
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.535358906081407,
      "learning_rate": 4.698758757562703e-06,
      "loss": 0.3832,
      "step": 3312
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.946944298110637,
      "learning_rate": 4.696153809751964e-06,
      "loss": 0.3951,
      "step": 3313
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2181003488789302,
      "learning_rate": 4.693548944717305e-06,
      "loss": 0.4613,
      "step": 3314
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.844560734263103,
      "learning_rate": 4.69094416316836e-06,
      "loss": 0.4224,
      "step": 3315
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.921751409937989,
      "learning_rate": 4.688339465814744e-06,
      "loss": 0.423,
      "step": 3316
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.631048747557401,
      "learning_rate": 4.685734853366049e-06,
      "loss": 0.4565,
      "step": 3317
    },
    {
      "epoch": 0.53,
      "grad_norm": 9.519475638284781,
      "learning_rate": 4.683130326531842e-06,
      "loss": 0.4706,
      "step": 3318
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.428074653653947,
      "learning_rate": 4.6805258860216675e-06,
      "loss": 0.4374,
      "step": 3319
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.313694257207128,
      "learning_rate": 4.677921532545052e-06,
      "loss": 0.456,
      "step": 3320
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.849316593749629,
      "learning_rate": 4.675317266811485e-06,
      "loss": 0.4398,
      "step": 3321
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.025838354173722,
      "learning_rate": 4.672713089530445e-06,
      "loss": 0.3967,
      "step": 3322
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.310205119422886,
      "learning_rate": 4.670109001411383e-06,
      "loss": 0.4866,
      "step": 3323
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.444862323027851,
      "learning_rate": 4.6675050031637216e-06,
      "loss": 0.4797,
      "step": 3324
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.9917143365153684,
      "learning_rate": 4.664901095496863e-06,
      "loss": 0.4421,
      "step": 3325
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.68857145607652,
      "learning_rate": 4.662297279120181e-06,
      "loss": 0.467,
      "step": 3326
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.450331140259409,
      "learning_rate": 4.659693554743032e-06,
      "loss": 0.4523,
      "step": 3327
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.138056485232844,
      "learning_rate": 4.657089923074737e-06,
      "loss": 0.4127,
      "step": 3328
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.353099062623064,
      "learning_rate": 4.6544863848246e-06,
      "loss": 0.3113,
      "step": 3329
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.573266462098007,
      "learning_rate": 4.651882940701897e-06,
      "loss": 0.4299,
      "step": 3330
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.986807043582329,
      "learning_rate": 4.649279591415876e-06,
      "loss": 0.4611,
      "step": 3331
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.456177625235982,
      "learning_rate": 4.646676337675762e-06,
      "loss": 0.488,
      "step": 3332
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3643161604762688,
      "learning_rate": 4.644073180190753e-06,
      "loss": 0.4455,
      "step": 3333
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.343416304766303,
      "learning_rate": 4.64147011967002e-06,
      "loss": 0.4093,
      "step": 3334
    },
    {
      "epoch": 0.54,
      "grad_norm": 9.36574093433074,
      "learning_rate": 4.638867156822711e-06,
      "loss": 0.3843,
      "step": 3335
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.067264187857168,
      "learning_rate": 4.636264292357943e-06,
      "loss": 0.4281,
      "step": 3336
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.982734015017322,
      "learning_rate": 4.633661526984808e-06,
      "loss": 0.468,
      "step": 3337
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.214565995373139,
      "learning_rate": 4.63105886141237e-06,
      "loss": 0.4038,
      "step": 3338
    },
    {
      "epoch": 0.54,
      "grad_norm": 16.32759485837536,
      "learning_rate": 4.628456296349669e-06,
      "loss": 0.498,
      "step": 3339
    },
    {
      "epoch": 0.54,
      "grad_norm": 9.408589100677231,
      "learning_rate": 4.6258538325057145e-06,
      "loss": 0.448,
      "step": 3340
    },
    {
      "epoch": 0.54,
      "grad_norm": 10.100266373879446,
      "learning_rate": 4.623251470589488e-06,
      "loss": 0.4797,
      "step": 3341
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.934670092737477,
      "learning_rate": 4.62064921130995e-06,
      "loss": 0.4749,
      "step": 3342
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.080513029104854,
      "learning_rate": 4.618047055376019e-06,
      "loss": 0.426,
      "step": 3343
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.996952381641377,
      "learning_rate": 4.615445003496602e-06,
      "loss": 0.3745,
      "step": 3344
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.581790745784809,
      "learning_rate": 4.612843056380564e-06,
      "loss": 0.4046,
      "step": 3345
    },
    {
      "epoch": 0.54,
      "grad_norm": 14.960138212332955,
      "learning_rate": 4.610241214736751e-06,
      "loss": 0.3596,
      "step": 3346
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.305584947863702,
      "learning_rate": 4.607639479273976e-06,
      "loss": 0.4126,
      "step": 3347
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.981537767765301,
      "learning_rate": 4.605037850701019e-06,
      "loss": 0.4994,
      "step": 3348
    },
    {
      "epoch": 0.54,
      "grad_norm": 12.499253997744956,
      "learning_rate": 4.602436329726643e-06,
      "loss": 0.4995,
      "step": 3349
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.02415974888144,
      "learning_rate": 4.599834917059565e-06,
      "loss": 0.5045,
      "step": 3350
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.64220047727567,
      "learning_rate": 4.597233613408488e-06,
      "loss": 0.4166,
      "step": 3351
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.884072875711894,
      "learning_rate": 4.5946324194820795e-06,
      "loss": 0.4831,
      "step": 3352
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.279347768775072,
      "learning_rate": 4.592031335988971e-06,
      "loss": 0.4899,
      "step": 3353
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.185293329294756,
      "learning_rate": 4.589430363637773e-06,
      "loss": 0.4395,
      "step": 3354
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.638959561914513,
      "learning_rate": 4.5868295031370625e-06,
      "loss": 0.4438,
      "step": 3355
    },
    {
      "epoch": 0.54,
      "grad_norm": 11.493372569380933,
      "learning_rate": 4.584228755195382e-06,
      "loss": 0.4356,
      "step": 3356
    },
    {
      "epoch": 0.54,
      "grad_norm": 11.552985044818614,
      "learning_rate": 4.581628120521251e-06,
      "loss": 0.4326,
      "step": 3357
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.160875125254548,
      "learning_rate": 4.579027599823153e-06,
      "loss": 0.4143,
      "step": 3358
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.011579050794526,
      "learning_rate": 4.5764271938095405e-06,
      "loss": 0.39,
      "step": 3359
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.23525210595496,
      "learning_rate": 4.573826903188836e-06,
      "loss": 0.5036,
      "step": 3360
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1281370541847753,
      "learning_rate": 4.571226728669432e-06,
      "loss": 0.4627,
      "step": 3361
    },
    {
      "epoch": 0.54,
      "grad_norm": 15.609299410102988,
      "learning_rate": 4.568626670959684e-06,
      "loss": 0.4318,
      "step": 3362
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.581426143431007,
      "learning_rate": 4.566026730767922e-06,
      "loss": 0.4761,
      "step": 3363
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.624384769865726,
      "learning_rate": 4.5634269088024434e-06,
      "loss": 0.4591,
      "step": 3364
    },
    {
      "epoch": 0.54,
      "grad_norm": 10.770952137295978,
      "learning_rate": 4.560827205771506e-06,
      "loss": 0.4322,
      "step": 3365
    },
    {
      "epoch": 0.54,
      "grad_norm": 11.361493142898524,
      "learning_rate": 4.558227622383345e-06,
      "loss": 0.4367,
      "step": 3366
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.583502725660896,
      "learning_rate": 4.555628159346158e-06,
      "loss": 0.4636,
      "step": 3367
    },
    {
      "epoch": 0.54,
      "grad_norm": 10.275416916262637,
      "learning_rate": 4.55302881736811e-06,
      "loss": 0.4436,
      "step": 3368
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.94653162161526,
      "learning_rate": 4.550429597157331e-06,
      "loss": 0.4219,
      "step": 3369
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.707972573054583,
      "learning_rate": 4.547830499421921e-06,
      "loss": 0.3876,
      "step": 3370
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.056059756574957,
      "learning_rate": 4.545231524869952e-06,
      "loss": 0.3804,
      "step": 3371
    },
    {
      "epoch": 0.54,
      "grad_norm": 9.768783473244461,
      "learning_rate": 4.542632674209447e-06,
      "loss": 0.3962,
      "step": 3372
    },
    {
      "epoch": 0.54,
      "grad_norm": 9.453477678226667,
      "learning_rate": 4.54003394814841e-06,
      "loss": 0.3495,
      "step": 3373
    },
    {
      "epoch": 0.54,
      "grad_norm": 9.438729644808358,
      "learning_rate": 4.537435347394807e-06,
      "loss": 0.4212,
      "step": 3374
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.837001294294776,
      "learning_rate": 4.534836872656563e-06,
      "loss": 0.4371,
      "step": 3375
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.415510685959797,
      "learning_rate": 4.532238524641578e-06,
      "loss": 0.4048,
      "step": 3376
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.753633627335367,
      "learning_rate": 4.529640304057714e-06,
      "loss": 0.4105,
      "step": 3377
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.635286562765643,
      "learning_rate": 4.527042211612796e-06,
      "loss": 0.4637,
      "step": 3378
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.027102779401307,
      "learning_rate": 4.524444248014615e-06,
      "loss": 0.4621,
      "step": 3379
    },
    {
      "epoch": 0.54,
      "grad_norm": 37.19150062910144,
      "learning_rate": 4.521846413970932e-06,
      "loss": 0.4799,
      "step": 3380
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.501783972594417,
      "learning_rate": 4.519248710189465e-06,
      "loss": 0.4457,
      "step": 3381
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2070566048443694,
      "learning_rate": 4.516651137377902e-06,
      "loss": 0.5095,
      "step": 3382
    },
    {
      "epoch": 0.55,
      "grad_norm": 48.83425910903912,
      "learning_rate": 4.514053696243893e-06,
      "loss": 0.4635,
      "step": 3383
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.784641510771142,
      "learning_rate": 4.511456387495052e-06,
      "loss": 0.4801,
      "step": 3384
    },
    {
      "epoch": 0.55,
      "grad_norm": 9.417376346266119,
      "learning_rate": 4.508859211838955e-06,
      "loss": 0.4022,
      "step": 3385
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.5602398297054405,
      "learning_rate": 4.50626216998315e-06,
      "loss": 0.4091,
      "step": 3386
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.46714993901338,
      "learning_rate": 4.503665262635141e-06,
      "loss": 0.3541,
      "step": 3387
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.049172970818485,
      "learning_rate": 4.501068490502394e-06,
      "loss": 0.4323,
      "step": 3388
    },
    {
      "epoch": 0.55,
      "grad_norm": 11.120056323171253,
      "learning_rate": 4.498471854292344e-06,
      "loss": 0.4153,
      "step": 3389
    },
    {
      "epoch": 0.55,
      "grad_norm": 13.08604013280653,
      "learning_rate": 4.495875354712386e-06,
      "loss": 0.4173,
      "step": 3390
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.692906690975171,
      "learning_rate": 4.493278992469877e-06,
      "loss": 0.5367,
      "step": 3391
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.723231631289586,
      "learning_rate": 4.4906827682721375e-06,
      "loss": 0.4141,
      "step": 3392
    },
    {
      "epoch": 0.55,
      "grad_norm": 40.89502386868,
      "learning_rate": 4.488086682826454e-06,
      "loss": 0.4422,
      "step": 3393
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.459800567126894,
      "learning_rate": 4.485490736840066e-06,
      "loss": 0.4705,
      "step": 3394
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3608876906379102,
      "learning_rate": 4.482894931020185e-06,
      "loss": 0.4635,
      "step": 3395
    },
    {
      "epoch": 0.55,
      "grad_norm": 13.571433855326525,
      "learning_rate": 4.48029926607398e-06,
      "loss": 0.4585,
      "step": 3396
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.06698804186974,
      "learning_rate": 4.477703742708579e-06,
      "loss": 0.4198,
      "step": 3397
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.014277717239853,
      "learning_rate": 4.475108361631076e-06,
      "loss": 0.4256,
      "step": 3398
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.443084231751913,
      "learning_rate": 4.472513123548525e-06,
      "loss": 0.4285,
      "step": 3399
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.027008386119307,
      "learning_rate": 4.469918029167939e-06,
      "loss": 0.4071,
      "step": 3400
    },
    {
      "epoch": 0.55,
      "grad_norm": 9.905486381752056,
      "learning_rate": 4.467323079196292e-06,
      "loss": 0.4614,
      "step": 3401
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.515826442824798,
      "learning_rate": 4.464728274340525e-06,
      "loss": 0.4052,
      "step": 3402
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.666701458144494,
      "learning_rate": 4.462133615307527e-06,
      "loss": 0.3932,
      "step": 3403
    },
    {
      "epoch": 0.55,
      "grad_norm": 15.644816529029521,
      "learning_rate": 4.45953910280416e-06,
      "loss": 0.457,
      "step": 3404
    },
    {
      "epoch": 0.55,
      "grad_norm": 32.281226202294235,
      "learning_rate": 4.4569447375372396e-06,
      "loss": 0.4086,
      "step": 3405
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.6559834994105085,
      "learning_rate": 4.454350520213543e-06,
      "loss": 0.4498,
      "step": 3406
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0882312619900265,
      "learning_rate": 4.451756451539804e-06,
      "loss": 0.4791,
      "step": 3407
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.794203687745213,
      "learning_rate": 4.44916253222272e-06,
      "loss": 0.397,
      "step": 3408
    },
    {
      "epoch": 0.55,
      "grad_norm": 10.373612177381668,
      "learning_rate": 4.446568762968947e-06,
      "loss": 0.4609,
      "step": 3409
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.846940997076772,
      "learning_rate": 4.443975144485099e-06,
      "loss": 0.4563,
      "step": 3410
    },
    {
      "epoch": 0.55,
      "grad_norm": 9.351003833760304,
      "learning_rate": 4.441381677477748e-06,
      "loss": 0.4521,
      "step": 3411
    },
    {
      "epoch": 0.55,
      "grad_norm": 12.57071861357142,
      "learning_rate": 4.438788362653426e-06,
      "loss": 0.4169,
      "step": 3412
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.93254865414392,
      "learning_rate": 4.436195200718625e-06,
      "loss": 0.4434,
      "step": 3413
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.282090971033764,
      "learning_rate": 4.43360219237979e-06,
      "loss": 0.3893,
      "step": 3414
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.5814903779140925,
      "learning_rate": 4.431009338343335e-06,
      "loss": 0.448,
      "step": 3415
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.187124031880494,
      "learning_rate": 4.428416639315616e-06,
      "loss": 0.4619,
      "step": 3416
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.120725231807672,
      "learning_rate": 4.425824096002962e-06,
      "loss": 0.4654,
      "step": 3417
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.8285222512380135,
      "learning_rate": 4.423231709111653e-06,
      "loss": 0.4827,
      "step": 3418
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.878570546858172,
      "learning_rate": 4.420639479347924e-06,
      "loss": 0.4019,
      "step": 3419
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.514084877046083,
      "learning_rate": 4.4180474074179705e-06,
      "loss": 0.4901,
      "step": 3420
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.038843681951046,
      "learning_rate": 4.415455494027946e-06,
      "loss": 0.4234,
      "step": 3421
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.117266795578791,
      "learning_rate": 4.412863739883958e-06,
      "loss": 0.4576,
      "step": 3422
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.061870633799726,
      "learning_rate": 4.410272145692069e-06,
      "loss": 0.4392,
      "step": 3423
    },
    {
      "epoch": 0.55,
      "grad_norm": 3.8862794940787184,
      "learning_rate": 4.407680712158308e-06,
      "loss": 0.4954,
      "step": 3424
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3002227617701827,
      "learning_rate": 4.405089439988645e-06,
      "loss": 0.5078,
      "step": 3425
    },
    {
      "epoch": 0.55,
      "grad_norm": 10.696716583327925,
      "learning_rate": 4.402498329889018e-06,
      "loss": 0.4282,
      "step": 3426
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.5085646349168575,
      "learning_rate": 4.399907382565316e-06,
      "loss": 0.4796,
      "step": 3427
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.293745563617675,
      "learning_rate": 4.397316598723385e-06,
      "loss": 0.3997,
      "step": 3428
    },
    {
      "epoch": 0.55,
      "grad_norm": 10.471068981386365,
      "learning_rate": 4.394725979069026e-06,
      "loss": 0.4171,
      "step": 3429
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.180680695010223,
      "learning_rate": 4.392135524307993e-06,
      "loss": 0.4627,
      "step": 3430
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.9907302300562355,
      "learning_rate": 4.389545235146003e-06,
      "loss": 0.4042,
      "step": 3431
    },
    {
      "epoch": 0.55,
      "grad_norm": 10.59490501434163,
      "learning_rate": 4.386955112288714e-06,
      "loss": 0.4856,
      "step": 3432
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.423085616935056,
      "learning_rate": 4.384365156441752e-06,
      "loss": 0.4092,
      "step": 3433
    },
    {
      "epoch": 0.55,
      "grad_norm": 14.756201447757219,
      "learning_rate": 4.381775368310694e-06,
      "loss": 0.3825,
      "step": 3434
    },
    {
      "epoch": 0.55,
      "grad_norm": 11.118937096646885,
      "learning_rate": 4.3791857486010655e-06,
      "loss": 0.439,
      "step": 3435
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.128043504056167,
      "learning_rate": 4.3765962980183515e-06,
      "loss": 0.4997,
      "step": 3436
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.1122657749917595,
      "learning_rate": 4.374007017267991e-06,
      "loss": 0.3894,
      "step": 3437
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.612449101145293,
      "learning_rate": 4.371417907055373e-06,
      "loss": 0.4304,
      "step": 3438
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.458107820241239,
      "learning_rate": 4.3688289680858456e-06,
      "loss": 0.437,
      "step": 3439
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.067725231643833,
      "learning_rate": 4.366240201064705e-06,
      "loss": 0.462,
      "step": 3440
    },
    {
      "epoch": 0.55,
      "grad_norm": 13.330828103307006,
      "learning_rate": 4.363651606697204e-06,
      "loss": 0.4231,
      "step": 3441
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.38924920255093,
      "learning_rate": 4.361063185688546e-06,
      "loss": 0.3892,
      "step": 3442
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.27568819112588,
      "learning_rate": 4.358474938743889e-06,
      "loss": 0.4501,
      "step": 3443
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.729684600793394,
      "learning_rate": 4.355886866568342e-06,
      "loss": 0.4815,
      "step": 3444
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.51861301174476,
      "learning_rate": 4.353298969866966e-06,
      "loss": 0.4226,
      "step": 3445
    },
    {
      "epoch": 0.56,
      "grad_norm": 10.03624715246827,
      "learning_rate": 4.350711249344778e-06,
      "loss": 0.4262,
      "step": 3446
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.839333530457556,
      "learning_rate": 4.348123705706745e-06,
      "loss": 0.4446,
      "step": 3447
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.812619094320759,
      "learning_rate": 4.345536339657783e-06,
      "loss": 0.4303,
      "step": 3448
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.796860758313819,
      "learning_rate": 4.342949151902761e-06,
      "loss": 0.4355,
      "step": 3449
    },
    {
      "epoch": 0.56,
      "grad_norm": 14.59777829572145,
      "learning_rate": 4.340362143146504e-06,
      "loss": 0.3922,
      "step": 3450
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.5269446099193065,
      "learning_rate": 4.337775314093781e-06,
      "loss": 0.4762,
      "step": 3451
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.315530577146882,
      "learning_rate": 4.335188665449316e-06,
      "loss": 0.4256,
      "step": 3452
    },
    {
      "epoch": 0.56,
      "grad_norm": 10.739329546124404,
      "learning_rate": 4.3326021979177865e-06,
      "loss": 0.4646,
      "step": 3453
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.0886515069476905,
      "learning_rate": 4.330015912203812e-06,
      "loss": 0.4297,
      "step": 3454
    },
    {
      "epoch": 0.56,
      "grad_norm": 15.019102731368733,
      "learning_rate": 4.327429809011973e-06,
      "loss": 0.4657,
      "step": 3455
    },
    {
      "epoch": 0.56,
      "grad_norm": 10.034397946411353,
      "learning_rate": 4.324843889046795e-06,
      "loss": 0.4372,
      "step": 3456
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.529004925936166,
      "learning_rate": 4.322258153012751e-06,
      "loss": 0.419,
      "step": 3457
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.218066916873486,
      "learning_rate": 4.319672601614268e-06,
      "loss": 0.4342,
      "step": 3458
    },
    {
      "epoch": 0.56,
      "grad_norm": 19.71388694159525,
      "learning_rate": 4.317087235555722e-06,
      "loss": 0.4315,
      "step": 3459
    },
    {
      "epoch": 0.56,
      "grad_norm": 13.3807502071221,
      "learning_rate": 4.314502055541437e-06,
      "loss": 0.4532,
      "step": 3460
    },
    {
      "epoch": 0.56,
      "grad_norm": 19.345401133326384,
      "learning_rate": 4.311917062275688e-06,
      "loss": 0.439,
      "step": 3461
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.458860808757126,
      "learning_rate": 4.309332256462699e-06,
      "loss": 0.4399,
      "step": 3462
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.206606286746489,
      "learning_rate": 4.306747638806641e-06,
      "loss": 0.4857,
      "step": 3463
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.266032133966359,
      "learning_rate": 4.304163210011636e-06,
      "loss": 0.4637,
      "step": 3464
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.976606062540379,
      "learning_rate": 4.301578970781753e-06,
      "loss": 0.4199,
      "step": 3465
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.488918401229766,
      "learning_rate": 4.2989949218210125e-06,
      "loss": 0.4068,
      "step": 3466
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.253583678162672,
      "learning_rate": 4.2964110638333755e-06,
      "loss": 0.4572,
      "step": 3467
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.703991693692433,
      "learning_rate": 4.29382739752276e-06,
      "loss": 0.5107,
      "step": 3468
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.672809763223897,
      "learning_rate": 4.2912439235930305e-06,
      "loss": 0.38,
      "step": 3469
    },
    {
      "epoch": 0.56,
      "grad_norm": 9.510942203346927,
      "learning_rate": 4.2886606427479905e-06,
      "loss": 0.4241,
      "step": 3470
    },
    {
      "epoch": 0.56,
      "grad_norm": 16.088609560307717,
      "learning_rate": 4.2860775556913995e-06,
      "loss": 0.447,
      "step": 3471
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.313219202335723,
      "learning_rate": 4.283494663126962e-06,
      "loss": 0.4282,
      "step": 3472
    },
    {
      "epoch": 0.56,
      "grad_norm": 20.741036901517163,
      "learning_rate": 4.28091196575833e-06,
      "loss": 0.482,
      "step": 3473
    },
    {
      "epoch": 0.56,
      "grad_norm": 10.307669045433087,
      "learning_rate": 4.278329464289098e-06,
      "loss": 0.4429,
      "step": 3474
    },
    {
      "epoch": 0.56,
      "grad_norm": 12.812807370177131,
      "learning_rate": 4.275747159422815e-06,
      "loss": 0.413,
      "step": 3475
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.630313500494156,
      "learning_rate": 4.273165051862969e-06,
      "loss": 0.38,
      "step": 3476
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.678087343871618,
      "learning_rate": 4.270583142312998e-06,
      "loss": 0.478,
      "step": 3477
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.224841752898653,
      "learning_rate": 4.268001431476286e-06,
      "loss": 0.4302,
      "step": 3478
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.6365634153621205,
      "learning_rate": 4.265419920056162e-06,
      "loss": 0.4574,
      "step": 3479
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.086752174009546,
      "learning_rate": 4.262838608755899e-06,
      "loss": 0.4332,
      "step": 3480
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.107151771630165,
      "learning_rate": 4.2602574982787216e-06,
      "loss": 0.429,
      "step": 3481
    },
    {
      "epoch": 0.56,
      "grad_norm": 26.058421604216008,
      "learning_rate": 4.257676589327791e-06,
      "loss": 0.3935,
      "step": 3482
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.247612679805015,
      "learning_rate": 4.255095882606219e-06,
      "loss": 0.4171,
      "step": 3483
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.225492598127672,
      "learning_rate": 4.252515378817062e-06,
      "loss": 0.4745,
      "step": 3484
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.13817567781495,
      "learning_rate": 4.2499350786633235e-06,
      "loss": 0.4459,
      "step": 3485
    },
    {
      "epoch": 0.56,
      "grad_norm": 10.965874592245191,
      "learning_rate": 4.247354982847942e-06,
      "loss": 0.4498,
      "step": 3486
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.278921015524402,
      "learning_rate": 4.244775092073811e-06,
      "loss": 0.3276,
      "step": 3487
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.49106739202001,
      "learning_rate": 4.242195407043763e-06,
      "loss": 0.4687,
      "step": 3488
    },
    {
      "epoch": 0.56,
      "grad_norm": 15.68109419366356,
      "learning_rate": 4.239615928460574e-06,
      "loss": 0.4427,
      "step": 3489
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1467590083741845,
      "learning_rate": 4.2370366570269676e-06,
      "loss": 0.4577,
      "step": 3490
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.521048383500448,
      "learning_rate": 4.234457593445608e-06,
      "loss": 0.3743,
      "step": 3491
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.212378853894835,
      "learning_rate": 4.2318787384191e-06,
      "loss": 0.4271,
      "step": 3492
    },
    {
      "epoch": 0.56,
      "grad_norm": 10.819709265294971,
      "learning_rate": 4.229300092649997e-06,
      "loss": 0.4334,
      "step": 3493
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.686945184776075,
      "learning_rate": 4.226721656840796e-06,
      "loss": 0.4402,
      "step": 3494
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0970547333872147,
      "learning_rate": 4.224143431693929e-06,
      "loss": 0.469,
      "step": 3495
    },
    {
      "epoch": 0.56,
      "grad_norm": 9.165600042630059,
      "learning_rate": 4.2215654179117765e-06,
      "loss": 0.4681,
      "step": 3496
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.2856202376375405,
      "learning_rate": 4.2189876161966655e-06,
      "loss": 0.4358,
      "step": 3497
    },
    {
      "epoch": 0.56,
      "grad_norm": 11.53513932988826,
      "learning_rate": 4.216410027250853e-06,
      "loss": 0.4265,
      "step": 3498
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.63913120081179,
      "learning_rate": 4.213832651776551e-06,
      "loss": 0.4702,
      "step": 3499
    },
    {
      "epoch": 0.56,
      "grad_norm": 13.709006707528753,
      "learning_rate": 4.211255490475905e-06,
      "loss": 0.3962,
      "step": 3500
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.545058952118208,
      "learning_rate": 4.208678544051005e-06,
      "loss": 0.5057,
      "step": 3501
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.427371715233713,
      "learning_rate": 4.2061018132038825e-06,
      "loss": 0.49,
      "step": 3502
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.511280198755118,
      "learning_rate": 4.203525298636512e-06,
      "loss": 0.493,
      "step": 3503
    },
    {
      "epoch": 0.56,
      "grad_norm": 22.72715837920541,
      "learning_rate": 4.2009490010508026e-06,
      "loss": 0.4759,
      "step": 3504
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.860979531284946,
      "learning_rate": 4.19837292114861e-06,
      "loss": 0.3932,
      "step": 3505
    },
    {
      "epoch": 0.56,
      "grad_norm": 10.76229700344235,
      "learning_rate": 4.195797059631733e-06,
      "loss": 0.3596,
      "step": 3506
    },
    {
      "epoch": 0.57,
      "grad_norm": 14.02573135336524,
      "learning_rate": 4.1932214172019056e-06,
      "loss": 0.4198,
      "step": 3507
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.302398856684944,
      "learning_rate": 4.190645994560802e-06,
      "loss": 0.4939,
      "step": 3508
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.776301037506045,
      "learning_rate": 4.188070792410039e-06,
      "loss": 0.4322,
      "step": 3509
    },
    {
      "epoch": 0.57,
      "grad_norm": 10.515926041735389,
      "learning_rate": 4.185495811451175e-06,
      "loss": 0.4693,
      "step": 3510
    },
    {
      "epoch": 0.57,
      "grad_norm": 14.517266946668926,
      "learning_rate": 4.182921052385702e-06,
      "loss": 0.3717,
      "step": 3511
    },
    {
      "epoch": 0.57,
      "grad_norm": 27.524687356786455,
      "learning_rate": 4.180346515915057e-06,
      "loss": 0.4285,
      "step": 3512
    },
    {
      "epoch": 0.57,
      "grad_norm": 10.271454648682413,
      "learning_rate": 4.177772202740617e-06,
      "loss": 0.417,
      "step": 3513
    },
    {
      "epoch": 0.57,
      "grad_norm": 17.271864560305875,
      "learning_rate": 4.175198113563692e-06,
      "loss": 0.4566,
      "step": 3514
    },
    {
      "epoch": 0.57,
      "grad_norm": 9.41885435842443,
      "learning_rate": 4.172624249085537e-06,
      "loss": 0.3763,
      "step": 3515
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.5452418473603355,
      "learning_rate": 4.170050610007344e-06,
      "loss": 0.3519,
      "step": 3516
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.9420470448266345,
      "learning_rate": 4.167477197030242e-06,
      "loss": 0.4104,
      "step": 3517
    },
    {
      "epoch": 0.57,
      "grad_norm": 26.840609168874533,
      "learning_rate": 4.164904010855299e-06,
      "loss": 0.4371,
      "step": 3518
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.912170424630667,
      "learning_rate": 4.162331052183526e-06,
      "loss": 0.438,
      "step": 3519
    },
    {
      "epoch": 0.57,
      "grad_norm": 9.059705334613804,
      "learning_rate": 4.159758321715862e-06,
      "loss": 0.4381,
      "step": 3520
    },
    {
      "epoch": 0.57,
      "grad_norm": 75.06967341796643,
      "learning_rate": 4.157185820153193e-06,
      "loss": 0.4531,
      "step": 3521
    },
    {
      "epoch": 0.57,
      "grad_norm": 9.899915320127883,
      "learning_rate": 4.154613548196341e-06,
      "loss": 0.4134,
      "step": 3522
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.0126014489231965,
      "learning_rate": 4.15204150654606e-06,
      "loss": 0.401,
      "step": 3523
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.309410053753508,
      "learning_rate": 4.149469695903047e-06,
      "loss": 0.4295,
      "step": 3524
    },
    {
      "epoch": 0.57,
      "grad_norm": 10.555110408120894,
      "learning_rate": 4.146898116967932e-06,
      "loss": 0.4364,
      "step": 3525
    },
    {
      "epoch": 0.57,
      "grad_norm": 11.826402512941181,
      "learning_rate": 4.1443267704412895e-06,
      "loss": 0.4023,
      "step": 3526
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.578437344884836,
      "learning_rate": 4.141755657023618e-06,
      "loss": 0.3779,
      "step": 3527
    },
    {
      "epoch": 0.57,
      "grad_norm": 14.557961660842366,
      "learning_rate": 4.139184777415365e-06,
      "loss": 0.3378,
      "step": 3528
    },
    {
      "epoch": 0.57,
      "grad_norm": 9.733268122144805,
      "learning_rate": 4.1366141323169076e-06,
      "loss": 0.4716,
      "step": 3529
    },
    {
      "epoch": 0.57,
      "grad_norm": 12.596412647396354,
      "learning_rate": 4.1340437224285594e-06,
      "loss": 0.4076,
      "step": 3530
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.0968260606884,
      "learning_rate": 4.131473548450571e-06,
      "loss": 0.4456,
      "step": 3531
    },
    {
      "epoch": 0.57,
      "grad_norm": 13.551312628942533,
      "learning_rate": 4.12890361108313e-06,
      "loss": 0.4205,
      "step": 3532
    },
    {
      "epoch": 0.57,
      "grad_norm": 9.738172538955002,
      "learning_rate": 4.126333911026357e-06,
      "loss": 0.5139,
      "step": 3533
    },
    {
      "epoch": 0.57,
      "grad_norm": 14.431413856628142,
      "learning_rate": 4.123764448980308e-06,
      "loss": 0.4078,
      "step": 3534
    },
    {
      "epoch": 0.57,
      "grad_norm": 13.331344611886424,
      "learning_rate": 4.1211952256449796e-06,
      "loss": 0.4638,
      "step": 3535
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.328944083678772,
      "learning_rate": 4.118626241720293e-06,
      "loss": 0.3746,
      "step": 3536
    },
    {
      "epoch": 0.57,
      "grad_norm": 11.60294032970051,
      "learning_rate": 4.116057497906114e-06,
      "loss": 0.4362,
      "step": 3537
    },
    {
      "epoch": 0.57,
      "grad_norm": 69.1378934821045,
      "learning_rate": 4.11348899490224e-06,
      "loss": 0.4514,
      "step": 3538
    },
    {
      "epoch": 0.57,
      "grad_norm": 12.145953436160097,
      "learning_rate": 4.1109207334084e-06,
      "loss": 0.4029,
      "step": 3539
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.98389370525785,
      "learning_rate": 4.1083527141242594e-06,
      "loss": 0.3766,
      "step": 3540
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.446197390375823,
      "learning_rate": 4.105784937749419e-06,
      "loss": 0.3926,
      "step": 3541
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.324769313439205,
      "learning_rate": 4.103217404983409e-06,
      "loss": 0.4735,
      "step": 3542
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.32542526181782,
      "learning_rate": 4.100650116525698e-06,
      "loss": 0.4214,
      "step": 3543
    },
    {
      "epoch": 0.57,
      "grad_norm": 9.257213289099845,
      "learning_rate": 4.098083073075686e-06,
      "loss": 0.4767,
      "step": 3544
    },
    {
      "epoch": 0.57,
      "grad_norm": 27.627490570146293,
      "learning_rate": 4.0955162753327085e-06,
      "loss": 0.3947,
      "step": 3545
    },
    {
      "epoch": 0.57,
      "grad_norm": 10.05256553881891,
      "learning_rate": 4.092949723996028e-06,
      "loss": 0.4547,
      "step": 3546
    },
    {
      "epoch": 0.57,
      "grad_norm": 13.535300601371363,
      "learning_rate": 4.0903834197648444e-06,
      "loss": 0.4685,
      "step": 3547
    },
    {
      "epoch": 0.57,
      "grad_norm": 12.990737528339785,
      "learning_rate": 4.087817363338294e-06,
      "loss": 0.4393,
      "step": 3548
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.0860485279130945,
      "learning_rate": 4.085251555415437e-06,
      "loss": 0.4167,
      "step": 3549
    },
    {
      "epoch": 0.57,
      "grad_norm": 111.97826905788796,
      "learning_rate": 4.08268599669527e-06,
      "loss": 0.4819,
      "step": 3550
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.503513543788104,
      "learning_rate": 4.080120687876726e-06,
      "loss": 0.424,
      "step": 3551
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.378768617660748,
      "learning_rate": 4.077555629658662e-06,
      "loss": 0.4493,
      "step": 3552
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.263087740304019,
      "learning_rate": 4.074990822739871e-06,
      "loss": 0.3973,
      "step": 3553
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.388060489075584,
      "learning_rate": 4.072426267819081e-06,
      "loss": 0.3926,
      "step": 3554
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.834880253448762,
      "learning_rate": 4.069861965594941e-06,
      "loss": 0.4343,
      "step": 3555
    },
    {
      "epoch": 0.57,
      "grad_norm": 13.445803755529155,
      "learning_rate": 4.067297916766042e-06,
      "loss": 0.4413,
      "step": 3556
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.0794359054678475,
      "learning_rate": 4.0647341220309024e-06,
      "loss": 0.4638,
      "step": 3557
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.242683554494942,
      "learning_rate": 4.062170582087965e-06,
      "loss": 0.4664,
      "step": 3558
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.2844353857897,
      "learning_rate": 4.059607297635615e-06,
      "loss": 0.4416,
      "step": 3559
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.282764265667226,
      "learning_rate": 4.057044269372159e-06,
      "loss": 0.4805,
      "step": 3560
    },
    {
      "epoch": 0.57,
      "grad_norm": 9.170877191508653,
      "learning_rate": 4.054481497995836e-06,
      "loss": 0.5217,
      "step": 3561
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.756305433392412,
      "learning_rate": 4.051918984204817e-06,
      "loss": 0.42,
      "step": 3562
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.251699506887296,
      "learning_rate": 4.049356728697199e-06,
      "loss": 0.5017,
      "step": 3563
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1735564664472435,
      "learning_rate": 4.046794732171016e-06,
      "loss": 0.4323,
      "step": 3564
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.352716009411997,
      "learning_rate": 4.04423299532422e-06,
      "loss": 0.4437,
      "step": 3565
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.829489997019559,
      "learning_rate": 4.041671518854704e-06,
      "loss": 0.4353,
      "step": 3566
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.626290683031584,
      "learning_rate": 4.0391103034602826e-06,
      "loss": 0.3558,
      "step": 3567
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.6146929760792785,
      "learning_rate": 4.036549349838701e-06,
      "loss": 0.4686,
      "step": 3568
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.953461616613704,
      "learning_rate": 4.033988658687634e-06,
      "loss": 0.3941,
      "step": 3569
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.033827666490349,
      "learning_rate": 4.031428230704686e-06,
      "loss": 0.4012,
      "step": 3570
    },
    {
      "epoch": 0.58,
      "grad_norm": 8.90893041001692,
      "learning_rate": 4.028868066587384e-06,
      "loss": 0.4304,
      "step": 3571
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.065832794765413,
      "learning_rate": 4.026308167033191e-06,
      "loss": 0.4751,
      "step": 3572
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2474893456415863,
      "learning_rate": 4.023748532739496e-06,
      "loss": 0.4998,
      "step": 3573
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.562580237692714,
      "learning_rate": 4.0211891644036085e-06,
      "loss": 0.4726,
      "step": 3574
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1031163096751515,
      "learning_rate": 4.018630062722774e-06,
      "loss": 0.5107,
      "step": 3575
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.596053069325565,
      "learning_rate": 4.016071228394164e-06,
      "loss": 0.4099,
      "step": 3576
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1788368610921687,
      "learning_rate": 4.013512662114871e-06,
      "loss": 0.5138,
      "step": 3577
    },
    {
      "epoch": 0.58,
      "grad_norm": 8.172498121023418,
      "learning_rate": 4.010954364581921e-06,
      "loss": 0.4949,
      "step": 3578
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.132733494317439,
      "learning_rate": 4.008396336492269e-06,
      "loss": 0.5088,
      "step": 3579
    },
    {
      "epoch": 0.58,
      "grad_norm": 11.229850964266946,
      "learning_rate": 4.005838578542785e-06,
      "loss": 0.4317,
      "step": 3580
    },
    {
      "epoch": 0.58,
      "grad_norm": 15.63283983738339,
      "learning_rate": 4.003281091430279e-06,
      "loss": 0.4385,
      "step": 3581
    },
    {
      "epoch": 0.58,
      "grad_norm": 8.50555505850098,
      "learning_rate": 4.000723875851477e-06,
      "loss": 0.4804,
      "step": 3582
    },
    {
      "epoch": 0.58,
      "grad_norm": 10.886756622461682,
      "learning_rate": 3.998166932503037e-06,
      "loss": 0.4978,
      "step": 3583
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.2214590196898065,
      "learning_rate": 3.99561026208154e-06,
      "loss": 0.4153,
      "step": 3584
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.531539260526917,
      "learning_rate": 3.993053865283492e-06,
      "loss": 0.4438,
      "step": 3585
    },
    {
      "epoch": 0.58,
      "grad_norm": 10.939185047921445,
      "learning_rate": 3.9904977428053325e-06,
      "loss": 0.3712,
      "step": 3586
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.366524620525083,
      "learning_rate": 3.9879418953434105e-06,
      "loss": 0.3978,
      "step": 3587
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.759678829013252,
      "learning_rate": 3.9853863235940145e-06,
      "loss": 0.414,
      "step": 3588
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.846520661650958,
      "learning_rate": 3.982831028253354e-06,
      "loss": 0.4103,
      "step": 3589
    },
    {
      "epoch": 0.58,
      "grad_norm": 9.782937194919242,
      "learning_rate": 3.980276010017559e-06,
      "loss": 0.3835,
      "step": 3590
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.823158158254266,
      "learning_rate": 3.977721269582686e-06,
      "loss": 0.3856,
      "step": 3591
    },
    {
      "epoch": 0.58,
      "grad_norm": 9.378426233049174,
      "learning_rate": 3.975166807644721e-06,
      "loss": 0.4506,
      "step": 3592
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.613035828092862,
      "learning_rate": 3.972612624899566e-06,
      "loss": 0.4306,
      "step": 3593
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.138319673242159,
      "learning_rate": 3.97005872204305e-06,
      "loss": 0.4452,
      "step": 3594
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.9077022756240645,
      "learning_rate": 3.967505099770932e-06,
      "loss": 0.4414,
      "step": 3595
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.288992368732283,
      "learning_rate": 3.964951758778885e-06,
      "loss": 0.4859,
      "step": 3596
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.59060967051074,
      "learning_rate": 3.962398699762508e-06,
      "loss": 0.4628,
      "step": 3597
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.474342405635066,
      "learning_rate": 3.95984592341733e-06,
      "loss": 0.4654,
      "step": 3598
    },
    {
      "epoch": 0.58,
      "grad_norm": 9.972404209306013,
      "learning_rate": 3.957293430438792e-06,
      "loss": 0.3722,
      "step": 3599
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.3179493197170125,
      "learning_rate": 3.954741221522266e-06,
      "loss": 0.4535,
      "step": 3600
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.716858181155409,
      "learning_rate": 3.952189297363047e-06,
      "loss": 0.4634,
      "step": 3601
    },
    {
      "epoch": 0.58,
      "grad_norm": 12.653989121672188,
      "learning_rate": 3.949637658656343e-06,
      "loss": 0.4516,
      "step": 3602
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.553174554910584,
      "learning_rate": 3.947086306097295e-06,
      "loss": 0.4967,
      "step": 3603
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.925315165983608,
      "learning_rate": 3.944535240380961e-06,
      "loss": 0.42,
      "step": 3604
    },
    {
      "epoch": 0.58,
      "grad_norm": 15.240486803150265,
      "learning_rate": 3.941984462202323e-06,
      "loss": 0.4267,
      "step": 3605
    },
    {
      "epoch": 0.58,
      "grad_norm": 10.080074671543342,
      "learning_rate": 3.939433972256281e-06,
      "loss": 0.47,
      "step": 3606
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.687731477483454,
      "learning_rate": 3.936883771237658e-06,
      "loss": 0.4054,
      "step": 3607
    },
    {
      "epoch": 0.58,
      "grad_norm": 8.141404927612301,
      "learning_rate": 3.934333859841204e-06,
      "loss": 0.4495,
      "step": 3608
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.513827117719276,
      "learning_rate": 3.931784238761579e-06,
      "loss": 0.4799,
      "step": 3609
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.947672218301246,
      "learning_rate": 3.929234908693373e-06,
      "loss": 0.4671,
      "step": 3610
    },
    {
      "epoch": 0.58,
      "grad_norm": 8.710804898855605,
      "learning_rate": 3.9266858703310965e-06,
      "loss": 0.4179,
      "step": 3611
    },
    {
      "epoch": 0.58,
      "grad_norm": 9.28626211905646,
      "learning_rate": 3.924137124369172e-06,
      "loss": 0.4372,
      "step": 3612
    },
    {
      "epoch": 0.58,
      "grad_norm": 8.629412425029125,
      "learning_rate": 3.9215886715019525e-06,
      "loss": 0.4221,
      "step": 3613
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.857351383153173,
      "learning_rate": 3.919040512423706e-06,
      "loss": 0.5064,
      "step": 3614
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.093414447538241,
      "learning_rate": 3.91649264782862e-06,
      "loss": 0.3801,
      "step": 3615
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.681772215490231,
      "learning_rate": 3.913945078410802e-06,
      "loss": 0.4956,
      "step": 3616
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.062223572954884,
      "learning_rate": 3.911397804864285e-06,
      "loss": 0.4402,
      "step": 3617
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.60473819491873,
      "learning_rate": 3.908850827883012e-06,
      "loss": 0.4225,
      "step": 3618
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.671705785598382,
      "learning_rate": 3.90630414816085e-06,
      "loss": 0.4841,
      "step": 3619
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3120089614250947,
      "learning_rate": 3.9037577663915885e-06,
      "loss": 0.4581,
      "step": 3620
    },
    {
      "epoch": 0.58,
      "grad_norm": 9.600982976314652,
      "learning_rate": 3.901211683268928e-06,
      "loss": 0.509,
      "step": 3621
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.7183008735063,
      "learning_rate": 3.898665899486493e-06,
      "loss": 0.3722,
      "step": 3622
    },
    {
      "epoch": 0.58,
      "grad_norm": 9.976191138077947,
      "learning_rate": 3.896120415737825e-06,
      "loss": 0.4613,
      "step": 3623
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.266893597870025,
      "learning_rate": 3.893575232716387e-06,
      "loss": 0.4501,
      "step": 3624
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.762090639064801,
      "learning_rate": 3.891030351115552e-06,
      "loss": 0.4326,
      "step": 3625
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.506032424269474,
      "learning_rate": 3.888485771628618e-06,
      "loss": 0.4586,
      "step": 3626
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.912987618128784,
      "learning_rate": 3.885941494948802e-06,
      "loss": 0.426,
      "step": 3627
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.437431704510847,
      "learning_rate": 3.8833975217692285e-06,
      "loss": 0.4345,
      "step": 3628
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.550368720017152,
      "learning_rate": 3.88085385278295e-06,
      "loss": 0.4994,
      "step": 3629
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.332374809278879,
      "learning_rate": 3.878310488682934e-06,
      "loss": 0.4562,
      "step": 3630
    },
    {
      "epoch": 0.59,
      "grad_norm": 8.230840107510328,
      "learning_rate": 3.875767430162058e-06,
      "loss": 0.4732,
      "step": 3631
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.83231383853078,
      "learning_rate": 3.873224677913124e-06,
      "loss": 0.4721,
      "step": 3632
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.178892473458497,
      "learning_rate": 3.870682232628848e-06,
      "loss": 0.4718,
      "step": 3633
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.496459612358705,
      "learning_rate": 3.8681400950018615e-06,
      "loss": 0.4771,
      "step": 3634
    },
    {
      "epoch": 0.59,
      "grad_norm": 11.923952159695997,
      "learning_rate": 3.865598265724713e-06,
      "loss": 0.4633,
      "step": 3635
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.026335488207349,
      "learning_rate": 3.8630567454898676e-06,
      "loss": 0.4344,
      "step": 3636
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.504905440483758,
      "learning_rate": 3.860515534989704e-06,
      "loss": 0.4192,
      "step": 3637
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.014858828256301,
      "learning_rate": 3.857974634916517e-06,
      "loss": 0.4509,
      "step": 3638
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.754091087243154,
      "learning_rate": 3.8554340459625235e-06,
      "loss": 0.4549,
      "step": 3639
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.859093904586176,
      "learning_rate": 3.852893768819843e-06,
      "loss": 0.4449,
      "step": 3640
    },
    {
      "epoch": 0.59,
      "grad_norm": 29.64419447681958,
      "learning_rate": 3.85035380418052e-06,
      "loss": 0.4291,
      "step": 3641
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.210152238570601,
      "learning_rate": 3.847814152736512e-06,
      "loss": 0.3387,
      "step": 3642
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.035947572784318,
      "learning_rate": 3.84527481517969e-06,
      "loss": 0.4258,
      "step": 3643
    },
    {
      "epoch": 0.59,
      "grad_norm": 19.787646272183938,
      "learning_rate": 3.842735792201837e-06,
      "loss": 0.4243,
      "step": 3644
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.731788367437117,
      "learning_rate": 3.840197084494653e-06,
      "loss": 0.4366,
      "step": 3645
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.918817686068713,
      "learning_rate": 3.8376586927497565e-06,
      "loss": 0.4773,
      "step": 3646
    },
    {
      "epoch": 0.59,
      "grad_norm": 8.885571734136057,
      "learning_rate": 3.835120617658669e-06,
      "loss": 0.4162,
      "step": 3647
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1830208672898663,
      "learning_rate": 3.832582859912838e-06,
      "loss": 0.475,
      "step": 3648
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.498436191571576,
      "learning_rate": 3.8300454202036155e-06,
      "loss": 0.3808,
      "step": 3649
    },
    {
      "epoch": 0.59,
      "grad_norm": 17.98603487343265,
      "learning_rate": 3.82750829922227e-06,
      "loss": 0.4493,
      "step": 3650
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.909395338043835,
      "learning_rate": 3.824971497659983e-06,
      "loss": 0.4034,
      "step": 3651
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.22223668236618,
      "learning_rate": 3.8224350162078526e-06,
      "loss": 0.4316,
      "step": 3652
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.01308389590162,
      "learning_rate": 3.81989885555688e-06,
      "loss": 0.4191,
      "step": 3653
    },
    {
      "epoch": 0.59,
      "grad_norm": 8.056124664503239,
      "learning_rate": 3.81736301639799e-06,
      "loss": 0.4459,
      "step": 3654
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.641203411304238,
      "learning_rate": 3.814827499422016e-06,
      "loss": 0.4235,
      "step": 3655
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.3396006412407315,
      "learning_rate": 3.8122923053196984e-06,
      "loss": 0.4509,
      "step": 3656
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.164843058392669,
      "learning_rate": 3.809757434781697e-06,
      "loss": 0.4646,
      "step": 3657
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.3734534295823,
      "learning_rate": 3.8072228884985803e-06,
      "loss": 0.4449,
      "step": 3658
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.976982389694024,
      "learning_rate": 3.8046886671608264e-06,
      "loss": 0.4129,
      "step": 3659
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.270447451515192,
      "learning_rate": 3.802154771458828e-06,
      "loss": 0.4785,
      "step": 3660
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.101905286580596,
      "learning_rate": 3.7996212020828915e-06,
      "loss": 0.4864,
      "step": 3661
    },
    {
      "epoch": 0.59,
      "grad_norm": 10.55494672430918,
      "learning_rate": 3.797087959723225e-06,
      "loss": 0.3858,
      "step": 3662
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.505594070812139,
      "learning_rate": 3.7945550450699585e-06,
      "loss": 0.4776,
      "step": 3663
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1903808073395685,
      "learning_rate": 3.7920224588131256e-06,
      "loss": 0.4844,
      "step": 3664
    },
    {
      "epoch": 0.59,
      "grad_norm": 8.78739447726393,
      "learning_rate": 3.7894902016426738e-06,
      "loss": 0.4583,
      "step": 3665
    },
    {
      "epoch": 0.59,
      "grad_norm": 10.233824708198869,
      "learning_rate": 3.786958274248458e-06,
      "loss": 0.3974,
      "step": 3666
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.480623375921674,
      "learning_rate": 3.7844266773202448e-06,
      "loss": 0.3707,
      "step": 3667
    },
    {
      "epoch": 0.59,
      "grad_norm": 12.381672442360758,
      "learning_rate": 3.7818954115477158e-06,
      "loss": 0.4492,
      "step": 3668
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.055208882728673,
      "learning_rate": 3.7793644776204503e-06,
      "loss": 0.431,
      "step": 3669
    },
    {
      "epoch": 0.59,
      "grad_norm": 8.815205364987651,
      "learning_rate": 3.7768338762279493e-06,
      "loss": 0.4042,
      "step": 3670
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.029064235347674,
      "learning_rate": 3.7743036080596184e-06,
      "loss": 0.4121,
      "step": 3671
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.087642521099984,
      "learning_rate": 3.77177367380477e-06,
      "loss": 0.493,
      "step": 3672
    },
    {
      "epoch": 0.59,
      "grad_norm": 8.63607252462849,
      "learning_rate": 3.7692440741526293e-06,
      "loss": 0.4406,
      "step": 3673
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.567058702018806,
      "learning_rate": 3.76671480979233e-06,
      "loss": 0.4102,
      "step": 3674
    },
    {
      "epoch": 0.59,
      "grad_norm": 15.337530033380453,
      "learning_rate": 3.7641858814129093e-06,
      "loss": 0.4147,
      "step": 3675
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.03743627688545,
      "learning_rate": 3.7616572897033223e-06,
      "loss": 0.3592,
      "step": 3676
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.525252334033553,
      "learning_rate": 3.7591290353524247e-06,
      "loss": 0.3393,
      "step": 3677
    },
    {
      "epoch": 0.59,
      "grad_norm": 13.131548968249211,
      "learning_rate": 3.7566011190489815e-06,
      "loss": 0.3776,
      "step": 3678
    },
    {
      "epoch": 0.59,
      "grad_norm": 22.71700553053672,
      "learning_rate": 3.7540735414816685e-06,
      "loss": 0.422,
      "step": 3679
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.265438864383878,
      "learning_rate": 3.7515463033390676e-06,
      "loss": 0.3941,
      "step": 3680
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.237391676604251,
      "learning_rate": 3.7490194053096668e-06,
      "loss": 0.3593,
      "step": 3681
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.86745225627778,
      "learning_rate": 3.7464928480818623e-06,
      "loss": 0.377,
      "step": 3682
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.455762457490691,
      "learning_rate": 3.7439666323439603e-06,
      "loss": 0.4798,
      "step": 3683
    },
    {
      "epoch": 0.59,
      "grad_norm": 16.159513827959582,
      "learning_rate": 3.741440758784172e-06,
      "loss": 0.4854,
      "step": 3684
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.562778439397837,
      "learning_rate": 3.738915228090611e-06,
      "loss": 0.4538,
      "step": 3685
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.474535527182887,
      "learning_rate": 3.736390040951304e-06,
      "loss": 0.4426,
      "step": 3686
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.779301552270303,
      "learning_rate": 3.733865198054184e-06,
      "loss": 0.4613,
      "step": 3687
    },
    {
      "epoch": 0.59,
      "grad_norm": 209.31357563799193,
      "learning_rate": 3.7313407000870826e-06,
      "loss": 0.409,
      "step": 3688
    },
    {
      "epoch": 0.59,
      "grad_norm": 22.335873580769764,
      "learning_rate": 3.728816547737745e-06,
      "loss": 0.4536,
      "step": 3689
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.755604488359913,
      "learning_rate": 3.7262927416938234e-06,
      "loss": 0.4706,
      "step": 3690
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.885297154441947,
      "learning_rate": 3.7237692826428662e-06,
      "loss": 0.4022,
      "step": 3691
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.27689791820019,
      "learning_rate": 3.721246171272336e-06,
      "loss": 0.4637,
      "step": 3692
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.930415162452213,
      "learning_rate": 3.718723408269599e-06,
      "loss": 0.4238,
      "step": 3693
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.1200510782945,
      "learning_rate": 3.7162009943219234e-06,
      "loss": 0.3634,
      "step": 3694
    },
    {
      "epoch": 0.6,
      "grad_norm": 9.665935309207324,
      "learning_rate": 3.7136789301164854e-06,
      "loss": 0.3866,
      "step": 3695
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.9151545218963095,
      "learning_rate": 3.7111572163403653e-06,
      "loss": 0.3851,
      "step": 3696
    },
    {
      "epoch": 0.6,
      "grad_norm": 9.083363945087907,
      "learning_rate": 3.7086358536805455e-06,
      "loss": 0.4504,
      "step": 3697
    },
    {
      "epoch": 0.6,
      "grad_norm": 10.244944004402264,
      "learning_rate": 3.7061148428239147e-06,
      "loss": 0.4392,
      "step": 3698
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.887578774537082,
      "learning_rate": 3.7035941844572687e-06,
      "loss": 0.3745,
      "step": 3699
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.858388654261098,
      "learning_rate": 3.701073879267302e-06,
      "loss": 0.4248,
      "step": 3700
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.518290630923201,
      "learning_rate": 3.698553927940615e-06,
      "loss": 0.3484,
      "step": 3701
    },
    {
      "epoch": 0.6,
      "grad_norm": 10.056516156579661,
      "learning_rate": 3.6960343311637132e-06,
      "loss": 0.4709,
      "step": 3702
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.847072226433738,
      "learning_rate": 3.6935150896230045e-06,
      "loss": 0.4833,
      "step": 3703
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.722526327995164,
      "learning_rate": 3.6909962040047964e-06,
      "loss": 0.4651,
      "step": 3704
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.928061030520333,
      "learning_rate": 3.6884776749953065e-06,
      "loss": 0.4839,
      "step": 3705
    },
    {
      "epoch": 0.6,
      "grad_norm": 9.031987960935293,
      "learning_rate": 3.6859595032806518e-06,
      "loss": 0.4651,
      "step": 3706
    },
    {
      "epoch": 0.6,
      "grad_norm": 12.69415506332248,
      "learning_rate": 3.683441689546849e-06,
      "loss": 0.4707,
      "step": 3707
    },
    {
      "epoch": 0.6,
      "grad_norm": 10.81833911102138,
      "learning_rate": 3.6809242344798207e-06,
      "loss": 0.4088,
      "step": 3708
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.8777602129088375,
      "learning_rate": 3.6784071387653926e-06,
      "loss": 0.4653,
      "step": 3709
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.51814568631765,
      "learning_rate": 3.675890403089289e-06,
      "loss": 0.3622,
      "step": 3710
    },
    {
      "epoch": 0.6,
      "grad_norm": 9.133428919448342,
      "learning_rate": 3.6733740281371377e-06,
      "loss": 0.4229,
      "step": 3711
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.249489178248336,
      "learning_rate": 3.670858014594473e-06,
      "loss": 0.4353,
      "step": 3712
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.052782825718367,
      "learning_rate": 3.66834236314672e-06,
      "loss": 0.3568,
      "step": 3713
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.437251849986943,
      "learning_rate": 3.665827074479215e-06,
      "loss": 0.4409,
      "step": 3714
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.262738580210845,
      "learning_rate": 3.6633121492771933e-06,
      "loss": 0.4165,
      "step": 3715
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2157340975349153,
      "learning_rate": 3.6607975882257875e-06,
      "loss": 0.4564,
      "step": 3716
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.504310033689151,
      "learning_rate": 3.6582833920100336e-06,
      "loss": 0.41,
      "step": 3717
    },
    {
      "epoch": 0.6,
      "grad_norm": 16.238393245347265,
      "learning_rate": 3.655769561314869e-06,
      "loss": 0.3976,
      "step": 3718
    },
    {
      "epoch": 0.6,
      "grad_norm": 14.699000444193242,
      "learning_rate": 3.653256096825129e-06,
      "loss": 0.4226,
      "step": 3719
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.032966775559035,
      "learning_rate": 3.6507429992255503e-06,
      "loss": 0.4245,
      "step": 3720
    },
    {
      "epoch": 0.6,
      "grad_norm": 9.009929203715936,
      "learning_rate": 3.648230269200775e-06,
      "loss": 0.4338,
      "step": 3721
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.506451985134628,
      "learning_rate": 3.645717907435332e-06,
      "loss": 0.4466,
      "step": 3722
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.828082705762483,
      "learning_rate": 3.6432059146136633e-06,
      "loss": 0.4234,
      "step": 3723
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.307577630534607,
      "learning_rate": 3.6406942914201045e-06,
      "loss": 0.4295,
      "step": 3724
    },
    {
      "epoch": 0.6,
      "grad_norm": 14.160162269904504,
      "learning_rate": 3.6381830385388907e-06,
      "loss": 0.3971,
      "step": 3725
    },
    {
      "epoch": 0.6,
      "grad_norm": 15.336906778733216,
      "learning_rate": 3.635672156654154e-06,
      "loss": 0.4407,
      "step": 3726
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.45320870741854,
      "learning_rate": 3.6331616464499297e-06,
      "loss": 0.3912,
      "step": 3727
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.706710924609895,
      "learning_rate": 3.6306515086101522e-06,
      "loss": 0.4802,
      "step": 3728
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.196630551758835,
      "learning_rate": 3.6281417438186484e-06,
      "loss": 0.3593,
      "step": 3729
    },
    {
      "epoch": 0.6,
      "grad_norm": 10.700947146404365,
      "learning_rate": 3.6256323527591496e-06,
      "loss": 0.3572,
      "step": 3730
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.769946956062393,
      "learning_rate": 3.623123336115284e-06,
      "loss": 0.521,
      "step": 3731
    },
    {
      "epoch": 0.6,
      "grad_norm": 13.089161791675759,
      "learning_rate": 3.6206146945705735e-06,
      "loss": 0.4652,
      "step": 3732
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.575411642801652,
      "learning_rate": 3.6181064288084423e-06,
      "loss": 0.4438,
      "step": 3733
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.746309761261565,
      "learning_rate": 3.6155985395122157e-06,
      "loss": 0.3801,
      "step": 3734
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.448202337907825,
      "learning_rate": 3.613091027365104e-06,
      "loss": 0.4199,
      "step": 3735
    },
    {
      "epoch": 0.6,
      "grad_norm": 10.070241965256509,
      "learning_rate": 3.610583893050229e-06,
      "loss": 0.4652,
      "step": 3736
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.887941479824022,
      "learning_rate": 3.6080771372506017e-06,
      "loss": 0.368,
      "step": 3737
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.91163129134379,
      "learning_rate": 3.6055707606491297e-06,
      "loss": 0.4361,
      "step": 3738
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.090419071065636,
      "learning_rate": 3.6030647639286196e-06,
      "loss": 0.3958,
      "step": 3739
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.230466721776446,
      "learning_rate": 3.6005591477717766e-06,
      "loss": 0.3909,
      "step": 3740
    },
    {
      "epoch": 0.6,
      "grad_norm": 13.87991254856921,
      "learning_rate": 3.598053912861196e-06,
      "loss": 0.4375,
      "step": 3741
    },
    {
      "epoch": 0.6,
      "grad_norm": 10.747180010743921,
      "learning_rate": 3.5955490598793734e-06,
      "loss": 0.3797,
      "step": 3742
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.6880895423569475,
      "learning_rate": 3.5930445895087017e-06,
      "loss": 0.4134,
      "step": 3743
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.093335667846363,
      "learning_rate": 3.5905405024314683e-06,
      "loss": 0.4798,
      "step": 3744
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.294738453695441,
      "learning_rate": 3.588036799329853e-06,
      "loss": 0.4043,
      "step": 3745
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.054470459013915,
      "learning_rate": 3.585533480885934e-06,
      "loss": 0.3901,
      "step": 3746
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.195631491838864,
      "learning_rate": 3.5830305477816863e-06,
      "loss": 0.4064,
      "step": 3747
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.745841675948784,
      "learning_rate": 3.580528000698975e-06,
      "loss": 0.3911,
      "step": 3748
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3528458823370044,
      "learning_rate": 3.5780258403195635e-06,
      "loss": 0.4904,
      "step": 3749
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.981756348983047,
      "learning_rate": 3.5755240673251125e-06,
      "loss": 0.3938,
      "step": 3750
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.195900763060852,
      "learning_rate": 3.5730226823971693e-06,
      "loss": 0.4209,
      "step": 3751
    },
    {
      "epoch": 0.6,
      "grad_norm": 23.919364099506623,
      "learning_rate": 3.5705216862171823e-06,
      "loss": 0.4268,
      "step": 3752
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.82379589992891,
      "learning_rate": 3.568021079466494e-06,
      "loss": 0.4276,
      "step": 3753
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.239583233597651,
      "learning_rate": 3.5655208628263345e-06,
      "loss": 0.4136,
      "step": 3754
    },
    {
      "epoch": 0.61,
      "grad_norm": 9.361350699378562,
      "learning_rate": 3.563021036977834e-06,
      "loss": 0.3942,
      "step": 3755
    },
    {
      "epoch": 0.61,
      "grad_norm": 9.241565549986962,
      "learning_rate": 3.560521602602014e-06,
      "loss": 0.4162,
      "step": 3756
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.338414108567371,
      "learning_rate": 3.5580225603797873e-06,
      "loss": 0.4303,
      "step": 3757
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.900660440844053,
      "learning_rate": 3.5555239109919647e-06,
      "loss": 0.4236,
      "step": 3758
    },
    {
      "epoch": 0.61,
      "grad_norm": 8.103109907220519,
      "learning_rate": 3.5530256551192467e-06,
      "loss": 0.4397,
      "step": 3759
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.665997272125521,
      "learning_rate": 3.5505277934422254e-06,
      "loss": 0.4133,
      "step": 3760
    },
    {
      "epoch": 0.61,
      "grad_norm": 8.054224710856591,
      "learning_rate": 3.5480303266413884e-06,
      "loss": 0.3622,
      "step": 3761
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.155547702029373,
      "learning_rate": 3.545533255397112e-06,
      "loss": 0.3229,
      "step": 3762
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.610373875386152,
      "learning_rate": 3.5430365803896736e-06,
      "loss": 0.4054,
      "step": 3763
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.375255654294126,
      "learning_rate": 3.540540302299229e-06,
      "loss": 0.4455,
      "step": 3764
    },
    {
      "epoch": 0.61,
      "grad_norm": 8.10028533181914,
      "learning_rate": 3.5380444218058374e-06,
      "loss": 0.5043,
      "step": 3765
    },
    {
      "epoch": 0.61,
      "grad_norm": 10.052960716124757,
      "learning_rate": 3.5355489395894448e-06,
      "loss": 0.425,
      "step": 3766
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.516449082340731,
      "learning_rate": 3.533053856329889e-06,
      "loss": 0.409,
      "step": 3767
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.942930605904436,
      "learning_rate": 3.5305591727068984e-06,
      "loss": 0.4446,
      "step": 3768
    },
    {
      "epoch": 0.61,
      "grad_norm": 8.297245342851765,
      "learning_rate": 3.5280648894000957e-06,
      "loss": 0.3697,
      "step": 3769
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.916684787761617,
      "learning_rate": 3.5255710070889903e-06,
      "loss": 0.4373,
      "step": 3770
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.292164430953637,
      "learning_rate": 3.5230775264529837e-06,
      "loss": 0.5054,
      "step": 3771
    },
    {
      "epoch": 0.61,
      "grad_norm": 10.339643447197412,
      "learning_rate": 3.520584448171375e-06,
      "loss": 0.4004,
      "step": 3772
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.186865972558311,
      "learning_rate": 3.518091772923339e-06,
      "loss": 0.4523,
      "step": 3773
    },
    {
      "epoch": 0.61,
      "grad_norm": 13.36042797616664,
      "learning_rate": 3.515599501387954e-06,
      "loss": 0.5128,
      "step": 3774
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.551796063271116,
      "learning_rate": 3.5131076342441838e-06,
      "loss": 0.4716,
      "step": 3775
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0742615000135156,
      "learning_rate": 3.5106161721708797e-06,
      "loss": 0.4477,
      "step": 3776
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2253269186874773,
      "learning_rate": 3.508125115846785e-06,
      "loss": 0.4614,
      "step": 3777
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.183087179518292,
      "learning_rate": 3.5056344659505335e-06,
      "loss": 0.4108,
      "step": 3778
    },
    {
      "epoch": 0.61,
      "grad_norm": 10.248108339690626,
      "learning_rate": 3.503144223160644e-06,
      "loss": 0.4677,
      "step": 3779
    },
    {
      "epoch": 0.61,
      "grad_norm": 14.770455236119721,
      "learning_rate": 3.5006543881555304e-06,
      "loss": 0.3876,
      "step": 3780
    },
    {
      "epoch": 0.61,
      "grad_norm": 9.143019175180838,
      "learning_rate": 3.4981649616134912e-06,
      "loss": 0.4941,
      "step": 3781
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.69436401486207,
      "learning_rate": 3.495675944212715e-06,
      "loss": 0.3964,
      "step": 3782
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.754006586841706,
      "learning_rate": 3.4931873366312785e-06,
      "loss": 0.4148,
      "step": 3783
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.979897179747272,
      "learning_rate": 3.490699139547146e-06,
      "loss": 0.3954,
      "step": 3784
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.676939567287066,
      "learning_rate": 3.4882113536381744e-06,
      "loss": 0.4091,
      "step": 3785
    },
    {
      "epoch": 0.61,
      "grad_norm": 14.382115901387778,
      "learning_rate": 3.4857239795821003e-06,
      "loss": 0.4739,
      "step": 3786
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.6734561881557335,
      "learning_rate": 3.483237018056556e-06,
      "loss": 0.3917,
      "step": 3787
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.35804523611172,
      "learning_rate": 3.480750469739059e-06,
      "loss": 0.4823,
      "step": 3788
    },
    {
      "epoch": 0.61,
      "grad_norm": 428.8467476679283,
      "learning_rate": 3.478264335307011e-06,
      "loss": 0.4766,
      "step": 3789
    },
    {
      "epoch": 0.61,
      "grad_norm": 13.674578244281337,
      "learning_rate": 3.475778615437706e-06,
      "loss": 0.4235,
      "step": 3790
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.99967296003406,
      "learning_rate": 3.4732933108083218e-06,
      "loss": 0.4681,
      "step": 3791
    },
    {
      "epoch": 0.61,
      "grad_norm": 12.5668404229954,
      "learning_rate": 3.470808422095923e-06,
      "loss": 0.5383,
      "step": 3792
    },
    {
      "epoch": 0.61,
      "grad_norm": 14.801378649572548,
      "learning_rate": 3.4683239499774606e-06,
      "loss": 0.4073,
      "step": 3793
    },
    {
      "epoch": 0.61,
      "grad_norm": 12.397697395657023,
      "learning_rate": 3.465839895129779e-06,
      "loss": 0.3943,
      "step": 3794
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.652280886539598,
      "learning_rate": 3.463356258229596e-06,
      "loss": 0.4095,
      "step": 3795
    },
    {
      "epoch": 0.61,
      "grad_norm": 24.702249208867922,
      "learning_rate": 3.4608730399535273e-06,
      "loss": 0.3952,
      "step": 3796
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.632785379963838,
      "learning_rate": 3.4583902409780693e-06,
      "loss": 0.445,
      "step": 3797
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.171934297374592,
      "learning_rate": 3.4559078619796036e-06,
      "loss": 0.4495,
      "step": 3798
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2998072634038784,
      "learning_rate": 3.4534259036343996e-06,
      "loss": 0.5017,
      "step": 3799
    },
    {
      "epoch": 0.61,
      "grad_norm": 18.305652232070777,
      "learning_rate": 3.450944366618613e-06,
      "loss": 0.3674,
      "step": 3800
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.686535922367812,
      "learning_rate": 3.4484632516082784e-06,
      "loss": 0.4571,
      "step": 3801
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.006503850121893,
      "learning_rate": 3.445982559279322e-06,
      "loss": 0.4502,
      "step": 3802
    },
    {
      "epoch": 0.61,
      "grad_norm": 17.829982394009967,
      "learning_rate": 3.4435022903075536e-06,
      "loss": 0.4442,
      "step": 3803
    },
    {
      "epoch": 0.61,
      "grad_norm": 11.590574273880994,
      "learning_rate": 3.441022445368668e-06,
      "loss": 0.4958,
      "step": 3804
    },
    {
      "epoch": 0.61,
      "grad_norm": 11.262082922828691,
      "learning_rate": 3.4385430251382407e-06,
      "loss": 0.3915,
      "step": 3805
    },
    {
      "epoch": 0.61,
      "grad_norm": 8.018203507209071,
      "learning_rate": 3.4360640302917353e-06,
      "loss": 0.3981,
      "step": 3806
    },
    {
      "epoch": 0.61,
      "grad_norm": 8.202660890425134,
      "learning_rate": 3.433585461504499e-06,
      "loss": 0.3819,
      "step": 3807
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.93954200337496,
      "learning_rate": 3.4311073194517596e-06,
      "loss": 0.4655,
      "step": 3808
    },
    {
      "epoch": 0.61,
      "grad_norm": 9.129071776646366,
      "learning_rate": 3.4286296048086343e-06,
      "loss": 0.3867,
      "step": 3809
    },
    {
      "epoch": 0.61,
      "grad_norm": 13.246350404321706,
      "learning_rate": 3.42615231825012e-06,
      "loss": 0.4572,
      "step": 3810
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.240479734912556,
      "learning_rate": 3.423675460451097e-06,
      "loss": 0.4101,
      "step": 3811
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.29477836530849,
      "learning_rate": 3.4211990320863307e-06,
      "loss": 0.437,
      "step": 3812
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.405405629330422,
      "learning_rate": 3.4187230338304684e-06,
      "loss": 0.388,
      "step": 3813
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.236224966620971,
      "learning_rate": 3.416247466358039e-06,
      "loss": 0.5006,
      "step": 3814
    },
    {
      "epoch": 0.61,
      "grad_norm": 17.041243045413403,
      "learning_rate": 3.413772330343455e-06,
      "loss": 0.4239,
      "step": 3815
    },
    {
      "epoch": 0.61,
      "grad_norm": 12.480223488697884,
      "learning_rate": 3.4112976264610167e-06,
      "loss": 0.3808,
      "step": 3816
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.5494025940608145,
      "learning_rate": 3.408823355384894e-06,
      "loss": 0.4783,
      "step": 3817
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.707765716198661,
      "learning_rate": 3.406349517789151e-06,
      "loss": 0.4165,
      "step": 3818
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.420873159377349,
      "learning_rate": 3.4038761143477296e-06,
      "loss": 0.489,
      "step": 3819
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.120507430164611,
      "learning_rate": 3.4014031457344517e-06,
      "loss": 0.4016,
      "step": 3820
    },
    {
      "epoch": 0.62,
      "grad_norm": 13.344251288601486,
      "learning_rate": 3.3989306126230226e-06,
      "loss": 0.4624,
      "step": 3821
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.859040899404327,
      "learning_rate": 3.3964585156870267e-06,
      "loss": 0.462,
      "step": 3822
    },
    {
      "epoch": 0.62,
      "grad_norm": 21.486218302671194,
      "learning_rate": 3.393986855599936e-06,
      "loss": 0.4103,
      "step": 3823
    },
    {
      "epoch": 0.62,
      "grad_norm": 8.991755064636807,
      "learning_rate": 3.391515633035093e-06,
      "loss": 0.4447,
      "step": 3824
    },
    {
      "epoch": 0.62,
      "grad_norm": 8.176807277648642,
      "learning_rate": 3.389044848665731e-06,
      "loss": 0.4207,
      "step": 3825
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.282797818435638,
      "learning_rate": 3.3865745031649595e-06,
      "loss": 0.4313,
      "step": 3826
    },
    {
      "epoch": 0.62,
      "grad_norm": 11.243219379621953,
      "learning_rate": 3.3841045972057663e-06,
      "loss": 0.4216,
      "step": 3827
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3200055250596865,
      "learning_rate": 3.3816351314610235e-06,
      "loss": 0.5029,
      "step": 3828
    },
    {
      "epoch": 0.62,
      "grad_norm": 14.619968733439679,
      "learning_rate": 3.379166106603482e-06,
      "loss": 0.508,
      "step": 3829
    },
    {
      "epoch": 0.62,
      "grad_norm": 18.874955412737965,
      "learning_rate": 3.3766975233057715e-06,
      "loss": 0.4382,
      "step": 3830
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.676890388709013,
      "learning_rate": 3.3742293822404005e-06,
      "loss": 0.4436,
      "step": 3831
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.789412793873831,
      "learning_rate": 3.371761684079763e-06,
      "loss": 0.4341,
      "step": 3832
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.423450605314177,
      "learning_rate": 3.369294429496124e-06,
      "loss": 0.4295,
      "step": 3833
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.961196154707833,
      "learning_rate": 3.366827619161632e-06,
      "loss": 0.3611,
      "step": 3834
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.36499251201153,
      "learning_rate": 3.364361253748318e-06,
      "loss": 0.459,
      "step": 3835
    },
    {
      "epoch": 0.62,
      "grad_norm": 9.380048467270422,
      "learning_rate": 3.361895333928083e-06,
      "loss": 0.4167,
      "step": 3836
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.59706560788016,
      "learning_rate": 3.3594298603727126e-06,
      "loss": 0.4643,
      "step": 3837
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.863650471995008,
      "learning_rate": 3.356964833753875e-06,
      "loss": 0.4243,
      "step": 3838
    },
    {
      "epoch": 0.62,
      "grad_norm": 9.107274109106763,
      "learning_rate": 3.3545002547431034e-06,
      "loss": 0.4839,
      "step": 3839
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.431758593392631,
      "learning_rate": 3.3520361240118216e-06,
      "loss": 0.409,
      "step": 3840
    },
    {
      "epoch": 0.62,
      "grad_norm": 11.011473926008383,
      "learning_rate": 3.3495724422313262e-06,
      "loss": 0.3431,
      "step": 3841
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.191946366407205,
      "learning_rate": 3.347109210072793e-06,
      "loss": 0.404,
      "step": 3842
    },
    {
      "epoch": 0.62,
      "grad_norm": 11.026464907581602,
      "learning_rate": 3.3446464282072723e-06,
      "loss": 0.4647,
      "step": 3843
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.913846886798677,
      "learning_rate": 3.3421840973056935e-06,
      "loss": 0.4091,
      "step": 3844
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.506033097245198,
      "learning_rate": 3.3397222180388677e-06,
      "loss": 0.4427,
      "step": 3845
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.6779302935728735,
      "learning_rate": 3.3372607910774726e-06,
      "loss": 0.4282,
      "step": 3846
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.323013515910669,
      "learning_rate": 3.3347998170920724e-06,
      "loss": 0.4865,
      "step": 3847
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.391155649929585,
      "learning_rate": 3.3323392967531043e-06,
      "loss": 0.4322,
      "step": 3848
    },
    {
      "epoch": 0.62,
      "grad_norm": 12.864544592814104,
      "learning_rate": 3.32987923073088e-06,
      "loss": 0.4345,
      "step": 3849
    },
    {
      "epoch": 0.62,
      "grad_norm": 8.392522288107367,
      "learning_rate": 3.327419619695591e-06,
      "loss": 0.4896,
      "step": 3850
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.465387259429603,
      "learning_rate": 3.3249604643173037e-06,
      "loss": 0.487,
      "step": 3851
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.074298185066563,
      "learning_rate": 3.3225017652659577e-06,
      "loss": 0.3377,
      "step": 3852
    },
    {
      "epoch": 0.62,
      "grad_norm": 9.399331281596357,
      "learning_rate": 3.3200435232113694e-06,
      "loss": 0.432,
      "step": 3853
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.315498840291,
      "learning_rate": 3.3175857388232376e-06,
      "loss": 0.3759,
      "step": 3854
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.302053744739064,
      "learning_rate": 3.3151284127711227e-06,
      "loss": 0.3961,
      "step": 3855
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.77417220119939,
      "learning_rate": 3.312671545724474e-06,
      "loss": 0.3881,
      "step": 3856
    },
    {
      "epoch": 0.62,
      "grad_norm": 11.077963968298171,
      "learning_rate": 3.3102151383526077e-06,
      "loss": 0.4117,
      "step": 3857
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.02025396257588,
      "learning_rate": 3.3077591913247166e-06,
      "loss": 0.4116,
      "step": 3858
    },
    {
      "epoch": 0.62,
      "grad_norm": 8.035588478737553,
      "learning_rate": 3.305303705309868e-06,
      "loss": 0.4428,
      "step": 3859
    },
    {
      "epoch": 0.62,
      "grad_norm": 8.161759837477927,
      "learning_rate": 3.3028486809770046e-06,
      "loss": 0.4482,
      "step": 3860
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.542874776982548,
      "learning_rate": 3.300394118994944e-06,
      "loss": 0.4085,
      "step": 3861
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.103733324244878,
      "learning_rate": 3.297940020032374e-06,
      "loss": 0.4382,
      "step": 3862
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.204240749145806,
      "learning_rate": 3.295486384757861e-06,
      "loss": 0.4924,
      "step": 3863
    },
    {
      "epoch": 0.62,
      "grad_norm": 24.438732780196176,
      "learning_rate": 3.2930332138398422e-06,
      "loss": 0.3813,
      "step": 3864
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.095960342518633,
      "learning_rate": 3.2905805079466284e-06,
      "loss": 0.4399,
      "step": 3865
    },
    {
      "epoch": 0.62,
      "grad_norm": 13.231829863412326,
      "learning_rate": 3.2881282677464034e-06,
      "loss": 0.4536,
      "step": 3866
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.483480643881775,
      "learning_rate": 3.2856764939072294e-06,
      "loss": 0.3819,
      "step": 3867
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1071807986244537,
      "learning_rate": 3.283225187097031e-06,
      "loss": 0.4808,
      "step": 3868
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.464704668664216,
      "learning_rate": 3.2807743479836155e-06,
      "loss": 0.3978,
      "step": 3869
    },
    {
      "epoch": 0.62,
      "grad_norm": 9.297451467508541,
      "learning_rate": 3.27832397723466e-06,
      "loss": 0.4092,
      "step": 3870
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.227927326257817,
      "learning_rate": 3.27587407551771e-06,
      "loss": 0.4667,
      "step": 3871
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.645441072236689,
      "learning_rate": 3.273424643500187e-06,
      "loss": 0.4388,
      "step": 3872
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.832650695736625,
      "learning_rate": 3.2709756818493867e-06,
      "loss": 0.4105,
      "step": 3873
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.263621282172071,
      "learning_rate": 3.26852719123247e-06,
      "loss": 0.434,
      "step": 3874
    },
    {
      "epoch": 0.62,
      "grad_norm": 55.43658253287473,
      "learning_rate": 3.266079172316473e-06,
      "loss": 0.411,
      "step": 3875
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.620225794016527,
      "learning_rate": 3.263631625768309e-06,
      "loss": 0.4445,
      "step": 3876
    },
    {
      "epoch": 0.62,
      "grad_norm": 8.40532565964153,
      "learning_rate": 3.2611845522547503e-06,
      "loss": 0.425,
      "step": 3877
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.454488073777527,
      "learning_rate": 3.2587379524424513e-06,
      "loss": 0.4532,
      "step": 3878
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.984955103541048,
      "learning_rate": 3.2562918269979334e-06,
      "loss": 0.4105,
      "step": 3879
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.221844218672381,
      "learning_rate": 3.253846176587586e-06,
      "loss": 0.458,
      "step": 3880
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.837717736597488,
      "learning_rate": 3.251401001877673e-06,
      "loss": 0.5099,
      "step": 3881
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4254505434014106,
      "learning_rate": 3.2489563035343276e-06,
      "loss": 0.4804,
      "step": 3882
    },
    {
      "epoch": 0.63,
      "grad_norm": 9.259000450344796,
      "learning_rate": 3.246512082223555e-06,
      "loss": 0.3999,
      "step": 3883
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.7410027747933725,
      "learning_rate": 3.2440683386112238e-06,
      "loss": 0.4594,
      "step": 3884
    },
    {
      "epoch": 0.63,
      "grad_norm": 11.116035129808385,
      "learning_rate": 3.24162507336308e-06,
      "loss": 0.3986,
      "step": 3885
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.969200842735214,
      "learning_rate": 3.2391822871447377e-06,
      "loss": 0.3892,
      "step": 3886
    },
    {
      "epoch": 0.63,
      "grad_norm": 9.349340535556742,
      "learning_rate": 3.2367399806216765e-06,
      "loss": 0.4179,
      "step": 3887
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2729133500196799,
      "learning_rate": 3.234298154459249e-06,
      "loss": 0.4769,
      "step": 3888
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.140624857035874,
      "learning_rate": 3.231856809322677e-06,
      "loss": 0.3546,
      "step": 3889
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.9891059664066155,
      "learning_rate": 3.229415945877048e-06,
      "loss": 0.349,
      "step": 3890
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.826321831394866,
      "learning_rate": 3.226975564787322e-06,
      "loss": 0.4171,
      "step": 3891
    },
    {
      "epoch": 0.63,
      "grad_norm": 8.16171566729037,
      "learning_rate": 3.224535666718327e-06,
      "loss": 0.3866,
      "step": 3892
    },
    {
      "epoch": 0.63,
      "grad_norm": 9.359787585511752,
      "learning_rate": 3.2220962523347567e-06,
      "loss": 0.4196,
      "step": 3893
    },
    {
      "epoch": 0.63,
      "grad_norm": 10.413002510828044,
      "learning_rate": 3.219657322301175e-06,
      "loss": 0.4043,
      "step": 3894
    },
    {
      "epoch": 0.63,
      "grad_norm": 9.185744630369937,
      "learning_rate": 3.2172188772820154e-06,
      "loss": 0.4139,
      "step": 3895
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.649074456561787,
      "learning_rate": 3.214780917941575e-06,
      "loss": 0.3456,
      "step": 3896
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.444724143890425,
      "learning_rate": 3.212343444944022e-06,
      "loss": 0.4098,
      "step": 3897
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.08327011452638,
      "learning_rate": 3.209906458953394e-06,
      "loss": 0.4812,
      "step": 3898
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.938854776164191,
      "learning_rate": 3.207469960633588e-06,
      "loss": 0.4074,
      "step": 3899
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.325535071971464,
      "learning_rate": 3.2050339506483774e-06,
      "loss": 0.398,
      "step": 3900
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.2926373815058145,
      "learning_rate": 3.2025984296613965e-06,
      "loss": 0.4667,
      "step": 3901
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.915290186431194,
      "learning_rate": 3.200163398336151e-06,
      "loss": 0.4143,
      "step": 3902
    },
    {
      "epoch": 0.63,
      "grad_norm": 14.470093075954667,
      "learning_rate": 3.1977288573360064e-06,
      "loss": 0.3875,
      "step": 3903
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.0085829274895435,
      "learning_rate": 3.1952948073242006e-06,
      "loss": 0.4297,
      "step": 3904
    },
    {
      "epoch": 0.63,
      "grad_norm": 11.077590310644805,
      "learning_rate": 3.19286124896384e-06,
      "loss": 0.4757,
      "step": 3905
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.812140673266212,
      "learning_rate": 3.190428182917885e-06,
      "loss": 0.4312,
      "step": 3906
    },
    {
      "epoch": 0.63,
      "grad_norm": 10.613660991900307,
      "learning_rate": 3.187995609849176e-06,
      "loss": 0.4225,
      "step": 3907
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.314089279811991,
      "learning_rate": 3.1855635304204113e-06,
      "loss": 0.3917,
      "step": 3908
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.001666234912583,
      "learning_rate": 3.1831319452941557e-06,
      "loss": 0.4231,
      "step": 3909
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.595272503287338,
      "learning_rate": 3.1807008551328407e-06,
      "loss": 0.4178,
      "step": 3910
    },
    {
      "epoch": 0.63,
      "grad_norm": 16.558727964983014,
      "learning_rate": 3.1782702605987623e-06,
      "loss": 0.3813,
      "step": 3911
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.94579978481947,
      "learning_rate": 3.175840162354081e-06,
      "loss": 0.5287,
      "step": 3912
    },
    {
      "epoch": 0.63,
      "grad_norm": 12.6798435583847,
      "learning_rate": 3.1734105610608213e-06,
      "loss": 0.4233,
      "step": 3913
    },
    {
      "epoch": 0.63,
      "grad_norm": 14.5699831367811,
      "learning_rate": 3.1709814573808766e-06,
      "loss": 0.3999,
      "step": 3914
    },
    {
      "epoch": 0.63,
      "grad_norm": 8.218165217062538,
      "learning_rate": 3.168552851976e-06,
      "loss": 0.382,
      "step": 3915
    },
    {
      "epoch": 0.63,
      "grad_norm": 11.811377247396477,
      "learning_rate": 3.1661247455078097e-06,
      "loss": 0.4672,
      "step": 3916
    },
    {
      "epoch": 0.63,
      "grad_norm": 8.592689211836243,
      "learning_rate": 3.163697138637791e-06,
      "loss": 0.4739,
      "step": 3917
    },
    {
      "epoch": 0.63,
      "grad_norm": 10.725786428568863,
      "learning_rate": 3.161270032027289e-06,
      "loss": 0.503,
      "step": 3918
    },
    {
      "epoch": 0.63,
      "grad_norm": 8.493780431419871,
      "learning_rate": 3.1588434263375146e-06,
      "loss": 0.4439,
      "step": 3919
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2410384633187128,
      "learning_rate": 3.156417322229543e-06,
      "loss": 0.4685,
      "step": 3920
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.98844777471595,
      "learning_rate": 3.153991720364313e-06,
      "loss": 0.4306,
      "step": 3921
    },
    {
      "epoch": 0.63,
      "grad_norm": 11.036505276721325,
      "learning_rate": 3.151566621402622e-06,
      "loss": 0.5004,
      "step": 3922
    },
    {
      "epoch": 0.63,
      "grad_norm": 11.231747102755259,
      "learning_rate": 3.1491420260051362e-06,
      "loss": 0.4135,
      "step": 3923
    },
    {
      "epoch": 0.63,
      "grad_norm": 8.11558368618105,
      "learning_rate": 3.146717934832383e-06,
      "loss": 0.4053,
      "step": 3924
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.633279402436404,
      "learning_rate": 3.1442943485447493e-06,
      "loss": 0.416,
      "step": 3925
    },
    {
      "epoch": 0.63,
      "grad_norm": 10.104233276438467,
      "learning_rate": 3.1418712678024866e-06,
      "loss": 0.4342,
      "step": 3926
    },
    {
      "epoch": 0.63,
      "grad_norm": 8.910542264331065,
      "learning_rate": 3.1394486932657133e-06,
      "loss": 0.3737,
      "step": 3927
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2150810229222804,
      "learning_rate": 3.137026625594399e-06,
      "loss": 0.491,
      "step": 3928
    },
    {
      "epoch": 0.63,
      "grad_norm": 8.00445592365204,
      "learning_rate": 3.1346050654483867e-06,
      "loss": 0.3584,
      "step": 3929
    },
    {
      "epoch": 0.63,
      "grad_norm": 13.498654365312982,
      "learning_rate": 3.132184013487375e-06,
      "loss": 0.4198,
      "step": 3930
    },
    {
      "epoch": 0.63,
      "grad_norm": 10.36515321305887,
      "learning_rate": 3.129763470370924e-06,
      "loss": 0.3742,
      "step": 3931
    },
    {
      "epoch": 0.63,
      "grad_norm": 16.03945945439608,
      "learning_rate": 3.1273434367584567e-06,
      "loss": 0.4737,
      "step": 3932
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.92103714372606,
      "learning_rate": 3.124923913309259e-06,
      "loss": 0.4227,
      "step": 3933
    },
    {
      "epoch": 0.63,
      "grad_norm": 9.31115060656964,
      "learning_rate": 3.1225049006824724e-06,
      "loss": 0.4625,
      "step": 3934
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.868615355852885,
      "learning_rate": 3.1200863995371035e-06,
      "loss": 0.4283,
      "step": 3935
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.044001908186342,
      "learning_rate": 3.1176684105320208e-06,
      "loss": 0.405,
      "step": 3936
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.718611906311166,
      "learning_rate": 3.1152509343259494e-06,
      "loss": 0.3942,
      "step": 3937
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.90730884931657,
      "learning_rate": 3.112833971577478e-06,
      "loss": 0.4438,
      "step": 3938
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.528634175574992,
      "learning_rate": 3.1104175229450517e-06,
      "loss": 0.4525,
      "step": 3939
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.242032131667083,
      "learning_rate": 3.1080015890869796e-06,
      "loss": 0.4349,
      "step": 3940
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.186320444867246,
      "learning_rate": 3.1055861706614264e-06,
      "loss": 0.3913,
      "step": 3941
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.684996804129319,
      "learning_rate": 3.1031712683264204e-06,
      "loss": 0.3745,
      "step": 3942
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.057233039378118,
      "learning_rate": 3.1007568827398495e-06,
      "loss": 0.3881,
      "step": 3943
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.267704559789786,
      "learning_rate": 3.0983430145594547e-06,
      "loss": 0.3886,
      "step": 3944
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.302044610716438,
      "learning_rate": 3.0959296644428427e-06,
      "loss": 0.3984,
      "step": 3945
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.549173992354037,
      "learning_rate": 3.0935168330474763e-06,
      "loss": 0.4953,
      "step": 3946
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.603657268953786,
      "learning_rate": 3.0911045210306767e-06,
      "loss": 0.3674,
      "step": 3947
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.1604961072220465,
      "learning_rate": 3.088692729049624e-06,
      "loss": 0.4192,
      "step": 3948
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.08077714788108,
      "learning_rate": 3.0862814577613598e-06,
      "loss": 0.4532,
      "step": 3949
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3097702487268492,
      "learning_rate": 3.083870707822776e-06,
      "loss": 0.4987,
      "step": 3950
    },
    {
      "epoch": 0.64,
      "grad_norm": 11.437046106916698,
      "learning_rate": 3.08146047989063e-06,
      "loss": 0.4311,
      "step": 3951
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.133323429468513,
      "learning_rate": 3.079050774621536e-06,
      "loss": 0.503,
      "step": 3952
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.50629168000756,
      "learning_rate": 3.0766415926719606e-06,
      "loss": 0.4408,
      "step": 3953
    },
    {
      "epoch": 0.64,
      "grad_norm": 18.933211029018103,
      "learning_rate": 3.074232934698234e-06,
      "loss": 0.4829,
      "step": 3954
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.972536101586685,
      "learning_rate": 3.07182480135654e-06,
      "loss": 0.4056,
      "step": 3955
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.2784843992299715,
      "learning_rate": 3.06941719330292e-06,
      "loss": 0.4231,
      "step": 3956
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.038697659989545,
      "learning_rate": 3.067010111193272e-06,
      "loss": 0.4833,
      "step": 3957
    },
    {
      "epoch": 0.64,
      "grad_norm": 10.645402990183074,
      "learning_rate": 3.0646035556833563e-06,
      "loss": 0.4302,
      "step": 3958
    },
    {
      "epoch": 0.64,
      "grad_norm": 10.634149823846363,
      "learning_rate": 3.0621975274287784e-06,
      "loss": 0.4277,
      "step": 3959
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.424690088210346,
      "learning_rate": 3.059792027085011e-06,
      "loss": 0.4327,
      "step": 3960
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.553280557506933,
      "learning_rate": 3.0573870553073776e-06,
      "loss": 0.4139,
      "step": 3961
    },
    {
      "epoch": 0.64,
      "grad_norm": 10.681732186006151,
      "learning_rate": 3.0549826127510595e-06,
      "loss": 0.4596,
      "step": 3962
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.385413164021655,
      "learning_rate": 3.0525787000710915e-06,
      "loss": 0.347,
      "step": 3963
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.152752065287386,
      "learning_rate": 3.0501753179223657e-06,
      "loss": 0.3941,
      "step": 3964
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.953582916236869,
      "learning_rate": 3.0477724669596326e-06,
      "loss": 0.4578,
      "step": 3965
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.498910858175644,
      "learning_rate": 3.045370147837492e-06,
      "loss": 0.4675,
      "step": 3966
    },
    {
      "epoch": 0.64,
      "grad_norm": 10.060906483355321,
      "learning_rate": 3.042968361210403e-06,
      "loss": 0.4214,
      "step": 3967
    },
    {
      "epoch": 0.64,
      "grad_norm": 10.71558569204994,
      "learning_rate": 3.040567107732679e-06,
      "loss": 0.475,
      "step": 3968
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.157128913000169,
      "learning_rate": 3.0381663880584855e-06,
      "loss": 0.3907,
      "step": 3969
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.5998373731786595,
      "learning_rate": 3.0357662028418455e-06,
      "loss": 0.4336,
      "step": 3970
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.418040781831431,
      "learning_rate": 3.0333665527366394e-06,
      "loss": 0.4271,
      "step": 3971
    },
    {
      "epoch": 0.64,
      "grad_norm": 12.156264782261951,
      "learning_rate": 3.0309674383965915e-06,
      "loss": 0.431,
      "step": 3972
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.216031073810308,
      "learning_rate": 3.0285688604752916e-06,
      "loss": 0.4591,
      "step": 3973
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.262753958624208,
      "learning_rate": 3.026170819626178e-06,
      "loss": 0.5165,
      "step": 3974
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.654005532859886,
      "learning_rate": 3.0237733165025408e-06,
      "loss": 0.4638,
      "step": 3975
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.048303571177509,
      "learning_rate": 3.021376351757527e-06,
      "loss": 0.4399,
      "step": 3976
    },
    {
      "epoch": 0.64,
      "grad_norm": 10.179051111259673,
      "learning_rate": 3.0189799260441377e-06,
      "loss": 0.4978,
      "step": 3977
    },
    {
      "epoch": 0.64,
      "grad_norm": 11.254755996971404,
      "learning_rate": 3.0165840400152218e-06,
      "loss": 0.4282,
      "step": 3978
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.758616489528508,
      "learning_rate": 3.014188694323486e-06,
      "loss": 0.4649,
      "step": 3979
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.131247225133208,
      "learning_rate": 3.0117938896214904e-06,
      "loss": 0.4255,
      "step": 3980
    },
    {
      "epoch": 0.64,
      "grad_norm": 13.186642966683491,
      "learning_rate": 3.0093996265616447e-06,
      "loss": 0.4065,
      "step": 3981
    },
    {
      "epoch": 0.64,
      "grad_norm": 10.123510012404244,
      "learning_rate": 3.007005905796212e-06,
      "loss": 0.4442,
      "step": 3982
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.047122857490186,
      "learning_rate": 3.0046127279773067e-06,
      "loss": 0.4492,
      "step": 3983
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.592404601568773,
      "learning_rate": 3.002220093756899e-06,
      "loss": 0.5083,
      "step": 3984
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.218807573454606,
      "learning_rate": 2.999828003786806e-06,
      "loss": 0.4177,
      "step": 3985
    },
    {
      "epoch": 0.64,
      "grad_norm": 10.038340043835284,
      "learning_rate": 2.9974364587186988e-06,
      "loss": 0.4441,
      "step": 3986
    },
    {
      "epoch": 0.64,
      "grad_norm": 11.697258013265095,
      "learning_rate": 2.995045459204104e-06,
      "loss": 0.4243,
      "step": 3987
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.5521696703458385,
      "learning_rate": 2.9926550058943905e-06,
      "loss": 0.4101,
      "step": 3988
    },
    {
      "epoch": 0.64,
      "grad_norm": 12.55235533345434,
      "learning_rate": 2.9902650994407867e-06,
      "loss": 0.483,
      "step": 3989
    },
    {
      "epoch": 0.64,
      "grad_norm": 15.623612132289999,
      "learning_rate": 2.9878757404943694e-06,
      "loss": 0.4687,
      "step": 3990
    },
    {
      "epoch": 0.64,
      "grad_norm": 11.771551649035427,
      "learning_rate": 2.985486929706064e-06,
      "loss": 0.4644,
      "step": 3991
    },
    {
      "epoch": 0.64,
      "grad_norm": 10.948094345839378,
      "learning_rate": 2.9830986677266495e-06,
      "loss": 0.3678,
      "step": 3992
    },
    {
      "epoch": 0.64,
      "grad_norm": 13.72985317510009,
      "learning_rate": 2.980710955206755e-06,
      "loss": 0.4085,
      "step": 3993
    },
    {
      "epoch": 0.64,
      "grad_norm": 9.261450208768267,
      "learning_rate": 2.9783237927968567e-06,
      "loss": 0.3972,
      "step": 3994
    },
    {
      "epoch": 0.64,
      "grad_norm": 10.76624855982844,
      "learning_rate": 2.9759371811472857e-06,
      "loss": 0.4219,
      "step": 3995
    },
    {
      "epoch": 0.64,
      "grad_norm": 12.306752379919638,
      "learning_rate": 2.9735511209082213e-06,
      "loss": 0.4408,
      "step": 3996
    },
    {
      "epoch": 0.64,
      "grad_norm": 10.812893461065988,
      "learning_rate": 2.9711656127296895e-06,
      "loss": 0.4182,
      "step": 3997
    },
    {
      "epoch": 0.64,
      "grad_norm": 12.090564397952443,
      "learning_rate": 2.968780657261571e-06,
      "loss": 0.4029,
      "step": 3998
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.88600991784268,
      "learning_rate": 2.96639625515359e-06,
      "loss": 0.4332,
      "step": 3999
    },
    {
      "epoch": 0.64,
      "grad_norm": 14.26690256779112,
      "learning_rate": 2.9640124070553296e-06,
      "loss": 0.4561,
      "step": 4000
    },
    {
      "epoch": 0.64,
      "grad_norm": 12.92138213722854,
      "learning_rate": 2.961629113616209e-06,
      "loss": 0.4174,
      "step": 4001
    },
    {
      "epoch": 0.64,
      "grad_norm": 11.860403783133458,
      "learning_rate": 2.959246375485506e-06,
      "loss": 0.3936,
      "step": 4002
    },
    {
      "epoch": 0.64,
      "grad_norm": 12.89227905917057,
      "learning_rate": 2.9568641933123456e-06,
      "loss": 0.4895,
      "step": 4003
    },
    {
      "epoch": 0.65,
      "grad_norm": 10.61796083498606,
      "learning_rate": 2.954482567745697e-06,
      "loss": 0.4174,
      "step": 4004
    },
    {
      "epoch": 0.65,
      "grad_norm": 9.500493743748446,
      "learning_rate": 2.9521014994343823e-06,
      "loss": 0.4561,
      "step": 4005
    },
    {
      "epoch": 0.65,
      "grad_norm": 11.933944810350585,
      "learning_rate": 2.9497209890270704e-06,
      "loss": 0.3906,
      "step": 4006
    },
    {
      "epoch": 0.65,
      "grad_norm": 33.31742745070077,
      "learning_rate": 2.947341037172277e-06,
      "loss": 0.424,
      "step": 4007
    },
    {
      "epoch": 0.65,
      "grad_norm": 10.277650679488188,
      "learning_rate": 2.944961644518366e-06,
      "loss": 0.4421,
      "step": 4008
    },
    {
      "epoch": 0.65,
      "grad_norm": 10.494944992363877,
      "learning_rate": 2.942582811713553e-06,
      "loss": 0.4046,
      "step": 4009
    },
    {
      "epoch": 0.65,
      "grad_norm": 24.258999854553792,
      "learning_rate": 2.940204539405892e-06,
      "loss": 0.4198,
      "step": 4010
    },
    {
      "epoch": 0.65,
      "grad_norm": 17.093710182987405,
      "learning_rate": 2.937826828243294e-06,
      "loss": 0.39,
      "step": 4011
    },
    {
      "epoch": 0.65,
      "grad_norm": 13.89987432976777,
      "learning_rate": 2.9354496788735145e-06,
      "loss": 0.4146,
      "step": 4012
    },
    {
      "epoch": 0.65,
      "grad_norm": 11.784108394930481,
      "learning_rate": 2.9330730919441498e-06,
      "loss": 0.4099,
      "step": 4013
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.166655027292675,
      "learning_rate": 2.9306970681026503e-06,
      "loss": 0.437,
      "step": 4014
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.730180386052663,
      "learning_rate": 2.9283216079963108e-06,
      "loss": 0.4202,
      "step": 4015
    },
    {
      "epoch": 0.65,
      "grad_norm": 10.081434382911809,
      "learning_rate": 2.9259467122722705e-06,
      "loss": 0.383,
      "step": 4016
    },
    {
      "epoch": 0.65,
      "grad_norm": 73.12482241273136,
      "learning_rate": 2.9235723815775167e-06,
      "loss": 0.4303,
      "step": 4017
    },
    {
      "epoch": 0.65,
      "grad_norm": 21.97171023440878,
      "learning_rate": 2.9211986165588856e-06,
      "loss": 0.4342,
      "step": 4018
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.246003768055832,
      "learning_rate": 2.9188254178630526e-06,
      "loss": 0.4349,
      "step": 4019
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.053694361853395,
      "learning_rate": 2.916452786136542e-06,
      "loss": 0.4068,
      "step": 4020
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.257174461020268,
      "learning_rate": 2.914080722025728e-06,
      "loss": 0.4788,
      "step": 4021
    },
    {
      "epoch": 0.65,
      "grad_norm": 24.18150139444251,
      "learning_rate": 2.9117092261768247e-06,
      "loss": 0.4285,
      "step": 4022
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.720649052335874,
      "learning_rate": 2.9093382992358897e-06,
      "loss": 0.3947,
      "step": 4023
    },
    {
      "epoch": 0.65,
      "grad_norm": 26.162742092696202,
      "learning_rate": 2.906967941848834e-06,
      "loss": 0.3844,
      "step": 4024
    },
    {
      "epoch": 0.65,
      "grad_norm": 11.461626079351024,
      "learning_rate": 2.9045981546614057e-06,
      "loss": 0.4242,
      "step": 4025
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.475126822089058,
      "learning_rate": 2.9022289383191972e-06,
      "loss": 0.4235,
      "step": 4026
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.044076016056755,
      "learning_rate": 2.899860293467652e-06,
      "loss": 0.3326,
      "step": 4027
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.861613646683157,
      "learning_rate": 2.897492220752057e-06,
      "loss": 0.4364,
      "step": 4028
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2582715157697517,
      "learning_rate": 2.8951247208175337e-06,
      "loss": 0.4621,
      "step": 4029
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.210951359886305,
      "learning_rate": 2.8927577943090574e-06,
      "loss": 0.4227,
      "step": 4030
    },
    {
      "epoch": 0.65,
      "grad_norm": 10.49034146872277,
      "learning_rate": 2.890391441871446e-06,
      "loss": 0.4019,
      "step": 4031
    },
    {
      "epoch": 0.65,
      "grad_norm": 19.47035804206972,
      "learning_rate": 2.8880256641493582e-06,
      "loss": 0.381,
      "step": 4032
    },
    {
      "epoch": 0.65,
      "grad_norm": 10.213800843593058,
      "learning_rate": 2.885660461787294e-06,
      "loss": 0.4149,
      "step": 4033
    },
    {
      "epoch": 0.65,
      "grad_norm": 10.920550682368328,
      "learning_rate": 2.8832958354296048e-06,
      "loss": 0.4335,
      "step": 4034
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.728176713723922,
      "learning_rate": 2.880931785720474e-06,
      "loss": 0.3801,
      "step": 4035
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.380266568796461,
      "learning_rate": 2.8785683133039404e-06,
      "loss": 0.4631,
      "step": 4036
    },
    {
      "epoch": 0.65,
      "grad_norm": 12.318999960644637,
      "learning_rate": 2.876205418823875e-06,
      "loss": 0.3883,
      "step": 4037
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.542671164814548,
      "learning_rate": 2.8738431029239954e-06,
      "loss": 0.4403,
      "step": 4038
    },
    {
      "epoch": 0.65,
      "grad_norm": 11.752702610731474,
      "learning_rate": 2.8714813662478633e-06,
      "loss": 0.4053,
      "step": 4039
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.301703294975158,
      "learning_rate": 2.869120209438879e-06,
      "loss": 0.4629,
      "step": 4040
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.562987473029538,
      "learning_rate": 2.8667596331402892e-06,
      "loss": 0.4491,
      "step": 4041
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.962089496456033,
      "learning_rate": 2.864399637995176e-06,
      "loss": 0.3567,
      "step": 4042
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0472967021520163,
      "learning_rate": 2.8620402246464717e-06,
      "loss": 0.442,
      "step": 4043
    },
    {
      "epoch": 0.65,
      "grad_norm": 16.701923541017162,
      "learning_rate": 2.8596813937369437e-06,
      "loss": 0.3999,
      "step": 4044
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.337274947531368,
      "learning_rate": 2.8573231459092007e-06,
      "loss": 0.4207,
      "step": 4045
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1493757880078077,
      "learning_rate": 2.854965481805697e-06,
      "loss": 0.4299,
      "step": 4046
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.302567736593238,
      "learning_rate": 2.852608402068725e-06,
      "loss": 0.4168,
      "step": 4047
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.885929386480481,
      "learning_rate": 2.8502519073404155e-06,
      "loss": 0.3532,
      "step": 4048
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.912332766439919,
      "learning_rate": 2.847895998262744e-06,
      "loss": 0.4568,
      "step": 4049
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.190630551580049,
      "learning_rate": 2.8455406754775305e-06,
      "loss": 0.4863,
      "step": 4050
    },
    {
      "epoch": 0.65,
      "grad_norm": 12.775629984632445,
      "learning_rate": 2.8431859396264203e-06,
      "loss": 0.4041,
      "step": 4051
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.701422672429989,
      "learning_rate": 2.8408317913509137e-06,
      "loss": 0.37,
      "step": 4052
    },
    {
      "epoch": 0.65,
      "grad_norm": 10.62754919406745,
      "learning_rate": 2.8384782312923466e-06,
      "loss": 0.3765,
      "step": 4053
    },
    {
      "epoch": 0.65,
      "grad_norm": 25.284005747120446,
      "learning_rate": 2.836125260091892e-06,
      "loss": 0.4581,
      "step": 4054
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.238369254751749,
      "learning_rate": 2.8337728783905618e-06,
      "loss": 0.4433,
      "step": 4055
    },
    {
      "epoch": 0.65,
      "grad_norm": 13.346183507852489,
      "learning_rate": 2.8314210868292145e-06,
      "loss": 0.4889,
      "step": 4056
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.968675860088378,
      "learning_rate": 2.82906988604854e-06,
      "loss": 0.38,
      "step": 4057
    },
    {
      "epoch": 0.65,
      "grad_norm": 10.293968643609517,
      "learning_rate": 2.8267192766890685e-06,
      "loss": 0.4259,
      "step": 4058
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.523391278492276,
      "learning_rate": 2.824369259391173e-06,
      "loss": 0.4074,
      "step": 4059
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2710245453089868,
      "learning_rate": 2.822019834795065e-06,
      "loss": 0.4766,
      "step": 4060
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.873122286942166,
      "learning_rate": 2.8196710035407905e-06,
      "loss": 0.4019,
      "step": 4061
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.6436522698014455,
      "learning_rate": 2.8173227662682332e-06,
      "loss": 0.3839,
      "step": 4062
    },
    {
      "epoch": 0.65,
      "grad_norm": 11.158823722136578,
      "learning_rate": 2.814975123617123e-06,
      "loss": 0.4136,
      "step": 4063
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.983087691082732,
      "learning_rate": 2.8126280762270177e-06,
      "loss": 0.4207,
      "step": 4064
    },
    {
      "epoch": 0.65,
      "grad_norm": 9.023024845041666,
      "learning_rate": 2.810281624737321e-06,
      "loss": 0.394,
      "step": 4065
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.848112526262543,
      "learning_rate": 2.8079357697872696e-06,
      "loss": 0.4071,
      "step": 4066
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.063030811260548,
      "learning_rate": 2.8055905120159375e-06,
      "loss": 0.4352,
      "step": 4067
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.661607244408412,
      "learning_rate": 2.803245852062241e-06,
      "loss": 0.382,
      "step": 4068
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.797194907618332,
      "learning_rate": 2.8009017905649283e-06,
      "loss": 0.441,
      "step": 4069
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.380711409934143,
      "learning_rate": 2.7985583281625844e-06,
      "loss": 0.4258,
      "step": 4070
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.609573554014029,
      "learning_rate": 2.7962154654936352e-06,
      "loss": 0.4529,
      "step": 4071
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.397561185306046,
      "learning_rate": 2.7938732031963443e-06,
      "loss": 0.4127,
      "step": 4072
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.8024437283177654,
      "learning_rate": 2.791531541908802e-06,
      "loss": 0.4632,
      "step": 4073
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.980770748406464,
      "learning_rate": 2.7891904822689443e-06,
      "loss": 0.489,
      "step": 4074
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.727460521582941,
      "learning_rate": 2.7868500249145414e-06,
      "loss": 0.3897,
      "step": 4075
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.7826803324300835,
      "learning_rate": 2.784510170483199e-06,
      "loss": 0.3998,
      "step": 4076
    },
    {
      "epoch": 0.66,
      "grad_norm": 27.228951585944433,
      "learning_rate": 2.7821709196123535e-06,
      "loss": 0.4231,
      "step": 4077
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.155809327604749,
      "learning_rate": 2.779832272939285e-06,
      "loss": 0.4571,
      "step": 4078
    },
    {
      "epoch": 0.66,
      "grad_norm": 9.903641862402326,
      "learning_rate": 2.7774942311011082e-06,
      "loss": 0.4137,
      "step": 4079
    },
    {
      "epoch": 0.66,
      "grad_norm": 15.644873383883255,
      "learning_rate": 2.7751567947347624e-06,
      "loss": 0.4626,
      "step": 4080
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.029165368558605,
      "learning_rate": 2.772819964477035e-06,
      "loss": 0.4429,
      "step": 4081
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.962690628802151,
      "learning_rate": 2.7704837409645425e-06,
      "loss": 0.3702,
      "step": 4082
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.490858912493868,
      "learning_rate": 2.768148124833736e-06,
      "loss": 0.4769,
      "step": 4083
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.019911488950765,
      "learning_rate": 2.765813116720901e-06,
      "loss": 0.5053,
      "step": 4084
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.376564849420182,
      "learning_rate": 2.7634787172621593e-06,
      "loss": 0.4235,
      "step": 4085
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.838580686097576,
      "learning_rate": 2.761144927093464e-06,
      "loss": 0.4535,
      "step": 4086
    },
    {
      "epoch": 0.66,
      "grad_norm": 13.232388174686443,
      "learning_rate": 2.7588117468506064e-06,
      "loss": 0.4325,
      "step": 4087
    },
    {
      "epoch": 0.66,
      "grad_norm": 11.874232325541847,
      "learning_rate": 2.756479177169208e-06,
      "loss": 0.3909,
      "step": 4088
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.954044799090756,
      "learning_rate": 2.7541472186847224e-06,
      "loss": 0.4425,
      "step": 4089
    },
    {
      "epoch": 0.66,
      "grad_norm": 17.299557497791668,
      "learning_rate": 2.751815872032444e-06,
      "loss": 0.3656,
      "step": 4090
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.251069936834028,
      "learning_rate": 2.7494851378474936e-06,
      "loss": 0.477,
      "step": 4091
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.450700149587626,
      "learning_rate": 2.7471550167648255e-06,
      "loss": 0.3715,
      "step": 4092
    },
    {
      "epoch": 0.66,
      "grad_norm": 10.96852987781002,
      "learning_rate": 2.744825509419231e-06,
      "loss": 0.403,
      "step": 4093
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.258096251162194,
      "learning_rate": 2.742496616445335e-06,
      "loss": 0.4288,
      "step": 4094
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.670768883870506,
      "learning_rate": 2.740168338477587e-06,
      "loss": 0.377,
      "step": 4095
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.096608459555834,
      "learning_rate": 2.7378406761502747e-06,
      "loss": 0.3798,
      "step": 4096
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.823029148873207,
      "learning_rate": 2.7355136300975214e-06,
      "loss": 0.392,
      "step": 4097
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.442626769692936,
      "learning_rate": 2.733187200953276e-06,
      "loss": 0.5114,
      "step": 4098
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.937613130670315,
      "learning_rate": 2.7308613893513205e-06,
      "loss": 0.464,
      "step": 4099
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2974110953311102,
      "learning_rate": 2.7285361959252716e-06,
      "loss": 0.4829,
      "step": 4100
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.017500787809507,
      "learning_rate": 2.72621162130858e-06,
      "loss": 0.3992,
      "step": 4101
    },
    {
      "epoch": 0.66,
      "grad_norm": 20.578963576315946,
      "learning_rate": 2.723887666134516e-06,
      "loss": 0.3943,
      "step": 4102
    },
    {
      "epoch": 0.66,
      "grad_norm": 10.076796391928223,
      "learning_rate": 2.721564331036194e-06,
      "loss": 0.3603,
      "step": 4103
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.7410652564437,
      "learning_rate": 2.719241616646555e-06,
      "loss": 0.4251,
      "step": 4104
    },
    {
      "epoch": 0.66,
      "grad_norm": 15.370693187680478,
      "learning_rate": 2.716919523598369e-06,
      "loss": 0.3538,
      "step": 4105
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.458244242672271,
      "learning_rate": 2.7145980525242367e-06,
      "loss": 0.4043,
      "step": 4106
    },
    {
      "epoch": 0.66,
      "grad_norm": 12.695412633924612,
      "learning_rate": 2.712277204056594e-06,
      "loss": 0.4154,
      "step": 4107
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.944815976082589,
      "learning_rate": 2.7099569788277023e-06,
      "loss": 0.4163,
      "step": 4108
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2742473074045384,
      "learning_rate": 2.7076373774696527e-06,
      "loss": 0.469,
      "step": 4109
    },
    {
      "epoch": 0.66,
      "grad_norm": 10.480603220247222,
      "learning_rate": 2.705318400614374e-06,
      "loss": 0.5074,
      "step": 4110
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.542335896165337,
      "learning_rate": 2.703000048893613e-06,
      "loss": 0.4193,
      "step": 4111
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.216860934977393,
      "learning_rate": 2.7006823229389588e-06,
      "loss": 0.3678,
      "step": 4112
    },
    {
      "epoch": 0.66,
      "grad_norm": 11.700580206054259,
      "learning_rate": 2.6983652233818205e-06,
      "loss": 0.4373,
      "step": 4113
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.132194919353882,
      "learning_rate": 2.6960487508534383e-06,
      "loss": 0.4861,
      "step": 4114
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.44334421058753,
      "learning_rate": 2.693732905984885e-06,
      "loss": 0.4153,
      "step": 4115
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.792853524770847,
      "learning_rate": 2.6914176894070644e-06,
      "loss": 0.414,
      "step": 4116
    },
    {
      "epoch": 0.66,
      "grad_norm": 13.170838629033128,
      "learning_rate": 2.6891031017506986e-06,
      "loss": 0.4608,
      "step": 4117
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.219435461805495,
      "learning_rate": 2.686789143646347e-06,
      "loss": 0.4224,
      "step": 4118
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.995582395523539,
      "learning_rate": 2.6844758157244e-06,
      "loss": 0.3665,
      "step": 4119
    },
    {
      "epoch": 0.66,
      "grad_norm": 9.581149161719905,
      "learning_rate": 2.6821631186150676e-06,
      "loss": 0.4009,
      "step": 4120
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.263796594158633,
      "learning_rate": 2.6798510529483913e-06,
      "loss": 0.381,
      "step": 4121
    },
    {
      "epoch": 0.66,
      "grad_norm": 11.857085782382196,
      "learning_rate": 2.6775396193542436e-06,
      "loss": 0.4209,
      "step": 4122
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.336189884862522,
      "learning_rate": 2.675228818462327e-06,
      "loss": 0.4072,
      "step": 4123
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.768104492487895,
      "learning_rate": 2.6729186509021587e-06,
      "loss": 0.4567,
      "step": 4124
    },
    {
      "epoch": 0.66,
      "grad_norm": 9.614551832363667,
      "learning_rate": 2.670609117303096e-06,
      "loss": 0.5031,
      "step": 4125
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.668503350971288,
      "learning_rate": 2.668300218294322e-06,
      "loss": 0.3761,
      "step": 4126
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.431839544530715,
      "learning_rate": 2.6659919545048424e-06,
      "loss": 0.4532,
      "step": 4127
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.121424885586337,
      "learning_rate": 2.663684326563489e-06,
      "loss": 0.461,
      "step": 4128
    },
    {
      "epoch": 0.67,
      "grad_norm": 8.361358820383945,
      "learning_rate": 2.6613773350989293e-06,
      "loss": 0.4436,
      "step": 4129
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.402754343318197,
      "learning_rate": 2.6590709807396464e-06,
      "loss": 0.4158,
      "step": 4130
    },
    {
      "epoch": 0.67,
      "grad_norm": 10.995060547517761,
      "learning_rate": 2.656765264113955e-06,
      "loss": 0.443,
      "step": 4131
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.632179340798159,
      "learning_rate": 2.65446018585e-06,
      "loss": 0.4339,
      "step": 4132
    },
    {
      "epoch": 0.67,
      "grad_norm": 11.316089583514849,
      "learning_rate": 2.652155746575743e-06,
      "loss": 0.4186,
      "step": 4133
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.550851383671837,
      "learning_rate": 2.6498519469189814e-06,
      "loss": 0.388,
      "step": 4134
    },
    {
      "epoch": 0.67,
      "grad_norm": 12.196477300352107,
      "learning_rate": 2.6475487875073318e-06,
      "loss": 0.4677,
      "step": 4135
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.904051962711692,
      "learning_rate": 2.6452462689682364e-06,
      "loss": 0.4143,
      "step": 4136
    },
    {
      "epoch": 0.67,
      "grad_norm": 8.285040720370983,
      "learning_rate": 2.6429443919289676e-06,
      "loss": 0.4171,
      "step": 4137
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.425512741652333,
      "learning_rate": 2.640643157016618e-06,
      "loss": 0.3893,
      "step": 4138
    },
    {
      "epoch": 0.67,
      "grad_norm": 8.364313201013164,
      "learning_rate": 2.63834256485811e-06,
      "loss": 0.4263,
      "step": 4139
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.290199349108534,
      "learning_rate": 2.636042616080185e-06,
      "loss": 0.4161,
      "step": 4140
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.829513383505944,
      "learning_rate": 2.6337433113094164e-06,
      "loss": 0.4198,
      "step": 4141
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.603742934833703,
      "learning_rate": 2.6314446511721957e-06,
      "loss": 0.4185,
      "step": 4142
    },
    {
      "epoch": 0.67,
      "grad_norm": 18.506489664743384,
      "learning_rate": 2.6291466362947392e-06,
      "loss": 0.3287,
      "step": 4143
    },
    {
      "epoch": 0.67,
      "grad_norm": 10.00200850338193,
      "learning_rate": 2.626849267303093e-06,
      "loss": 0.4395,
      "step": 4144
    },
    {
      "epoch": 0.67,
      "grad_norm": 10.055796381446534,
      "learning_rate": 2.624552544823126e-06,
      "loss": 0.3724,
      "step": 4145
    },
    {
      "epoch": 0.67,
      "grad_norm": 16.61425303145286,
      "learning_rate": 2.622256469480522e-06,
      "loss": 0.4729,
      "step": 4146
    },
    {
      "epoch": 0.67,
      "grad_norm": 10.841218608861091,
      "learning_rate": 2.6199610419007983e-06,
      "loss": 0.4358,
      "step": 4147
    },
    {
      "epoch": 0.67,
      "grad_norm": 8.704822476419505,
      "learning_rate": 2.617666262709294e-06,
      "loss": 0.435,
      "step": 4148
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.107318093381502,
      "learning_rate": 2.61537213253117e-06,
      "loss": 0.3948,
      "step": 4149
    },
    {
      "epoch": 0.67,
      "grad_norm": 11.090248887518076,
      "learning_rate": 2.6130786519914074e-06,
      "loss": 0.4849,
      "step": 4150
    },
    {
      "epoch": 0.67,
      "grad_norm": 13.79445673411347,
      "learning_rate": 2.6107858217148185e-06,
      "loss": 0.4053,
      "step": 4151
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.989762365147147,
      "learning_rate": 2.6084936423260303e-06,
      "loss": 0.4142,
      "step": 4152
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.510794379134416,
      "learning_rate": 2.606202114449495e-06,
      "loss": 0.4327,
      "step": 4153
    },
    {
      "epoch": 0.67,
      "grad_norm": 9.272543144518421,
      "learning_rate": 2.60391123870949e-06,
      "loss": 0.4053,
      "step": 4154
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.3434844864508495,
      "learning_rate": 2.601621015730111e-06,
      "loss": 0.4287,
      "step": 4155
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.148290235405018,
      "learning_rate": 2.5993314461352802e-06,
      "loss": 0.4021,
      "step": 4156
    },
    {
      "epoch": 0.67,
      "grad_norm": 16.498149077975754,
      "learning_rate": 2.597042530548737e-06,
      "loss": 0.4485,
      "step": 4157
    },
    {
      "epoch": 0.67,
      "grad_norm": 8.350037216800475,
      "learning_rate": 2.5947542695940485e-06,
      "loss": 0.3529,
      "step": 4158
    },
    {
      "epoch": 0.67,
      "grad_norm": 9.697050795135292,
      "learning_rate": 2.5924666638945973e-06,
      "loss": 0.435,
      "step": 4159
    },
    {
      "epoch": 0.67,
      "grad_norm": 19.11128432461532,
      "learning_rate": 2.5901797140735895e-06,
      "loss": 0.4305,
      "step": 4160
    },
    {
      "epoch": 0.67,
      "grad_norm": 10.487376953269806,
      "learning_rate": 2.5878934207540564e-06,
      "loss": 0.4047,
      "step": 4161
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.570922357007311,
      "learning_rate": 2.5856077845588433e-06,
      "loss": 0.4987,
      "step": 4162
    },
    {
      "epoch": 0.67,
      "grad_norm": 9.676241594060867,
      "learning_rate": 2.5833228061106253e-06,
      "loss": 0.4801,
      "step": 4163
    },
    {
      "epoch": 0.67,
      "grad_norm": 9.628210521812516,
      "learning_rate": 2.5810384860318904e-06,
      "loss": 0.4144,
      "step": 4164
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.466535377421348,
      "learning_rate": 2.57875482494495e-06,
      "loss": 0.4123,
      "step": 4165
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.752183075233021,
      "learning_rate": 2.576471823471938e-06,
      "loss": 0.3831,
      "step": 4166
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.491606804773458,
      "learning_rate": 2.5741894822348073e-06,
      "loss": 0.416,
      "step": 4167
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.814097043970005,
      "learning_rate": 2.571907801855327e-06,
      "loss": 0.4349,
      "step": 4168
    },
    {
      "epoch": 0.67,
      "grad_norm": 9.078920456171158,
      "learning_rate": 2.5696267829550926e-06,
      "loss": 0.3544,
      "step": 4169
    },
    {
      "epoch": 0.67,
      "grad_norm": 8.513759767746317,
      "learning_rate": 2.5673464261555174e-06,
      "loss": 0.3955,
      "step": 4170
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.481594369983413,
      "learning_rate": 2.5650667320778333e-06,
      "loss": 0.3898,
      "step": 4171
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.707875209016076,
      "learning_rate": 2.562787701343088e-06,
      "loss": 0.4974,
      "step": 4172
    },
    {
      "epoch": 0.67,
      "grad_norm": 10.053459390609397,
      "learning_rate": 2.5605093345721564e-06,
      "loss": 0.4082,
      "step": 4173
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.93480955505699,
      "learning_rate": 2.558231632385728e-06,
      "loss": 0.3907,
      "step": 4174
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.520104317005605,
      "learning_rate": 2.5559545954043086e-06,
      "loss": 0.3437,
      "step": 4175
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.470803186495222,
      "learning_rate": 2.5536782242482294e-06,
      "loss": 0.3647,
      "step": 4176
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.835280092168652,
      "learning_rate": 2.5514025195376336e-06,
      "loss": 0.3836,
      "step": 4177
    },
    {
      "epoch": 0.67,
      "grad_norm": 8.929742443683036,
      "learning_rate": 2.5491274818924893e-06,
      "loss": 0.4718,
      "step": 4178
    },
    {
      "epoch": 0.67,
      "grad_norm": 12.267339532483499,
      "learning_rate": 2.5468531119325756e-06,
      "loss": 0.4525,
      "step": 4179
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.687945762334257,
      "learning_rate": 2.544579410277498e-06,
      "loss": 0.3754,
      "step": 4180
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2681804608139207,
      "learning_rate": 2.542306377546673e-06,
      "loss": 0.4831,
      "step": 4181
    },
    {
      "epoch": 0.67,
      "grad_norm": 12.806330170006689,
      "learning_rate": 2.540034014359335e-06,
      "loss": 0.3569,
      "step": 4182
    },
    {
      "epoch": 0.67,
      "grad_norm": 12.223113619149723,
      "learning_rate": 2.537762321334543e-06,
      "loss": 0.3795,
      "step": 4183
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2712539655738475,
      "learning_rate": 2.535491299091165e-06,
      "loss": 0.4972,
      "step": 4184
    },
    {
      "epoch": 0.67,
      "grad_norm": 9.46751671481261,
      "learning_rate": 2.533220948247893e-06,
      "loss": 0.3998,
      "step": 4185
    },
    {
      "epoch": 0.67,
      "grad_norm": 13.974185073279031,
      "learning_rate": 2.530951269423233e-06,
      "loss": 0.4172,
      "step": 4186
    },
    {
      "epoch": 0.67,
      "grad_norm": 15.806406716413644,
      "learning_rate": 2.528682263235504e-06,
      "loss": 0.4536,
      "step": 4187
    },
    {
      "epoch": 0.67,
      "grad_norm": 9.2876175864752,
      "learning_rate": 2.526413930302851e-06,
      "loss": 0.4463,
      "step": 4188
    },
    {
      "epoch": 0.67,
      "grad_norm": 21.148403140033174,
      "learning_rate": 2.5241462712432284e-06,
      "loss": 0.3989,
      "step": 4189
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.583779532711976,
      "learning_rate": 2.5218792866744068e-06,
      "loss": 0.4018,
      "step": 4190
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.576099076048648,
      "learning_rate": 2.5196129772139766e-06,
      "loss": 0.4609,
      "step": 4191
    },
    {
      "epoch": 0.68,
      "grad_norm": 12.629992210999397,
      "learning_rate": 2.517347343479346e-06,
      "loss": 0.4352,
      "step": 4192
    },
    {
      "epoch": 0.68,
      "grad_norm": 10.50223287720631,
      "learning_rate": 2.5150823860877345e-06,
      "loss": 0.4289,
      "step": 4193
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.675515643351629,
      "learning_rate": 2.5128181056561755e-06,
      "loss": 0.4864,
      "step": 4194
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.932977894506335,
      "learning_rate": 2.5105545028015266e-06,
      "loss": 0.4698,
      "step": 4195
    },
    {
      "epoch": 0.68,
      "grad_norm": 10.419095394884001,
      "learning_rate": 2.508291578140453e-06,
      "loss": 0.4309,
      "step": 4196
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.956752727037793,
      "learning_rate": 2.5060293322894353e-06,
      "loss": 0.3392,
      "step": 4197
    },
    {
      "epoch": 0.68,
      "grad_norm": 9.946544990209416,
      "learning_rate": 2.503767765864774e-06,
      "loss": 0.4457,
      "step": 4198
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.547597349860176,
      "learning_rate": 2.5015068794825853e-06,
      "loss": 0.4196,
      "step": 4199
    },
    {
      "epoch": 0.68,
      "grad_norm": 22.39995093987481,
      "learning_rate": 2.4992466737587932e-06,
      "loss": 0.3684,
      "step": 4200
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.566526287942026,
      "learning_rate": 2.4969871493091398e-06,
      "loss": 0.4152,
      "step": 4201
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.177585170492367,
      "learning_rate": 2.494728306749184e-06,
      "loss": 0.3991,
      "step": 4202
    },
    {
      "epoch": 0.68,
      "grad_norm": 10.969985854549728,
      "learning_rate": 2.4924701466942957e-06,
      "loss": 0.4845,
      "step": 4203
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.5601311135489535,
      "learning_rate": 2.490212669759658e-06,
      "loss": 0.3901,
      "step": 4204
    },
    {
      "epoch": 0.68,
      "grad_norm": 10.958832086517287,
      "learning_rate": 2.487955876560274e-06,
      "loss": 0.4489,
      "step": 4205
    },
    {
      "epoch": 0.68,
      "grad_norm": 45.57462788417418,
      "learning_rate": 2.485699767710952e-06,
      "loss": 0.4994,
      "step": 4206
    },
    {
      "epoch": 0.68,
      "grad_norm": 19.15410495400765,
      "learning_rate": 2.4834443438263224e-06,
      "loss": 0.4083,
      "step": 4207
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.714075635194001,
      "learning_rate": 2.4811896055208227e-06,
      "loss": 0.3931,
      "step": 4208
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.44363030723576,
      "learning_rate": 2.4789355534087038e-06,
      "loss": 0.4449,
      "step": 4209
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.566071689595343,
      "learning_rate": 2.476682188104036e-06,
      "loss": 0.3679,
      "step": 4210
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.6310567158629805,
      "learning_rate": 2.4744295102206954e-06,
      "loss": 0.3644,
      "step": 4211
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.102317223454744,
      "learning_rate": 2.4721775203723726e-06,
      "loss": 0.4196,
      "step": 4212
    },
    {
      "epoch": 0.68,
      "grad_norm": 12.864610115518843,
      "learning_rate": 2.4699262191725726e-06,
      "loss": 0.44,
      "step": 4213
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.208630654219187,
      "learning_rate": 2.467675607234615e-06,
      "loss": 0.4774,
      "step": 4214
    },
    {
      "epoch": 0.68,
      "grad_norm": 13.03107045254372,
      "learning_rate": 2.465425685171625e-06,
      "loss": 0.4378,
      "step": 4215
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.167309614176898,
      "learning_rate": 2.463176453596543e-06,
      "loss": 0.4377,
      "step": 4216
    },
    {
      "epoch": 0.68,
      "grad_norm": 18.203050138435987,
      "learning_rate": 2.4609279131221243e-06,
      "loss": 0.4813,
      "step": 4217
    },
    {
      "epoch": 0.68,
      "grad_norm": 11.147093600544459,
      "learning_rate": 2.4586800643609326e-06,
      "loss": 0.5236,
      "step": 4218
    },
    {
      "epoch": 0.68,
      "grad_norm": 12.872399660138699,
      "learning_rate": 2.456432907925341e-06,
      "loss": 0.4374,
      "step": 4219
    },
    {
      "epoch": 0.68,
      "grad_norm": 10.272396814350538,
      "learning_rate": 2.454186444427539e-06,
      "loss": 0.3242,
      "step": 4220
    },
    {
      "epoch": 0.68,
      "grad_norm": 9.546138198627961,
      "learning_rate": 2.4519406744795275e-06,
      "loss": 0.444,
      "step": 4221
    },
    {
      "epoch": 0.68,
      "grad_norm": 18.384124383050438,
      "learning_rate": 2.4496955986931143e-06,
      "loss": 0.4414,
      "step": 4222
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4359102126699177,
      "learning_rate": 2.447451217679917e-06,
      "loss": 0.4951,
      "step": 4223
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1472710225447738,
      "learning_rate": 2.445207532051372e-06,
      "loss": 0.4334,
      "step": 4224
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.969045245651673,
      "learning_rate": 2.442964542418718e-06,
      "loss": 0.4177,
      "step": 4225
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.716336055746022,
      "learning_rate": 2.4407222493930063e-06,
      "loss": 0.4607,
      "step": 4226
    },
    {
      "epoch": 0.68,
      "grad_norm": 13.721660088307331,
      "learning_rate": 2.438480653585102e-06,
      "loss": 0.4887,
      "step": 4227
    },
    {
      "epoch": 0.68,
      "grad_norm": 15.918198999025421,
      "learning_rate": 2.436239755605675e-06,
      "loss": 0.4972,
      "step": 4228
    },
    {
      "epoch": 0.68,
      "grad_norm": 12.033352755178402,
      "learning_rate": 2.433999556065211e-06,
      "loss": 0.3561,
      "step": 4229
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.940825603100634,
      "learning_rate": 2.4317600555739997e-06,
      "loss": 0.4122,
      "step": 4230
    },
    {
      "epoch": 0.68,
      "grad_norm": 9.881922386110519,
      "learning_rate": 2.4295212547421415e-06,
      "loss": 0.4111,
      "step": 4231
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.011183680846465,
      "learning_rate": 2.427283154179551e-06,
      "loss": 0.3911,
      "step": 4232
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.490130899541349,
      "learning_rate": 2.4250457544959465e-06,
      "loss": 0.4313,
      "step": 4233
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.888261885601563,
      "learning_rate": 2.422809056300856e-06,
      "loss": 0.4195,
      "step": 4234
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.944924068723493,
      "learning_rate": 2.4205730602036194e-06,
      "loss": 0.3277,
      "step": 4235
    },
    {
      "epoch": 0.68,
      "grad_norm": 10.000713296361035,
      "learning_rate": 2.4183377668133843e-06,
      "loss": 0.4979,
      "step": 4236
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.390518916921966,
      "learning_rate": 2.4161031767391067e-06,
      "loss": 0.4311,
      "step": 4237
    },
    {
      "epoch": 0.68,
      "grad_norm": 11.74555507583681,
      "learning_rate": 2.4138692905895473e-06,
      "loss": 0.4027,
      "step": 4238
    },
    {
      "epoch": 0.68,
      "grad_norm": 10.818241007379687,
      "learning_rate": 2.4116361089732815e-06,
      "loss": 0.4332,
      "step": 4239
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.85162674830752,
      "learning_rate": 2.4094036324986888e-06,
      "loss": 0.4321,
      "step": 4240
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.69905985392428,
      "learning_rate": 2.4071718617739542e-06,
      "loss": 0.3611,
      "step": 4241
    },
    {
      "epoch": 0.68,
      "grad_norm": 9.429108670270677,
      "learning_rate": 2.404940797407077e-06,
      "loss": 0.4433,
      "step": 4242
    },
    {
      "epoch": 0.68,
      "grad_norm": 10.07399175230551,
      "learning_rate": 2.4027104400058615e-06,
      "loss": 0.4623,
      "step": 4243
    },
    {
      "epoch": 0.68,
      "grad_norm": 12.336888452965798,
      "learning_rate": 2.4004807901779164e-06,
      "loss": 0.4372,
      "step": 4244
    },
    {
      "epoch": 0.68,
      "grad_norm": 17.669999404381336,
      "learning_rate": 2.3982518485306587e-06,
      "loss": 0.4197,
      "step": 4245
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.945442763952645,
      "learning_rate": 2.3960236156713175e-06,
      "loss": 0.4582,
      "step": 4246
    },
    {
      "epoch": 0.68,
      "grad_norm": 9.985244862608376,
      "learning_rate": 2.3937960922069213e-06,
      "loss": 0.4334,
      "step": 4247
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.291570835060414,
      "learning_rate": 2.391569278744309e-06,
      "loss": 0.3636,
      "step": 4248
    },
    {
      "epoch": 0.68,
      "grad_norm": 22.862092734354505,
      "learning_rate": 2.3893431758901285e-06,
      "loss": 0.4398,
      "step": 4249
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.942843579207113,
      "learning_rate": 2.387117784250828e-06,
      "loss": 0.4308,
      "step": 4250
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.29159159174315,
      "learning_rate": 2.38489310443267e-06,
      "loss": 0.4219,
      "step": 4251
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.557277868188023,
      "learning_rate": 2.382669137041716e-06,
      "loss": 0.4848,
      "step": 4252
    },
    {
      "epoch": 0.69,
      "grad_norm": 23.652767009991326,
      "learning_rate": 2.3804458826838337e-06,
      "loss": 0.4173,
      "step": 4253
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.665393625173859,
      "learning_rate": 2.3782233419647043e-06,
      "loss": 0.4268,
      "step": 4254
    },
    {
      "epoch": 0.69,
      "grad_norm": 12.162902201870216,
      "learning_rate": 2.376001515489806e-06,
      "loss": 0.3976,
      "step": 4255
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.588250893036699,
      "learning_rate": 2.3737804038644236e-06,
      "loss": 0.4202,
      "step": 4256
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2522265499211311,
      "learning_rate": 2.371560007693651e-06,
      "loss": 0.482,
      "step": 4257
    },
    {
      "epoch": 0.69,
      "grad_norm": 11.498976358712579,
      "learning_rate": 2.3693403275823883e-06,
      "loss": 0.3906,
      "step": 4258
    },
    {
      "epoch": 0.69,
      "grad_norm": 8.683671503424353,
      "learning_rate": 2.367121364135335e-06,
      "loss": 0.5123,
      "step": 4259
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.081374392957123,
      "learning_rate": 2.3649031179569954e-06,
      "loss": 0.4233,
      "step": 4260
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.622890565700507,
      "learning_rate": 2.3626855896516855e-06,
      "loss": 0.4297,
      "step": 4261
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.52548499686504,
      "learning_rate": 2.360468779823519e-06,
      "loss": 0.4318,
      "step": 4262
    },
    {
      "epoch": 0.69,
      "grad_norm": 10.034246698944754,
      "learning_rate": 2.3582526890764135e-06,
      "loss": 0.4315,
      "step": 4263
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.866283895757253,
      "learning_rate": 2.356037318014096e-06,
      "loss": 0.3836,
      "step": 4264
    },
    {
      "epoch": 0.69,
      "grad_norm": 19.185538034135106,
      "learning_rate": 2.3538226672400982e-06,
      "loss": 0.401,
      "step": 4265
    },
    {
      "epoch": 0.69,
      "grad_norm": 8.10584373606494,
      "learning_rate": 2.3516087373577447e-06,
      "loss": 0.4181,
      "step": 4266
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.3430698032377135,
      "learning_rate": 2.3493955289701744e-06,
      "loss": 0.3782,
      "step": 4267
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.916309275386448,
      "learning_rate": 2.347183042680328e-06,
      "loss": 0.3547,
      "step": 4268
    },
    {
      "epoch": 0.69,
      "grad_norm": 9.728209659123351,
      "learning_rate": 2.3449712790909465e-06,
      "loss": 0.3828,
      "step": 4269
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.5629660848165114,
      "learning_rate": 2.3427602388045723e-06,
      "loss": 0.4831,
      "step": 4270
    },
    {
      "epoch": 0.69,
      "grad_norm": 11.368515832158197,
      "learning_rate": 2.3405499224235583e-06,
      "loss": 0.3903,
      "step": 4271
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0545311408577567,
      "learning_rate": 2.3383403305500523e-06,
      "loss": 0.4192,
      "step": 4272
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.2148256341141925,
      "learning_rate": 2.3361314637860104e-06,
      "loss": 0.4566,
      "step": 4273
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.747737630643661,
      "learning_rate": 2.3339233227331887e-06,
      "loss": 0.4221,
      "step": 4274
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.779715322704246,
      "learning_rate": 2.331715907993142e-06,
      "loss": 0.4068,
      "step": 4275
    },
    {
      "epoch": 0.69,
      "grad_norm": 9.289331868300978,
      "learning_rate": 2.329509220167236e-06,
      "loss": 0.3873,
      "step": 4276
    },
    {
      "epoch": 0.69,
      "grad_norm": 14.460206163385553,
      "learning_rate": 2.327303259856629e-06,
      "loss": 0.377,
      "step": 4277
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.07713686507014,
      "learning_rate": 2.3250980276622893e-06,
      "loss": 0.4238,
      "step": 4278
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.259662583051834,
      "learning_rate": 2.322893524184979e-06,
      "loss": 0.4151,
      "step": 4279
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.049557259364586,
      "learning_rate": 2.320689750025269e-06,
      "loss": 0.3609,
      "step": 4280
    },
    {
      "epoch": 0.69,
      "grad_norm": 8.715552592894973,
      "learning_rate": 2.3184867057835274e-06,
      "loss": 0.4026,
      "step": 4281
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.715970778066933,
      "learning_rate": 2.316284392059921e-06,
      "loss": 0.4176,
      "step": 4282
    },
    {
      "epoch": 0.69,
      "grad_norm": 20.689659325880644,
      "learning_rate": 2.3140828094544253e-06,
      "loss": 0.4429,
      "step": 4283
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.475431675691197,
      "learning_rate": 2.3118819585668106e-06,
      "loss": 0.3778,
      "step": 4284
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.243658626819774,
      "learning_rate": 2.3096818399966465e-06,
      "loss": 0.4704,
      "step": 4285
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.819868414277102,
      "learning_rate": 2.3074824543433084e-06,
      "loss": 0.4149,
      "step": 4286
    },
    {
      "epoch": 0.69,
      "grad_norm": 8.110554820545692,
      "learning_rate": 2.305283802205973e-06,
      "loss": 0.4344,
      "step": 4287
    },
    {
      "epoch": 0.69,
      "grad_norm": 15.914223475516382,
      "learning_rate": 2.3030858841836063e-06,
      "loss": 0.4246,
      "step": 4288
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.791008571397075,
      "learning_rate": 2.300888700874986e-06,
      "loss": 0.4,
      "step": 4289
    },
    {
      "epoch": 0.69,
      "grad_norm": 10.946221525894051,
      "learning_rate": 2.2986922528786865e-06,
      "loss": 0.4186,
      "step": 4290
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.546902497052178,
      "learning_rate": 2.2964965407930796e-06,
      "loss": 0.4845,
      "step": 4291
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.9330668102116535,
      "learning_rate": 2.2943015652163355e-06,
      "loss": 0.3907,
      "step": 4292
    },
    {
      "epoch": 0.69,
      "grad_norm": 8.737545100421132,
      "learning_rate": 2.29210732674643e-06,
      "loss": 0.4769,
      "step": 4293
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.08812055315629,
      "learning_rate": 2.289913825981132e-06,
      "loss": 0.3932,
      "step": 4294
    },
    {
      "epoch": 0.69,
      "grad_norm": 9.491288418838039,
      "learning_rate": 2.2877210635180098e-06,
      "loss": 0.4429,
      "step": 4295
    },
    {
      "epoch": 0.69,
      "grad_norm": 8.21891713053102,
      "learning_rate": 2.2855290399544346e-06,
      "loss": 0.4669,
      "step": 4296
    },
    {
      "epoch": 0.69,
      "grad_norm": 24.380215040880355,
      "learning_rate": 2.2833377558875754e-06,
      "loss": 0.4421,
      "step": 4297
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.706997440132937,
      "learning_rate": 2.281147211914397e-06,
      "loss": 0.424,
      "step": 4298
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.107377579565903,
      "learning_rate": 2.278957408631662e-06,
      "loss": 0.3966,
      "step": 4299
    },
    {
      "epoch": 0.69,
      "grad_norm": 8.686099994283008,
      "learning_rate": 2.276768346635936e-06,
      "loss": 0.3957,
      "step": 4300
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.382977801372351,
      "learning_rate": 2.2745800265235773e-06,
      "loss": 0.4269,
      "step": 4301
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.375887189397306,
      "learning_rate": 2.2723924488907477e-06,
      "loss": 0.3965,
      "step": 4302
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.116808655408611,
      "learning_rate": 2.2702056143334025e-06,
      "loss": 0.4561,
      "step": 4303
    },
    {
      "epoch": 0.69,
      "grad_norm": 9.51378427638696,
      "learning_rate": 2.268019523447292e-06,
      "loss": 0.4539,
      "step": 4304
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.102955431802788,
      "learning_rate": 2.265834176827974e-06,
      "loss": 0.384,
      "step": 4305
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.878846197043499,
      "learning_rate": 2.263649575070794e-06,
      "loss": 0.3734,
      "step": 4306
    },
    {
      "epoch": 0.69,
      "grad_norm": 10.466505311472522,
      "learning_rate": 2.261465718770895e-06,
      "loss": 0.4128,
      "step": 4307
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.208216704117919,
      "learning_rate": 2.259282608523223e-06,
      "loss": 0.4142,
      "step": 4308
    },
    {
      "epoch": 0.69,
      "grad_norm": 17.332765945447687,
      "learning_rate": 2.25710024492252e-06,
      "loss": 0.4113,
      "step": 4309
    },
    {
      "epoch": 0.69,
      "grad_norm": 10.062647172743207,
      "learning_rate": 2.254918628563315e-06,
      "loss": 0.4189,
      "step": 4310
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2589073642177175,
      "learning_rate": 2.2527377600399446e-06,
      "loss": 0.4748,
      "step": 4311
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.323459328808105,
      "learning_rate": 2.2505576399465384e-06,
      "loss": 0.4489,
      "step": 4312
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.27444860343837,
      "learning_rate": 2.2483782688770208e-06,
      "loss": 0.4468,
      "step": 4313
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.084049623859299,
      "learning_rate": 2.246199647425109e-06,
      "loss": 0.4168,
      "step": 4314
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.057735738937002,
      "learning_rate": 2.2440217761843244e-06,
      "loss": 0.4081,
      "step": 4315
    },
    {
      "epoch": 0.7,
      "grad_norm": 13.58330657614081,
      "learning_rate": 2.241844655747977e-06,
      "loss": 0.3706,
      "step": 4316
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.237682929272237,
      "learning_rate": 2.2396682867091723e-06,
      "loss": 0.4705,
      "step": 4317
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.480521774762631,
      "learning_rate": 2.237492669660815e-06,
      "loss": 0.3738,
      "step": 4318
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.49088626287101,
      "learning_rate": 2.235317805195606e-06,
      "loss": 0.3698,
      "step": 4319
    },
    {
      "epoch": 0.7,
      "grad_norm": 13.123959096964384,
      "learning_rate": 2.2331436939060354e-06,
      "loss": 0.3924,
      "step": 4320
    },
    {
      "epoch": 0.7,
      "grad_norm": 8.392184629746076,
      "learning_rate": 2.2309703363843893e-06,
      "loss": 0.4228,
      "step": 4321
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.891337291051181,
      "learning_rate": 2.2287977332227543e-06,
      "loss": 0.4519,
      "step": 4322
    },
    {
      "epoch": 0.7,
      "grad_norm": 16.824704923606525,
      "learning_rate": 2.2266258850130055e-06,
      "loss": 0.4186,
      "step": 4323
    },
    {
      "epoch": 0.7,
      "grad_norm": 9.037476434418672,
      "learning_rate": 2.224454792346813e-06,
      "loss": 0.4054,
      "step": 4324
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.7766535240179255,
      "learning_rate": 2.2222844558156444e-06,
      "loss": 0.4313,
      "step": 4325
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.468536075371333,
      "learning_rate": 2.2201148760107568e-06,
      "loss": 0.4252,
      "step": 4326
    },
    {
      "epoch": 0.7,
      "grad_norm": 9.609264793868432,
      "learning_rate": 2.217946053523207e-06,
      "loss": 0.3615,
      "step": 4327
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.617137266075924,
      "learning_rate": 2.2157779889438393e-06,
      "loss": 0.4466,
      "step": 4328
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.578073814287037,
      "learning_rate": 2.213610682863293e-06,
      "loss": 0.3968,
      "step": 4329
    },
    {
      "epoch": 0.7,
      "grad_norm": 25.05186132050751,
      "learning_rate": 2.2114441358720046e-06,
      "loss": 0.4156,
      "step": 4330
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9951237829930111,
      "learning_rate": 2.209278348560204e-06,
      "loss": 0.4194,
      "step": 4331
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.829220574271433,
      "learning_rate": 2.2071133215179043e-06,
      "loss": 0.4362,
      "step": 4332
    },
    {
      "epoch": 0.7,
      "grad_norm": 18.61106585225155,
      "learning_rate": 2.204949055334922e-06,
      "loss": 0.4415,
      "step": 4333
    },
    {
      "epoch": 0.7,
      "grad_norm": 11.266827887612756,
      "learning_rate": 2.202785550600865e-06,
      "loss": 0.4267,
      "step": 4334
    },
    {
      "epoch": 0.7,
      "grad_norm": 12.566342604335452,
      "learning_rate": 2.20062280790513e-06,
      "loss": 0.4889,
      "step": 4335
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.444505810988872,
      "learning_rate": 2.198460827836905e-06,
      "loss": 0.4501,
      "step": 4336
    },
    {
      "epoch": 0.7,
      "grad_norm": 8.602908245085976,
      "learning_rate": 2.1962996109851757e-06,
      "loss": 0.4335,
      "step": 4337
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.893448333339696,
      "learning_rate": 2.1941391579387204e-06,
      "loss": 0.4398,
      "step": 4338
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.340960631113362,
      "learning_rate": 2.1919794692860992e-06,
      "loss": 0.4122,
      "step": 4339
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.70550507972833,
      "learning_rate": 2.189820545615674e-06,
      "loss": 0.4215,
      "step": 4340
    },
    {
      "epoch": 0.7,
      "grad_norm": 12.935221105651625,
      "learning_rate": 2.187662387515597e-06,
      "loss": 0.3941,
      "step": 4341
    },
    {
      "epoch": 0.7,
      "grad_norm": 10.681118190790285,
      "learning_rate": 2.1855049955738077e-06,
      "loss": 0.3887,
      "step": 4342
    },
    {
      "epoch": 0.7,
      "grad_norm": 8.052530977979147,
      "learning_rate": 2.1833483703780383e-06,
      "loss": 0.3858,
      "step": 4343
    },
    {
      "epoch": 0.7,
      "grad_norm": 17.1225656366262,
      "learning_rate": 2.1811925125158154e-06,
      "loss": 0.4049,
      "step": 4344
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.361990125940465,
      "learning_rate": 2.1790374225744537e-06,
      "loss": 0.421,
      "step": 4345
    },
    {
      "epoch": 0.7,
      "grad_norm": 26.97258920318783,
      "learning_rate": 2.1768831011410567e-06,
      "loss": 0.4283,
      "step": 4346
    },
    {
      "epoch": 0.7,
      "grad_norm": 28.408495272774765,
      "learning_rate": 2.174729548802524e-06,
      "loss": 0.3844,
      "step": 4347
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.232498793267007,
      "learning_rate": 2.1725767661455393e-06,
      "loss": 0.4354,
      "step": 4348
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9692155791690763,
      "learning_rate": 2.170424753756584e-06,
      "loss": 0.4064,
      "step": 4349
    },
    {
      "epoch": 0.7,
      "grad_norm": 8.832123278917905,
      "learning_rate": 2.1682735122219236e-06,
      "loss": 0.4232,
      "step": 4350
    },
    {
      "epoch": 0.7,
      "grad_norm": 11.013858088253174,
      "learning_rate": 2.1661230421276134e-06,
      "loss": 0.4566,
      "step": 4351
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.120908347879851,
      "learning_rate": 2.163973344059503e-06,
      "loss": 0.36,
      "step": 4352
    },
    {
      "epoch": 0.7,
      "grad_norm": 9.82044553928464,
      "learning_rate": 2.1618244186032328e-06,
      "loss": 0.474,
      "step": 4353
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.786781018903208,
      "learning_rate": 2.159676266344222e-06,
      "loss": 0.4267,
      "step": 4354
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.799309770591843,
      "learning_rate": 2.15752888786769e-06,
      "loss": 0.4363,
      "step": 4355
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.947235692468703,
      "learning_rate": 2.155382283758644e-06,
      "loss": 0.4084,
      "step": 4356
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0861305413983438,
      "learning_rate": 2.153236454601877e-06,
      "loss": 0.4142,
      "step": 4357
    },
    {
      "epoch": 0.7,
      "grad_norm": 10.7635693601465,
      "learning_rate": 2.1510914009819684e-06,
      "loss": 0.4156,
      "step": 4358
    },
    {
      "epoch": 0.7,
      "grad_norm": 9.668108685312125,
      "learning_rate": 2.148947123483293e-06,
      "loss": 0.4009,
      "step": 4359
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.197390316900548,
      "learning_rate": 2.146803622690015e-06,
      "loss": 0.4384,
      "step": 4360
    },
    {
      "epoch": 0.7,
      "grad_norm": 20.949508982337,
      "learning_rate": 2.144660899186075e-06,
      "loss": 0.3825,
      "step": 4361
    },
    {
      "epoch": 0.7,
      "grad_norm": 8.312990175828494,
      "learning_rate": 2.1425189535552143e-06,
      "loss": 0.4103,
      "step": 4362
    },
    {
      "epoch": 0.7,
      "grad_norm": 9.78545262823547,
      "learning_rate": 2.1403777863809594e-06,
      "loss": 0.475,
      "step": 4363
    },
    {
      "epoch": 0.7,
      "grad_norm": 8.777424862452708,
      "learning_rate": 2.1382373982466213e-06,
      "loss": 0.4163,
      "step": 4364
    },
    {
      "epoch": 0.7,
      "grad_norm": 29.540356303456907,
      "learning_rate": 2.1360977897352992e-06,
      "loss": 0.5227,
      "step": 4365
    },
    {
      "epoch": 0.7,
      "grad_norm": 13.879500482463184,
      "learning_rate": 2.1339589614298844e-06,
      "loss": 0.3996,
      "step": 4366
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.281780099592318,
      "learning_rate": 2.131820913913052e-06,
      "loss": 0.485,
      "step": 4367
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.858714216975375,
      "learning_rate": 2.129683647767262e-06,
      "loss": 0.4537,
      "step": 4368
    },
    {
      "epoch": 0.7,
      "grad_norm": 12.852052781324238,
      "learning_rate": 2.1275471635747695e-06,
      "loss": 0.3989,
      "step": 4369
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.835472331949467,
      "learning_rate": 2.125411461917607e-06,
      "loss": 0.3965,
      "step": 4370
    },
    {
      "epoch": 0.7,
      "grad_norm": 10.2985632096115,
      "learning_rate": 2.1232765433776033e-06,
      "loss": 0.3937,
      "step": 4371
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.9953277142207035,
      "learning_rate": 2.121142408536366e-06,
      "loss": 0.4512,
      "step": 4372
    },
    {
      "epoch": 0.7,
      "grad_norm": 15.010428276545154,
      "learning_rate": 2.1190090579752908e-06,
      "loss": 0.4713,
      "step": 4373
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.799952640573696,
      "learning_rate": 2.116876492275565e-06,
      "loss": 0.3791,
      "step": 4374
    },
    {
      "epoch": 0.7,
      "grad_norm": 9.731776170892486,
      "learning_rate": 2.114744712018155e-06,
      "loss": 0.4667,
      "step": 4375
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.29983685170601,
      "learning_rate": 2.1126137177838197e-06,
      "loss": 0.4766,
      "step": 4376
    },
    {
      "epoch": 0.71,
      "grad_norm": 10.954109319055911,
      "learning_rate": 2.110483510153097e-06,
      "loss": 0.408,
      "step": 4377
    },
    {
      "epoch": 0.71,
      "grad_norm": 9.903235534310632,
      "learning_rate": 2.108354089706318e-06,
      "loss": 0.4359,
      "step": 4378
    },
    {
      "epoch": 0.71,
      "grad_norm": 9.591783186171359,
      "learning_rate": 2.1062254570235934e-06,
      "loss": 0.4451,
      "step": 4379
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.298285921846783,
      "learning_rate": 2.1040976126848208e-06,
      "loss": 0.4084,
      "step": 4380
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.358330758524463,
      "learning_rate": 2.1019705572696836e-06,
      "loss": 0.423,
      "step": 4381
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.43978143728766,
      "learning_rate": 2.099844291357655e-06,
      "loss": 0.4333,
      "step": 4382
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.60983741110548,
      "learning_rate": 2.0977188155279815e-06,
      "loss": 0.4287,
      "step": 4383
    },
    {
      "epoch": 0.71,
      "grad_norm": 19.48883112623078,
      "learning_rate": 2.095594130359704e-06,
      "loss": 0.3816,
      "step": 4384
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.539522298499108,
      "learning_rate": 2.093470236431648e-06,
      "loss": 0.4432,
      "step": 4385
    },
    {
      "epoch": 0.71,
      "grad_norm": 9.683954062576332,
      "learning_rate": 2.0913471343224183e-06,
      "loss": 0.4199,
      "step": 4386
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.279885964523098,
      "learning_rate": 2.0892248246104047e-06,
      "loss": 0.4267,
      "step": 4387
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.223336448097532,
      "learning_rate": 2.0871033078737875e-06,
      "loss": 0.3935,
      "step": 4388
    },
    {
      "epoch": 0.71,
      "grad_norm": 11.759195162361502,
      "learning_rate": 2.0849825846905223e-06,
      "loss": 0.4184,
      "step": 4389
    },
    {
      "epoch": 0.71,
      "grad_norm": 10.933543550089196,
      "learning_rate": 2.0828626556383534e-06,
      "loss": 0.3982,
      "step": 4390
    },
    {
      "epoch": 0.71,
      "grad_norm": 17.80300884195896,
      "learning_rate": 2.0807435212948094e-06,
      "loss": 0.3717,
      "step": 4391
    },
    {
      "epoch": 0.71,
      "grad_norm": 19.334584862210274,
      "learning_rate": 2.078625182237199e-06,
      "loss": 0.4278,
      "step": 4392
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.96169224266698,
      "learning_rate": 2.0765076390426187e-06,
      "loss": 0.3852,
      "step": 4393
    },
    {
      "epoch": 0.71,
      "grad_norm": 10.090401119719939,
      "learning_rate": 2.074390892287944e-06,
      "loss": 0.3973,
      "step": 4394
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.789922024115263,
      "learning_rate": 2.0722749425498332e-06,
      "loss": 0.4515,
      "step": 4395
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.485738012982616,
      "learning_rate": 2.0701597904047332e-06,
      "loss": 0.4441,
      "step": 4396
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.904543206912847,
      "learning_rate": 2.068045436428866e-06,
      "loss": 0.4806,
      "step": 4397
    },
    {
      "epoch": 0.71,
      "grad_norm": 22.833858964293672,
      "learning_rate": 2.0659318811982433e-06,
      "loss": 0.397,
      "step": 4398
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.470022395882339,
      "learning_rate": 2.0638191252886526e-06,
      "loss": 0.4333,
      "step": 4399
    },
    {
      "epoch": 0.71,
      "grad_norm": 39.47916050038513,
      "learning_rate": 2.06170716927567e-06,
      "loss": 0.4291,
      "step": 4400
    },
    {
      "epoch": 0.71,
      "grad_norm": 10.27179095436662,
      "learning_rate": 2.0595960137346494e-06,
      "loss": 0.428,
      "step": 4401
    },
    {
      "epoch": 0.71,
      "grad_norm": 11.484540351984919,
      "learning_rate": 2.0574856592407255e-06,
      "loss": 0.4026,
      "step": 4402
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.351660530253886,
      "learning_rate": 2.0553761063688204e-06,
      "loss": 0.4211,
      "step": 4403
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.819512721404266,
      "learning_rate": 2.0532673556936333e-06,
      "loss": 0.423,
      "step": 4404
    },
    {
      "epoch": 0.71,
      "grad_norm": 9.449572983887284,
      "learning_rate": 2.051159407789644e-06,
      "loss": 0.4245,
      "step": 4405
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.505960540720057,
      "learning_rate": 2.0490522632311173e-06,
      "loss": 0.3922,
      "step": 4406
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.9630812562857,
      "learning_rate": 2.0469459225920987e-06,
      "loss": 0.3867,
      "step": 4407
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.901076205271426,
      "learning_rate": 2.0448403864464123e-06,
      "loss": 0.4438,
      "step": 4408
    },
    {
      "epoch": 0.71,
      "grad_norm": 9.848871198251322,
      "learning_rate": 2.0427356553676625e-06,
      "loss": 0.451,
      "step": 4409
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.545526471702682,
      "learning_rate": 2.0406317299292394e-06,
      "loss": 0.4027,
      "step": 4410
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.359238054637132,
      "learning_rate": 2.0385286107043086e-06,
      "loss": 0.4747,
      "step": 4411
    },
    {
      "epoch": 0.71,
      "grad_norm": 12.146205661525215,
      "learning_rate": 2.0364262982658155e-06,
      "loss": 0.3972,
      "step": 4412
    },
    {
      "epoch": 0.71,
      "grad_norm": 10.466800750346675,
      "learning_rate": 2.0343247931864927e-06,
      "loss": 0.4294,
      "step": 4413
    },
    {
      "epoch": 0.71,
      "grad_norm": 12.994073279549283,
      "learning_rate": 2.0322240960388436e-06,
      "loss": 0.4178,
      "step": 4414
    },
    {
      "epoch": 0.71,
      "grad_norm": 10.059384276326798,
      "learning_rate": 2.030124207395159e-06,
      "loss": 0.4575,
      "step": 4415
    },
    {
      "epoch": 0.71,
      "grad_norm": 45.525884299980575,
      "learning_rate": 2.0280251278275043e-06,
      "loss": 0.4917,
      "step": 4416
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.153790147113465,
      "learning_rate": 2.0259268579077295e-06,
      "loss": 0.3704,
      "step": 4417
    },
    {
      "epoch": 0.71,
      "grad_norm": 12.183936839186744,
      "learning_rate": 2.0238293982074593e-06,
      "loss": 0.3563,
      "step": 4418
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.9656865714321485,
      "learning_rate": 2.0217327492980977e-06,
      "loss": 0.4435,
      "step": 4419
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.34617681880783,
      "learning_rate": 2.0196369117508336e-06,
      "loss": 0.419,
      "step": 4420
    },
    {
      "epoch": 0.71,
      "grad_norm": 10.48706800501729,
      "learning_rate": 2.017541886136627e-06,
      "loss": 0.4419,
      "step": 4421
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.881699617762127,
      "learning_rate": 2.0154476730262244e-06,
      "loss": 0.4754,
      "step": 4422
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.649901093244667,
      "learning_rate": 2.013354272990145e-06,
      "loss": 0.4315,
      "step": 4423
    },
    {
      "epoch": 0.71,
      "grad_norm": 14.790581769198985,
      "learning_rate": 2.011261686598688e-06,
      "loss": 0.4143,
      "step": 4424
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.82473973200117,
      "learning_rate": 2.009169914421934e-06,
      "loss": 0.534,
      "step": 4425
    },
    {
      "epoch": 0.71,
      "grad_norm": 9.922066664136056,
      "learning_rate": 2.0070789570297377e-06,
      "loss": 0.3887,
      "step": 4426
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.22300780706498,
      "learning_rate": 2.004988814991732e-06,
      "loss": 0.5028,
      "step": 4427
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.736755804491099,
      "learning_rate": 2.002899488877332e-06,
      "loss": 0.4131,
      "step": 4428
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.824948514582528,
      "learning_rate": 2.000810979255728e-06,
      "loss": 0.4051,
      "step": 4429
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.173186971412607,
      "learning_rate": 1.9987232866958863e-06,
      "loss": 0.4336,
      "step": 4430
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.905065773763113,
      "learning_rate": 1.9966364117665505e-06,
      "loss": 0.3984,
      "step": 4431
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.643772538989016,
      "learning_rate": 1.9945503550362453e-06,
      "loss": 0.4667,
      "step": 4432
    },
    {
      "epoch": 0.71,
      "grad_norm": 13.382679799362768,
      "learning_rate": 1.99246511707327e-06,
      "loss": 0.4113,
      "step": 4433
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.383399680160455,
      "learning_rate": 1.990380698445698e-06,
      "loss": 0.3519,
      "step": 4434
    },
    {
      "epoch": 0.71,
      "grad_norm": 11.37449709402352,
      "learning_rate": 1.988297099721384e-06,
      "loss": 0.4117,
      "step": 4435
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.956839813475648,
      "learning_rate": 1.9862143214679606e-06,
      "loss": 0.3807,
      "step": 4436
    },
    {
      "epoch": 0.71,
      "grad_norm": 31.986037238045043,
      "learning_rate": 1.984132364252831e-06,
      "loss": 0.4028,
      "step": 4437
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2946333337660927,
      "learning_rate": 1.9820512286431764e-06,
      "loss": 0.4806,
      "step": 4438
    },
    {
      "epoch": 0.72,
      "grad_norm": 10.355038409587957,
      "learning_rate": 1.979970915205959e-06,
      "loss": 0.4028,
      "step": 4439
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.498570633005586,
      "learning_rate": 1.9778914245079127e-06,
      "loss": 0.429,
      "step": 4440
    },
    {
      "epoch": 0.72,
      "grad_norm": 26.988962682738865,
      "learning_rate": 1.975812757115545e-06,
      "loss": 0.4141,
      "step": 4441
    },
    {
      "epoch": 0.72,
      "grad_norm": 11.195857178876848,
      "learning_rate": 1.9737349135951463e-06,
      "loss": 0.4787,
      "step": 4442
    },
    {
      "epoch": 0.72,
      "grad_norm": 16.36707806183333,
      "learning_rate": 1.971657894512775e-06,
      "loss": 0.4428,
      "step": 4443
    },
    {
      "epoch": 0.72,
      "grad_norm": 10.947922100241565,
      "learning_rate": 1.9695817004342715e-06,
      "loss": 0.4366,
      "step": 4444
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.415022766398078,
      "learning_rate": 1.9675063319252467e-06,
      "loss": 0.4384,
      "step": 4445
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.658875228761434,
      "learning_rate": 1.9654317895510867e-06,
      "loss": 0.3381,
      "step": 4446
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.35630093613522,
      "learning_rate": 1.9633580738769577e-06,
      "loss": 0.331,
      "step": 4447
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.232887352381377,
      "learning_rate": 1.961285185467794e-06,
      "loss": 0.4452,
      "step": 4448
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.218960516738631,
      "learning_rate": 1.959213124888307e-06,
      "loss": 0.3543,
      "step": 4449
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.093506284498239,
      "learning_rate": 1.957141892702985e-06,
      "loss": 0.381,
      "step": 4450
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.817138287826958,
      "learning_rate": 1.9550714894760913e-06,
      "loss": 0.5023,
      "step": 4451
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.965315079851673,
      "learning_rate": 1.9530019157716547e-06,
      "loss": 0.4398,
      "step": 4452
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1823351385561875,
      "learning_rate": 1.9509331721534873e-06,
      "loss": 0.4668,
      "step": 4453
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.683122395410382,
      "learning_rate": 1.9488652591851732e-06,
      "loss": 0.4181,
      "step": 4454
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.041355545559082,
      "learning_rate": 1.9467981774300676e-06,
      "loss": 0.3437,
      "step": 4455
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.677519039490286,
      "learning_rate": 1.944731927451299e-06,
      "loss": 0.4202,
      "step": 4456
    },
    {
      "epoch": 0.72,
      "grad_norm": 11.205096180352168,
      "learning_rate": 1.942666509811772e-06,
      "loss": 0.4245,
      "step": 4457
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.782078116243902,
      "learning_rate": 1.940601925074165e-06,
      "loss": 0.477,
      "step": 4458
    },
    {
      "epoch": 0.72,
      "grad_norm": 10.345226605845056,
      "learning_rate": 1.9385381738009273e-06,
      "loss": 0.4779,
      "step": 4459
    },
    {
      "epoch": 0.72,
      "grad_norm": 13.330755324419297,
      "learning_rate": 1.936475256554279e-06,
      "loss": 0.4002,
      "step": 4460
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.0847911725588215,
      "learning_rate": 1.9344131738962183e-06,
      "loss": 0.359,
      "step": 4461
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.581007533990149,
      "learning_rate": 1.932351926388513e-06,
      "loss": 0.4669,
      "step": 4462
    },
    {
      "epoch": 0.72,
      "grad_norm": 16.170267287444368,
      "learning_rate": 1.9302915145927015e-06,
      "loss": 0.4115,
      "step": 4463
    },
    {
      "epoch": 0.72,
      "grad_norm": 9.384153108782398,
      "learning_rate": 1.9282319390701e-06,
      "loss": 0.4721,
      "step": 4464
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.340382604716182,
      "learning_rate": 1.92617320038179e-06,
      "loss": 0.4244,
      "step": 4465
    },
    {
      "epoch": 0.72,
      "grad_norm": 13.073009077945114,
      "learning_rate": 1.924115299088633e-06,
      "loss": 0.4686,
      "step": 4466
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.202668765068819,
      "learning_rate": 1.9220582357512555e-06,
      "loss": 0.493,
      "step": 4467
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.011181770403301,
      "learning_rate": 1.9200020109300567e-06,
      "loss": 0.4265,
      "step": 4468
    },
    {
      "epoch": 0.72,
      "grad_norm": 16.99522640329488,
      "learning_rate": 1.917946625185213e-06,
      "loss": 0.3617,
      "step": 4469
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.165239241897893,
      "learning_rate": 1.9158920790766657e-06,
      "loss": 0.4047,
      "step": 4470
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.131780283687418,
      "learning_rate": 1.913838373164128e-06,
      "loss": 0.3522,
      "step": 4471
    },
    {
      "epoch": 0.72,
      "grad_norm": 9.758083127974773,
      "learning_rate": 1.911785508007088e-06,
      "loss": 0.3787,
      "step": 4472
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.582123736196778,
      "learning_rate": 1.9097334841648064e-06,
      "loss": 0.4153,
      "step": 4473
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.437957406193051,
      "learning_rate": 1.9076823021963036e-06,
      "loss": 0.4083,
      "step": 4474
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.6840378327340435,
      "learning_rate": 1.9056319626603826e-06,
      "loss": 0.356,
      "step": 4475
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.145750917524555,
      "learning_rate": 1.9035824661156128e-06,
      "loss": 0.4746,
      "step": 4476
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.799775287498655,
      "learning_rate": 1.9015338131203325e-06,
      "loss": 0.3714,
      "step": 4477
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.773571653964158,
      "learning_rate": 1.8994860042326501e-06,
      "loss": 0.4117,
      "step": 4478
    },
    {
      "epoch": 0.72,
      "grad_norm": 20.62146403561568,
      "learning_rate": 1.8974390400104458e-06,
      "loss": 0.3832,
      "step": 4479
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.721476026797257,
      "learning_rate": 1.895392921011373e-06,
      "loss": 0.3824,
      "step": 4480
    },
    {
      "epoch": 0.72,
      "grad_norm": 9.683811683500084,
      "learning_rate": 1.8933476477928447e-06,
      "loss": 0.4467,
      "step": 4481
    },
    {
      "epoch": 0.72,
      "grad_norm": 12.6321430043805,
      "learning_rate": 1.8913032209120519e-06,
      "loss": 0.4748,
      "step": 4482
    },
    {
      "epoch": 0.72,
      "grad_norm": 10.034303712241616,
      "learning_rate": 1.8892596409259556e-06,
      "loss": 0.4836,
      "step": 4483
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.595006005249415,
      "learning_rate": 1.8872169083912806e-06,
      "loss": 0.406,
      "step": 4484
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.728369207340053,
      "learning_rate": 1.8851750238645228e-06,
      "loss": 0.4063,
      "step": 4485
    },
    {
      "epoch": 0.72,
      "grad_norm": 9.358594417108787,
      "learning_rate": 1.8831339879019511e-06,
      "loss": 0.4951,
      "step": 4486
    },
    {
      "epoch": 0.72,
      "grad_norm": 9.22701153043282,
      "learning_rate": 1.881093801059596e-06,
      "loss": 0.4751,
      "step": 4487
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.52387648903579,
      "learning_rate": 1.879054463893264e-06,
      "loss": 0.4961,
      "step": 4488
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.114330291485143,
      "learning_rate": 1.8770159769585261e-06,
      "loss": 0.4012,
      "step": 4489
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.002696353087496,
      "learning_rate": 1.8749783408107192e-06,
      "loss": 0.4164,
      "step": 4490
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.806856826435281,
      "learning_rate": 1.8729415560049563e-06,
      "loss": 0.4295,
      "step": 4491
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.47002588598355,
      "learning_rate": 1.8709056230961115e-06,
      "loss": 0.5099,
      "step": 4492
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.6005650858730185,
      "learning_rate": 1.8688705426388264e-06,
      "loss": 0.4637,
      "step": 4493
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.3836870835541095,
      "learning_rate": 1.8668363151875163e-06,
      "loss": 0.4292,
      "step": 4494
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.39085257807561,
      "learning_rate": 1.864802941296362e-06,
      "loss": 0.3879,
      "step": 4495
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.139721817600888,
      "learning_rate": 1.8627704215193082e-06,
      "loss": 0.442,
      "step": 4496
    },
    {
      "epoch": 0.72,
      "grad_norm": 9.419559157773179,
      "learning_rate": 1.8607387564100681e-06,
      "loss": 0.434,
      "step": 4497
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.147661855859265,
      "learning_rate": 1.858707946522127e-06,
      "loss": 0.4354,
      "step": 4498
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.939034758512586,
      "learning_rate": 1.8566779924087315e-06,
      "loss": 0.4035,
      "step": 4499
    },
    {
      "epoch": 0.73,
      "grad_norm": 10.078699321026132,
      "learning_rate": 1.8546488946228952e-06,
      "loss": 0.3746,
      "step": 4500
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.43603794232887,
      "learning_rate": 1.8526206537174019e-06,
      "loss": 0.4804,
      "step": 4501
    },
    {
      "epoch": 0.73,
      "grad_norm": 9.917698533050066,
      "learning_rate": 1.8505932702448042e-06,
      "loss": 0.4768,
      "step": 4502
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.185670272098717,
      "learning_rate": 1.8485667447574101e-06,
      "loss": 0.4751,
      "step": 4503
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.465843839930071,
      "learning_rate": 1.8465410778073046e-06,
      "loss": 0.4428,
      "step": 4504
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.798474733860775,
      "learning_rate": 1.8445162699463365e-06,
      "loss": 0.397,
      "step": 4505
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.614205703576717,
      "learning_rate": 1.842492321726118e-06,
      "loss": 0.39,
      "step": 4506
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1790116772121924,
      "learning_rate": 1.8404692336980272e-06,
      "loss": 0.4923,
      "step": 4507
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.198082523076538,
      "learning_rate": 1.838447006413211e-06,
      "loss": 0.4902,
      "step": 4508
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.663941045909598,
      "learning_rate": 1.8364256404225777e-06,
      "loss": 0.4443,
      "step": 4509
    },
    {
      "epoch": 0.73,
      "grad_norm": 15.554781014486947,
      "learning_rate": 1.834405136276806e-06,
      "loss": 0.4409,
      "step": 4510
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.3112079446470934,
      "learning_rate": 1.8323854945263354e-06,
      "loss": 0.4226,
      "step": 4511
    },
    {
      "epoch": 0.73,
      "grad_norm": 14.151324814611046,
      "learning_rate": 1.83036671572137e-06,
      "loss": 0.4842,
      "step": 4512
    },
    {
      "epoch": 0.73,
      "grad_norm": 13.693167290152273,
      "learning_rate": 1.828348800411885e-06,
      "loss": 0.4029,
      "step": 4513
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.224978524836273,
      "learning_rate": 1.8263317491476112e-06,
      "loss": 0.416,
      "step": 4514
    },
    {
      "epoch": 0.73,
      "grad_norm": 8.352885364369113,
      "learning_rate": 1.824315562478054e-06,
      "loss": 0.3888,
      "step": 4515
    },
    {
      "epoch": 0.73,
      "grad_norm": 20.383188767531394,
      "learning_rate": 1.8223002409524736e-06,
      "loss": 0.4296,
      "step": 4516
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.359239147858997,
      "learning_rate": 1.8202857851199034e-06,
      "loss": 0.3238,
      "step": 4517
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.379114089786967,
      "learning_rate": 1.818272195529135e-06,
      "loss": 0.3931,
      "step": 4518
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.778534179894923,
      "learning_rate": 1.816259472728723e-06,
      "loss": 0.4643,
      "step": 4519
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.000699179034113,
      "learning_rate": 1.8142476172669932e-06,
      "loss": 0.4028,
      "step": 4520
    },
    {
      "epoch": 0.73,
      "grad_norm": 8.440199071135973,
      "learning_rate": 1.812236629692028e-06,
      "loss": 0.395,
      "step": 4521
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.815255094572453,
      "learning_rate": 1.8102265105516741e-06,
      "loss": 0.3779,
      "step": 4522
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.690480123903951,
      "learning_rate": 1.8082172603935455e-06,
      "loss": 0.4091,
      "step": 4523
    },
    {
      "epoch": 0.73,
      "grad_norm": 11.253687091057483,
      "learning_rate": 1.8062088797650208e-06,
      "loss": 0.4604,
      "step": 4524
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.272150845229022,
      "learning_rate": 1.804201369213231e-06,
      "loss": 0.3761,
      "step": 4525
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.090587509423968,
      "learning_rate": 1.8021947292850811e-06,
      "loss": 0.3645,
      "step": 4526
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.474965390855698,
      "learning_rate": 1.8001889605272366e-06,
      "loss": 0.3573,
      "step": 4527
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.159616526586958,
      "learning_rate": 1.7981840634861225e-06,
      "loss": 0.4322,
      "step": 4528
    },
    {
      "epoch": 0.73,
      "grad_norm": 9.51375722735744,
      "learning_rate": 1.7961800387079265e-06,
      "loss": 0.3999,
      "step": 4529
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.217687486172273,
      "learning_rate": 1.794176886738604e-06,
      "loss": 0.3783,
      "step": 4530
    },
    {
      "epoch": 0.73,
      "grad_norm": 10.646457711630864,
      "learning_rate": 1.7921746081238656e-06,
      "loss": 0.3591,
      "step": 4531
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.018200423395607,
      "learning_rate": 1.7901732034091867e-06,
      "loss": 0.4616,
      "step": 4532
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.02769591009989,
      "learning_rate": 1.788172673139807e-06,
      "loss": 0.4683,
      "step": 4533
    },
    {
      "epoch": 0.73,
      "grad_norm": 8.833755381411487,
      "learning_rate": 1.7861730178607274e-06,
      "loss": 0.453,
      "step": 4534
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.253192148843246,
      "learning_rate": 1.784174238116707e-06,
      "loss": 0.4201,
      "step": 4535
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1414014713680958,
      "learning_rate": 1.7821763344522675e-06,
      "loss": 0.4773,
      "step": 4536
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.659071949856766,
      "learning_rate": 1.7801793074116958e-06,
      "loss": 0.4204,
      "step": 4537
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.608664291690792,
      "learning_rate": 1.778183157539034e-06,
      "loss": 0.4432,
      "step": 4538
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.268885163844621,
      "learning_rate": 1.7761878853780918e-06,
      "loss": 0.3616,
      "step": 4539
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.7821856706517964,
      "learning_rate": 1.774193491472434e-06,
      "loss": 0.48,
      "step": 4540
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.765691912077277,
      "learning_rate": 1.772199976365388e-06,
      "loss": 0.4096,
      "step": 4541
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.9870936730716675,
      "learning_rate": 1.7702073406000447e-06,
      "loss": 0.4464,
      "step": 4542
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.378926483478102,
      "learning_rate": 1.768215584719251e-06,
      "loss": 0.4041,
      "step": 4543
    },
    {
      "epoch": 0.73,
      "grad_norm": 8.365345549858345,
      "learning_rate": 1.7662247092656159e-06,
      "loss": 0.4149,
      "step": 4544
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.595828027065605,
      "learning_rate": 1.7642347147815091e-06,
      "loss": 0.407,
      "step": 4545
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.671693620156142,
      "learning_rate": 1.7622456018090638e-06,
      "loss": 0.4202,
      "step": 4546
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.099633766858797,
      "learning_rate": 1.7602573708901627e-06,
      "loss": 0.3776,
      "step": 4547
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.890540105613058,
      "learning_rate": 1.7582700225664574e-06,
      "loss": 0.3792,
      "step": 4548
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.787750814843247,
      "learning_rate": 1.7562835573793585e-06,
      "loss": 0.4217,
      "step": 4549
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.80485350028781,
      "learning_rate": 1.7542979758700323e-06,
      "loss": 0.4209,
      "step": 4550
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.308267665365734,
      "learning_rate": 1.752313278579404e-06,
      "loss": 0.4253,
      "step": 4551
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.744100388949374,
      "learning_rate": 1.7503294660481635e-06,
      "loss": 0.4005,
      "step": 4552
    },
    {
      "epoch": 0.73,
      "grad_norm": 14.917344706595149,
      "learning_rate": 1.7483465388167542e-06,
      "loss": 0.3972,
      "step": 4553
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.036256385290395,
      "learning_rate": 1.746364497425378e-06,
      "loss": 0.417,
      "step": 4554
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.48479127490291,
      "learning_rate": 1.744383342414e-06,
      "loss": 0.3427,
      "step": 4555
    },
    {
      "epoch": 0.73,
      "grad_norm": 8.665985720515634,
      "learning_rate": 1.7424030743223424e-06,
      "loss": 0.4963,
      "step": 4556
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.934116805528895,
      "learning_rate": 1.7404236936898837e-06,
      "loss": 0.394,
      "step": 4557
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.130312903469824,
      "learning_rate": 1.7384452010558605e-06,
      "loss": 0.4399,
      "step": 4558
    },
    {
      "epoch": 0.73,
      "grad_norm": 17.145545285700436,
      "learning_rate": 1.736467596959271e-06,
      "loss": 0.4122,
      "step": 4559
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.8221860966117145,
      "learning_rate": 1.7344908819388678e-06,
      "loss": 0.4287,
      "step": 4560
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.527892503959112,
      "learning_rate": 1.7325150565331612e-06,
      "loss": 0.3714,
      "step": 4561
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.3962441730490545,
      "learning_rate": 1.7305401212804235e-06,
      "loss": 0.3884,
      "step": 4562
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.915484726784404,
      "learning_rate": 1.7285660767186778e-06,
      "loss": 0.4514,
      "step": 4563
    },
    {
      "epoch": 0.74,
      "grad_norm": 14.954831485611054,
      "learning_rate": 1.7265929233857125e-06,
      "loss": 0.4523,
      "step": 4564
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.688720378880985,
      "learning_rate": 1.7246206618190676e-06,
      "loss": 0.4659,
      "step": 4565
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.179100415248463,
      "learning_rate": 1.7226492925560385e-06,
      "loss": 0.4128,
      "step": 4566
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.304398041990792,
      "learning_rate": 1.7206788161336825e-06,
      "loss": 0.4277,
      "step": 4567
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.301338457177896,
      "learning_rate": 1.7187092330888155e-06,
      "loss": 0.3907,
      "step": 4568
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.310549089673422,
      "learning_rate": 1.7167405439579994e-06,
      "loss": 0.4849,
      "step": 4569
    },
    {
      "epoch": 0.74,
      "grad_norm": 11.533623461543193,
      "learning_rate": 1.714772749277563e-06,
      "loss": 0.4775,
      "step": 4570
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0781386720030601,
      "learning_rate": 1.7128058495835886e-06,
      "loss": 0.4635,
      "step": 4571
    },
    {
      "epoch": 0.74,
      "grad_norm": 14.06992913712214,
      "learning_rate": 1.7108398454119134e-06,
      "loss": 0.4523,
      "step": 4572
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.339645809984674,
      "learning_rate": 1.7088747372981285e-06,
      "loss": 0.4367,
      "step": 4573
    },
    {
      "epoch": 0.74,
      "grad_norm": 15.20227216952294,
      "learning_rate": 1.7069105257775859e-06,
      "loss": 0.4487,
      "step": 4574
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.46212923264766,
      "learning_rate": 1.7049472113853937e-06,
      "loss": 0.425,
      "step": 4575
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.501462220732155,
      "learning_rate": 1.7029847946564066e-06,
      "loss": 0.3756,
      "step": 4576
    },
    {
      "epoch": 0.74,
      "grad_norm": 14.066646895580233,
      "learning_rate": 1.7010232761252438e-06,
      "loss": 0.4154,
      "step": 4577
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.787512291620692,
      "learning_rate": 1.6990626563262797e-06,
      "loss": 0.3682,
      "step": 4578
    },
    {
      "epoch": 0.74,
      "grad_norm": 17.15749661157707,
      "learning_rate": 1.6971029357936385e-06,
      "loss": 0.5007,
      "step": 4579
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.156608984134489,
      "learning_rate": 1.6951441150612008e-06,
      "loss": 0.3622,
      "step": 4580
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.790379115869285,
      "learning_rate": 1.6931861946626066e-06,
      "loss": 0.3853,
      "step": 4581
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.259066103804772,
      "learning_rate": 1.6912291751312465e-06,
      "loss": 0.4324,
      "step": 4582
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.230729953696585,
      "learning_rate": 1.6892730570002635e-06,
      "loss": 0.3687,
      "step": 4583
    },
    {
      "epoch": 0.74,
      "grad_norm": 9.990300105168071,
      "learning_rate": 1.6873178408025625e-06,
      "loss": 0.3828,
      "step": 4584
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.468122764424567,
      "learning_rate": 1.6853635270707947e-06,
      "loss": 0.4229,
      "step": 4585
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.922696195350371,
      "learning_rate": 1.6834101163373716e-06,
      "loss": 0.3407,
      "step": 4586
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.373762495046375,
      "learning_rate": 1.6814576091344559e-06,
      "loss": 0.412,
      "step": 4587
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.180575190525728,
      "learning_rate": 1.6795060059939617e-06,
      "loss": 0.4528,
      "step": 4588
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.992521934988659,
      "learning_rate": 1.6775553074475631e-06,
      "loss": 0.4216,
      "step": 4589
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.396207231397725,
      "learning_rate": 1.6756055140266826e-06,
      "loss": 0.4631,
      "step": 4590
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.890771369152579,
      "learning_rate": 1.6736566262624959e-06,
      "loss": 0.4589,
      "step": 4591
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.366268060889483,
      "learning_rate": 1.6717086446859354e-06,
      "loss": 0.3915,
      "step": 4592
    },
    {
      "epoch": 0.74,
      "grad_norm": 25.601613250263814,
      "learning_rate": 1.6697615698276864e-06,
      "loss": 0.3934,
      "step": 4593
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.484650623798676,
      "learning_rate": 1.6678154022181842e-06,
      "loss": 0.3976,
      "step": 4594
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.173629134663297,
      "learning_rate": 1.6658701423876172e-06,
      "loss": 0.4425,
      "step": 4595
    },
    {
      "epoch": 0.74,
      "grad_norm": 12.479155403553335,
      "learning_rate": 1.6639257908659294e-06,
      "loss": 0.4609,
      "step": 4596
    },
    {
      "epoch": 0.74,
      "grad_norm": 9.955595612285999,
      "learning_rate": 1.661982348182819e-06,
      "loss": 0.502,
      "step": 4597
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.1711472517913455,
      "learning_rate": 1.660039814867726e-06,
      "loss": 0.4114,
      "step": 4598
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.650007136229758,
      "learning_rate": 1.6580981914498545e-06,
      "loss": 0.4686,
      "step": 4599
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.086713512739285,
      "learning_rate": 1.6561574784581574e-06,
      "loss": 0.3929,
      "step": 4600
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.773929173316987,
      "learning_rate": 1.6542176764213364e-06,
      "loss": 0.4218,
      "step": 4601
    },
    {
      "epoch": 0.74,
      "grad_norm": 12.665841262605005,
      "learning_rate": 1.6522787858678463e-06,
      "loss": 0.4952,
      "step": 4602
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.245531199524228,
      "learning_rate": 1.6503408073258964e-06,
      "loss": 0.3787,
      "step": 4603
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.202749093500418,
      "learning_rate": 1.648403741323445e-06,
      "loss": 0.4217,
      "step": 4604
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.916055203248557,
      "learning_rate": 1.6464675883881998e-06,
      "loss": 0.3377,
      "step": 4605
    },
    {
      "epoch": 0.74,
      "grad_norm": 21.10247271389902,
      "learning_rate": 1.6445323490476255e-06,
      "loss": 0.476,
      "step": 4606
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.241044742479016,
      "learning_rate": 1.6425980238289324e-06,
      "loss": 0.4573,
      "step": 4607
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.327067079670294,
      "learning_rate": 1.6406646132590865e-06,
      "loss": 0.397,
      "step": 4608
    },
    {
      "epoch": 0.74,
      "grad_norm": 10.60781185986753,
      "learning_rate": 1.6387321178648003e-06,
      "loss": 0.4876,
      "step": 4609
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.903810281500996,
      "learning_rate": 1.6368005381725372e-06,
      "loss": 0.4214,
      "step": 4610
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.571707061212956,
      "learning_rate": 1.6348698747085168e-06,
      "loss": 0.4597,
      "step": 4611
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.106116272818694,
      "learning_rate": 1.6329401279987023e-06,
      "loss": 0.4273,
      "step": 4612
    },
    {
      "epoch": 0.74,
      "grad_norm": 12.613522793385254,
      "learning_rate": 1.6310112985688093e-06,
      "loss": 0.3722,
      "step": 4613
    },
    {
      "epoch": 0.74,
      "grad_norm": 9.457216530778723,
      "learning_rate": 1.629083386944305e-06,
      "loss": 0.4171,
      "step": 4614
    },
    {
      "epoch": 0.74,
      "grad_norm": 9.407565057138966,
      "learning_rate": 1.6271563936504082e-06,
      "loss": 0.4286,
      "step": 4615
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.592314554626736,
      "learning_rate": 1.6252303192120821e-06,
      "loss": 0.4036,
      "step": 4616
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.577679933583275,
      "learning_rate": 1.6233051641540415e-06,
      "loss": 0.4331,
      "step": 4617
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.5187452043022205,
      "learning_rate": 1.6213809290007554e-06,
      "loss": 0.4153,
      "step": 4618
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.8223210786395025,
      "learning_rate": 1.6194576142764363e-06,
      "loss": 0.4471,
      "step": 4619
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.259815969873206,
      "learning_rate": 1.617535220505046e-06,
      "loss": 0.3662,
      "step": 4620
    },
    {
      "epoch": 0.74,
      "grad_norm": 10.13696228355579,
      "learning_rate": 1.6156137482103006e-06,
      "loss": 0.4518,
      "step": 4621
    },
    {
      "epoch": 0.74,
      "grad_norm": 57.16343160844678,
      "learning_rate": 1.6136931979156628e-06,
      "loss": 0.3889,
      "step": 4622
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.21886681893156,
      "learning_rate": 1.6117735701443416e-06,
      "loss": 0.3747,
      "step": 4623
    },
    {
      "epoch": 0.75,
      "grad_norm": 13.636813767408224,
      "learning_rate": 1.6098548654192958e-06,
      "loss": 0.3645,
      "step": 4624
    },
    {
      "epoch": 0.75,
      "grad_norm": 18.248806230269267,
      "learning_rate": 1.6079370842632358e-06,
      "loss": 0.4661,
      "step": 4625
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.001766067642663,
      "learning_rate": 1.606020227198617e-06,
      "loss": 0.3831,
      "step": 4626
    },
    {
      "epoch": 0.75,
      "grad_norm": 45.56941218779736,
      "learning_rate": 1.604104294747642e-06,
      "loss": 0.4539,
      "step": 4627
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1998878029432454,
      "learning_rate": 1.6021892874322675e-06,
      "loss": 0.4419,
      "step": 4628
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.874921802741356,
      "learning_rate": 1.6002752057741906e-06,
      "loss": 0.4457,
      "step": 4629
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.765032231765931,
      "learning_rate": 1.5983620502948632e-06,
      "loss": 0.3657,
      "step": 4630
    },
    {
      "epoch": 0.75,
      "grad_norm": 11.345891485526218,
      "learning_rate": 1.59644982151548e-06,
      "loss": 0.392,
      "step": 4631
    },
    {
      "epoch": 0.75,
      "grad_norm": 12.252487669741566,
      "learning_rate": 1.5945385199569836e-06,
      "loss": 0.384,
      "step": 4632
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.377527005330427,
      "learning_rate": 1.5926281461400673e-06,
      "loss": 0.4566,
      "step": 4633
    },
    {
      "epoch": 0.75,
      "grad_norm": 10.542549984466994,
      "learning_rate": 1.5907187005851676e-06,
      "loss": 0.4099,
      "step": 4634
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2084658991101749,
      "learning_rate": 1.588810183812473e-06,
      "loss": 0.4608,
      "step": 4635
    },
    {
      "epoch": 0.75,
      "grad_norm": 10.288202998465433,
      "learning_rate": 1.5869025963419122e-06,
      "loss": 0.4617,
      "step": 4636
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.244407638174307,
      "learning_rate": 1.584995938693169e-06,
      "loss": 0.4433,
      "step": 4637
    },
    {
      "epoch": 0.75,
      "grad_norm": 8.125490995155578,
      "learning_rate": 1.5830902113856667e-06,
      "loss": 0.3993,
      "step": 4638
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.094428339158442,
      "learning_rate": 1.581185414938577e-06,
      "loss": 0.5109,
      "step": 4639
    },
    {
      "epoch": 0.75,
      "grad_norm": 8.932827421215514,
      "learning_rate": 1.579281549870822e-06,
      "loss": 0.4661,
      "step": 4640
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.1413025803030195,
      "learning_rate": 1.5773786167010657e-06,
      "loss": 0.4754,
      "step": 4641
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.075568966356319,
      "learning_rate": 1.5754766159477174e-06,
      "loss": 0.4084,
      "step": 4642
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.09319808041101,
      "learning_rate": 1.5735755481289371e-06,
      "loss": 0.345,
      "step": 4643
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2118788375193417,
      "learning_rate": 1.5716754137626284e-06,
      "loss": 0.4573,
      "step": 4644
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.130568093279297,
      "learning_rate": 1.5697762133664396e-06,
      "loss": 0.4089,
      "step": 4645
    },
    {
      "epoch": 0.75,
      "grad_norm": 10.077398697389578,
      "learning_rate": 1.5678779474577633e-06,
      "loss": 0.4617,
      "step": 4646
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.503558335218649,
      "learning_rate": 1.5659806165537428e-06,
      "loss": 0.4145,
      "step": 4647
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2038104881067744,
      "learning_rate": 1.5640842211712614e-06,
      "loss": 0.4575,
      "step": 4648
    },
    {
      "epoch": 0.75,
      "grad_norm": 10.987019251152837,
      "learning_rate": 1.5621887618269481e-06,
      "loss": 0.4637,
      "step": 4649
    },
    {
      "epoch": 0.75,
      "grad_norm": 14.278256622617576,
      "learning_rate": 1.5602942390371817e-06,
      "loss": 0.4053,
      "step": 4650
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.8201013658936915,
      "learning_rate": 1.558400653318079e-06,
      "loss": 0.3921,
      "step": 4651
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.053688884360372,
      "learning_rate": 1.556508005185508e-06,
      "loss": 0.4357,
      "step": 4652
    },
    {
      "epoch": 0.75,
      "grad_norm": 13.942405752874604,
      "learning_rate": 1.5546162951550759e-06,
      "loss": 0.4588,
      "step": 4653
    },
    {
      "epoch": 0.75,
      "grad_norm": 8.193451270663555,
      "learning_rate": 1.5527255237421384e-06,
      "loss": 0.385,
      "step": 4654
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.981863504654614,
      "learning_rate": 1.5508356914617933e-06,
      "loss": 0.37,
      "step": 4655
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0994779119821592,
      "learning_rate": 1.5489467988288809e-06,
      "loss": 0.4224,
      "step": 4656
    },
    {
      "epoch": 0.75,
      "grad_norm": 12.958297403644622,
      "learning_rate": 1.5470588463579906e-06,
      "loss": 0.4947,
      "step": 4657
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.752223097478931,
      "learning_rate": 1.5451718345634503e-06,
      "loss": 0.4114,
      "step": 4658
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.1607250437165,
      "learning_rate": 1.5432857639593362e-06,
      "loss": 0.3766,
      "step": 4659
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.430045775903346,
      "learning_rate": 1.5414006350594658e-06,
      "loss": 0.3773,
      "step": 4660
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.098351910611733,
      "learning_rate": 1.5395164483773966e-06,
      "loss": 0.4551,
      "step": 4661
    },
    {
      "epoch": 0.75,
      "grad_norm": 11.405661443815728,
      "learning_rate": 1.537633204426438e-06,
      "loss": 0.3602,
      "step": 4662
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.37381676410755,
      "learning_rate": 1.535750903719635e-06,
      "loss": 0.4099,
      "step": 4663
    },
    {
      "epoch": 0.75,
      "grad_norm": 15.793761840575337,
      "learning_rate": 1.5338695467697767e-06,
      "loss": 0.3894,
      "step": 4664
    },
    {
      "epoch": 0.75,
      "grad_norm": 9.325148263663642,
      "learning_rate": 1.531989134089399e-06,
      "loss": 0.4038,
      "step": 4665
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.796158938602956,
      "learning_rate": 1.5301096661907782e-06,
      "loss": 0.3829,
      "step": 4666
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.0753382101938875,
      "learning_rate": 1.5282311435859325e-06,
      "loss": 0.4524,
      "step": 4667
    },
    {
      "epoch": 0.75,
      "grad_norm": 18.313548427033474,
      "learning_rate": 1.526353566786622e-06,
      "loss": 0.42,
      "step": 4668
    },
    {
      "epoch": 0.75,
      "grad_norm": 9.4989438793356,
      "learning_rate": 1.5244769363043527e-06,
      "loss": 0.3953,
      "step": 4669
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.5374863936432295,
      "learning_rate": 1.5226012526503698e-06,
      "loss": 0.4192,
      "step": 4670
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.079716538407558,
      "learning_rate": 1.5207265163356588e-06,
      "loss": 0.3841,
      "step": 4671
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.6692461619279335,
      "learning_rate": 1.5188527278709514e-06,
      "loss": 0.4285,
      "step": 4672
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3422580863913138,
      "learning_rate": 1.5169798877667207e-06,
      "loss": 0.4582,
      "step": 4673
    },
    {
      "epoch": 0.75,
      "grad_norm": 10.199541682369334,
      "learning_rate": 1.5151079965331788e-06,
      "loss": 0.4424,
      "step": 4674
    },
    {
      "epoch": 0.75,
      "grad_norm": 10.975200998632868,
      "learning_rate": 1.5132370546802794e-06,
      "loss": 0.4698,
      "step": 4675
    },
    {
      "epoch": 0.75,
      "grad_norm": 10.150392570091952,
      "learning_rate": 1.5113670627177202e-06,
      "loss": 0.4784,
      "step": 4676
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.791925786554734,
      "learning_rate": 1.5094980211549382e-06,
      "loss": 0.4379,
      "step": 4677
    },
    {
      "epoch": 0.75,
      "grad_norm": 9.781364996027921,
      "learning_rate": 1.5076299305011095e-06,
      "loss": 0.4149,
      "step": 4678
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.758523227175734,
      "learning_rate": 1.5057627912651574e-06,
      "loss": 0.3857,
      "step": 4679
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.54993798202104,
      "learning_rate": 1.5038966039557384e-06,
      "loss": 0.4054,
      "step": 4680
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.095553493776552,
      "learning_rate": 1.502031369081257e-06,
      "loss": 0.3795,
      "step": 4681
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.819217148493869,
      "learning_rate": 1.5001670871498524e-06,
      "loss": 0.3972,
      "step": 4682
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.63342726126234,
      "learning_rate": 1.4983037586694055e-06,
      "loss": 0.4214,
      "step": 4683
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.701235856709712,
      "learning_rate": 1.4964413841475412e-06,
      "loss": 0.4362,
      "step": 4684
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.341167161807946,
      "learning_rate": 1.4945799640916198e-06,
      "loss": 0.3852,
      "step": 4685
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.648424099530535,
      "learning_rate": 1.4927194990087418e-06,
      "loss": 0.364,
      "step": 4686
    },
    {
      "epoch": 0.76,
      "grad_norm": 16.097428509509562,
      "learning_rate": 1.4908599894057512e-06,
      "loss": 0.4178,
      "step": 4687
    },
    {
      "epoch": 0.76,
      "grad_norm": 9.255350769204757,
      "learning_rate": 1.489001435789233e-06,
      "loss": 0.4116,
      "step": 4688
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.06905511979481,
      "learning_rate": 1.4871438386655018e-06,
      "loss": 0.4294,
      "step": 4689
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.220965583094371,
      "learning_rate": 1.4852871985406208e-06,
      "loss": 0.4179,
      "step": 4690
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.017250768930354,
      "learning_rate": 1.4834315159203927e-06,
      "loss": 0.4308,
      "step": 4691
    },
    {
      "epoch": 0.76,
      "grad_norm": 9.09828115301685,
      "learning_rate": 1.481576791310354e-06,
      "loss": 0.4503,
      "step": 4692
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.980873503916682,
      "learning_rate": 1.4797230252157818e-06,
      "loss": 0.414,
      "step": 4693
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.953017388628192,
      "learning_rate": 1.4778702181416949e-06,
      "loss": 0.3822,
      "step": 4694
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.656739404093136,
      "learning_rate": 1.4760183705928499e-06,
      "loss": 0.3832,
      "step": 4695
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.867230419798012,
      "learning_rate": 1.4741674830737396e-06,
      "loss": 0.4147,
      "step": 4696
    },
    {
      "epoch": 0.76,
      "grad_norm": 10.037414286459864,
      "learning_rate": 1.472317556088596e-06,
      "loss": 0.4386,
      "step": 4697
    },
    {
      "epoch": 0.76,
      "grad_norm": 14.169953079567211,
      "learning_rate": 1.4704685901413928e-06,
      "loss": 0.4598,
      "step": 4698
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.604438551997105,
      "learning_rate": 1.468620585735837e-06,
      "loss": 0.4007,
      "step": 4699
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.338202814470881,
      "learning_rate": 1.466773543375376e-06,
      "loss": 0.3017,
      "step": 4700
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.0901971823117504,
      "learning_rate": 1.4649274635631972e-06,
      "loss": 0.4974,
      "step": 4701
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.784070095862422,
      "learning_rate": 1.4630823468022204e-06,
      "loss": 0.3955,
      "step": 4702
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.265811291867935,
      "learning_rate": 1.4612381935951093e-06,
      "loss": 0.3747,
      "step": 4703
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.5309677228438945,
      "learning_rate": 1.4593950044442612e-06,
      "loss": 0.453,
      "step": 4704
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.109056826410635,
      "learning_rate": 1.4575527798518097e-06,
      "loss": 0.4385,
      "step": 4705
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.384068069497013,
      "learning_rate": 1.4557115203196304e-06,
      "loss": 0.4287,
      "step": 4706
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.047456726143926,
      "learning_rate": 1.453871226349332e-06,
      "loss": 0.4402,
      "step": 4707
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.250867589098619,
      "learning_rate": 1.45203189844226e-06,
      "loss": 0.4889,
      "step": 4708
    },
    {
      "epoch": 0.76,
      "grad_norm": 10.581721365328761,
      "learning_rate": 1.4501935370994985e-06,
      "loss": 0.3741,
      "step": 4709
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.988287206988314,
      "learning_rate": 1.4483561428218717e-06,
      "loss": 0.4316,
      "step": 4710
    },
    {
      "epoch": 0.76,
      "grad_norm": 9.098809711178268,
      "learning_rate": 1.4465197161099305e-06,
      "loss": 0.4455,
      "step": 4711
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.335500236476351,
      "learning_rate": 1.4446842574639708e-06,
      "loss": 0.4337,
      "step": 4712
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.7634684848111855,
      "learning_rate": 1.4428497673840235e-06,
      "loss": 0.4347,
      "step": 4713
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.34945801270178,
      "learning_rate": 1.441016246369853e-06,
      "loss": 0.4275,
      "step": 4714
    },
    {
      "epoch": 0.76,
      "grad_norm": 11.212341327535952,
      "learning_rate": 1.4391836949209597e-06,
      "loss": 0.3857,
      "step": 4715
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.596761707332899,
      "learning_rate": 1.437352113536582e-06,
      "loss": 0.429,
      "step": 4716
    },
    {
      "epoch": 0.76,
      "grad_norm": 9.958990627516943,
      "learning_rate": 1.435521502715697e-06,
      "loss": 0.4115,
      "step": 4717
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.940055691048675,
      "learning_rate": 1.4336918629570069e-06,
      "loss": 0.4426,
      "step": 4718
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.061893793042824,
      "learning_rate": 1.431863194758959e-06,
      "loss": 0.4412,
      "step": 4719
    },
    {
      "epoch": 0.76,
      "grad_norm": 18.28261646655348,
      "learning_rate": 1.4300354986197345e-06,
      "loss": 0.4034,
      "step": 4720
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.260865529445302,
      "learning_rate": 1.4282087750372475e-06,
      "loss": 0.4316,
      "step": 4721
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.258700967367357,
      "learning_rate": 1.4263830245091454e-06,
      "loss": 0.4195,
      "step": 4722
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.597806041817395,
      "learning_rate": 1.4245582475328156e-06,
      "loss": 0.4429,
      "step": 4723
    },
    {
      "epoch": 0.76,
      "grad_norm": 10.99264609316521,
      "learning_rate": 1.4227344446053759e-06,
      "loss": 0.4597,
      "step": 4724
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.7895136763495865,
      "learning_rate": 1.420911616223683e-06,
      "loss": 0.4535,
      "step": 4725
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.04661784543088,
      "learning_rate": 1.4190897628843242e-06,
      "loss": 0.4754,
      "step": 4726
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.959644756315961,
      "learning_rate": 1.4172688850836202e-06,
      "loss": 0.4291,
      "step": 4727
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.185760476766221,
      "learning_rate": 1.4154489833176322e-06,
      "loss": 0.3277,
      "step": 4728
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.563011587331211,
      "learning_rate": 1.41363005808215e-06,
      "loss": 0.4263,
      "step": 4729
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3517934854711944,
      "learning_rate": 1.4118121098726972e-06,
      "loss": 0.4419,
      "step": 4730
    },
    {
      "epoch": 0.76,
      "grad_norm": 10.96599313418365,
      "learning_rate": 1.4099951391845345e-06,
      "loss": 0.3903,
      "step": 4731
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.847731651684053,
      "learning_rate": 1.408179146512657e-06,
      "loss": 0.3482,
      "step": 4732
    },
    {
      "epoch": 0.76,
      "grad_norm": 15.370965521557725,
      "learning_rate": 1.4063641323517886e-06,
      "loss": 0.4742,
      "step": 4733
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.825540222288062,
      "learning_rate": 1.4045500971963882e-06,
      "loss": 0.4253,
      "step": 4734
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.531820966558123,
      "learning_rate": 1.4027370415406528e-06,
      "loss": 0.4135,
      "step": 4735
    },
    {
      "epoch": 0.76,
      "grad_norm": 11.505491077657233,
      "learning_rate": 1.4009249658785058e-06,
      "loss": 0.4026,
      "step": 4736
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.39851432561194,
      "learning_rate": 1.399113870703605e-06,
      "loss": 0.3939,
      "step": 4737
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.1147202540359205,
      "learning_rate": 1.3973037565093455e-06,
      "loss": 0.4186,
      "step": 4738
    },
    {
      "epoch": 0.76,
      "grad_norm": 25.473576998314215,
      "learning_rate": 1.395494623788855e-06,
      "loss": 0.3992,
      "step": 4739
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.988597560782217,
      "learning_rate": 1.3936864730349842e-06,
      "loss": 0.4547,
      "step": 4740
    },
    {
      "epoch": 0.76,
      "grad_norm": 11.842046519064832,
      "learning_rate": 1.3918793047403268e-06,
      "loss": 0.4367,
      "step": 4741
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.544780679844971,
      "learning_rate": 1.3900731193972073e-06,
      "loss": 0.4236,
      "step": 4742
    },
    {
      "epoch": 0.76,
      "grad_norm": 9.482266476532116,
      "learning_rate": 1.3882679174976777e-06,
      "loss": 0.3975,
      "step": 4743
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.047525965063067,
      "learning_rate": 1.386463699533524e-06,
      "loss": 0.4418,
      "step": 4744
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.250636395846284,
      "learning_rate": 1.3846604659962676e-06,
      "loss": 0.4321,
      "step": 4745
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.241619148739895,
      "learning_rate": 1.3828582173771576e-06,
      "loss": 0.4084,
      "step": 4746
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.177054175648749,
      "learning_rate": 1.3810569541671754e-06,
      "loss": 0.3787,
      "step": 4747
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.439514156092396,
      "learning_rate": 1.3792566768570364e-06,
      "loss": 0.3511,
      "step": 4748
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.09834036145748,
      "learning_rate": 1.3774573859371842e-06,
      "loss": 0.4139,
      "step": 4749
    },
    {
      "epoch": 0.77,
      "grad_norm": 9.762571078952499,
      "learning_rate": 1.3756590818977972e-06,
      "loss": 0.427,
      "step": 4750
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0794038150859964,
      "learning_rate": 1.37386176522878e-06,
      "loss": 0.439,
      "step": 4751
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.647485064078477,
      "learning_rate": 1.372065436419775e-06,
      "loss": 0.4598,
      "step": 4752
    },
    {
      "epoch": 0.77,
      "grad_norm": 10.02934074901242,
      "learning_rate": 1.3702700959601483e-06,
      "loss": 0.3638,
      "step": 4753
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.362510397145931,
      "learning_rate": 1.368475744339003e-06,
      "loss": 0.4542,
      "step": 4754
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.389912186454759,
      "learning_rate": 1.366682382045168e-06,
      "loss": 0.4376,
      "step": 4755
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.170886566018786,
      "learning_rate": 1.364890009567204e-06,
      "loss": 0.4524,
      "step": 4756
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.3128007962098565,
      "learning_rate": 1.3630986273934054e-06,
      "loss": 0.4304,
      "step": 4757
    },
    {
      "epoch": 0.77,
      "grad_norm": 9.08799522117687,
      "learning_rate": 1.3613082360117924e-06,
      "loss": 0.4185,
      "step": 4758
    },
    {
      "epoch": 0.77,
      "grad_norm": 17.866289081018913,
      "learning_rate": 1.3595188359101152e-06,
      "loss": 0.4351,
      "step": 4759
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.457582559212692,
      "learning_rate": 1.357730427575858e-06,
      "loss": 0.3738,
      "step": 4760
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.9616537972812,
      "learning_rate": 1.3559430114962345e-06,
      "loss": 0.4459,
      "step": 4761
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.52768059100778,
      "learning_rate": 1.3541565881581815e-06,
      "loss": 0.4266,
      "step": 4762
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2110562917915075,
      "learning_rate": 1.3523711580483717e-06,
      "loss": 0.45,
      "step": 4763
    },
    {
      "epoch": 0.77,
      "grad_norm": 14.546734886259859,
      "learning_rate": 1.350586721653207e-06,
      "loss": 0.392,
      "step": 4764
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.7114285612021725,
      "learning_rate": 1.3488032794588168e-06,
      "loss": 0.4357,
      "step": 4765
    },
    {
      "epoch": 0.77,
      "grad_norm": 11.25814414495884,
      "learning_rate": 1.347020831951057e-06,
      "loss": 0.4871,
      "step": 4766
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.19639877710852,
      "learning_rate": 1.3452393796155194e-06,
      "loss": 0.4209,
      "step": 4767
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.07613534542183,
      "learning_rate": 1.343458922937519e-06,
      "loss": 0.354,
      "step": 4768
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.652272743659418,
      "learning_rate": 1.341679462402099e-06,
      "loss": 0.4296,
      "step": 4769
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.967775619002554,
      "learning_rate": 1.3399009984940376e-06,
      "loss": 0.4239,
      "step": 4770
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.129126482678892,
      "learning_rate": 1.338123531697834e-06,
      "loss": 0.435,
      "step": 4771
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.649837485305931,
      "learning_rate": 1.3363470624977221e-06,
      "loss": 0.4006,
      "step": 4772
    },
    {
      "epoch": 0.77,
      "grad_norm": 14.366720829431518,
      "learning_rate": 1.3345715913776575e-06,
      "loss": 0.392,
      "step": 4773
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.119457539215147,
      "learning_rate": 1.3327971188213317e-06,
      "loss": 0.3707,
      "step": 4774
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.5648236873714,
      "learning_rate": 1.3310236453121562e-06,
      "loss": 0.385,
      "step": 4775
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.8181693412224975,
      "learning_rate": 1.3292511713332767e-06,
      "loss": 0.4953,
      "step": 4776
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.6061573522521755,
      "learning_rate": 1.3274796973675629e-06,
      "loss": 0.4324,
      "step": 4777
    },
    {
      "epoch": 0.77,
      "grad_norm": 13.825480751326745,
      "learning_rate": 1.3257092238976122e-06,
      "loss": 0.3591,
      "step": 4778
    },
    {
      "epoch": 0.77,
      "grad_norm": 19.25773862154062,
      "learning_rate": 1.3239397514057523e-06,
      "loss": 0.4209,
      "step": 4779
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.060051817070309,
      "learning_rate": 1.3221712803740356e-06,
      "loss": 0.4133,
      "step": 4780
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.073113500616108,
      "learning_rate": 1.3204038112842404e-06,
      "loss": 0.4025,
      "step": 4781
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.614316324453301,
      "learning_rate": 1.3186373446178757e-06,
      "loss": 0.4339,
      "step": 4782
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.815605939948523,
      "learning_rate": 1.3168718808561793e-06,
      "loss": 0.3755,
      "step": 4783
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.3773254226286555,
      "learning_rate": 1.3151074204801046e-06,
      "loss": 0.3894,
      "step": 4784
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.579724402021042,
      "learning_rate": 1.313343963970344e-06,
      "loss": 0.363,
      "step": 4785
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.069779306259264,
      "learning_rate": 1.3115815118073116e-06,
      "loss": 0.371,
      "step": 4786
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.751869174939817,
      "learning_rate": 1.3098200644711478e-06,
      "loss": 0.4794,
      "step": 4787
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.845005176100715,
      "learning_rate": 1.3080596224417174e-06,
      "loss": 0.3874,
      "step": 4788
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.796120520823414,
      "learning_rate": 1.3063001861986162e-06,
      "loss": 0.3719,
      "step": 4789
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.795757597625095,
      "learning_rate": 1.3045417562211616e-06,
      "loss": 0.4322,
      "step": 4790
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.42140988415866,
      "learning_rate": 1.3027843329883972e-06,
      "loss": 0.3743,
      "step": 4791
    },
    {
      "epoch": 0.77,
      "grad_norm": 9.967291438561222,
      "learning_rate": 1.3010279169790947e-06,
      "loss": 0.4008,
      "step": 4792
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.453848656225191,
      "learning_rate": 1.2992725086717518e-06,
      "loss": 0.4656,
      "step": 4793
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.257711428115967,
      "learning_rate": 1.2975181085445887e-06,
      "loss": 0.4076,
      "step": 4794
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.044788233291754,
      "learning_rate": 1.2957647170755504e-06,
      "loss": 0.4564,
      "step": 4795
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.613923480801501,
      "learning_rate": 1.2940123347423133e-06,
      "loss": 0.4283,
      "step": 4796
    },
    {
      "epoch": 0.77,
      "grad_norm": 9.054242363437876,
      "learning_rate": 1.292260962022272e-06,
      "loss": 0.4662,
      "step": 4797
    },
    {
      "epoch": 0.77,
      "grad_norm": 9.741412496945943,
      "learning_rate": 1.2905105993925477e-06,
      "loss": 0.4206,
      "step": 4798
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.763334296659333,
      "learning_rate": 1.2887612473299905e-06,
      "loss": 0.4709,
      "step": 4799
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.457734338729853,
      "learning_rate": 1.2870129063111685e-06,
      "loss": 0.4196,
      "step": 4800
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.5119562863606175,
      "learning_rate": 1.2852655768123811e-06,
      "loss": 0.3866,
      "step": 4801
    },
    {
      "epoch": 0.77,
      "grad_norm": 12.733457095218865,
      "learning_rate": 1.2835192593096485e-06,
      "loss": 0.4194,
      "step": 4802
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.946065069622087,
      "learning_rate": 1.2817739542787134e-06,
      "loss": 0.4093,
      "step": 4803
    },
    {
      "epoch": 0.77,
      "grad_norm": 10.880314887596017,
      "learning_rate": 1.2800296621950463e-06,
      "loss": 0.3792,
      "step": 4804
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.511249416942179,
      "learning_rate": 1.2782863835338444e-06,
      "loss": 0.5258,
      "step": 4805
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.86039024105612,
      "learning_rate": 1.2765441187700179e-06,
      "loss": 0.4578,
      "step": 4806
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.851119683417882,
      "learning_rate": 1.2748028683782115e-06,
      "loss": 0.3771,
      "step": 4807
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.794361005822668,
      "learning_rate": 1.2730626328327906e-06,
      "loss": 0.3474,
      "step": 4808
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.2513150652853176,
      "learning_rate": 1.2713234126078423e-06,
      "loss": 0.4734,
      "step": 4809
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.663512757648,
      "learning_rate": 1.2695852081771758e-06,
      "loss": 0.4235,
      "step": 4810
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.661108650185705,
      "learning_rate": 1.267848020014329e-06,
      "loss": 0.3803,
      "step": 4811
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.145690485357668,
      "learning_rate": 1.266111848592561e-06,
      "loss": 0.3994,
      "step": 4812
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.661461533529053,
      "learning_rate": 1.2643766943848484e-06,
      "loss": 0.4569,
      "step": 4813
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.814108637481711,
      "learning_rate": 1.2626425578638973e-06,
      "loss": 0.4524,
      "step": 4814
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.430298968976859,
      "learning_rate": 1.2609094395021354e-06,
      "loss": 0.366,
      "step": 4815
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.375367103054462,
      "learning_rate": 1.259177339771711e-06,
      "loss": 0.3775,
      "step": 4816
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2287926248963557,
      "learning_rate": 1.257446259144494e-06,
      "loss": 0.4245,
      "step": 4817
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.350105195055138,
      "learning_rate": 1.2557161980920824e-06,
      "loss": 0.4589,
      "step": 4818
    },
    {
      "epoch": 0.78,
      "grad_norm": 9.270269980420554,
      "learning_rate": 1.2539871570857893e-06,
      "loss": 0.4408,
      "step": 4819
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.273158463216462,
      "learning_rate": 1.252259136596653e-06,
      "loss": 0.4001,
      "step": 4820
    },
    {
      "epoch": 0.78,
      "grad_norm": 9.43957974473868,
      "learning_rate": 1.250532137095436e-06,
      "loss": 0.3991,
      "step": 4821
    },
    {
      "epoch": 0.78,
      "grad_norm": 9.569199359459839,
      "learning_rate": 1.2488061590526185e-06,
      "loss": 0.4416,
      "step": 4822
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.716206736622448,
      "learning_rate": 1.2470812029384071e-06,
      "loss": 0.3839,
      "step": 4823
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.599728007522551,
      "learning_rate": 1.2453572692227257e-06,
      "loss": 0.4449,
      "step": 4824
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.634701682554734,
      "learning_rate": 1.2436343583752197e-06,
      "loss": 0.4027,
      "step": 4825
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.532351873203318,
      "learning_rate": 1.2419124708652607e-06,
      "loss": 0.417,
      "step": 4826
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.043082239316731,
      "learning_rate": 1.2401916071619374e-06,
      "loss": 0.4196,
      "step": 4827
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.390813896748056,
      "learning_rate": 1.2384717677340585e-06,
      "loss": 0.4674,
      "step": 4828
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.168855612728432,
      "learning_rate": 1.2367529530501571e-06,
      "loss": 0.4384,
      "step": 4829
    },
    {
      "epoch": 0.78,
      "grad_norm": 9.087948860162664,
      "learning_rate": 1.2350351635784875e-06,
      "loss": 0.4515,
      "step": 4830
    },
    {
      "epoch": 0.78,
      "grad_norm": 9.49881774896016,
      "learning_rate": 1.2333183997870207e-06,
      "loss": 0.4665,
      "step": 4831
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.606645952295898,
      "learning_rate": 1.2316026621434502e-06,
      "loss": 0.3427,
      "step": 4832
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.296504266617278,
      "learning_rate": 1.2298879511151906e-06,
      "loss": 0.4123,
      "step": 4833
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.954155165172583,
      "learning_rate": 1.2281742671693798e-06,
      "loss": 0.412,
      "step": 4834
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.030820647092895,
      "learning_rate": 1.2264616107728666e-06,
      "loss": 0.4213,
      "step": 4835
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.226509670338716,
      "learning_rate": 1.2247499823922287e-06,
      "loss": 0.4184,
      "step": 4836
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.055801030254379,
      "learning_rate": 1.2230393824937631e-06,
      "loss": 0.4271,
      "step": 4837
    },
    {
      "epoch": 0.78,
      "grad_norm": 14.778853863372987,
      "learning_rate": 1.2213298115434812e-06,
      "loss": 0.4056,
      "step": 4838
    },
    {
      "epoch": 0.78,
      "grad_norm": 10.875500736448283,
      "learning_rate": 1.219621270007118e-06,
      "loss": 0.4807,
      "step": 4839
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.812803412095989,
      "learning_rate": 1.2179137583501282e-06,
      "loss": 0.463,
      "step": 4840
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.0145360970099695,
      "learning_rate": 1.2162072770376848e-06,
      "loss": 0.3725,
      "step": 4841
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.82177435095611,
      "learning_rate": 1.2145018265346786e-06,
      "loss": 0.4113,
      "step": 4842
    },
    {
      "epoch": 0.78,
      "grad_norm": 12.324405086185829,
      "learning_rate": 1.2127974073057241e-06,
      "loss": 0.4018,
      "step": 4843
    },
    {
      "epoch": 0.78,
      "grad_norm": 9.224705225362557,
      "learning_rate": 1.2110940198151489e-06,
      "loss": 0.4489,
      "step": 4844
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.532376202257181,
      "learning_rate": 1.2093916645270066e-06,
      "loss": 0.3946,
      "step": 4845
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.534333913007494,
      "learning_rate": 1.2076903419050629e-06,
      "loss": 0.4825,
      "step": 4846
    },
    {
      "epoch": 0.78,
      "grad_norm": 12.747471279876997,
      "learning_rate": 1.2059900524128048e-06,
      "loss": 0.3488,
      "step": 4847
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.369645366493282,
      "learning_rate": 1.2042907965134404e-06,
      "loss": 0.4659,
      "step": 4848
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.836203920138631,
      "learning_rate": 1.2025925746698918e-06,
      "loss": 0.4487,
      "step": 4849
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.162206563773072,
      "learning_rate": 1.200895387344801e-06,
      "loss": 0.3704,
      "step": 4850
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.877802731151928,
      "learning_rate": 1.1991992350005294e-06,
      "loss": 0.4521,
      "step": 4851
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.813674888979266,
      "learning_rate": 1.1975041180991576e-06,
      "loss": 0.4337,
      "step": 4852
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.597605002150404,
      "learning_rate": 1.19581003710248e-06,
      "loss": 0.4217,
      "step": 4853
    },
    {
      "epoch": 0.78,
      "grad_norm": 24.20871815898599,
      "learning_rate": 1.1941169924720103e-06,
      "loss": 0.4004,
      "step": 4854
    },
    {
      "epoch": 0.78,
      "grad_norm": 9.861936868830862,
      "learning_rate": 1.1924249846689835e-06,
      "loss": 0.3839,
      "step": 4855
    },
    {
      "epoch": 0.78,
      "grad_norm": 16.127468078524796,
      "learning_rate": 1.1907340141543466e-06,
      "loss": 0.4094,
      "step": 4856
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.604542676373059,
      "learning_rate": 1.189044081388766e-06,
      "loss": 0.4602,
      "step": 4857
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0427336029682848,
      "learning_rate": 1.1873551868326272e-06,
      "loss": 0.4342,
      "step": 4858
    },
    {
      "epoch": 0.78,
      "grad_norm": 10.791213059539333,
      "learning_rate": 1.185667330946033e-06,
      "loss": 0.4099,
      "step": 4859
    },
    {
      "epoch": 0.78,
      "grad_norm": 34.67555047690918,
      "learning_rate": 1.1839805141888012e-06,
      "loss": 0.3813,
      "step": 4860
    },
    {
      "epoch": 0.78,
      "grad_norm": 16.213908656291117,
      "learning_rate": 1.1822947370204647e-06,
      "loss": 0.36,
      "step": 4861
    },
    {
      "epoch": 0.78,
      "grad_norm": 12.967771159934538,
      "learning_rate": 1.180609999900279e-06,
      "loss": 0.3874,
      "step": 4862
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.493203190032089,
      "learning_rate": 1.1789263032872112e-06,
      "loss": 0.4315,
      "step": 4863
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.478590041519976,
      "learning_rate": 1.1772436476399456e-06,
      "loss": 0.3658,
      "step": 4864
    },
    {
      "epoch": 0.78,
      "grad_norm": 9.337306400351714,
      "learning_rate": 1.1755620334168866e-06,
      "loss": 0.5091,
      "step": 4865
    },
    {
      "epoch": 0.78,
      "grad_norm": 10.8222973441737,
      "learning_rate": 1.1738814610761496e-06,
      "loss": 0.4442,
      "step": 4866
    },
    {
      "epoch": 0.78,
      "grad_norm": 16.8350004410709,
      "learning_rate": 1.1722019310755717e-06,
      "loss": 0.4507,
      "step": 4867
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.2404060652488225,
      "learning_rate": 1.1705234438727015e-06,
      "loss": 0.3829,
      "step": 4868
    },
    {
      "epoch": 0.78,
      "grad_norm": 11.951481349415708,
      "learning_rate": 1.1688459999248042e-06,
      "loss": 0.4379,
      "step": 4869
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.542107477687765,
      "learning_rate": 1.1671695996888637e-06,
      "loss": 0.423,
      "step": 4870
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.173045720650641,
      "learning_rate": 1.165494243621576e-06,
      "loss": 0.3366,
      "step": 4871
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.817230880108993,
      "learning_rate": 1.1638199321793563e-06,
      "loss": 0.387,
      "step": 4872
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1257645849375972,
      "learning_rate": 1.1621466658183306e-06,
      "loss": 0.4826,
      "step": 4873
    },
    {
      "epoch": 0.79,
      "grad_norm": 12.143604211494988,
      "learning_rate": 1.1604744449943455e-06,
      "loss": 0.4169,
      "step": 4874
    },
    {
      "epoch": 0.79,
      "grad_norm": 9.033032469971031,
      "learning_rate": 1.1588032701629592e-06,
      "loss": 0.3437,
      "step": 4875
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.933810233315558,
      "learning_rate": 1.1571331417794435e-06,
      "loss": 0.4569,
      "step": 4876
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.270906845343537,
      "learning_rate": 1.1554640602987905e-06,
      "loss": 0.417,
      "step": 4877
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.107057029200767,
      "learning_rate": 1.153796026175703e-06,
      "loss": 0.408,
      "step": 4878
    },
    {
      "epoch": 0.79,
      "grad_norm": 13.046430969569801,
      "learning_rate": 1.1521290398645978e-06,
      "loss": 0.392,
      "step": 4879
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.039660356518825,
      "learning_rate": 1.150463101819609e-06,
      "loss": 0.4277,
      "step": 4880
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.057185128188726,
      "learning_rate": 1.1487982124945861e-06,
      "loss": 0.3533,
      "step": 4881
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.721321919283583,
      "learning_rate": 1.147134372343089e-06,
      "loss": 0.3886,
      "step": 4882
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.565969246659852,
      "learning_rate": 1.1454715818183927e-06,
      "loss": 0.4732,
      "step": 4883
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.19155752636563,
      "learning_rate": 1.1438098413734888e-06,
      "loss": 0.3264,
      "step": 4884
    },
    {
      "epoch": 0.79,
      "grad_norm": 16.138667175847573,
      "learning_rate": 1.142149151461081e-06,
      "loss": 0.4115,
      "step": 4885
    },
    {
      "epoch": 0.79,
      "grad_norm": 13.811759906429646,
      "learning_rate": 1.1404895125335859e-06,
      "loss": 0.4213,
      "step": 4886
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.23119607835735,
      "learning_rate": 1.1388309250431363e-06,
      "loss": 0.4176,
      "step": 4887
    },
    {
      "epoch": 0.79,
      "grad_norm": 18.30188683578474,
      "learning_rate": 1.1371733894415748e-06,
      "loss": 0.3778,
      "step": 4888
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.695343968996843,
      "learning_rate": 1.1355169061804632e-06,
      "loss": 0.4512,
      "step": 4889
    },
    {
      "epoch": 0.79,
      "grad_norm": 12.477559586149049,
      "learning_rate": 1.1338614757110706e-06,
      "loss": 0.4541,
      "step": 4890
    },
    {
      "epoch": 0.79,
      "grad_norm": 15.872409182259652,
      "learning_rate": 1.1322070984843837e-06,
      "loss": 0.4145,
      "step": 4891
    },
    {
      "epoch": 0.79,
      "grad_norm": 9.598142688364083,
      "learning_rate": 1.1305537749510993e-06,
      "loss": 0.4578,
      "step": 4892
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.174761032103409,
      "learning_rate": 1.128901505561627e-06,
      "loss": 0.4353,
      "step": 4893
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.258146760132579,
      "learning_rate": 1.1272502907660937e-06,
      "loss": 0.3567,
      "step": 4894
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.113047766642349,
      "learning_rate": 1.1256001310143327e-06,
      "loss": 0.4309,
      "step": 4895
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1136941523228963,
      "learning_rate": 1.1239510267558962e-06,
      "loss": 0.4268,
      "step": 4896
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.169937547317659,
      "learning_rate": 1.1223029784400436e-06,
      "loss": 0.432,
      "step": 4897
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.219536882294408,
      "learning_rate": 1.120655986515748e-06,
      "loss": 0.3921,
      "step": 4898
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.828703023916113,
      "learning_rate": 1.1190100514316977e-06,
      "loss": 0.4172,
      "step": 4899
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.69754409550244,
      "learning_rate": 1.1173651736362889e-06,
      "loss": 0.3658,
      "step": 4900
    },
    {
      "epoch": 0.79,
      "grad_norm": 13.589375433301951,
      "learning_rate": 1.1157213535776312e-06,
      "loss": 0.3768,
      "step": 4901
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.138009918242657,
      "learning_rate": 1.114078591703548e-06,
      "loss": 0.4149,
      "step": 4902
    },
    {
      "epoch": 0.79,
      "grad_norm": 12.98675754904074,
      "learning_rate": 1.1124368884615748e-06,
      "loss": 0.3853,
      "step": 4903
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.215564516546042,
      "learning_rate": 1.1107962442989518e-06,
      "loss": 0.4116,
      "step": 4904
    },
    {
      "epoch": 0.79,
      "grad_norm": 17.307138561534714,
      "learning_rate": 1.1091566596626384e-06,
      "loss": 0.3917,
      "step": 4905
    },
    {
      "epoch": 0.79,
      "grad_norm": 9.83477257736823,
      "learning_rate": 1.1075181349993042e-06,
      "loss": 0.4296,
      "step": 4906
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.867720886453453,
      "learning_rate": 1.1058806707553266e-06,
      "loss": 0.4034,
      "step": 4907
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.0926177512602955,
      "learning_rate": 1.104244267376795e-06,
      "loss": 0.4437,
      "step": 4908
    },
    {
      "epoch": 0.79,
      "grad_norm": 13.626745599928466,
      "learning_rate": 1.1026089253095134e-06,
      "loss": 0.4291,
      "step": 4909
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.2030915909672855,
      "learning_rate": 1.1009746449989916e-06,
      "loss": 0.4646,
      "step": 4910
    },
    {
      "epoch": 0.79,
      "grad_norm": 10.15358081378589,
      "learning_rate": 1.0993414268904552e-06,
      "loss": 0.3852,
      "step": 4911
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.900883586711025,
      "learning_rate": 1.0977092714288345e-06,
      "loss": 0.4327,
      "step": 4912
    },
    {
      "epoch": 0.79,
      "grad_norm": 72.6580849504174,
      "learning_rate": 1.0960781790587776e-06,
      "loss": 0.4626,
      "step": 4913
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.72585593304791,
      "learning_rate": 1.0944481502246368e-06,
      "loss": 0.3884,
      "step": 4914
    },
    {
      "epoch": 0.79,
      "grad_norm": 30.363817516976347,
      "learning_rate": 1.0928191853704757e-06,
      "loss": 0.4103,
      "step": 4915
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.780525590379177,
      "learning_rate": 1.0911912849400712e-06,
      "loss": 0.452,
      "step": 4916
    },
    {
      "epoch": 0.79,
      "grad_norm": 15.30028644570836,
      "learning_rate": 1.089564449376907e-06,
      "loss": 0.3628,
      "step": 4917
    },
    {
      "epoch": 0.79,
      "grad_norm": 12.452722242899878,
      "learning_rate": 1.0879386791241797e-06,
      "loss": 0.4274,
      "step": 4918
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.7821004256668616,
      "learning_rate": 1.086313974624793e-06,
      "loss": 0.4,
      "step": 4919
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.786521716995293,
      "learning_rate": 1.0846903363213595e-06,
      "loss": 0.3846,
      "step": 4920
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.547791064923995,
      "learning_rate": 1.083067764656206e-06,
      "loss": 0.3836,
      "step": 4921
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3809455618444775,
      "learning_rate": 1.0814462600713642e-06,
      "loss": 0.4703,
      "step": 4922
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.379113198955013,
      "learning_rate": 1.0798258230085756e-06,
      "loss": 0.4598,
      "step": 4923
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.964939935856226,
      "learning_rate": 1.078206453909293e-06,
      "loss": 0.4321,
      "step": 4924
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.940795913640375,
      "learning_rate": 1.0765881532146793e-06,
      "loss": 0.3879,
      "step": 4925
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.74925577751493,
      "learning_rate": 1.0749709213656001e-06,
      "loss": 0.4159,
      "step": 4926
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.313000682212276,
      "learning_rate": 1.0733547588026355e-06,
      "loss": 0.4708,
      "step": 4927
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.199337850968073,
      "learning_rate": 1.071739665966075e-06,
      "loss": 0.4357,
      "step": 4928
    },
    {
      "epoch": 0.79,
      "grad_norm": 9.858182101830659,
      "learning_rate": 1.0701256432959123e-06,
      "loss": 0.3574,
      "step": 4929
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.826240907848948,
      "learning_rate": 1.0685126912318511e-06,
      "loss": 0.4226,
      "step": 4930
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.013713067614893,
      "learning_rate": 1.0669008102133044e-06,
      "loss": 0.4422,
      "step": 4931
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.081971722930161,
      "learning_rate": 1.0652900006793953e-06,
      "loss": 0.3924,
      "step": 4932
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.943272492713196,
      "learning_rate": 1.0636802630689508e-06,
      "loss": 0.4615,
      "step": 4933
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.5140519934195265,
      "learning_rate": 1.0620715978205066e-06,
      "loss": 0.3525,
      "step": 4934
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.067346893747588,
      "learning_rate": 1.0604640053723098e-06,
      "loss": 0.4371,
      "step": 4935
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.667967905546646,
      "learning_rate": 1.058857486162312e-06,
      "loss": 0.3647,
      "step": 4936
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.2920299974096565,
      "learning_rate": 1.0572520406281716e-06,
      "loss": 0.3445,
      "step": 4937
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.956711707048635,
      "learning_rate": 1.0556476692072598e-06,
      "loss": 0.3427,
      "step": 4938
    },
    {
      "epoch": 0.8,
      "grad_norm": 12.600431628391135,
      "learning_rate": 1.0540443723366478e-06,
      "loss": 0.4162,
      "step": 4939
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.489760134334394,
      "learning_rate": 1.0524421504531212e-06,
      "loss": 0.4186,
      "step": 4940
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.139154625046737,
      "learning_rate": 1.0508410039931683e-06,
      "loss": 0.4075,
      "step": 4941
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.192614250626416,
      "learning_rate": 1.0492409333929833e-06,
      "loss": 0.3863,
      "step": 4942
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.8984561597926985,
      "learning_rate": 1.0476419390884723e-06,
      "loss": 0.3916,
      "step": 4943
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.2364041240399715,
      "learning_rate": 1.046044021515245e-06,
      "loss": 0.363,
      "step": 4944
    },
    {
      "epoch": 0.8,
      "grad_norm": 15.079074923930056,
      "learning_rate": 1.0444471811086166e-06,
      "loss": 0.3714,
      "step": 4945
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.656039669805464,
      "learning_rate": 1.0428514183036109e-06,
      "loss": 0.3978,
      "step": 4946
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2374127938961095,
      "learning_rate": 1.0412567335349616e-06,
      "loss": 0.4888,
      "step": 4947
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.830543773477803,
      "learning_rate": 1.0396631272370982e-06,
      "loss": 0.3774,
      "step": 4948
    },
    {
      "epoch": 0.8,
      "grad_norm": 12.974909796855309,
      "learning_rate": 1.0380705998441664e-06,
      "loss": 0.3599,
      "step": 4949
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.90625739868267,
      "learning_rate": 1.0364791517900164e-06,
      "loss": 0.4047,
      "step": 4950
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.4773383034245455,
      "learning_rate": 1.0348887835082e-06,
      "loss": 0.3973,
      "step": 4951
    },
    {
      "epoch": 0.8,
      "grad_norm": 9.863954438828529,
      "learning_rate": 1.0332994954319763e-06,
      "loss": 0.4799,
      "step": 4952
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.146777214949624,
      "learning_rate": 1.031711287994313e-06,
      "loss": 0.4244,
      "step": 4953
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.369826997485884,
      "learning_rate": 1.0301241616278845e-06,
      "loss": 0.4147,
      "step": 4954
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.849554090027531,
      "learning_rate": 1.0285381167650615e-06,
      "loss": 0.3987,
      "step": 4955
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.095824091148038,
      "learning_rate": 1.0269531538379295e-06,
      "loss": 0.4238,
      "step": 4956
    },
    {
      "epoch": 0.8,
      "grad_norm": 20.55153736765723,
      "learning_rate": 1.0253692732782778e-06,
      "loss": 0.3951,
      "step": 4957
    },
    {
      "epoch": 0.8,
      "grad_norm": 9.279900706244042,
      "learning_rate": 1.0237864755175969e-06,
      "loss": 0.4272,
      "step": 4958
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.260168225815382,
      "learning_rate": 1.022204760987084e-06,
      "loss": 0.401,
      "step": 4959
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.032808876788408,
      "learning_rate": 1.0206241301176439e-06,
      "loss": 0.4949,
      "step": 4960
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.836601342135245,
      "learning_rate": 1.0190445833398814e-06,
      "loss": 0.4721,
      "step": 4961
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.477879073364938,
      "learning_rate": 1.0174661210841119e-06,
      "loss": 0.4417,
      "step": 4962
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.353823630239038,
      "learning_rate": 1.0158887437803499e-06,
      "loss": 0.438,
      "step": 4963
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.861081314770807,
      "learning_rate": 1.0143124518583158e-06,
      "loss": 0.4406,
      "step": 4964
    },
    {
      "epoch": 0.8,
      "grad_norm": 9.334349305032633,
      "learning_rate": 1.012737245747437e-06,
      "loss": 0.4935,
      "step": 4965
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.167322926874983,
      "learning_rate": 1.0111631258768416e-06,
      "loss": 0.4175,
      "step": 4966
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.953860508967381,
      "learning_rate": 1.0095900926753632e-06,
      "loss": 0.3807,
      "step": 4967
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.197595117450276,
      "learning_rate": 1.0080181465715394e-06,
      "loss": 0.3851,
      "step": 4968
    },
    {
      "epoch": 0.8,
      "grad_norm": 10.924122292778346,
      "learning_rate": 1.0064472879936132e-06,
      "loss": 0.4282,
      "step": 4969
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.531869034942238,
      "learning_rate": 1.0048775173695285e-06,
      "loss": 0.3676,
      "step": 4970
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.307874892284786,
      "learning_rate": 1.0033088351269338e-06,
      "loss": 0.4513,
      "step": 4971
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.287532284042153,
      "learning_rate": 1.0017412416931826e-06,
      "loss": 0.3795,
      "step": 4972
    },
    {
      "epoch": 0.8,
      "grad_norm": 15.654009692165715,
      "learning_rate": 1.0001747374953297e-06,
      "loss": 0.3887,
      "step": 4973
    },
    {
      "epoch": 0.8,
      "grad_norm": 10.775942144550207,
      "learning_rate": 9.986093229601328e-07,
      "loss": 0.3329,
      "step": 4974
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.487499343942186,
      "learning_rate": 9.970449985140557e-07,
      "loss": 0.3832,
      "step": 4975
    },
    {
      "epoch": 0.8,
      "grad_norm": 10.42239110313061,
      "learning_rate": 9.954817645832654e-07,
      "loss": 0.4377,
      "step": 4976
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.963052878324259,
      "learning_rate": 9.939196215936253e-07,
      "loss": 0.4652,
      "step": 4977
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.882870542554774,
      "learning_rate": 9.923585699707084e-07,
      "loss": 0.4309,
      "step": 4978
    },
    {
      "epoch": 0.8,
      "grad_norm": 17.696632571582906,
      "learning_rate": 9.907986101397898e-07,
      "loss": 0.456,
      "step": 4979
    },
    {
      "epoch": 0.8,
      "grad_norm": 10.098344269277957,
      "learning_rate": 9.892397425258437e-07,
      "loss": 0.4707,
      "step": 4980
    },
    {
      "epoch": 0.8,
      "grad_norm": 16.544295198139,
      "learning_rate": 9.876819675535477e-07,
      "loss": 0.4684,
      "step": 4981
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.965667646841321,
      "learning_rate": 9.861252856472857e-07,
      "loss": 0.3916,
      "step": 4982
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.193253063214609,
      "learning_rate": 9.845696972311385e-07,
      "loss": 0.4676,
      "step": 4983
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.036223267656382,
      "learning_rate": 9.830152027288907e-07,
      "loss": 0.4002,
      "step": 4984
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.938859508741169,
      "learning_rate": 9.81461802564032e-07,
      "loss": 0.3744,
      "step": 4985
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.290840342282463,
      "learning_rate": 9.799094971597483e-07,
      "loss": 0.4288,
      "step": 4986
    },
    {
      "epoch": 0.8,
      "grad_norm": 25.701346448766785,
      "learning_rate": 9.783582869389336e-07,
      "loss": 0.4139,
      "step": 4987
    },
    {
      "epoch": 0.8,
      "grad_norm": 11.838085518267414,
      "learning_rate": 9.768081723241785e-07,
      "loss": 0.393,
      "step": 4988
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.870090570475528,
      "learning_rate": 9.752591537377758e-07,
      "loss": 0.4114,
      "step": 4989
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.032666404941196,
      "learning_rate": 9.737112316017221e-07,
      "loss": 0.3744,
      "step": 4990
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.9707821947421165,
      "learning_rate": 9.72164406337716e-07,
      "loss": 0.4239,
      "step": 4991
    },
    {
      "epoch": 0.8,
      "grad_norm": 9.060884011168355,
      "learning_rate": 9.706186783671535e-07,
      "loss": 0.4425,
      "step": 4992
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.996254019001918,
      "learning_rate": 9.69074048111132e-07,
      "loss": 0.4072,
      "step": 4993
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.58636588938138,
      "learning_rate": 9.675305159904546e-07,
      "loss": 0.3688,
      "step": 4994
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.58696816776118,
      "learning_rate": 9.659880824256202e-07,
      "loss": 0.4528,
      "step": 4995
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.1081586607191145,
      "learning_rate": 9.644467478368286e-07,
      "loss": 0.4519,
      "step": 4996
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.281218821644457,
      "learning_rate": 9.629065126439842e-07,
      "loss": 0.4273,
      "step": 4997
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.869212687187768,
      "learning_rate": 9.61367377266691e-07,
      "loss": 0.4843,
      "step": 4998
    },
    {
      "epoch": 0.81,
      "grad_norm": 9.437692765761465,
      "learning_rate": 9.59829342124247e-07,
      "loss": 0.4287,
      "step": 4999
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.8777397953456045,
      "learning_rate": 9.582924076356587e-07,
      "loss": 0.4853,
      "step": 5000
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.337724926224135,
      "learning_rate": 9.5675657421963e-07,
      "loss": 0.4358,
      "step": 5001
    },
    {
      "epoch": 0.81,
      "grad_norm": 10.23514949433835,
      "learning_rate": 9.552218422945636e-07,
      "loss": 0.4233,
      "step": 5002
    },
    {
      "epoch": 0.81,
      "grad_norm": 12.399285657233223,
      "learning_rate": 9.536882122785602e-07,
      "loss": 0.3559,
      "step": 5003
    },
    {
      "epoch": 0.81,
      "grad_norm": 103.07473343182457,
      "learning_rate": 9.52155684589427e-07,
      "loss": 0.3986,
      "step": 5004
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.24667607287334,
      "learning_rate": 9.506242596446641e-07,
      "loss": 0.3972,
      "step": 5005
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.704459868219285,
      "learning_rate": 9.490939378614739e-07,
      "loss": 0.4332,
      "step": 5006
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.909994027098494,
      "learning_rate": 9.4756471965676e-07,
      "loss": 0.4195,
      "step": 5007
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1908633225652752,
      "learning_rate": 9.46036605447121e-07,
      "loss": 0.4562,
      "step": 5008
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.898239638696525,
      "learning_rate": 9.445095956488604e-07,
      "loss": 0.4203,
      "step": 5009
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.931213696842268,
      "learning_rate": 9.42983690677975e-07,
      "loss": 0.459,
      "step": 5010
    },
    {
      "epoch": 0.81,
      "grad_norm": 11.746113045100664,
      "learning_rate": 9.414588909501654e-07,
      "loss": 0.4417,
      "step": 5011
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.110872701715413,
      "learning_rate": 9.399351968808285e-07,
      "loss": 0.4614,
      "step": 5012
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.282669694218209,
      "learning_rate": 9.384126088850592e-07,
      "loss": 0.5,
      "step": 5013
    },
    {
      "epoch": 0.81,
      "grad_norm": 10.611356955787391,
      "learning_rate": 9.368911273776543e-07,
      "loss": 0.3931,
      "step": 5014
    },
    {
      "epoch": 0.81,
      "grad_norm": 9.486513387758569,
      "learning_rate": 9.35370752773106e-07,
      "loss": 0.4794,
      "step": 5015
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.956353836422622,
      "learning_rate": 9.338514854856073e-07,
      "loss": 0.3784,
      "step": 5016
    },
    {
      "epoch": 0.81,
      "grad_norm": 11.38250462575223,
      "learning_rate": 9.323333259290484e-07,
      "loss": 0.4483,
      "step": 5017
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.924234592942446,
      "learning_rate": 9.308162745170163e-07,
      "loss": 0.3826,
      "step": 5018
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.271223702875999,
      "learning_rate": 9.293003316627985e-07,
      "loss": 0.3633,
      "step": 5019
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.939777435546889,
      "learning_rate": 9.277854977793827e-07,
      "loss": 0.4074,
      "step": 5020
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.543684184766702,
      "learning_rate": 9.262717732794457e-07,
      "loss": 0.438,
      "step": 5021
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.478895814218908,
      "learning_rate": 9.247591585753707e-07,
      "loss": 0.4307,
      "step": 5022
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.114989081342317,
      "learning_rate": 9.232476540792367e-07,
      "loss": 0.4465,
      "step": 5023
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.845936716756069,
      "learning_rate": 9.217372602028185e-07,
      "loss": 0.4219,
      "step": 5024
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.77658401705603,
      "learning_rate": 9.202279773575873e-07,
      "loss": 0.3496,
      "step": 5025
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.26252364972767,
      "learning_rate": 9.187198059547153e-07,
      "loss": 0.4356,
      "step": 5026
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.138575439222402,
      "learning_rate": 9.172127464050701e-07,
      "loss": 0.4114,
      "step": 5027
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.469923901436861,
      "learning_rate": 9.157067991192137e-07,
      "loss": 0.472,
      "step": 5028
    },
    {
      "epoch": 0.81,
      "grad_norm": 10.216767559556537,
      "learning_rate": 9.14201964507409e-07,
      "loss": 0.4154,
      "step": 5029
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.9130901693400455,
      "learning_rate": 9.126982429796172e-07,
      "loss": 0.4365,
      "step": 5030
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.246565335193918,
      "learning_rate": 9.111956349454904e-07,
      "loss": 0.3882,
      "step": 5031
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.706807199624973,
      "learning_rate": 9.0969414081438e-07,
      "loss": 0.4016,
      "step": 5032
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.071067235671073,
      "learning_rate": 9.081937609953367e-07,
      "loss": 0.3895,
      "step": 5033
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.314847151289053,
      "learning_rate": 9.066944958971046e-07,
      "loss": 0.4585,
      "step": 5034
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.808177914660189,
      "learning_rate": 9.051963459281232e-07,
      "loss": 0.394,
      "step": 5035
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.361053525014949,
      "learning_rate": 9.03699311496532e-07,
      "loss": 0.449,
      "step": 5036
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.768545414900244,
      "learning_rate": 9.022033930101625e-07,
      "loss": 0.4506,
      "step": 5037
    },
    {
      "epoch": 0.81,
      "grad_norm": 17.874350752070075,
      "learning_rate": 9.007085908765467e-07,
      "loss": 0.4147,
      "step": 5038
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.2758915111608795,
      "learning_rate": 8.992149055029081e-07,
      "loss": 0.3686,
      "step": 5039
    },
    {
      "epoch": 0.81,
      "grad_norm": 41.76492998447651,
      "learning_rate": 8.977223372961663e-07,
      "loss": 0.3433,
      "step": 5040
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.987419195651066,
      "learning_rate": 8.962308866629416e-07,
      "loss": 0.4251,
      "step": 5041
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.777584928992443,
      "learning_rate": 8.947405540095444e-07,
      "loss": 0.4039,
      "step": 5042
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.187751934916821,
      "learning_rate": 8.932513397419812e-07,
      "loss": 0.4226,
      "step": 5043
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.730972702504362,
      "learning_rate": 8.917632442659563e-07,
      "loss": 0.3469,
      "step": 5044
    },
    {
      "epoch": 0.81,
      "grad_norm": 9.530501604764343,
      "learning_rate": 8.902762679868687e-07,
      "loss": 0.4028,
      "step": 5045
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.25284876830902,
      "learning_rate": 8.887904113098111e-07,
      "loss": 0.5188,
      "step": 5046
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.38209678098137,
      "learning_rate": 8.873056746395703e-07,
      "loss": 0.4911,
      "step": 5047
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.289130904746449,
      "learning_rate": 8.858220583806309e-07,
      "loss": 0.5184,
      "step": 5048
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.307090914903794,
      "learning_rate": 8.843395629371738e-07,
      "loss": 0.4796,
      "step": 5049
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.4389697513109665,
      "learning_rate": 8.828581887130655e-07,
      "loss": 0.3963,
      "step": 5050
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.864259839253838,
      "learning_rate": 8.813779361118763e-07,
      "loss": 0.4268,
      "step": 5051
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.783937020930577,
      "learning_rate": 8.79898805536869e-07,
      "loss": 0.3773,
      "step": 5052
    },
    {
      "epoch": 0.81,
      "grad_norm": 9.811138935871762,
      "learning_rate": 8.784207973909986e-07,
      "loss": 0.4421,
      "step": 5053
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.266262088179931,
      "learning_rate": 8.769439120769135e-07,
      "loss": 0.4213,
      "step": 5054
    },
    {
      "epoch": 0.81,
      "grad_norm": 33.33006529592601,
      "learning_rate": 8.754681499969608e-07,
      "loss": 0.4191,
      "step": 5055
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.2625330347314,
      "learning_rate": 8.739935115531772e-07,
      "loss": 0.478,
      "step": 5056
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.600008122911344,
      "learning_rate": 8.725199971472942e-07,
      "loss": 0.3498,
      "step": 5057
    },
    {
      "epoch": 0.81,
      "grad_norm": 13.994978585331703,
      "learning_rate": 8.7104760718074e-07,
      "loss": 0.3388,
      "step": 5058
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.982346453941434,
      "learning_rate": 8.69576342054631e-07,
      "loss": 0.396,
      "step": 5059
    },
    {
      "epoch": 0.82,
      "grad_norm": 11.120155084141487,
      "learning_rate": 8.681062021697839e-07,
      "loss": 0.4331,
      "step": 5060
    },
    {
      "epoch": 0.82,
      "grad_norm": 11.85996426607785,
      "learning_rate": 8.666371879267038e-07,
      "loss": 0.43,
      "step": 5061
    },
    {
      "epoch": 0.82,
      "grad_norm": 13.23469206285628,
      "learning_rate": 8.65169299725589e-07,
      "loss": 0.4509,
      "step": 5062
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.889834146422762,
      "learning_rate": 8.637025379663355e-07,
      "loss": 0.4901,
      "step": 5063
    },
    {
      "epoch": 0.82,
      "grad_norm": 9.136873429072761,
      "learning_rate": 8.622369030485283e-07,
      "loss": 0.3732,
      "step": 5064
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.858097165859271,
      "learning_rate": 8.607723953714453e-07,
      "loss": 0.445,
      "step": 5065
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.123284154519073,
      "learning_rate": 8.593090153340611e-07,
      "loss": 0.3695,
      "step": 5066
    },
    {
      "epoch": 0.82,
      "grad_norm": 9.005781728346276,
      "learning_rate": 8.578467633350407e-07,
      "loss": 0.4059,
      "step": 5067
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.880776543816584,
      "learning_rate": 8.56385639772741e-07,
      "loss": 0.3374,
      "step": 5068
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.990039657052862,
      "learning_rate": 8.549256450452109e-07,
      "loss": 0.433,
      "step": 5069
    },
    {
      "epoch": 0.82,
      "grad_norm": 58.3536616355321,
      "learning_rate": 8.534667795501955e-07,
      "loss": 0.3737,
      "step": 5070
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.283607640655086,
      "learning_rate": 8.520090436851314e-07,
      "loss": 0.3981,
      "step": 5071
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.622247180882688,
      "learning_rate": 8.505524378471408e-07,
      "loss": 0.426,
      "step": 5072
    },
    {
      "epoch": 0.82,
      "grad_norm": 9.219660306114157,
      "learning_rate": 8.490969624330469e-07,
      "loss": 0.4115,
      "step": 5073
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.284870532914952,
      "learning_rate": 8.476426178393621e-07,
      "loss": 0.3513,
      "step": 5074
    },
    {
      "epoch": 0.82,
      "grad_norm": 12.855142735261378,
      "learning_rate": 8.461894044622882e-07,
      "loss": 0.4067,
      "step": 5075
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.357532557638631,
      "learning_rate": 8.447373226977201e-07,
      "loss": 0.4784,
      "step": 5076
    },
    {
      "epoch": 0.82,
      "grad_norm": 9.555023108575012,
      "learning_rate": 8.432863729412466e-07,
      "loss": 0.4062,
      "step": 5077
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.916831346099286,
      "learning_rate": 8.418365555881458e-07,
      "loss": 0.3929,
      "step": 5078
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.419111076422727,
      "learning_rate": 8.403878710333868e-07,
      "loss": 0.4166,
      "step": 5079
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.644087443411669,
      "learning_rate": 8.389403196716328e-07,
      "loss": 0.4285,
      "step": 5080
    },
    {
      "epoch": 0.82,
      "grad_norm": 14.911990514687616,
      "learning_rate": 8.37493901897235e-07,
      "loss": 0.427,
      "step": 5081
    },
    {
      "epoch": 0.82,
      "grad_norm": 9.121109002864998,
      "learning_rate": 8.360486181042399e-07,
      "loss": 0.4209,
      "step": 5082
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.789623457700772,
      "learning_rate": 8.346044686863808e-07,
      "loss": 0.4171,
      "step": 5083
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1576800855573193,
      "learning_rate": 8.331614540370836e-07,
      "loss": 0.4864,
      "step": 5084
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.590676409719685,
      "learning_rate": 8.317195745494666e-07,
      "loss": 0.4181,
      "step": 5085
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.7462881976101965,
      "learning_rate": 8.302788306163373e-07,
      "loss": 0.4647,
      "step": 5086
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.749243857141998,
      "learning_rate": 8.288392226301917e-07,
      "loss": 0.431,
      "step": 5087
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.646232890597468,
      "learning_rate": 8.27400750983221e-07,
      "loss": 0.3512,
      "step": 5088
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.454838500654198,
      "learning_rate": 8.259634160673052e-07,
      "loss": 0.3768,
      "step": 5089
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.378151531920107,
      "learning_rate": 8.245272182740133e-07,
      "loss": 0.5171,
      "step": 5090
    },
    {
      "epoch": 0.82,
      "grad_norm": 20.47400090970667,
      "learning_rate": 8.230921579946027e-07,
      "loss": 0.3883,
      "step": 5091
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.033912900830468,
      "learning_rate": 8.21658235620027e-07,
      "loss": 0.3485,
      "step": 5092
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.148421388748061,
      "learning_rate": 8.202254515409247e-07,
      "loss": 0.4077,
      "step": 5093
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.079231207233739,
      "learning_rate": 8.187938061476242e-07,
      "loss": 0.4479,
      "step": 5094
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.634770125244585,
      "learning_rate": 8.173632998301468e-07,
      "loss": 0.4354,
      "step": 5095
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.236355670378655,
      "learning_rate": 8.159339329782023e-07,
      "loss": 0.4242,
      "step": 5096
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.270762262276684,
      "learning_rate": 8.145057059811895e-07,
      "loss": 0.4398,
      "step": 5097
    },
    {
      "epoch": 0.82,
      "grad_norm": 9.077991442575138,
      "learning_rate": 8.130786192281947e-07,
      "loss": 0.4107,
      "step": 5098
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.37305112777276,
      "learning_rate": 8.116526731079982e-07,
      "loss": 0.4508,
      "step": 5099
    },
    {
      "epoch": 0.82,
      "grad_norm": 19.233811337769104,
      "learning_rate": 8.102278680090664e-07,
      "loss": 0.3905,
      "step": 5100
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3087221307641297,
      "learning_rate": 8.088042043195538e-07,
      "loss": 0.4759,
      "step": 5101
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.541691314593341,
      "learning_rate": 8.073816824273078e-07,
      "loss": 0.3774,
      "step": 5102
    },
    {
      "epoch": 0.82,
      "grad_norm": 14.917923897365943,
      "learning_rate": 8.059603027198609e-07,
      "loss": 0.3637,
      "step": 5103
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.587423764554688,
      "learning_rate": 8.045400655844382e-07,
      "loss": 0.3576,
      "step": 5104
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.7038863566358975,
      "learning_rate": 8.031209714079496e-07,
      "loss": 0.3861,
      "step": 5105
    },
    {
      "epoch": 0.82,
      "grad_norm": 11.545903182540822,
      "learning_rate": 8.017030205769955e-07,
      "loss": 0.4432,
      "step": 5106
    },
    {
      "epoch": 0.82,
      "grad_norm": 13.324631043222343,
      "learning_rate": 8.002862134778661e-07,
      "loss": 0.4032,
      "step": 5107
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.142056457711467,
      "learning_rate": 7.988705504965372e-07,
      "loss": 0.4519,
      "step": 5108
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.953079242233721,
      "learning_rate": 7.974560320186759e-07,
      "loss": 0.4332,
      "step": 5109
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.05747456108488,
      "learning_rate": 7.960426584296338e-07,
      "loss": 0.3811,
      "step": 5110
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.718273671096098,
      "learning_rate": 7.946304301144564e-07,
      "loss": 0.3672,
      "step": 5111
    },
    {
      "epoch": 0.82,
      "grad_norm": 9.70772345914396,
      "learning_rate": 7.932193474578709e-07,
      "loss": 0.4407,
      "step": 5112
    },
    {
      "epoch": 0.82,
      "grad_norm": 9.558099997689615,
      "learning_rate": 7.918094108442947e-07,
      "loss": 0.498,
      "step": 5113
    },
    {
      "epoch": 0.82,
      "grad_norm": 12.489949043992755,
      "learning_rate": 7.904006206578358e-07,
      "loss": 0.4309,
      "step": 5114
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.260661081557242,
      "learning_rate": 7.889929772822857e-07,
      "loss": 0.3713,
      "step": 5115
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.52438405730543,
      "learning_rate": 7.875864811011247e-07,
      "loss": 0.4329,
      "step": 5116
    },
    {
      "epoch": 0.82,
      "grad_norm": 29.19908978996887,
      "learning_rate": 7.861811324975221e-07,
      "loss": 0.4499,
      "step": 5117
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.007335192045254,
      "learning_rate": 7.847769318543346e-07,
      "loss": 0.4334,
      "step": 5118
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.066842574742502,
      "learning_rate": 7.833738795541046e-07,
      "loss": 0.3913,
      "step": 5119
    },
    {
      "epoch": 0.82,
      "grad_norm": 11.800773451843927,
      "learning_rate": 7.819719759790606e-07,
      "loss": 0.4498,
      "step": 5120
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.36539528834278,
      "learning_rate": 7.805712215111216e-07,
      "loss": 0.425,
      "step": 5121
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.861032416313649,
      "learning_rate": 7.791716165318913e-07,
      "loss": 0.4139,
      "step": 5122
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.30968744525048,
      "learning_rate": 7.777731614226596e-07,
      "loss": 0.4057,
      "step": 5123
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.139607003145905,
      "learning_rate": 7.763758565644064e-07,
      "loss": 0.3865,
      "step": 5124
    },
    {
      "epoch": 0.83,
      "grad_norm": 12.307439492851042,
      "learning_rate": 7.749797023377931e-07,
      "loss": 0.4457,
      "step": 5125
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.708985016392934,
      "learning_rate": 7.735846991231738e-07,
      "loss": 0.4103,
      "step": 5126
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.370661339049936,
      "learning_rate": 7.721908473005829e-07,
      "loss": 0.3914,
      "step": 5127
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.286683666830271,
      "learning_rate": 7.707981472497467e-07,
      "loss": 0.3795,
      "step": 5128
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.59325971539908,
      "learning_rate": 7.694065993500732e-07,
      "loss": 0.4373,
      "step": 5129
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.043648530045134,
      "learning_rate": 7.680162039806588e-07,
      "loss": 0.3604,
      "step": 5130
    },
    {
      "epoch": 0.83,
      "grad_norm": 10.627896574575761,
      "learning_rate": 7.666269615202865e-07,
      "loss": 0.4544,
      "step": 5131
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.8887630781292755,
      "learning_rate": 7.652388723474224e-07,
      "loss": 0.4397,
      "step": 5132
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.048414509645428,
      "learning_rate": 7.638519368402225e-07,
      "loss": 0.3538,
      "step": 5133
    },
    {
      "epoch": 0.83,
      "grad_norm": 10.304033147877178,
      "learning_rate": 7.62466155376525e-07,
      "loss": 0.4329,
      "step": 5134
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.30889483830578,
      "learning_rate": 7.61081528333854e-07,
      "loss": 0.4773,
      "step": 5135
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.900192158664378,
      "learning_rate": 7.596980560894224e-07,
      "loss": 0.4574,
      "step": 5136
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.117761613069653,
      "learning_rate": 7.583157390201246e-07,
      "loss": 0.3957,
      "step": 5137
    },
    {
      "epoch": 0.83,
      "grad_norm": 17.90609966401349,
      "learning_rate": 7.56934577502541e-07,
      "loss": 0.3195,
      "step": 5138
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.19709239219677,
      "learning_rate": 7.555545719129398e-07,
      "loss": 0.3886,
      "step": 5139
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.339998278799756,
      "learning_rate": 7.541757226272744e-07,
      "loss": 0.3891,
      "step": 5140
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.427456854020218,
      "learning_rate": 7.527980300211762e-07,
      "loss": 0.364,
      "step": 5141
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.213020379867123,
      "learning_rate": 7.514214944699694e-07,
      "loss": 0.366,
      "step": 5142
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.037970929440187,
      "learning_rate": 7.500461163486616e-07,
      "loss": 0.4126,
      "step": 5143
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.096322028069207,
      "learning_rate": 7.486718960319428e-07,
      "loss": 0.4405,
      "step": 5144
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.453391240070106,
      "learning_rate": 7.472988338941861e-07,
      "loss": 0.4334,
      "step": 5145
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1601826509736752,
      "learning_rate": 7.459269303094552e-07,
      "loss": 0.4366,
      "step": 5146
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.390840621782635,
      "learning_rate": 7.445561856514916e-07,
      "loss": 0.4249,
      "step": 5147
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.691907359287466,
      "learning_rate": 7.431866002937254e-07,
      "loss": 0.3628,
      "step": 5148
    },
    {
      "epoch": 0.83,
      "grad_norm": 21.710205529708354,
      "learning_rate": 7.41818174609269e-07,
      "loss": 0.4319,
      "step": 5149
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.898975346621165,
      "learning_rate": 7.404509089709194e-07,
      "loss": 0.401,
      "step": 5150
    },
    {
      "epoch": 0.83,
      "grad_norm": 20.070586310102055,
      "learning_rate": 7.390848037511578e-07,
      "loss": 0.3985,
      "step": 5151
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.655898911910306,
      "learning_rate": 7.377198593221474e-07,
      "loss": 0.4698,
      "step": 5152
    },
    {
      "epoch": 0.83,
      "grad_norm": 9.951372286771235,
      "learning_rate": 7.363560760557392e-07,
      "loss": 0.4483,
      "step": 5153
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.298966659960592,
      "learning_rate": 7.349934543234621e-07,
      "loss": 0.4103,
      "step": 5154
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.386646055671122,
      "learning_rate": 7.336319944965353e-07,
      "loss": 0.4714,
      "step": 5155
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.873003923169134,
      "learning_rate": 7.32271696945856e-07,
      "loss": 0.4885,
      "step": 5156
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.411095983910647,
      "learning_rate": 7.30912562042006e-07,
      "loss": 0.4166,
      "step": 5157
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.944645019856956,
      "learning_rate": 7.295545901552536e-07,
      "loss": 0.4621,
      "step": 5158
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.405397586304533,
      "learning_rate": 7.281977816555463e-07,
      "loss": 0.4244,
      "step": 5159
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.284280832255569,
      "learning_rate": 7.268421369125145e-07,
      "loss": 0.4396,
      "step": 5160
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.43388570706495,
      "learning_rate": 7.254876562954755e-07,
      "loss": 0.4096,
      "step": 5161
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.492021713608688,
      "learning_rate": 7.241343401734285e-07,
      "loss": 0.426,
      "step": 5162
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.010980641146892,
      "learning_rate": 7.2278218891505e-07,
      "loss": 0.4339,
      "step": 5163
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.135272865110586,
      "learning_rate": 7.214312028887055e-07,
      "loss": 0.4223,
      "step": 5164
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3780039300140874,
      "learning_rate": 7.20081382462442e-07,
      "loss": 0.4925,
      "step": 5165
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.115186084121902,
      "learning_rate": 7.187327280039863e-07,
      "loss": 0.3585,
      "step": 5166
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.462073860132539,
      "learning_rate": 7.173852398807485e-07,
      "loss": 0.3935,
      "step": 5167
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.170810687069655,
      "learning_rate": 7.160389184598221e-07,
      "loss": 0.446,
      "step": 5168
    },
    {
      "epoch": 0.83,
      "grad_norm": 9.40846636279818,
      "learning_rate": 7.146937641079849e-07,
      "loss": 0.436,
      "step": 5169
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1604065529927903,
      "learning_rate": 7.133497771916886e-07,
      "loss": 0.4607,
      "step": 5170
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.439016556959326,
      "learning_rate": 7.120069580770755e-07,
      "loss": 0.4725,
      "step": 5171
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.566335676374294,
      "learning_rate": 7.10665307129968e-07,
      "loss": 0.4469,
      "step": 5172
    },
    {
      "epoch": 0.83,
      "grad_norm": 10.483125336935396,
      "learning_rate": 7.093248247158663e-07,
      "loss": 0.3794,
      "step": 5173
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.777562037756328,
      "learning_rate": 7.079855111999545e-07,
      "loss": 0.3974,
      "step": 5174
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.094239412490918,
      "learning_rate": 7.066473669471008e-07,
      "loss": 0.3842,
      "step": 5175
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.913868418828845,
      "learning_rate": 7.053103923218501e-07,
      "loss": 0.4481,
      "step": 5176
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.694489210691064,
      "learning_rate": 7.039745876884335e-07,
      "loss": 0.4061,
      "step": 5177
    },
    {
      "epoch": 0.83,
      "grad_norm": 9.869748455618549,
      "learning_rate": 7.026399534107603e-07,
      "loss": 0.3978,
      "step": 5178
    },
    {
      "epoch": 0.83,
      "grad_norm": 13.88981284597105,
      "learning_rate": 7.013064898524196e-07,
      "loss": 0.4508,
      "step": 5179
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.198039646114673,
      "learning_rate": 6.99974197376686e-07,
      "loss": 0.3819,
      "step": 5180
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.544678771266348,
      "learning_rate": 6.986430763465124e-07,
      "loss": 0.4694,
      "step": 5181
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.058448332166017,
      "learning_rate": 6.973131271245315e-07,
      "loss": 0.3997,
      "step": 5182
    },
    {
      "epoch": 0.84,
      "grad_norm": 8.839688502457097,
      "learning_rate": 6.95984350073059e-07,
      "loss": 0.4086,
      "step": 5183
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.586093286667702,
      "learning_rate": 6.94656745554092e-07,
      "loss": 0.3761,
      "step": 5184
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.781163485599814,
      "learning_rate": 6.933303139293035e-07,
      "loss": 0.3608,
      "step": 5185
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.431462725603388,
      "learning_rate": 6.92005055560051e-07,
      "loss": 0.4342,
      "step": 5186
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.075606158420179,
      "learning_rate": 6.906809708073736e-07,
      "loss": 0.4243,
      "step": 5187
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.0951292199482445,
      "learning_rate": 6.893580600319865e-07,
      "loss": 0.4082,
      "step": 5188
    },
    {
      "epoch": 0.84,
      "grad_norm": 10.932343625235042,
      "learning_rate": 6.880363235942861e-07,
      "loss": 0.4141,
      "step": 5189
    },
    {
      "epoch": 0.84,
      "grad_norm": 12.730457082527625,
      "learning_rate": 6.867157618543513e-07,
      "loss": 0.474,
      "step": 5190
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.543192585505034,
      "learning_rate": 6.853963751719417e-07,
      "loss": 0.4583,
      "step": 5191
    },
    {
      "epoch": 0.84,
      "grad_norm": 9.588366046957356,
      "learning_rate": 6.840781639064897e-07,
      "loss": 0.4319,
      "step": 5192
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.682578310155655,
      "learning_rate": 6.827611284171154e-07,
      "loss": 0.4709,
      "step": 5193
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.385630011329006,
      "learning_rate": 6.814452690626161e-07,
      "loss": 0.4122,
      "step": 5194
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.2765735964760685,
      "learning_rate": 6.801305862014667e-07,
      "loss": 0.4618,
      "step": 5195
    },
    {
      "epoch": 0.84,
      "grad_norm": 8.996336179152513,
      "learning_rate": 6.788170801918231e-07,
      "loss": 0.4121,
      "step": 5196
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0938053712995783,
      "learning_rate": 6.775047513915218e-07,
      "loss": 0.4822,
      "step": 5197
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.993674601956297,
      "learning_rate": 6.76193600158076e-07,
      "loss": 0.4328,
      "step": 5198
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.749437704899738,
      "learning_rate": 6.748836268486797e-07,
      "loss": 0.4052,
      "step": 5199
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.434455388949122,
      "learning_rate": 6.735748318202062e-07,
      "loss": 0.3911,
      "step": 5200
    },
    {
      "epoch": 0.84,
      "grad_norm": 8.000680899721354,
      "learning_rate": 6.722672154292065e-07,
      "loss": 0.3688,
      "step": 5201
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1828288561246265,
      "learning_rate": 6.709607780319133e-07,
      "loss": 0.4394,
      "step": 5202
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.647281907407361,
      "learning_rate": 6.696555199842347e-07,
      "loss": 0.4158,
      "step": 5203
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.537618367529139,
      "learning_rate": 6.683514416417574e-07,
      "loss": 0.4311,
      "step": 5204
    },
    {
      "epoch": 0.84,
      "grad_norm": 20.126615214102898,
      "learning_rate": 6.670485433597507e-07,
      "loss": 0.3932,
      "step": 5205
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.330010541672859,
      "learning_rate": 6.657468254931615e-07,
      "loss": 0.4206,
      "step": 5206
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.504147572911003,
      "learning_rate": 6.644462883966085e-07,
      "loss": 0.3979,
      "step": 5207
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.532142914990343,
      "learning_rate": 6.631469324243978e-07,
      "loss": 0.399,
      "step": 5208
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.566983490126192,
      "learning_rate": 6.618487579305089e-07,
      "loss": 0.4334,
      "step": 5209
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.97641712068898,
      "learning_rate": 6.605517652686005e-07,
      "loss": 0.3557,
      "step": 5210
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.940301658325301,
      "learning_rate": 6.59255954792008e-07,
      "loss": 0.4309,
      "step": 5211
    },
    {
      "epoch": 0.84,
      "grad_norm": 8.649855082632047,
      "learning_rate": 6.579613268537466e-07,
      "loss": 0.3905,
      "step": 5212
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.5507375327668145,
      "learning_rate": 6.566678818065108e-07,
      "loss": 0.4645,
      "step": 5213
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.967252404757335,
      "learning_rate": 6.553756200026668e-07,
      "loss": 0.4751,
      "step": 5214
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.898138413738658,
      "learning_rate": 6.540845417942637e-07,
      "loss": 0.3904,
      "step": 5215
    },
    {
      "epoch": 0.84,
      "grad_norm": 9.562978990480373,
      "learning_rate": 6.527946475330288e-07,
      "loss": 0.3843,
      "step": 5216
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.199447743700096,
      "learning_rate": 6.51505937570363e-07,
      "loss": 0.3851,
      "step": 5217
    },
    {
      "epoch": 0.84,
      "grad_norm": 12.900808448496539,
      "learning_rate": 6.502184122573457e-07,
      "loss": 0.4213,
      "step": 5218
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.055098217115098,
      "learning_rate": 6.489320719447367e-07,
      "loss": 0.4371,
      "step": 5219
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.072598732348102,
      "learning_rate": 6.476469169829691e-07,
      "loss": 0.4402,
      "step": 5220
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.925196170821637,
      "learning_rate": 6.463629477221533e-07,
      "loss": 0.4399,
      "step": 5221
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.314636921905483,
      "learning_rate": 6.450801645120808e-07,
      "loss": 0.3896,
      "step": 5222
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.786662543475865,
      "learning_rate": 6.43798567702214e-07,
      "loss": 0.4861,
      "step": 5223
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.715299785367704,
      "learning_rate": 6.425181576416978e-07,
      "loss": 0.3958,
      "step": 5224
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.311246426968363,
      "learning_rate": 6.412389346793507e-07,
      "loss": 0.4088,
      "step": 5225
    },
    {
      "epoch": 0.84,
      "grad_norm": 11.117707614321164,
      "learning_rate": 6.399608991636663e-07,
      "loss": 0.4347,
      "step": 5226
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.64619438614458,
      "learning_rate": 6.386840514428183e-07,
      "loss": 0.4163,
      "step": 5227
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.987053579448218,
      "learning_rate": 6.374083918646557e-07,
      "loss": 0.376,
      "step": 5228
    },
    {
      "epoch": 0.84,
      "grad_norm": 8.153410802065691,
      "learning_rate": 6.361339207767031e-07,
      "loss": 0.4637,
      "step": 5229
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.002782237204669,
      "learning_rate": 6.348606385261602e-07,
      "loss": 0.3521,
      "step": 5230
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.460430654341795,
      "learning_rate": 6.335885454599061e-07,
      "loss": 0.4933,
      "step": 5231
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.318259755349841,
      "learning_rate": 6.323176419244925e-07,
      "loss": 0.435,
      "step": 5232
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.396689682388432,
      "learning_rate": 6.310479282661485e-07,
      "loss": 0.4105,
      "step": 5233
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.633405190230257,
      "learning_rate": 6.297794048307798e-07,
      "loss": 0.4491,
      "step": 5234
    },
    {
      "epoch": 0.84,
      "grad_norm": 8.587733443088494,
      "learning_rate": 6.285120719639693e-07,
      "loss": 0.4547,
      "step": 5235
    },
    {
      "epoch": 0.84,
      "grad_norm": 26.494946319429015,
      "learning_rate": 6.27245930010969e-07,
      "loss": 0.453,
      "step": 5236
    },
    {
      "epoch": 0.84,
      "grad_norm": 8.969518436555928,
      "learning_rate": 6.259809793167127e-07,
      "loss": 0.4134,
      "step": 5237
    },
    {
      "epoch": 0.84,
      "grad_norm": 10.684680983005558,
      "learning_rate": 6.247172202258095e-07,
      "loss": 0.4037,
      "step": 5238
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.2207700234931105,
      "learning_rate": 6.234546530825408e-07,
      "loss": 0.4589,
      "step": 5239
    },
    {
      "epoch": 0.84,
      "grad_norm": 17.254429545347406,
      "learning_rate": 6.221932782308637e-07,
      "loss": 0.339,
      "step": 5240
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.640155389898209,
      "learning_rate": 6.209330960144139e-07,
      "loss": 0.4573,
      "step": 5241
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.563214768221484,
      "learning_rate": 6.196741067764977e-07,
      "loss": 0.4201,
      "step": 5242
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.586669227857651,
      "learning_rate": 6.184163108600988e-07,
      "loss": 0.3807,
      "step": 5243
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.161131667208156,
      "learning_rate": 6.171597086078767e-07,
      "loss": 0.3856,
      "step": 5244
    },
    {
      "epoch": 0.85,
      "grad_norm": 8.270947477998746,
      "learning_rate": 6.159043003621628e-07,
      "loss": 0.4205,
      "step": 5245
    },
    {
      "epoch": 0.85,
      "grad_norm": 9.022912117577617,
      "learning_rate": 6.146500864649668e-07,
      "loss": 0.4445,
      "step": 5246
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.534561715993588,
      "learning_rate": 6.133970672579693e-07,
      "loss": 0.4621,
      "step": 5247
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.370824609374356,
      "learning_rate": 6.121452430825287e-07,
      "loss": 0.4063,
      "step": 5248
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.865896705986213,
      "learning_rate": 6.108946142796757e-07,
      "loss": 0.418,
      "step": 5249
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.559630510588873,
      "learning_rate": 6.096451811901155e-07,
      "loss": 0.4283,
      "step": 5250
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.9704121019619345,
      "learning_rate": 6.083969441542298e-07,
      "loss": 0.4418,
      "step": 5251
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.097492463838492,
      "learning_rate": 6.071499035120703e-07,
      "loss": 0.3435,
      "step": 5252
    },
    {
      "epoch": 0.85,
      "grad_norm": 9.55162384945751,
      "learning_rate": 6.059040596033682e-07,
      "loss": 0.4491,
      "step": 5253
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.9052188533036984,
      "learning_rate": 6.046594127675242e-07,
      "loss": 0.4449,
      "step": 5254
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.286873179102048,
      "learning_rate": 6.034159633436132e-07,
      "loss": 0.367,
      "step": 5255
    },
    {
      "epoch": 0.85,
      "grad_norm": 10.066195808253005,
      "learning_rate": 6.021737116703868e-07,
      "loss": 0.4235,
      "step": 5256
    },
    {
      "epoch": 0.85,
      "grad_norm": 8.00092700847031,
      "learning_rate": 6.009326580862696e-07,
      "loss": 0.4327,
      "step": 5257
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.214885977049775,
      "learning_rate": 5.996928029293559e-07,
      "loss": 0.379,
      "step": 5258
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.6701913084493345,
      "learning_rate": 5.984541465374172e-07,
      "loss": 0.4222,
      "step": 5259
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.730498862416246,
      "learning_rate": 5.972166892478998e-07,
      "loss": 0.4175,
      "step": 5260
    },
    {
      "epoch": 0.85,
      "grad_norm": 10.033996435688959,
      "learning_rate": 5.959804313979195e-07,
      "loss": 0.3398,
      "step": 5261
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.894439461189259,
      "learning_rate": 5.947453733242659e-07,
      "loss": 0.397,
      "step": 5262
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.543333018152931,
      "learning_rate": 5.935115153634058e-07,
      "loss": 0.3796,
      "step": 5263
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.521751642396397,
      "learning_rate": 5.922788578514737e-07,
      "loss": 0.3892,
      "step": 5264
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.201298706929204,
      "learning_rate": 5.910474011242801e-07,
      "loss": 0.4028,
      "step": 5265
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.769529830326897,
      "learning_rate": 5.898171455173074e-07,
      "loss": 0.3664,
      "step": 5266
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.78138727972116,
      "learning_rate": 5.885880913657127e-07,
      "loss": 0.3678,
      "step": 5267
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.006605747945957,
      "learning_rate": 5.873602390043231e-07,
      "loss": 0.4777,
      "step": 5268
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.979367209414301,
      "learning_rate": 5.861335887676389e-07,
      "loss": 0.3887,
      "step": 5269
    },
    {
      "epoch": 0.85,
      "grad_norm": 19.90359480776006,
      "learning_rate": 5.84908140989835e-07,
      "loss": 0.404,
      "step": 5270
    },
    {
      "epoch": 0.85,
      "grad_norm": 10.954653367935228,
      "learning_rate": 5.836838960047558e-07,
      "loss": 0.4152,
      "step": 5271
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.183101663882465,
      "learning_rate": 5.824608541459192e-07,
      "loss": 0.4385,
      "step": 5272
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.861058308923739,
      "learning_rate": 5.812390157465169e-07,
      "loss": 0.3988,
      "step": 5273
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.214290779519238,
      "learning_rate": 5.8001838113941e-07,
      "loss": 0.4121,
      "step": 5274
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.587558386231922,
      "learning_rate": 5.78798950657134e-07,
      "loss": 0.3866,
      "step": 5275
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.64735138255537,
      "learning_rate": 5.775807246318954e-07,
      "loss": 0.3829,
      "step": 5276
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.901030518619207,
      "learning_rate": 5.76363703395571e-07,
      "loss": 0.4128,
      "step": 5277
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.377395250542658,
      "learning_rate": 5.751478872797128e-07,
      "loss": 0.4074,
      "step": 5278
    },
    {
      "epoch": 0.85,
      "grad_norm": 15.839005762580202,
      "learning_rate": 5.739332766155419e-07,
      "loss": 0.421,
      "step": 5279
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.784420757730078,
      "learning_rate": 5.727198717339511e-07,
      "loss": 0.4436,
      "step": 5280
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.901020614611662,
      "learning_rate": 5.715076729655056e-07,
      "loss": 0.4619,
      "step": 5281
    },
    {
      "epoch": 0.85,
      "grad_norm": 15.186678943229952,
      "learning_rate": 5.702966806404431e-07,
      "loss": 0.4333,
      "step": 5282
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.6226845446600535,
      "learning_rate": 5.690868950886702e-07,
      "loss": 0.3916,
      "step": 5283
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.127540306725197,
      "learning_rate": 5.67878316639765e-07,
      "loss": 0.4483,
      "step": 5284
    },
    {
      "epoch": 0.85,
      "grad_norm": 29.1759069348595,
      "learning_rate": 5.666709456229797e-07,
      "loss": 0.4417,
      "step": 5285
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.6249210971283485,
      "learning_rate": 5.654647823672337e-07,
      "loss": 0.4262,
      "step": 5286
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.030530317640041,
      "learning_rate": 5.642598272011196e-07,
      "loss": 0.4895,
      "step": 5287
    },
    {
      "epoch": 0.85,
      "grad_norm": 8.775071842926382,
      "learning_rate": 5.630560804528995e-07,
      "loss": 0.4445,
      "step": 5288
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.810718507449153,
      "learning_rate": 5.6185354245051e-07,
      "loss": 0.4329,
      "step": 5289
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.815670606153476,
      "learning_rate": 5.606522135215531e-07,
      "loss": 0.4077,
      "step": 5290
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.92539098905313,
      "learning_rate": 5.594520939933041e-07,
      "loss": 0.4639,
      "step": 5291
    },
    {
      "epoch": 0.85,
      "grad_norm": 8.408856401600202,
      "learning_rate": 5.582531841927097e-07,
      "loss": 0.4187,
      "step": 5292
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.4576385446173195,
      "learning_rate": 5.570554844463854e-07,
      "loss": 0.3915,
      "step": 5293
    },
    {
      "epoch": 0.85,
      "grad_norm": 27.394392984767407,
      "learning_rate": 5.558589950806164e-07,
      "loss": 0.3822,
      "step": 5294
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.24877921988408,
      "learning_rate": 5.546637164213625e-07,
      "loss": 0.452,
      "step": 5295
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.165858864977825,
      "learning_rate": 5.53469648794247e-07,
      "loss": 0.4082,
      "step": 5296
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.516831691363996,
      "learning_rate": 5.522767925245698e-07,
      "loss": 0.4061,
      "step": 5297
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.800234108625724,
      "learning_rate": 5.510851479372959e-07,
      "loss": 0.3538,
      "step": 5298
    },
    {
      "epoch": 0.85,
      "grad_norm": 9.581339603356755,
      "learning_rate": 5.498947153570622e-07,
      "loss": 0.3284,
      "step": 5299
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.58131831360925,
      "learning_rate": 5.487054951081772e-07,
      "loss": 0.4561,
      "step": 5300
    },
    {
      "epoch": 0.85,
      "grad_norm": 9.163680944832233,
      "learning_rate": 5.475174875146156e-07,
      "loss": 0.3986,
      "step": 5301
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.859099376781523,
      "learning_rate": 5.463306929000228e-07,
      "loss": 0.4695,
      "step": 5302
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.401345663289105,
      "learning_rate": 5.451451115877154e-07,
      "loss": 0.4027,
      "step": 5303
    },
    {
      "epoch": 0.85,
      "grad_norm": 13.239172666430148,
      "learning_rate": 5.439607439006795e-07,
      "loss": 0.3828,
      "step": 5304
    },
    {
      "epoch": 0.85,
      "grad_norm": 8.956289879751637,
      "learning_rate": 5.427775901615684e-07,
      "loss": 0.4772,
      "step": 5305
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.595138014754916,
      "learning_rate": 5.415956506927051e-07,
      "loss": 0.3764,
      "step": 5306
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.3624894142642585,
      "learning_rate": 5.404149258160835e-07,
      "loss": 0.4764,
      "step": 5307
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.66728557266445,
      "learning_rate": 5.392354158533658e-07,
      "loss": 0.4314,
      "step": 5308
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.053256277874953,
      "learning_rate": 5.380571211258811e-07,
      "loss": 0.4122,
      "step": 5309
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.034848403125363,
      "learning_rate": 5.36880041954631e-07,
      "loss": 0.4867,
      "step": 5310
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.917186721206175,
      "learning_rate": 5.357041786602851e-07,
      "loss": 0.4684,
      "step": 5311
    },
    {
      "epoch": 0.86,
      "grad_norm": 15.800665618559174,
      "learning_rate": 5.345295315631805e-07,
      "loss": 0.3881,
      "step": 5312
    },
    {
      "epoch": 0.86,
      "grad_norm": 15.165099391855884,
      "learning_rate": 5.33356100983321e-07,
      "loss": 0.4082,
      "step": 5313
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.823819267359406,
      "learning_rate": 5.32183887240385e-07,
      "loss": 0.4442,
      "step": 5314
    },
    {
      "epoch": 0.86,
      "grad_norm": 10.197485973182356,
      "learning_rate": 5.310128906537137e-07,
      "loss": 0.4084,
      "step": 5315
    },
    {
      "epoch": 0.86,
      "grad_norm": 24.47032414679811,
      "learning_rate": 5.298431115423186e-07,
      "loss": 0.4071,
      "step": 5316
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.920259189690462,
      "learning_rate": 5.286745502248819e-07,
      "loss": 0.4655,
      "step": 5317
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.544392882982514,
      "learning_rate": 5.27507207019749e-07,
      "loss": 0.4046,
      "step": 5318
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.578917239637161,
      "learning_rate": 5.263410822449388e-07,
      "loss": 0.4035,
      "step": 5319
    },
    {
      "epoch": 0.86,
      "grad_norm": 11.022203071199328,
      "learning_rate": 5.251761762181351e-07,
      "loss": 0.4126,
      "step": 5320
    },
    {
      "epoch": 0.86,
      "grad_norm": 19.64847858635474,
      "learning_rate": 5.240124892566895e-07,
      "loss": 0.3749,
      "step": 5321
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.244255691782633,
      "learning_rate": 5.228500216776239e-07,
      "loss": 0.3499,
      "step": 5322
    },
    {
      "epoch": 0.86,
      "grad_norm": 8.972797812331269,
      "learning_rate": 5.216887737976256e-07,
      "loss": 0.428,
      "step": 5323
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.428749845320174,
      "learning_rate": 5.205287459330499e-07,
      "loss": 0.4505,
      "step": 5324
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.399787409591956,
      "learning_rate": 5.193699383999213e-07,
      "loss": 0.4377,
      "step": 5325
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.789812068845749,
      "learning_rate": 5.182123515139315e-07,
      "loss": 0.3444,
      "step": 5326
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.2983560493318205,
      "learning_rate": 5.170559855904389e-07,
      "loss": 0.4078,
      "step": 5327
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.852153629855909,
      "learning_rate": 5.159008409444671e-07,
      "loss": 0.4321,
      "step": 5328
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.503261888489776,
      "learning_rate": 5.147469178907127e-07,
      "loss": 0.4311,
      "step": 5329
    },
    {
      "epoch": 0.86,
      "grad_norm": 74.52382112484217,
      "learning_rate": 5.135942167435342e-07,
      "loss": 0.4343,
      "step": 5330
    },
    {
      "epoch": 0.86,
      "grad_norm": 18.96210599058679,
      "learning_rate": 5.124427378169588e-07,
      "loss": 0.4209,
      "step": 5331
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.942747298502926,
      "learning_rate": 5.112924814246817e-07,
      "loss": 0.385,
      "step": 5332
    },
    {
      "epoch": 0.86,
      "grad_norm": 13.667452891848173,
      "learning_rate": 5.10143447880066e-07,
      "loss": 0.3988,
      "step": 5333
    },
    {
      "epoch": 0.86,
      "grad_norm": 8.825978089537543,
      "learning_rate": 5.089956374961386e-07,
      "loss": 0.4455,
      "step": 5334
    },
    {
      "epoch": 0.86,
      "grad_norm": 44.98588646909186,
      "learning_rate": 5.078490505855938e-07,
      "loss": 0.3822,
      "step": 5335
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.437944680339552,
      "learning_rate": 5.06703687460795e-07,
      "loss": 0.4725,
      "step": 5336
    },
    {
      "epoch": 0.86,
      "grad_norm": 16.626036690571258,
      "learning_rate": 5.055595484337705e-07,
      "loss": 0.4235,
      "step": 5337
    },
    {
      "epoch": 0.86,
      "grad_norm": 12.301427380854832,
      "learning_rate": 5.044166338162143e-07,
      "loss": 0.3771,
      "step": 5338
    },
    {
      "epoch": 0.86,
      "grad_norm": 9.17133501780956,
      "learning_rate": 5.03274943919489e-07,
      "loss": 0.4756,
      "step": 5339
    },
    {
      "epoch": 0.86,
      "grad_norm": 8.558840809623556,
      "learning_rate": 5.021344790546212e-07,
      "loss": 0.4177,
      "step": 5340
    },
    {
      "epoch": 0.86,
      "grad_norm": 12.89312055490241,
      "learning_rate": 5.009952395323065e-07,
      "loss": 0.4309,
      "step": 5341
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.360323063915149,
      "learning_rate": 4.998572256629047e-07,
      "loss": 0.484,
      "step": 5342
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.886018953131113,
      "learning_rate": 4.987204377564409e-07,
      "loss": 0.4222,
      "step": 5343
    },
    {
      "epoch": 0.86,
      "grad_norm": 9.026664722867963,
      "learning_rate": 4.975848761226088e-07,
      "loss": 0.4412,
      "step": 5344
    },
    {
      "epoch": 0.86,
      "grad_norm": 13.9126777291004,
      "learning_rate": 4.964505410707655e-07,
      "loss": 0.3429,
      "step": 5345
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.437184313701611,
      "learning_rate": 4.95317432909937e-07,
      "loss": 0.3848,
      "step": 5346
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.392627823546229,
      "learning_rate": 4.941855519488109e-07,
      "loss": 0.3559,
      "step": 5347
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.757502503976423,
      "learning_rate": 4.930548984957451e-07,
      "loss": 0.4044,
      "step": 5348
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.4593410509549125,
      "learning_rate": 4.919254728587591e-07,
      "loss": 0.4543,
      "step": 5349
    },
    {
      "epoch": 0.86,
      "grad_norm": 17.786174257561143,
      "learning_rate": 4.907972753455398e-07,
      "loss": 0.3413,
      "step": 5350
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.146808361338149,
      "learning_rate": 4.896703062634401e-07,
      "loss": 0.4458,
      "step": 5351
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.582402037250631,
      "learning_rate": 4.885445659194771e-07,
      "loss": 0.525,
      "step": 5352
    },
    {
      "epoch": 0.86,
      "grad_norm": 16.279801321200193,
      "learning_rate": 4.874200546203328e-07,
      "loss": 0.4178,
      "step": 5353
    },
    {
      "epoch": 0.86,
      "grad_norm": 8.872327605324376,
      "learning_rate": 4.86296772672355e-07,
      "loss": 0.5229,
      "step": 5354
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.4949892461922065,
      "learning_rate": 4.851747203815588e-07,
      "loss": 0.4124,
      "step": 5355
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.7456333789718,
      "learning_rate": 4.840538980536203e-07,
      "loss": 0.4055,
      "step": 5356
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.762640573618225,
      "learning_rate": 4.829343059938818e-07,
      "loss": 0.4165,
      "step": 5357
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.309000193762146,
      "learning_rate": 4.818159445073528e-07,
      "loss": 0.3481,
      "step": 5358
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.1651002833474635,
      "learning_rate": 4.806988138987051e-07,
      "loss": 0.3931,
      "step": 5359
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.766064247213585,
      "learning_rate": 4.795829144722752e-07,
      "loss": 0.3656,
      "step": 5360
    },
    {
      "epoch": 0.86,
      "grad_norm": 8.741023677040962,
      "learning_rate": 4.784682465320661e-07,
      "loss": 0.4296,
      "step": 5361
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.112649773777639,
      "learning_rate": 4.773548103817421e-07,
      "loss": 0.487,
      "step": 5362
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.938801122561541,
      "learning_rate": 4.762426063246367e-07,
      "loss": 0.4503,
      "step": 5363
    },
    {
      "epoch": 0.86,
      "grad_norm": 8.179647425183878,
      "learning_rate": 4.75131634663743e-07,
      "loss": 0.4095,
      "step": 5364
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.639280224955156,
      "learning_rate": 4.740218957017201e-07,
      "loss": 0.4691,
      "step": 5365
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.630583585791296,
      "learning_rate": 4.729133897408933e-07,
      "loss": 0.4169,
      "step": 5366
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.445101292974257,
      "learning_rate": 4.7180611708324776e-07,
      "loss": 0.4012,
      "step": 5367
    },
    {
      "epoch": 0.86,
      "grad_norm": 9.196392236634734,
      "learning_rate": 4.7070007803043714e-07,
      "loss": 0.4485,
      "step": 5368
    },
    {
      "epoch": 0.87,
      "grad_norm": 9.372825385616434,
      "learning_rate": 4.6959527288377493e-07,
      "loss": 0.4514,
      "step": 5369
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.199406521664843,
      "learning_rate": 4.684917019442431e-07,
      "loss": 0.4341,
      "step": 5370
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.9222169528871405,
      "learning_rate": 4.673893655124834e-07,
      "loss": 0.386,
      "step": 5371
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.726690736618884,
      "learning_rate": 4.6628826388880165e-07,
      "loss": 0.3834,
      "step": 5372
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.678756829787878,
      "learning_rate": 4.651883973731708e-07,
      "loss": 0.4837,
      "step": 5373
    },
    {
      "epoch": 0.87,
      "grad_norm": 9.666783151337423,
      "learning_rate": 4.640897662652227e-07,
      "loss": 0.3567,
      "step": 5374
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.35807639431972,
      "learning_rate": 4.629923708642542e-07,
      "loss": 0.4215,
      "step": 5375
    },
    {
      "epoch": 0.87,
      "grad_norm": 8.441488333315057,
      "learning_rate": 4.618962114692277e-07,
      "loss": 0.3784,
      "step": 5376
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.4704178387913025,
      "learning_rate": 4.608012883787688e-07,
      "loss": 0.3651,
      "step": 5377
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.218297444413399,
      "learning_rate": 4.5970760189116057e-07,
      "loss": 0.4418,
      "step": 5378
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.794336943082552,
      "learning_rate": 4.5861515230435525e-07,
      "loss": 0.395,
      "step": 5379
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.161564818104756,
      "learning_rate": 4.5752393991596754e-07,
      "loss": 0.4512,
      "step": 5380
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.332589886035079,
      "learning_rate": 4.5643396502327297e-07,
      "loss": 0.3552,
      "step": 5381
    },
    {
      "epoch": 0.87,
      "grad_norm": 8.256058574849284,
      "learning_rate": 4.55345227923209e-07,
      "loss": 0.4382,
      "step": 5382
    },
    {
      "epoch": 0.87,
      "grad_norm": 8.772645002317685,
      "learning_rate": 4.5425772891238065e-07,
      "loss": 0.442,
      "step": 5383
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.706827015891457,
      "learning_rate": 4.5317146828704973e-07,
      "loss": 0.3903,
      "step": 5384
    },
    {
      "epoch": 0.87,
      "grad_norm": 8.937386203510606,
      "learning_rate": 4.520864463431457e-07,
      "loss": 0.4243,
      "step": 5385
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9986632707986416,
      "learning_rate": 4.510026633762571e-07,
      "loss": 0.4653,
      "step": 5386
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.482633208710864,
      "learning_rate": 4.4992011968163775e-07,
      "loss": 0.4533,
      "step": 5387
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.593283144380061,
      "learning_rate": 4.488388155542012e-07,
      "loss": 0.4704,
      "step": 5388
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.352235569042049,
      "learning_rate": 4.47758751288524e-07,
      "loss": 0.3601,
      "step": 5389
    },
    {
      "epoch": 0.87,
      "grad_norm": 13.868612191346697,
      "learning_rate": 4.466799271788469e-07,
      "loss": 0.4203,
      "step": 5390
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.4114001384022865,
      "learning_rate": 4.4560234351906983e-07,
      "loss": 0.4019,
      "step": 5391
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.042674560771361,
      "learning_rate": 4.445260006027585e-07,
      "loss": 0.42,
      "step": 5392
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.729938661253006,
      "learning_rate": 4.4345089872313674e-07,
      "loss": 0.4238,
      "step": 5393
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.301088847798538,
      "learning_rate": 4.4237703817309073e-07,
      "loss": 0.4345,
      "step": 5394
    },
    {
      "epoch": 0.87,
      "grad_norm": 9.920558186486286,
      "learning_rate": 4.4130441924517263e-07,
      "loss": 0.3359,
      "step": 5395
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.141335568071695,
      "learning_rate": 4.4023304223159203e-07,
      "loss": 0.4834,
      "step": 5396
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.637868410280652,
      "learning_rate": 4.3916290742421986e-07,
      "loss": 0.4171,
      "step": 5397
    },
    {
      "epoch": 0.87,
      "grad_norm": 9.576498835788243,
      "learning_rate": 4.3809401511459237e-07,
      "loss": 0.4454,
      "step": 5398
    },
    {
      "epoch": 0.87,
      "grad_norm": 17.64865830466037,
      "learning_rate": 4.3702636559390667e-07,
      "loss": 0.4285,
      "step": 5399
    },
    {
      "epoch": 0.87,
      "grad_norm": 10.487336657771833,
      "learning_rate": 4.3595995915301614e-07,
      "loss": 0.434,
      "step": 5400
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.155011582092008,
      "learning_rate": 4.348947960824412e-07,
      "loss": 0.3712,
      "step": 5401
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.88295286378949,
      "learning_rate": 4.3383087667236254e-07,
      "loss": 0.4152,
      "step": 5402
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.0087228199859055,
      "learning_rate": 4.3276820121262053e-07,
      "loss": 0.3748,
      "step": 5403
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.933289816181155,
      "learning_rate": 4.3170676999271576e-07,
      "loss": 0.4071,
      "step": 5404
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.984943717751519,
      "learning_rate": 4.306465833018131e-07,
      "loss": 0.3457,
      "step": 5405
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.510480112242898,
      "learning_rate": 4.295876414287375e-07,
      "loss": 0.4196,
      "step": 5406
    },
    {
      "epoch": 0.87,
      "grad_norm": 10.677208689865212,
      "learning_rate": 4.2852994466197064e-07,
      "loss": 0.4234,
      "step": 5407
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.531107443400987,
      "learning_rate": 4.274734932896607e-07,
      "loss": 0.3452,
      "step": 5408
    },
    {
      "epoch": 0.87,
      "grad_norm": 8.937036726114478,
      "learning_rate": 4.264182875996142e-07,
      "loss": 0.4162,
      "step": 5409
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.2579553057063,
      "learning_rate": 4.253643278792974e-07,
      "loss": 0.3792,
      "step": 5410
    },
    {
      "epoch": 0.87,
      "grad_norm": 8.929406159334912,
      "learning_rate": 4.243116144158377e-07,
      "loss": 0.3967,
      "step": 5411
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.023669195069621,
      "learning_rate": 4.232601474960246e-07,
      "loss": 0.4112,
      "step": 5412
    },
    {
      "epoch": 0.87,
      "grad_norm": 39.19346853717111,
      "learning_rate": 4.2220992740630515e-07,
      "loss": 0.4079,
      "step": 5413
    },
    {
      "epoch": 0.87,
      "grad_norm": 11.521817440314724,
      "learning_rate": 4.211609544327894e-07,
      "loss": 0.4534,
      "step": 5414
    },
    {
      "epoch": 0.87,
      "grad_norm": 9.732753568406329,
      "learning_rate": 4.201132288612464e-07,
      "loss": 0.3942,
      "step": 5415
    },
    {
      "epoch": 0.87,
      "grad_norm": 13.014022336909237,
      "learning_rate": 4.1906675097710457e-07,
      "loss": 0.4312,
      "step": 5416
    },
    {
      "epoch": 0.87,
      "grad_norm": 11.198380202319221,
      "learning_rate": 4.180215210654548e-07,
      "loss": 0.4242,
      "step": 5417
    },
    {
      "epoch": 0.87,
      "grad_norm": 9.458794552423361,
      "learning_rate": 4.1697753941104533e-07,
      "loss": 0.4069,
      "step": 5418
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.869111662004919,
      "learning_rate": 4.159348062982849e-07,
      "loss": 0.3506,
      "step": 5419
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.2723683968509105,
      "learning_rate": 4.1489332201124443e-07,
      "loss": 0.4066,
      "step": 5420
    },
    {
      "epoch": 0.87,
      "grad_norm": 9.681952815930353,
      "learning_rate": 4.138530868336532e-07,
      "loss": 0.3979,
      "step": 5421
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.6253102245126305,
      "learning_rate": 4.1281410104889784e-07,
      "loss": 0.354,
      "step": 5422
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.578186864546129,
      "learning_rate": 4.1177636494002747e-07,
      "loss": 0.4411,
      "step": 5423
    },
    {
      "epoch": 0.87,
      "grad_norm": 10.040839160189126,
      "learning_rate": 4.107398787897515e-07,
      "loss": 0.4173,
      "step": 5424
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.980739229553199,
      "learning_rate": 4.097046428804363e-07,
      "loss": 0.4132,
      "step": 5425
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.1360496394444,
      "learning_rate": 4.0867065749410737e-07,
      "loss": 0.4636,
      "step": 5426
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.872259311922395,
      "learning_rate": 4.0763792291245274e-07,
      "loss": 0.3806,
      "step": 5427
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.535233556104301,
      "learning_rate": 4.066064394168184e-07,
      "loss": 0.4057,
      "step": 5428
    },
    {
      "epoch": 0.87,
      "grad_norm": 8.671308518902356,
      "learning_rate": 4.055762072882058e-07,
      "loss": 0.3552,
      "step": 5429
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.48708421757117,
      "learning_rate": 4.045472268072803e-07,
      "loss": 0.4629,
      "step": 5430
    },
    {
      "epoch": 0.88,
      "grad_norm": 32.987893833921135,
      "learning_rate": 4.0351949825436556e-07,
      "loss": 0.3902,
      "step": 5431
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.677609296824972,
      "learning_rate": 4.024930219094425e-07,
      "loss": 0.4101,
      "step": 5432
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.275223320415885,
      "learning_rate": 4.014677980521503e-07,
      "loss": 0.3787,
      "step": 5433
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.371291488025167,
      "learning_rate": 4.0044382696178994e-07,
      "loss": 0.3905,
      "step": 5434
    },
    {
      "epoch": 0.88,
      "grad_norm": 9.111739418832654,
      "learning_rate": 3.994211089173189e-07,
      "loss": 0.3932,
      "step": 5435
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.9055530480413685,
      "learning_rate": 3.9839964419735254e-07,
      "loss": 0.3884,
      "step": 5436
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.719362407858911,
      "learning_rate": 3.9737943308016726e-07,
      "loss": 0.4897,
      "step": 5437
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.723881868188386,
      "learning_rate": 3.963604758436962e-07,
      "loss": 0.4203,
      "step": 5438
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.323508223978595,
      "learning_rate": 3.953427727655323e-07,
      "loss": 0.3791,
      "step": 5439
    },
    {
      "epoch": 0.88,
      "grad_norm": 19.172766690550894,
      "learning_rate": 3.9432632412292504e-07,
      "loss": 0.4125,
      "step": 5440
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.253885545073005,
      "learning_rate": 3.933111301927828e-07,
      "loss": 0.3783,
      "step": 5441
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.224874946813128,
      "learning_rate": 3.922971912516721e-07,
      "loss": 0.413,
      "step": 5442
    },
    {
      "epoch": 0.88,
      "grad_norm": 14.85101369030795,
      "learning_rate": 3.912845075758209e-07,
      "loss": 0.3945,
      "step": 5443
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.573305520812423,
      "learning_rate": 3.9027307944110804e-07,
      "loss": 0.4342,
      "step": 5444
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.203101724932555,
      "learning_rate": 3.892629071230758e-07,
      "loss": 0.4151,
      "step": 5445
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.9868183058627915,
      "learning_rate": 3.8825399089692464e-07,
      "loss": 0.4192,
      "step": 5446
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.988996521107887,
      "learning_rate": 3.8724633103750964e-07,
      "loss": 0.4417,
      "step": 5447
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.908448389885545,
      "learning_rate": 3.862399278193446e-07,
      "loss": 0.4577,
      "step": 5448
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.847796516310014,
      "learning_rate": 3.852347815166024e-07,
      "loss": 0.416,
      "step": 5449
    },
    {
      "epoch": 0.88,
      "grad_norm": 37.32424538631784,
      "learning_rate": 3.8423089240311404e-07,
      "loss": 0.3622,
      "step": 5450
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.608093205080137,
      "learning_rate": 3.8322826075236295e-07,
      "loss": 0.4007,
      "step": 5451
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.008094350394244,
      "learning_rate": 3.822268868374962e-07,
      "loss": 0.4737,
      "step": 5452
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.985591573542547,
      "learning_rate": 3.8122677093131613e-07,
      "loss": 0.4111,
      "step": 5453
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.448830073636295,
      "learning_rate": 3.802279133062803e-07,
      "loss": 0.4319,
      "step": 5454
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.9140768824130205,
      "learning_rate": 3.7923031423450495e-07,
      "loss": 0.3814,
      "step": 5455
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2512554812928376,
      "learning_rate": 3.782339739877655e-07,
      "loss": 0.4572,
      "step": 5456
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.695170774842016,
      "learning_rate": 3.772388928374909e-07,
      "loss": 0.3633,
      "step": 5457
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.523170057426924,
      "learning_rate": 3.7624507105476826e-07,
      "loss": 0.4101,
      "step": 5458
    },
    {
      "epoch": 0.88,
      "grad_norm": 9.28645655818329,
      "learning_rate": 3.752525089103437e-07,
      "loss": 0.3745,
      "step": 5459
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.326758245641892,
      "learning_rate": 3.7426120667461706e-07,
      "loss": 0.3768,
      "step": 5460
    },
    {
      "epoch": 0.88,
      "grad_norm": 21.9191188755056,
      "learning_rate": 3.7327116461764737e-07,
      "loss": 0.4689,
      "step": 5461
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1328787617246863,
      "learning_rate": 3.7228238300914934e-07,
      "loss": 0.4346,
      "step": 5462
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.987198451785732,
      "learning_rate": 3.7129486211849307e-07,
      "loss": 0.4094,
      "step": 5463
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.9746696206509275,
      "learning_rate": 3.7030860221470787e-07,
      "loss": 0.4778,
      "step": 5464
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.554537967931254,
      "learning_rate": 3.6932360356647713e-07,
      "loss": 0.3647,
      "step": 5465
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.729118126587965,
      "learning_rate": 3.683398664421428e-07,
      "loss": 0.4044,
      "step": 5466
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.311153718031421,
      "learning_rate": 3.673573911097006e-07,
      "loss": 0.3506,
      "step": 5467
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.110406989310331,
      "learning_rate": 3.663761778368052e-07,
      "loss": 0.397,
      "step": 5468
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1286695981980963,
      "learning_rate": 3.653962268907657e-07,
      "loss": 0.4829,
      "step": 5469
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.7757287791021,
      "learning_rate": 3.6441753853854677e-07,
      "loss": 0.3001,
      "step": 5470
    },
    {
      "epoch": 0.88,
      "grad_norm": 10.257507230874316,
      "learning_rate": 3.6344011304677074e-07,
      "loss": 0.4251,
      "step": 5471
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.614687462690748,
      "learning_rate": 3.624639506817174e-07,
      "loss": 0.4134,
      "step": 5472
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.574919545999684,
      "learning_rate": 3.614890517093161e-07,
      "loss": 0.4507,
      "step": 5473
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.517129145556188,
      "learning_rate": 3.605154163951591e-07,
      "loss": 0.3863,
      "step": 5474
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.727385147495447,
      "learning_rate": 3.595430450044912e-07,
      "loss": 0.41,
      "step": 5475
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.21948050752002,
      "learning_rate": 3.585719378022129e-07,
      "loss": 0.3917,
      "step": 5476
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.981980428739117,
      "learning_rate": 3.576020950528797e-07,
      "loss": 0.4166,
      "step": 5477
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.361554366645686,
      "learning_rate": 3.566335170207053e-07,
      "loss": 0.3966,
      "step": 5478
    },
    {
      "epoch": 0.88,
      "grad_norm": 9.545832683699158,
      "learning_rate": 3.5566620396955573e-07,
      "loss": 0.3764,
      "step": 5479
    },
    {
      "epoch": 0.88,
      "grad_norm": 16.67168108490487,
      "learning_rate": 3.5470015616295315e-07,
      "loss": 0.3558,
      "step": 5480
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1356589802631945,
      "learning_rate": 3.537353738640775e-07,
      "loss": 0.4741,
      "step": 5481
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0998738254986102,
      "learning_rate": 3.527718573357597e-07,
      "loss": 0.4516,
      "step": 5482
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.639718197924721,
      "learning_rate": 3.518096068404908e-07,
      "loss": 0.4138,
      "step": 5483
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.50562551885508,
      "learning_rate": 3.508486226404123e-07,
      "loss": 0.4558,
      "step": 5484
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.989789786635936,
      "learning_rate": 3.4988890499732477e-07,
      "loss": 0.4399,
      "step": 5485
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.618643865636363,
      "learning_rate": 3.489304541726801e-07,
      "loss": 0.4128,
      "step": 5486
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.300461121221082,
      "learning_rate": 3.479732704275862e-07,
      "loss": 0.4088,
      "step": 5487
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.715432052846534,
      "learning_rate": 3.470173540228089e-07,
      "loss": 0.3819,
      "step": 5488
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.09393786445219,
      "learning_rate": 3.460627052187632e-07,
      "loss": 0.4202,
      "step": 5489
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2962622066295508,
      "learning_rate": 3.451093242755238e-07,
      "loss": 0.479,
      "step": 5490
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.380188669171899,
      "learning_rate": 3.4415721145281743e-07,
      "loss": 0.3514,
      "step": 5491
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.139686229850509,
      "learning_rate": 3.4320636701002497e-07,
      "loss": 0.4091,
      "step": 5492
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.7016491228908075,
      "learning_rate": 3.422567912061836e-07,
      "loss": 0.4037,
      "step": 5493
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.3058106533760725,
      "learning_rate": 3.4130848429998533e-07,
      "loss": 0.4441,
      "step": 5494
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.479731379543604,
      "learning_rate": 3.403614465497718e-07,
      "loss": 0.3756,
      "step": 5495
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.3533919790831614,
      "learning_rate": 3.3941567821354383e-07,
      "loss": 0.371,
      "step": 5496
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.236337118108129,
      "learning_rate": 3.3847117954895536e-07,
      "loss": 0.3874,
      "step": 5497
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.639016294539808,
      "learning_rate": 3.375279508133139e-07,
      "loss": 0.4702,
      "step": 5498
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.790078032133164,
      "learning_rate": 3.3658599226357944e-07,
      "loss": 0.467,
      "step": 5499
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4275112690455682,
      "learning_rate": 3.356453041563695e-07,
      "loss": 0.482,
      "step": 5500
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.192629278894666,
      "learning_rate": 3.347058867479519e-07,
      "loss": 0.3984,
      "step": 5501
    },
    {
      "epoch": 0.89,
      "grad_norm": 20.325624888121677,
      "learning_rate": 3.337677402942502e-07,
      "loss": 0.3643,
      "step": 5502
    },
    {
      "epoch": 0.89,
      "grad_norm": 14.783158208307313,
      "learning_rate": 3.3283086505084163e-07,
      "loss": 0.4456,
      "step": 5503
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.540932304429171,
      "learning_rate": 3.3189526127295644e-07,
      "loss": 0.4594,
      "step": 5504
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.714769276604025,
      "learning_rate": 3.3096092921547976e-07,
      "loss": 0.4267,
      "step": 5505
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.09451181999298,
      "learning_rate": 3.300278691329478e-07,
      "loss": 0.483,
      "step": 5506
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.420973659347821,
      "learning_rate": 3.290960812795535e-07,
      "loss": 0.4309,
      "step": 5507
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.622779209624776,
      "learning_rate": 3.2816556590914096e-07,
      "loss": 0.4088,
      "step": 5508
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.279418142128179,
      "learning_rate": 3.2723632327520795e-07,
      "loss": 0.4328,
      "step": 5509
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.999990854919739,
      "learning_rate": 3.2630835363090584e-07,
      "loss": 0.4532,
      "step": 5510
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.301159061123506,
      "learning_rate": 3.2538165722903913e-07,
      "loss": 0.3911,
      "step": 5511
    },
    {
      "epoch": 0.89,
      "grad_norm": 10.456957245504995,
      "learning_rate": 3.244562343220664e-07,
      "loss": 0.4198,
      "step": 5512
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.71878381693972,
      "learning_rate": 3.235320851620971e-07,
      "loss": 0.4129,
      "step": 5513
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.376724966639073,
      "learning_rate": 3.2260921000089483e-07,
      "loss": 0.3544,
      "step": 5514
    },
    {
      "epoch": 0.89,
      "grad_norm": 11.003656558271219,
      "learning_rate": 3.216876090898774e-07,
      "loss": 0.3624,
      "step": 5515
    },
    {
      "epoch": 0.89,
      "grad_norm": 11.086396283150378,
      "learning_rate": 3.207672826801139e-07,
      "loss": 0.479,
      "step": 5516
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.854438146062388,
      "learning_rate": 3.198482310223261e-07,
      "loss": 0.4268,
      "step": 5517
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.3969725068891,
      "learning_rate": 3.1893045436688863e-07,
      "loss": 0.4368,
      "step": 5518
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.670361452543856,
      "learning_rate": 3.1801395296383097e-07,
      "loss": 0.418,
      "step": 5519
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.372577489338434,
      "learning_rate": 3.1709872706283173e-07,
      "loss": 0.4046,
      "step": 5520
    },
    {
      "epoch": 0.89,
      "grad_norm": 10.359289475262425,
      "learning_rate": 3.161847769132237e-07,
      "loss": 0.3736,
      "step": 5521
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.042191481285021,
      "learning_rate": 3.152721027639932e-07,
      "loss": 0.3996,
      "step": 5522
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2575994245920827,
      "learning_rate": 3.1436070486377755e-07,
      "loss": 0.4433,
      "step": 5523
    },
    {
      "epoch": 0.89,
      "grad_norm": 15.277968502494566,
      "learning_rate": 3.134505834608653e-07,
      "loss": 0.4058,
      "step": 5524
    },
    {
      "epoch": 0.89,
      "grad_norm": 10.25046436677384,
      "learning_rate": 3.125417388031998e-07,
      "loss": 0.4633,
      "step": 5525
    },
    {
      "epoch": 0.89,
      "grad_norm": 11.497466934966727,
      "learning_rate": 3.116341711383758e-07,
      "loss": 0.4032,
      "step": 5526
    },
    {
      "epoch": 0.89,
      "grad_norm": 13.811747609706352,
      "learning_rate": 3.1072788071363936e-07,
      "loss": 0.3957,
      "step": 5527
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.9855802342414535,
      "learning_rate": 3.0982286777588745e-07,
      "loss": 0.4121,
      "step": 5528
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.58354653721065,
      "learning_rate": 3.089191325716728e-07,
      "loss": 0.4573,
      "step": 5529
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.899375440808218,
      "learning_rate": 3.0801667534719694e-07,
      "loss": 0.3894,
      "step": 5530
    },
    {
      "epoch": 0.89,
      "grad_norm": 9.363768946579198,
      "learning_rate": 3.0711549634831364e-07,
      "loss": 0.434,
      "step": 5531
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.588855976871044,
      "learning_rate": 3.062155958205293e-07,
      "loss": 0.44,
      "step": 5532
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.444901107272617,
      "learning_rate": 3.0531697400900117e-07,
      "loss": 0.5128,
      "step": 5533
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1376965762023412,
      "learning_rate": 3.044196311585401e-07,
      "loss": 0.4087,
      "step": 5534
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.552745013895243,
      "learning_rate": 3.0352356751360555e-07,
      "loss": 0.4275,
      "step": 5535
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.199416513287486,
      "learning_rate": 3.0262878331830946e-07,
      "loss": 0.3817,
      "step": 5536
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.610464811265526,
      "learning_rate": 3.017352788164174e-07,
      "loss": 0.3682,
      "step": 5537
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.781370166372906,
      "learning_rate": 3.0084305425134406e-07,
      "loss": 0.4601,
      "step": 5538
    },
    {
      "epoch": 0.89,
      "grad_norm": 9.457750131401145,
      "learning_rate": 2.999521098661551e-07,
      "loss": 0.3565,
      "step": 5539
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.748196291926938,
      "learning_rate": 2.990624459035685e-07,
      "loss": 0.4121,
      "step": 5540
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.338798291916017,
      "learning_rate": 2.981740626059548e-07,
      "loss": 0.3687,
      "step": 5541
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.839697376547755,
      "learning_rate": 2.972869602153333e-07,
      "loss": 0.4176,
      "step": 5542
    },
    {
      "epoch": 0.89,
      "grad_norm": 9.97373679300075,
      "learning_rate": 2.964011389733734e-07,
      "loss": 0.4286,
      "step": 5543
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.616848003760706,
      "learning_rate": 2.955165991214004e-07,
      "loss": 0.3551,
      "step": 5544
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.671203347313586,
      "learning_rate": 2.946333409003849e-07,
      "loss": 0.4266,
      "step": 5545
    },
    {
      "epoch": 0.89,
      "grad_norm": 9.588708477434107,
      "learning_rate": 2.93751364550951e-07,
      "loss": 0.4152,
      "step": 5546
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.116126028897421,
      "learning_rate": 2.928706703133738e-07,
      "loss": 0.3893,
      "step": 5547
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.349947313867396,
      "learning_rate": 2.919912584275797e-07,
      "loss": 0.4055,
      "step": 5548
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.078093493744602,
      "learning_rate": 2.911131291331437e-07,
      "loss": 0.4929,
      "step": 5549
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.081465285392764,
      "learning_rate": 2.902362826692923e-07,
      "loss": 0.465,
      "step": 5550
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.5943412451397,
      "learning_rate": 2.8936071927490317e-07,
      "loss": 0.3752,
      "step": 5551
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.974395079048967,
      "learning_rate": 2.884864391885034e-07,
      "loss": 0.4092,
      "step": 5552
    },
    {
      "epoch": 0.89,
      "grad_norm": 12.242571725196171,
      "learning_rate": 2.876134426482713e-07,
      "loss": 0.4513,
      "step": 5553
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.902770771407512,
      "learning_rate": 2.8674172989203497e-07,
      "loss": 0.4094,
      "step": 5554
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.56945427745902,
      "learning_rate": 2.858713011572728e-07,
      "loss": 0.4515,
      "step": 5555
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.094578927839349,
      "learning_rate": 2.850021566811145e-07,
      "loss": 0.3975,
      "step": 5556
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3269984400099661,
      "learning_rate": 2.8413429670033855e-07,
      "loss": 0.4496,
      "step": 5557
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.811283036584181,
      "learning_rate": 2.832677214513735e-07,
      "loss": 0.4666,
      "step": 5558
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.650471158491238,
      "learning_rate": 2.8240243117029885e-07,
      "loss": 0.3902,
      "step": 5559
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.415788916708207,
      "learning_rate": 2.815384260928433e-07,
      "loss": 0.3991,
      "step": 5560
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.053182638840294,
      "learning_rate": 2.8067570645438467e-07,
      "loss": 0.396,
      "step": 5561
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.757937813948898,
      "learning_rate": 2.7981427248995276e-07,
      "loss": 0.4013,
      "step": 5562
    },
    {
      "epoch": 0.9,
      "grad_norm": 13.535798367845265,
      "learning_rate": 2.7895412443422644e-07,
      "loss": 0.424,
      "step": 5563
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.573899018411119,
      "learning_rate": 2.780952625215333e-07,
      "loss": 0.4325,
      "step": 5564
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.394106107996087,
      "learning_rate": 2.772376869858495e-07,
      "loss": 0.4199,
      "step": 5565
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.26811469756973,
      "learning_rate": 2.7638139806080435e-07,
      "loss": 0.4142,
      "step": 5566
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.8975711230956875,
      "learning_rate": 2.755263959796739e-07,
      "loss": 0.4528,
      "step": 5567
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.881632462255824,
      "learning_rate": 2.746726809753836e-07,
      "loss": 0.4744,
      "step": 5568
    },
    {
      "epoch": 0.9,
      "grad_norm": 13.622670057496439,
      "learning_rate": 2.73820253280509e-07,
      "loss": 0.3588,
      "step": 5569
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.3511006473182485,
      "learning_rate": 2.72969113127276e-07,
      "loss": 0.3243,
      "step": 5570
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.5413959043765715,
      "learning_rate": 2.7211926074755755e-07,
      "loss": 0.4739,
      "step": 5571
    },
    {
      "epoch": 0.9,
      "grad_norm": 8.392800389466684,
      "learning_rate": 2.712706963728773e-07,
      "loss": 0.4265,
      "step": 5572
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.763077431925545,
      "learning_rate": 2.7042342023440803e-07,
      "loss": 0.3538,
      "step": 5573
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.968205827455832,
      "learning_rate": 2.6957743256297077e-07,
      "loss": 0.3907,
      "step": 5574
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.577269121955915,
      "learning_rate": 2.687327335890344e-07,
      "loss": 0.3777,
      "step": 5575
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1624599261141724,
      "learning_rate": 2.6788932354272104e-07,
      "loss": 0.4349,
      "step": 5576
    },
    {
      "epoch": 0.9,
      "grad_norm": 14.048962673973865,
      "learning_rate": 2.6704720265379626e-07,
      "loss": 0.3769,
      "step": 5577
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.474048595629797,
      "learning_rate": 2.662063711516788e-07,
      "loss": 0.4248,
      "step": 5578
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.50572960953007,
      "learning_rate": 2.653668292654338e-07,
      "loss": 0.3589,
      "step": 5579
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.593189811920075,
      "learning_rate": 2.645285772237743e-07,
      "loss": 0.4064,
      "step": 5580
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.179147139053955,
      "learning_rate": 2.636916152550656e-07,
      "loss": 0.3752,
      "step": 5581
    },
    {
      "epoch": 0.9,
      "grad_norm": 8.634652067822145,
      "learning_rate": 2.6285594358731736e-07,
      "loss": 0.3983,
      "step": 5582
    },
    {
      "epoch": 0.9,
      "grad_norm": 28.290569569023457,
      "learning_rate": 2.6202156244819035e-07,
      "loss": 0.3731,
      "step": 5583
    },
    {
      "epoch": 0.9,
      "grad_norm": 14.711132958500876,
      "learning_rate": 2.611884720649921e-07,
      "loss": 0.4772,
      "step": 5584
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.100438697344083,
      "learning_rate": 2.6035667266468166e-07,
      "loss": 0.4486,
      "step": 5585
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.558185883331002,
      "learning_rate": 2.595261644738617e-07,
      "loss": 0.3871,
      "step": 5586
    },
    {
      "epoch": 0.9,
      "grad_norm": 11.439006615067088,
      "learning_rate": 2.586969477187867e-07,
      "loss": 0.2952,
      "step": 5587
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.760567874843918,
      "learning_rate": 2.578690226253583e-07,
      "loss": 0.4017,
      "step": 5588
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.6509826784426185,
      "learning_rate": 2.570423894191254e-07,
      "loss": 0.3776,
      "step": 5589
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.419998145151117,
      "learning_rate": 2.5621704832528573e-07,
      "loss": 0.4632,
      "step": 5590
    },
    {
      "epoch": 0.9,
      "grad_norm": 8.641532690832065,
      "learning_rate": 2.553929995686849e-07,
      "loss": 0.3916,
      "step": 5591
    },
    {
      "epoch": 0.9,
      "grad_norm": 10.279192141379214,
      "learning_rate": 2.545702433738184e-07,
      "loss": 0.4417,
      "step": 5592
    },
    {
      "epoch": 0.9,
      "grad_norm": 10.920774811421737,
      "learning_rate": 2.537487799648247e-07,
      "loss": 0.3988,
      "step": 5593
    },
    {
      "epoch": 0.9,
      "grad_norm": 8.320572481407005,
      "learning_rate": 2.529286095654948e-07,
      "loss": 0.3717,
      "step": 5594
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.4727012963216595,
      "learning_rate": 2.5210973239926553e-07,
      "loss": 0.4002,
      "step": 5595
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4631861246146862,
      "learning_rate": 2.5129214868922227e-07,
      "loss": 0.482,
      "step": 5596
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.658722684089519,
      "learning_rate": 2.504758586580952e-07,
      "loss": 0.4273,
      "step": 5597
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.1104689381259725,
      "learning_rate": 2.4966086252826706e-07,
      "loss": 0.3916,
      "step": 5598
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.556154212533265,
      "learning_rate": 2.4884716052176237e-07,
      "loss": 0.5336,
      "step": 5599
    },
    {
      "epoch": 0.9,
      "grad_norm": 10.22787171734883,
      "learning_rate": 2.480347528602589e-07,
      "loss": 0.4152,
      "step": 5600
    },
    {
      "epoch": 0.9,
      "grad_norm": 8.332750596889245,
      "learning_rate": 2.4722363976507677e-07,
      "loss": 0.4157,
      "step": 5601
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.2104000268671555,
      "learning_rate": 2.4641382145718597e-07,
      "loss": 0.4159,
      "step": 5602
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.0060276755620965,
      "learning_rate": 2.456052981572038e-07,
      "loss": 0.4491,
      "step": 5603
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.590096276191707,
      "learning_rate": 2.447980700853941e-07,
      "loss": 0.4281,
      "step": 5604
    },
    {
      "epoch": 0.9,
      "grad_norm": 10.624694494160746,
      "learning_rate": 2.439921374616683e-07,
      "loss": 0.4455,
      "step": 5605
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.952641157708498,
      "learning_rate": 2.4318750050558336e-07,
      "loss": 0.4763,
      "step": 5606
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.050077799728524,
      "learning_rate": 2.423841594363469e-07,
      "loss": 0.4337,
      "step": 5607
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.364899562404551,
      "learning_rate": 2.415821144728092e-07,
      "loss": 0.3772,
      "step": 5608
    },
    {
      "epoch": 0.9,
      "grad_norm": 8.968789736601055,
      "learning_rate": 2.4078136583347014e-07,
      "loss": 0.3917,
      "step": 5609
    },
    {
      "epoch": 0.9,
      "grad_norm": 10.671810856381203,
      "learning_rate": 2.399819137364756e-07,
      "loss": 0.3443,
      "step": 5610
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.155547603754944,
      "learning_rate": 2.3918375839961927e-07,
      "loss": 0.38,
      "step": 5611
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.918086025206291,
      "learning_rate": 2.3838690004033826e-07,
      "loss": 0.3911,
      "step": 5612
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.83088684192959,
      "learning_rate": 2.3759133887572117e-07,
      "loss": 0.4189,
      "step": 5613
    },
    {
      "epoch": 0.9,
      "grad_norm": 11.84462985183511,
      "learning_rate": 2.3679707512250072e-07,
      "loss": 0.3947,
      "step": 5614
    },
    {
      "epoch": 0.9,
      "grad_norm": 14.446089985678176,
      "learning_rate": 2.3600410899705383e-07,
      "loss": 0.3836,
      "step": 5615
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.413138430978946,
      "learning_rate": 2.3521244071540883e-07,
      "loss": 0.418,
      "step": 5616
    },
    {
      "epoch": 0.91,
      "grad_norm": 15.603307346664275,
      "learning_rate": 2.3442207049323707e-07,
      "loss": 0.4165,
      "step": 5617
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.8440266939596635,
      "learning_rate": 2.3363299854585742e-07,
      "loss": 0.3487,
      "step": 5618
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.87770243045142,
      "learning_rate": 2.3284522508823403e-07,
      "loss": 0.4284,
      "step": 5619
    },
    {
      "epoch": 0.91,
      "grad_norm": 13.075798787105164,
      "learning_rate": 2.320587503349797e-07,
      "loss": 0.3929,
      "step": 5620
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.63692720012179,
      "learning_rate": 2.3127357450035025e-07,
      "loss": 0.3885,
      "step": 5621
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.028215367215198,
      "learning_rate": 2.304896977982496e-07,
      "loss": 0.4093,
      "step": 5622
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9939273095844205,
      "learning_rate": 2.2970712044222742e-07,
      "loss": 0.4121,
      "step": 5623
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.391065898999475,
      "learning_rate": 2.2892584264548046e-07,
      "loss": 0.4315,
      "step": 5624
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.5600712801803915,
      "learning_rate": 2.2814586462084897e-07,
      "loss": 0.4582,
      "step": 5625
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0951164069892068,
      "learning_rate": 2.2736718658082025e-07,
      "loss": 0.4513,
      "step": 5626
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.012808385793947,
      "learning_rate": 2.2658980873752902e-07,
      "loss": 0.3531,
      "step": 5627
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.8665867819509545,
      "learning_rate": 2.2581373130275307e-07,
      "loss": 0.4361,
      "step": 5628
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.1238007118596975,
      "learning_rate": 2.250389544879189e-07,
      "loss": 0.3674,
      "step": 5629
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.844578529625602,
      "learning_rate": 2.2426547850409598e-07,
      "loss": 0.3629,
      "step": 5630
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.820412043293247,
      "learning_rate": 2.234933035619996e-07,
      "loss": 0.444,
      "step": 5631
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.994468992140094,
      "learning_rate": 2.2272242987199379e-07,
      "loss": 0.4165,
      "step": 5632
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.119450590532243,
      "learning_rate": 2.219528576440838e-07,
      "loss": 0.4005,
      "step": 5633
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.390439593073815,
      "learning_rate": 2.2118458708792256e-07,
      "loss": 0.4145,
      "step": 5634
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.183710218944606,
      "learning_rate": 2.2041761841280873e-07,
      "loss": 0.3539,
      "step": 5635
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4930307966498106,
      "learning_rate": 2.196519518276874e-07,
      "loss": 0.5118,
      "step": 5636
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.589080595934432,
      "learning_rate": 2.1888758754114336e-07,
      "loss": 0.3842,
      "step": 5637
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.891383939730442,
      "learning_rate": 2.1812452576141285e-07,
      "loss": 0.4002,
      "step": 5638
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.613806238676052,
      "learning_rate": 2.1736276669637568e-07,
      "loss": 0.4374,
      "step": 5639
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.401610757722734,
      "learning_rate": 2.1660231055355473e-07,
      "loss": 0.3685,
      "step": 5640
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.537566396517372,
      "learning_rate": 2.1584315754011876e-07,
      "loss": 0.3972,
      "step": 5641
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.0473674999937455,
      "learning_rate": 2.1508530786288284e-07,
      "loss": 0.3839,
      "step": 5642
    },
    {
      "epoch": 0.91,
      "grad_norm": 11.07020816468202,
      "learning_rate": 2.143287617283074e-07,
      "loss": 0.4165,
      "step": 5643
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.168385323138658,
      "learning_rate": 2.135735193424937e-07,
      "loss": 0.3552,
      "step": 5644
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.24471549688324,
      "learning_rate": 2.1281958091119215e-07,
      "loss": 0.477,
      "step": 5645
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.607939437820756,
      "learning_rate": 2.1206694663979676e-07,
      "loss": 0.4038,
      "step": 5646
    },
    {
      "epoch": 0.91,
      "grad_norm": 16.831427855179,
      "learning_rate": 2.113156167333452e-07,
      "loss": 0.4615,
      "step": 5647
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.4601499721893925,
      "learning_rate": 2.105655913965199e-07,
      "loss": 0.3956,
      "step": 5648
    },
    {
      "epoch": 0.91,
      "grad_norm": 12.267416887269848,
      "learning_rate": 2.0981687083364955e-07,
      "loss": 0.3783,
      "step": 5649
    },
    {
      "epoch": 0.91,
      "grad_norm": 9.701809368497797,
      "learning_rate": 2.0906945524870493e-07,
      "loss": 0.398,
      "step": 5650
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.065051426680281,
      "learning_rate": 2.0832334484530424e-07,
      "loss": 0.4109,
      "step": 5651
    },
    {
      "epoch": 0.91,
      "grad_norm": 16.708565409964823,
      "learning_rate": 2.0757853982670773e-07,
      "loss": 0.481,
      "step": 5652
    },
    {
      "epoch": 0.91,
      "grad_norm": 9.664951354496948,
      "learning_rate": 2.0683504039582024e-07,
      "loss": 0.4597,
      "step": 5653
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.251462773099307,
      "learning_rate": 2.0609284675519315e-07,
      "loss": 0.4957,
      "step": 5654
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.621581606496009,
      "learning_rate": 2.053519591070191e-07,
      "loss": 0.32,
      "step": 5655
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.4525676078137,
      "learning_rate": 2.0461237765313612e-07,
      "loss": 0.3974,
      "step": 5656
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.599071881910322,
      "learning_rate": 2.0387410259502692e-07,
      "loss": 0.4761,
      "step": 5657
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.03062391755534,
      "learning_rate": 2.0313713413381952e-07,
      "loss": 0.4338,
      "step": 5658
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.4269472895707445,
      "learning_rate": 2.0240147247028164e-07,
      "loss": 0.3931,
      "step": 5659
    },
    {
      "epoch": 0.91,
      "grad_norm": 11.74083948579148,
      "learning_rate": 2.016671178048296e-07,
      "loss": 0.4064,
      "step": 5660
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.611220546341827,
      "learning_rate": 2.0093407033752178e-07,
      "loss": 0.3975,
      "step": 5661
    },
    {
      "epoch": 0.91,
      "grad_norm": 15.17178481068634,
      "learning_rate": 2.0020233026806003e-07,
      "loss": 0.3659,
      "step": 5662
    },
    {
      "epoch": 0.91,
      "grad_norm": 17.86489123164075,
      "learning_rate": 1.9947189779579047e-07,
      "loss": 0.3982,
      "step": 5663
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.903728722391923,
      "learning_rate": 1.9874277311970335e-07,
      "loss": 0.4124,
      "step": 5664
    },
    {
      "epoch": 0.91,
      "grad_norm": 11.292687402526482,
      "learning_rate": 1.9801495643843316e-07,
      "loss": 0.4633,
      "step": 5665
    },
    {
      "epoch": 0.91,
      "grad_norm": 9.153829829979024,
      "learning_rate": 1.972884479502557e-07,
      "loss": 0.3548,
      "step": 5666
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.193277159394358,
      "learning_rate": 1.965632478530921e-07,
      "loss": 0.364,
      "step": 5667
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.080483668032681,
      "learning_rate": 1.9583935634450768e-07,
      "loss": 0.4152,
      "step": 5668
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.315143910518079,
      "learning_rate": 1.9511677362171078e-07,
      "loss": 0.4389,
      "step": 5669
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.682125110626832,
      "learning_rate": 1.943954998815506e-07,
      "loss": 0.4371,
      "step": 5670
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.5242238784207585,
      "learning_rate": 1.9367553532052497e-07,
      "loss": 0.3862,
      "step": 5671
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.66512380749184,
      "learning_rate": 1.9295688013477031e-07,
      "loss": 0.3822,
      "step": 5672
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.677533531133708,
      "learning_rate": 1.9223953452006784e-07,
      "loss": 0.4665,
      "step": 5673
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.945319722257524,
      "learning_rate": 1.915234986718434e-07,
      "loss": 0.4004,
      "step": 5674
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.74825767323441,
      "learning_rate": 1.908087727851632e-07,
      "loss": 0.3902,
      "step": 5675
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.930941945445651,
      "learning_rate": 1.9009535705474091e-07,
      "loss": 0.466,
      "step": 5676
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.536868757526305,
      "learning_rate": 1.8938325167492888e-07,
      "loss": 0.4374,
      "step": 5677
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.84400853281233,
      "learning_rate": 1.8867245683972358e-07,
      "loss": 0.3935,
      "step": 5678
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.219624391224151,
      "learning_rate": 1.8796297274276677e-07,
      "loss": 0.4726,
      "step": 5679
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.225957038186819,
      "learning_rate": 1.8725479957734105e-07,
      "loss": 0.3698,
      "step": 5680
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.318578838339036,
      "learning_rate": 1.8654793753637156e-07,
      "loss": 0.4374,
      "step": 5681
    },
    {
      "epoch": 0.92,
      "grad_norm": 9.61787225320892,
      "learning_rate": 1.8584238681242705e-07,
      "loss": 0.4509,
      "step": 5682
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.194949426355194,
      "learning_rate": 1.8513814759772042e-07,
      "loss": 0.3756,
      "step": 5683
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.89789197224502,
      "learning_rate": 1.844352200841043e-07,
      "loss": 0.3682,
      "step": 5684
    },
    {
      "epoch": 0.92,
      "grad_norm": 10.599436218436441,
      "learning_rate": 1.8373360446307608e-07,
      "loss": 0.3818,
      "step": 5685
    },
    {
      "epoch": 0.92,
      "grad_norm": 20.47946827886676,
      "learning_rate": 1.8303330092577453e-07,
      "loss": 0.4035,
      "step": 5686
    },
    {
      "epoch": 0.92,
      "grad_norm": 12.848645435005375,
      "learning_rate": 1.823343096629837e-07,
      "loss": 0.4902,
      "step": 5687
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.11234515784128,
      "learning_rate": 1.8163663086512574e-07,
      "loss": 0.3471,
      "step": 5688
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.190382808486084,
      "learning_rate": 1.80940264722268e-07,
      "loss": 0.4705,
      "step": 5689
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.24348449545593,
      "learning_rate": 1.8024521142412155e-07,
      "loss": 0.4235,
      "step": 5690
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.322727403083543,
      "learning_rate": 1.7955147116003603e-07,
      "loss": 0.481,
      "step": 5691
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.047430117082018,
      "learning_rate": 1.7885904411900633e-07,
      "loss": 0.4241,
      "step": 5692
    },
    {
      "epoch": 0.92,
      "grad_norm": 11.59899323838557,
      "learning_rate": 1.7816793048966884e-07,
      "loss": 0.4988,
      "step": 5693
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.8043922269912915,
      "learning_rate": 1.7747813046030184e-07,
      "loss": 0.425,
      "step": 5694
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.594260174402231,
      "learning_rate": 1.7678964421882494e-07,
      "loss": 0.394,
      "step": 5695
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.238407288765113,
      "learning_rate": 1.7610247195280207e-07,
      "loss": 0.3868,
      "step": 5696
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.625979468931048,
      "learning_rate": 1.7541661384943676e-07,
      "loss": 0.4117,
      "step": 5697
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.6694017607711595,
      "learning_rate": 1.747320700955768e-07,
      "loss": 0.5149,
      "step": 5698
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.74752640227918,
      "learning_rate": 1.7404884087771023e-07,
      "loss": 0.3877,
      "step": 5699
    },
    {
      "epoch": 0.92,
      "grad_norm": 21.10747722192997,
      "learning_rate": 1.7336692638196706e-07,
      "loss": 0.4115,
      "step": 5700
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.666266737314029,
      "learning_rate": 1.726863267941209e-07,
      "loss": 0.4011,
      "step": 5701
    },
    {
      "epoch": 0.92,
      "grad_norm": 16.39343525880441,
      "learning_rate": 1.72007042299584e-07,
      "loss": 0.379,
      "step": 5702
    },
    {
      "epoch": 0.92,
      "grad_norm": 11.82139612384758,
      "learning_rate": 1.7132907308341383e-07,
      "loss": 0.4631,
      "step": 5703
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.481875989140034,
      "learning_rate": 1.7065241933030664e-07,
      "loss": 0.4138,
      "step": 5704
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.0392192920757095,
      "learning_rate": 1.6997708122460322e-07,
      "loss": 0.4292,
      "step": 5705
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.974780873464883,
      "learning_rate": 1.6930305895028255e-07,
      "loss": 0.4026,
      "step": 5706
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.763039421033171,
      "learning_rate": 1.6863035269096774e-07,
      "loss": 0.347,
      "step": 5707
    },
    {
      "epoch": 0.92,
      "grad_norm": 9.906384094260954,
      "learning_rate": 1.6795896262992162e-07,
      "loss": 0.3506,
      "step": 5708
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.002980847134111,
      "learning_rate": 1.672888889500518e-07,
      "loss": 0.3767,
      "step": 5709
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.940667328769693,
      "learning_rate": 1.6662013183390112e-07,
      "loss": 0.4258,
      "step": 5710
    },
    {
      "epoch": 0.92,
      "grad_norm": 18.654288103818917,
      "learning_rate": 1.6595269146366e-07,
      "loss": 0.4177,
      "step": 5711
    },
    {
      "epoch": 0.92,
      "grad_norm": 13.345879362058952,
      "learning_rate": 1.6528656802115794e-07,
      "loss": 0.4495,
      "step": 5712
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.763094233466857,
      "learning_rate": 1.6462176168786371e-07,
      "loss": 0.4351,
      "step": 5713
    },
    {
      "epoch": 0.92,
      "grad_norm": 12.300561476181244,
      "learning_rate": 1.6395827264488962e-07,
      "loss": 0.361,
      "step": 5714
    },
    {
      "epoch": 0.92,
      "grad_norm": 14.312788323776475,
      "learning_rate": 1.6329610107298833e-07,
      "loss": 0.3512,
      "step": 5715
    },
    {
      "epoch": 0.92,
      "grad_norm": 9.290958103442085,
      "learning_rate": 1.6263524715255442e-07,
      "loss": 0.4266,
      "step": 5716
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.692028736494481,
      "learning_rate": 1.6197571106362054e-07,
      "loss": 0.385,
      "step": 5717
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.973696118269885,
      "learning_rate": 1.6131749298586464e-07,
      "loss": 0.4409,
      "step": 5718
    },
    {
      "epoch": 0.92,
      "grad_norm": 11.989204208784603,
      "learning_rate": 1.6066059309860273e-07,
      "loss": 0.4621,
      "step": 5719
    },
    {
      "epoch": 0.92,
      "grad_norm": 10.87619009861215,
      "learning_rate": 1.6000501158079217e-07,
      "loss": 0.3863,
      "step": 5720
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.949451011883772,
      "learning_rate": 1.5935074861103183e-07,
      "loss": 0.3791,
      "step": 5721
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.216317036441988,
      "learning_rate": 1.5869780436756078e-07,
      "loss": 0.426,
      "step": 5722
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.149134998403504,
      "learning_rate": 1.5804617902825948e-07,
      "loss": 0.375,
      "step": 5723
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.221450749403776,
      "learning_rate": 1.573958727706476e-07,
      "loss": 0.4528,
      "step": 5724
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.105379535857221,
      "learning_rate": 1.567468857718879e-07,
      "loss": 0.3988,
      "step": 5725
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.074944905391083,
      "learning_rate": 1.560992182087806e-07,
      "loss": 0.4348,
      "step": 5726
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.979908624463976,
      "learning_rate": 1.554528702577701e-07,
      "loss": 0.4136,
      "step": 5727
    },
    {
      "epoch": 0.92,
      "grad_norm": 9.618657257238828,
      "learning_rate": 1.5480784209493837e-07,
      "loss": 0.4205,
      "step": 5728
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.251143012145175,
      "learning_rate": 1.5416413389600816e-07,
      "loss": 0.4069,
      "step": 5729
    },
    {
      "epoch": 0.92,
      "grad_norm": 23.746872150690024,
      "learning_rate": 1.5352174583634526e-07,
      "loss": 0.4506,
      "step": 5730
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.422963005313953,
      "learning_rate": 1.5288067809095196e-07,
      "loss": 0.4487,
      "step": 5731
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.232360640315047,
      "learning_rate": 1.5224093083447354e-07,
      "loss": 0.4942,
      "step": 5732
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.944194595825505,
      "learning_rate": 1.5160250424119505e-07,
      "loss": 0.3519,
      "step": 5733
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.282486211522995,
      "learning_rate": 1.509653984850412e-07,
      "loss": 0.3591,
      "step": 5734
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.743740952490462,
      "learning_rate": 1.5032961373957766e-07,
      "loss": 0.4343,
      "step": 5735
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.99238269707156,
      "learning_rate": 1.496951501780086e-07,
      "loss": 0.4543,
      "step": 5736
    },
    {
      "epoch": 0.92,
      "grad_norm": 9.001438292428004,
      "learning_rate": 1.490620079731808e-07,
      "loss": 0.4172,
      "step": 5737
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.906954103434556,
      "learning_rate": 1.4843018729757853e-07,
      "loss": 0.3922,
      "step": 5738
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.47791874659914,
      "learning_rate": 1.4779968832332737e-07,
      "loss": 0.4132,
      "step": 5739
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.025946833685009,
      "learning_rate": 1.4717051122219272e-07,
      "loss": 0.3783,
      "step": 5740
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.484472352686724,
      "learning_rate": 1.4654265616557973e-07,
      "loss": 0.3814,
      "step": 5741
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.450014671174384,
      "learning_rate": 1.459161233245343e-07,
      "loss": 0.3845,
      "step": 5742
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.856514252857716,
      "learning_rate": 1.4529091286973994e-07,
      "loss": 0.4347,
      "step": 5743
    },
    {
      "epoch": 0.93,
      "grad_norm": 10.236074033375482,
      "learning_rate": 1.4466702497152208e-07,
      "loss": 0.3948,
      "step": 5744
    },
    {
      "epoch": 0.93,
      "grad_norm": 20.737271150611626,
      "learning_rate": 1.4404445979984473e-07,
      "loss": 0.4728,
      "step": 5745
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.907427466254237,
      "learning_rate": 1.434232175243111e-07,
      "loss": 0.4026,
      "step": 5746
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.218725467008014,
      "learning_rate": 1.4280329831416585e-07,
      "loss": 0.4277,
      "step": 5747
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.544943070837199,
      "learning_rate": 1.4218470233829107e-07,
      "loss": 0.4163,
      "step": 5748
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.391966666981644,
      "learning_rate": 1.415674297652103e-07,
      "loss": 0.4179,
      "step": 5749
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.716714766436801,
      "learning_rate": 1.4095148076308518e-07,
      "loss": 0.4572,
      "step": 5750
    },
    {
      "epoch": 0.93,
      "grad_norm": 10.138736937389433,
      "learning_rate": 1.4033685549971643e-07,
      "loss": 0.3902,
      "step": 5751
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.13777654887846,
      "learning_rate": 1.397235541425468e-07,
      "loss": 0.4258,
      "step": 5752
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.806130950352292,
      "learning_rate": 1.3911157685865483e-07,
      "loss": 0.4422,
      "step": 5753
    },
    {
      "epoch": 0.93,
      "grad_norm": 56.00443958804668,
      "learning_rate": 1.3850092381476e-07,
      "loss": 0.4338,
      "step": 5754
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.2342355818159865,
      "learning_rate": 1.3789159517722138e-07,
      "loss": 0.4427,
      "step": 5755
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.283376878949043,
      "learning_rate": 1.3728359111203792e-07,
      "loss": 0.3929,
      "step": 5756
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.4032650526334365,
      "learning_rate": 1.3667691178484598e-07,
      "loss": 0.4017,
      "step": 5757
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.480814629253254,
      "learning_rate": 1.360715573609206e-07,
      "loss": 0.357,
      "step": 5758
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.146528631849833,
      "learning_rate": 1.354675280051787e-07,
      "loss": 0.3866,
      "step": 5759
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.164474112845113,
      "learning_rate": 1.348648238821737e-07,
      "loss": 0.3855,
      "step": 5760
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0382655116272952,
      "learning_rate": 1.3426344515609813e-07,
      "loss": 0.4254,
      "step": 5761
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2970577267304515,
      "learning_rate": 1.3366339199078538e-07,
      "loss": 0.4246,
      "step": 5762
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.75442760296835,
      "learning_rate": 1.3306466454970634e-07,
      "loss": 0.4346,
      "step": 5763
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.537737038293063,
      "learning_rate": 1.3246726299597058e-07,
      "loss": 0.475,
      "step": 5764
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.237304254787766,
      "learning_rate": 1.318711874923262e-07,
      "loss": 0.4411,
      "step": 5765
    },
    {
      "epoch": 0.93,
      "grad_norm": 15.920798167218466,
      "learning_rate": 1.312764382011611e-07,
      "loss": 0.4365,
      "step": 5766
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.756997851574939,
      "learning_rate": 1.3068301528450178e-07,
      "loss": 0.4368,
      "step": 5767
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.868328241856368,
      "learning_rate": 1.3009091890401226e-07,
      "loss": 0.3537,
      "step": 5768
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.354781934227097,
      "learning_rate": 1.2950014922099685e-07,
      "loss": 0.4488,
      "step": 5769
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.335241831777957,
      "learning_rate": 1.2891070639639625e-07,
      "loss": 0.3471,
      "step": 5770
    },
    {
      "epoch": 0.93,
      "grad_norm": 18.338255987382198,
      "learning_rate": 1.28322590590792e-07,
      "loss": 0.4087,
      "step": 5771
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.949105461969896,
      "learning_rate": 1.2773580196440262e-07,
      "loss": 0.4252,
      "step": 5772
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.518901631882311,
      "learning_rate": 1.271503406770852e-07,
      "loss": 0.3942,
      "step": 5773
    },
    {
      "epoch": 0.93,
      "grad_norm": 15.921139372478414,
      "learning_rate": 1.265662068883361e-07,
      "loss": 0.397,
      "step": 5774
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0867527144085891,
      "learning_rate": 1.2598340075728967e-07,
      "loss": 0.4128,
      "step": 5775
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.279062951916112,
      "learning_rate": 1.254019224427172e-07,
      "loss": 0.3784,
      "step": 5776
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.409989086022639,
      "learning_rate": 1.2482177210303036e-07,
      "loss": 0.4499,
      "step": 5777
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.967686677265175,
      "learning_rate": 1.242429498962783e-07,
      "loss": 0.4072,
      "step": 5778
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.269043266987582,
      "learning_rate": 1.23665455980147e-07,
      "loss": 0.375,
      "step": 5779
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.646761940051288,
      "learning_rate": 1.2308929051196296e-07,
      "loss": 0.4049,
      "step": 5780
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.888657230322561,
      "learning_rate": 1.2251445364868886e-07,
      "loss": 0.4077,
      "step": 5781
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1415596147112,
      "learning_rate": 1.2194094554692614e-07,
      "loss": 0.4514,
      "step": 5782
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.735622630760204,
      "learning_rate": 1.213687663629143e-07,
      "loss": 0.4083,
      "step": 5783
    },
    {
      "epoch": 0.93,
      "grad_norm": 10.085999038222868,
      "learning_rate": 1.2079791625253078e-07,
      "loss": 0.4123,
      "step": 5784
    },
    {
      "epoch": 0.93,
      "grad_norm": 11.992922934707742,
      "learning_rate": 1.2022839537129127e-07,
      "loss": 0.3911,
      "step": 5785
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.934844099779076,
      "learning_rate": 1.196602038743483e-07,
      "loss": 0.4428,
      "step": 5786
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.073061098502415,
      "learning_rate": 1.1909334191649302e-07,
      "loss": 0.3574,
      "step": 5787
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.763205650607409,
      "learning_rate": 1.1852780965215471e-07,
      "loss": 0.4088,
      "step": 5788
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.540506573593914,
      "learning_rate": 1.1796360723539956e-07,
      "loss": 0.4065,
      "step": 5789
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.781266709194573,
      "learning_rate": 1.174007348199313e-07,
      "loss": 0.3503,
      "step": 5790
    },
    {
      "epoch": 0.93,
      "grad_norm": 14.571636224652636,
      "learning_rate": 1.1683919255909337e-07,
      "loss": 0.3965,
      "step": 5791
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.199851786524772,
      "learning_rate": 1.1627898060586395e-07,
      "loss": 0.4313,
      "step": 5792
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.210104138994689,
      "learning_rate": 1.1572009911286097e-07,
      "loss": 0.4181,
      "step": 5793
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.622908778663875,
      "learning_rate": 1.151625482323393e-07,
      "loss": 0.4297,
      "step": 5794
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.27573984580691,
      "learning_rate": 1.1460632811619021e-07,
      "loss": 0.4822,
      "step": 5795
    },
    {
      "epoch": 0.93,
      "grad_norm": 12.533227596168032,
      "learning_rate": 1.1405143891594417e-07,
      "loss": 0.4847,
      "step": 5796
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.576708741714535,
      "learning_rate": 1.1349788078276857e-07,
      "loss": 0.422,
      "step": 5797
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.206835263603047,
      "learning_rate": 1.129456538674667e-07,
      "loss": 0.3712,
      "step": 5798
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.961290519001119,
      "learning_rate": 1.1239475832048152e-07,
      "loss": 0.442,
      "step": 5799
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.0001864632818,
      "learning_rate": 1.1184519429189245e-07,
      "loss": 0.4672,
      "step": 5800
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.141258510606905,
      "learning_rate": 1.1129696193141414e-07,
      "loss": 0.3983,
      "step": 5801
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.6674470744303855,
      "learning_rate": 1.107500613884016e-07,
      "loss": 0.4538,
      "step": 5802
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.114301779444167,
      "learning_rate": 1.1020449281184565e-07,
      "loss": 0.3953,
      "step": 5803
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.486918232967572,
      "learning_rate": 1.0966025635037403e-07,
      "loss": 0.3877,
      "step": 5804
    },
    {
      "epoch": 0.94,
      "grad_norm": 9.006151838538086,
      "learning_rate": 1.0911735215225095e-07,
      "loss": 0.3975,
      "step": 5805
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.485855712234616,
      "learning_rate": 1.0857578036537919e-07,
      "loss": 0.397,
      "step": 5806
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.333085687940403,
      "learning_rate": 1.0803554113729797e-07,
      "loss": 0.3897,
      "step": 5807
    },
    {
      "epoch": 0.94,
      "grad_norm": 9.402562581558769,
      "learning_rate": 1.0749663461518344e-07,
      "loss": 0.3487,
      "step": 5808
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.200716967468235,
      "learning_rate": 1.0695906094584763e-07,
      "loss": 0.385,
      "step": 5809
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.0716164350194015,
      "learning_rate": 1.0642282027574168e-07,
      "loss": 0.4614,
      "step": 5810
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.562872904319401,
      "learning_rate": 1.0588791275095156e-07,
      "loss": 0.4152,
      "step": 5811
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.382719579352955,
      "learning_rate": 1.0535433851720067e-07,
      "loss": 0.4263,
      "step": 5812
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.135548221215424,
      "learning_rate": 1.0482209771985052e-07,
      "loss": 0.4131,
      "step": 5813
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.925986057512551,
      "learning_rate": 1.0429119050389624e-07,
      "loss": 0.4009,
      "step": 5814
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.521372580819769,
      "learning_rate": 1.0376161701397325e-07,
      "loss": 0.3809,
      "step": 5815
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.258351875059061,
      "learning_rate": 1.0323337739435169e-07,
      "loss": 0.4214,
      "step": 5816
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.1623260769387125,
      "learning_rate": 1.0270647178893756e-07,
      "loss": 0.3965,
      "step": 5817
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.767774577862997,
      "learning_rate": 1.0218090034127603e-07,
      "loss": 0.3732,
      "step": 5818
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.209114100699132,
      "learning_rate": 1.0165666319454592e-07,
      "loss": 0.4681,
      "step": 5819
    },
    {
      "epoch": 0.94,
      "grad_norm": 26.165265312192133,
      "learning_rate": 1.011337604915641e-07,
      "loss": 0.3606,
      "step": 5820
    },
    {
      "epoch": 0.94,
      "grad_norm": 12.685863718973767,
      "learning_rate": 1.0061219237478382e-07,
      "loss": 0.3566,
      "step": 5821
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.286592878355289,
      "learning_rate": 1.0009195898629532e-07,
      "loss": 0.4823,
      "step": 5822
    },
    {
      "epoch": 0.94,
      "grad_norm": 9.583253894335648,
      "learning_rate": 9.957306046782411e-08,
      "loss": 0.3735,
      "step": 5823
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.836992136466474,
      "learning_rate": 9.905549696073213e-08,
      "loss": 0.3891,
      "step": 5824
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.8262920469017,
      "learning_rate": 9.85392686060177e-08,
      "loss": 0.4681,
      "step": 5825
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.65573669329764,
      "learning_rate": 9.802437554431665e-08,
      "loss": 0.3815,
      "step": 5826
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.335559481630403,
      "learning_rate": 9.7510817915899e-08,
      "loss": 0.4572,
      "step": 5827
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.226576757282827,
      "learning_rate": 9.699859586067229e-08,
      "loss": 0.426,
      "step": 5828
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.2284310102423746,
      "learning_rate": 9.648770951818098e-08,
      "loss": 0.4938,
      "step": 5829
    },
    {
      "epoch": 0.94,
      "grad_norm": 16.295223570527472,
      "learning_rate": 9.597815902760321e-08,
      "loss": 0.444,
      "step": 5830
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.891958484061828,
      "learning_rate": 9.54699445277546e-08,
      "loss": 0.4212,
      "step": 5831
    },
    {
      "epoch": 0.94,
      "grad_norm": 16.51412270214544,
      "learning_rate": 9.496306615708773e-08,
      "loss": 0.442,
      "step": 5832
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.652512388286109,
      "learning_rate": 9.44575240536899e-08,
      "loss": 0.3582,
      "step": 5833
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.67244200555198,
      "learning_rate": 9.395331835528431e-08,
      "loss": 0.4237,
      "step": 5834
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.067059554441094,
      "learning_rate": 9.345044919923108e-08,
      "loss": 0.4106,
      "step": 5835
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.6210532084480445,
      "learning_rate": 9.294891672252448e-08,
      "loss": 0.4206,
      "step": 5836
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.678608786024188,
      "learning_rate": 9.244872106179748e-08,
      "loss": 0.3936,
      "step": 5837
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.0062816596510284,
      "learning_rate": 9.194986235331604e-08,
      "loss": 0.3958,
      "step": 5838
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.972339549688348,
      "learning_rate": 9.145234073298314e-08,
      "loss": 0.3936,
      "step": 5839
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.548191144077434,
      "learning_rate": 9.095615633633814e-08,
      "loss": 0.3626,
      "step": 5840
    },
    {
      "epoch": 0.94,
      "grad_norm": 12.053233552367903,
      "learning_rate": 9.046130929855401e-08,
      "loss": 0.4655,
      "step": 5841
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.514570018198193,
      "learning_rate": 8.99677997544418e-08,
      "loss": 0.4338,
      "step": 5842
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.01981552612846,
      "learning_rate": 8.947562783844677e-08,
      "loss": 0.4357,
      "step": 5843
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.077244655508158,
      "learning_rate": 8.898479368464996e-08,
      "loss": 0.4475,
      "step": 5844
    },
    {
      "epoch": 0.94,
      "grad_norm": 9.963062713542467,
      "learning_rate": 8.849529742676887e-08,
      "loss": 0.4027,
      "step": 5845
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.142137641847229,
      "learning_rate": 8.800713919815407e-08,
      "loss": 0.4383,
      "step": 5846
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.180726440621798,
      "learning_rate": 8.752031913179527e-08,
      "loss": 0.3988,
      "step": 5847
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.331591741468932,
      "learning_rate": 8.70348373603147e-08,
      "loss": 0.4633,
      "step": 5848
    },
    {
      "epoch": 0.94,
      "grad_norm": 10.770742271464655,
      "learning_rate": 8.655069401597105e-08,
      "loss": 0.4182,
      "step": 5849
    },
    {
      "epoch": 0.94,
      "grad_norm": 11.587774022867231,
      "learning_rate": 8.606788923065824e-08,
      "loss": 0.3881,
      "step": 5850
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.058609342425633,
      "learning_rate": 8.558642313590603e-08,
      "loss": 0.3761,
      "step": 5851
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.825711552891778,
      "learning_rate": 8.510629586287844e-08,
      "loss": 0.3624,
      "step": 5852
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.264032076776157,
      "learning_rate": 8.462750754237581e-08,
      "loss": 0.4062,
      "step": 5853
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.5244532020392185,
      "learning_rate": 8.415005830483324e-08,
      "loss": 0.3587,
      "step": 5854
    },
    {
      "epoch": 0.94,
      "grad_norm": 11.225856190189385,
      "learning_rate": 8.367394828032116e-08,
      "loss": 0.4615,
      "step": 5855
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.573829437438821,
      "learning_rate": 8.319917759854413e-08,
      "loss": 0.3417,
      "step": 5856
    },
    {
      "epoch": 0.94,
      "grad_norm": 16.02296492438904,
      "learning_rate": 8.272574638884423e-08,
      "loss": 0.466,
      "step": 5857
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.79364953230151,
      "learning_rate": 8.225365478019664e-08,
      "loss": 0.4119,
      "step": 5858
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.846320201797515,
      "learning_rate": 8.178290290121127e-08,
      "loss": 0.3758,
      "step": 5859
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.725858630667987,
      "learning_rate": 8.131349088013495e-08,
      "loss": 0.5032,
      "step": 5860
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.234053807948427,
      "learning_rate": 8.08454188448482e-08,
      "loss": 0.4746,
      "step": 5861
    },
    {
      "epoch": 0.94,
      "grad_norm": 10.048455527480266,
      "learning_rate": 8.037868692286677e-08,
      "loss": 0.409,
      "step": 5862
    },
    {
      "epoch": 0.94,
      "grad_norm": 10.65406336088548,
      "learning_rate": 7.991329524134062e-08,
      "loss": 0.4167,
      "step": 5863
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.684963797557379,
      "learning_rate": 7.944924392705666e-08,
      "loss": 0.4379,
      "step": 5864
    },
    {
      "epoch": 0.94,
      "grad_norm": 11.321740926565184,
      "learning_rate": 7.898653310643378e-08,
      "loss": 0.4127,
      "step": 5865
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.667911571802254,
      "learning_rate": 7.852516290552781e-08,
      "loss": 0.4098,
      "step": 5866
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.479981074795613,
      "learning_rate": 7.80651334500293e-08,
      "loss": 0.4091,
      "step": 5867
    },
    {
      "epoch": 0.95,
      "grad_norm": 9.813719685254274,
      "learning_rate": 7.760644486526137e-08,
      "loss": 0.3635,
      "step": 5868
    },
    {
      "epoch": 0.95,
      "grad_norm": 14.118240563554707,
      "learning_rate": 7.714909727618569e-08,
      "loss": 0.4583,
      "step": 5869
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.4876422558385105,
      "learning_rate": 7.669309080739429e-08,
      "loss": 0.4245,
      "step": 5870
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.506215435477915,
      "learning_rate": 7.623842558311668e-08,
      "loss": 0.4258,
      "step": 5871
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.858537272584325,
      "learning_rate": 7.578510172721598e-08,
      "loss": 0.409,
      "step": 5872
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.141470147091458,
      "learning_rate": 7.533311936319121e-08,
      "loss": 0.3829,
      "step": 5873
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.657257389962846,
      "learning_rate": 7.48824786141733e-08,
      "loss": 0.4403,
      "step": 5874
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.344094551086219,
      "learning_rate": 7.443317960293018e-08,
      "loss": 0.4,
      "step": 5875
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.33474184075586,
      "learning_rate": 7.39852224518628e-08,
      "loss": 0.404,
      "step": 5876
    },
    {
      "epoch": 0.95,
      "grad_norm": 10.407355835805284,
      "learning_rate": 7.353860728300743e-08,
      "loss": 0.4065,
      "step": 5877
    },
    {
      "epoch": 0.95,
      "grad_norm": 20.59488469179399,
      "learning_rate": 7.3093334218034e-08,
      "loss": 0.4034,
      "step": 5878
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.22075818680981,
      "learning_rate": 7.264940337824767e-08,
      "loss": 0.4411,
      "step": 5879
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.6086331895969606,
      "learning_rate": 7.22068148845867e-08,
      "loss": 0.4129,
      "step": 5880
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.725880053857819,
      "learning_rate": 7.176556885762465e-08,
      "loss": 0.4562,
      "step": 5881
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.86585837054657,
      "learning_rate": 7.132566541756925e-08,
      "loss": 0.4635,
      "step": 5882
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.384710363284801,
      "learning_rate": 7.088710468426241e-08,
      "loss": 0.4583,
      "step": 5883
    },
    {
      "epoch": 0.95,
      "grad_norm": 12.045637904444138,
      "learning_rate": 7.044988677718023e-08,
      "loss": 0.3897,
      "step": 5884
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.2931893617965144,
      "learning_rate": 7.001401181543243e-08,
      "loss": 0.4356,
      "step": 5885
    },
    {
      "epoch": 0.95,
      "grad_norm": 23.042045191042856,
      "learning_rate": 6.957947991776403e-08,
      "loss": 0.4389,
      "step": 5886
    },
    {
      "epoch": 0.95,
      "grad_norm": 9.900109059428903,
      "learning_rate": 6.914629120255312e-08,
      "loss": 0.4223,
      "step": 5887
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.835897248992323,
      "learning_rate": 6.87144457878114e-08,
      "loss": 0.4284,
      "step": 5888
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.708066293394715,
      "learning_rate": 6.828394379118752e-08,
      "loss": 0.4358,
      "step": 5889
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.776882452938125,
      "learning_rate": 6.785478532995993e-08,
      "loss": 0.3985,
      "step": 5890
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.700469630526049,
      "learning_rate": 6.742697052104507e-08,
      "loss": 0.3593,
      "step": 5891
    },
    {
      "epoch": 0.95,
      "grad_norm": 9.260048895223225,
      "learning_rate": 6.70004994809903e-08,
      "loss": 0.4361,
      "step": 5892
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.654670722886092,
      "learning_rate": 6.657537232597766e-08,
      "loss": 0.4132,
      "step": 5893
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.49725715984542,
      "learning_rate": 6.615158917182507e-08,
      "loss": 0.389,
      "step": 5894
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.746975826596797,
      "learning_rate": 6.572915013398184e-08,
      "loss": 0.4404,
      "step": 5895
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.791142248871385,
      "learning_rate": 6.530805532753204e-08,
      "loss": 0.3724,
      "step": 5896
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.71437291283,
      "learning_rate": 6.488830486719333e-08,
      "loss": 0.4229,
      "step": 5897
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.822357403467221,
      "learning_rate": 6.446989886731758e-08,
      "loss": 0.4168,
      "step": 5898
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.262449623790251,
      "learning_rate": 6.405283744189027e-08,
      "loss": 0.4336,
      "step": 5899
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.701487737129059,
      "learning_rate": 6.363712070452999e-08,
      "loss": 0.4514,
      "step": 5900
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.290442548069577,
      "learning_rate": 6.322274876848944e-08,
      "loss": 0.4164,
      "step": 5901
    },
    {
      "epoch": 0.95,
      "grad_norm": 11.965618092778563,
      "learning_rate": 6.280972174665611e-08,
      "loss": 0.4178,
      "step": 5902
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.827712682324868,
      "learning_rate": 6.239803975154835e-08,
      "loss": 0.3925,
      "step": 5903
    },
    {
      "epoch": 0.95,
      "grad_norm": 10.917158579931378,
      "learning_rate": 6.19877028953203e-08,
      "loss": 0.373,
      "step": 5904
    },
    {
      "epoch": 0.95,
      "grad_norm": 13.140587831327757,
      "learning_rate": 6.157871128975923e-08,
      "loss": 0.4259,
      "step": 5905
    },
    {
      "epoch": 0.95,
      "grad_norm": 10.124391970048807,
      "learning_rate": 6.1171065046286e-08,
      "loss": 0.4616,
      "step": 5906
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.013872641561472,
      "learning_rate": 6.076476427595345e-08,
      "loss": 0.4711,
      "step": 5907
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.1520028372818825,
      "learning_rate": 6.035980908945027e-08,
      "loss": 0.4525,
      "step": 5908
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.280795576400204,
      "learning_rate": 5.995619959709764e-08,
      "loss": 0.4068,
      "step": 5909
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.071374327806699,
      "learning_rate": 5.9553935908848724e-08,
      "loss": 0.4113,
      "step": 5910
    },
    {
      "epoch": 0.95,
      "grad_norm": 15.592313663552922,
      "learning_rate": 5.915301813429197e-08,
      "loss": 0.3669,
      "step": 5911
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.7435192998579385,
      "learning_rate": 5.875344638264835e-08,
      "loss": 0.4452,
      "step": 5912
    },
    {
      "epoch": 0.95,
      "grad_norm": 23.5835348039031,
      "learning_rate": 5.835522076277189e-08,
      "loss": 0.3979,
      "step": 5913
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.786179809279504,
      "learning_rate": 5.795834138315137e-08,
      "loss": 0.4429,
      "step": 5914
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.909885190016353,
      "learning_rate": 5.756280835190586e-08,
      "loss": 0.4484,
      "step": 5915
    },
    {
      "epoch": 0.95,
      "grad_norm": 12.073537111284496,
      "learning_rate": 5.7168621776790836e-08,
      "loss": 0.4176,
      "step": 5916
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3233130574130547,
      "learning_rate": 5.6775781765193714e-08,
      "loss": 0.4256,
      "step": 5917
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.081024513234629,
      "learning_rate": 5.638428842413335e-08,
      "loss": 0.3803,
      "step": 5918
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.20612359915436,
      "learning_rate": 5.599414186026497e-08,
      "loss": 0.4613,
      "step": 5919
    },
    {
      "epoch": 0.95,
      "grad_norm": 9.012360383680965,
      "learning_rate": 5.5605342179874676e-08,
      "loss": 0.4371,
      "step": 5920
    },
    {
      "epoch": 0.95,
      "grad_norm": 11.770986569614706,
      "learning_rate": 5.5217889488882734e-08,
      "loss": 0.4802,
      "step": 5921
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.028635249280748,
      "learning_rate": 5.4831783892840275e-08,
      "loss": 0.403,
      "step": 5922
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.043275926565666,
      "learning_rate": 5.444702549693481e-08,
      "loss": 0.3939,
      "step": 5923
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.5037434684148785,
      "learning_rate": 5.406361440598529e-08,
      "loss": 0.3936,
      "step": 5924
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.1938582052948,
      "learning_rate": 5.368155072444148e-08,
      "loss": 0.3659,
      "step": 5925
    },
    {
      "epoch": 0.95,
      "grad_norm": 28.90732576803218,
      "learning_rate": 5.330083455638957e-08,
      "loss": 0.4069,
      "step": 5926
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.359009605500593,
      "learning_rate": 5.2921466005547706e-08,
      "loss": 0.3961,
      "step": 5927
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.13153873348958,
      "learning_rate": 5.2543445175264883e-08,
      "loss": 0.4379,
      "step": 5928
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.782696909653592,
      "learning_rate": 5.216677216852539e-08,
      "loss": 0.3796,
      "step": 5929
    },
    {
      "epoch": 0.96,
      "grad_norm": 13.692619169191119,
      "learning_rate": 5.179144708794437e-08,
      "loss": 0.4051,
      "step": 5930
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.08143424869347,
      "learning_rate": 5.141747003577224e-08,
      "loss": 0.3922,
      "step": 5931
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.4843549854584195,
      "learning_rate": 5.104484111388919e-08,
      "loss": 0.4962,
      "step": 5932
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.245580664891651,
      "learning_rate": 5.067356042381011e-08,
      "loss": 0.3699,
      "step": 5933
    },
    {
      "epoch": 0.96,
      "grad_norm": 12.121654555619664,
      "learning_rate": 5.0303628066681874e-08,
      "loss": 0.396,
      "step": 5934
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.493532646825219,
      "learning_rate": 4.9935044143284984e-08,
      "loss": 0.4032,
      "step": 5935
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.136654045646493,
      "learning_rate": 4.956780875403189e-08,
      "loss": 0.3645,
      "step": 5936
    },
    {
      "epoch": 0.96,
      "grad_norm": 16.803513505680726,
      "learning_rate": 4.920192199896645e-08,
      "loss": 0.3717,
      "step": 5937
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.894621661522535,
      "learning_rate": 4.883738397776727e-08,
      "loss": 0.4214,
      "step": 5938
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.902692149765108,
      "learning_rate": 4.84741947897438e-08,
      "loss": 0.3576,
      "step": 5939
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.696671894687354,
      "learning_rate": 4.8112354533839664e-08,
      "loss": 0.4199,
      "step": 5940
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.935088507076216,
      "learning_rate": 4.775186330862991e-08,
      "loss": 0.4368,
      "step": 5941
    },
    {
      "epoch": 0.96,
      "grad_norm": 11.359034999257965,
      "learning_rate": 4.7392721212322076e-08,
      "loss": 0.4076,
      "step": 5942
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.156012424957383,
      "learning_rate": 4.703492834275625e-08,
      "loss": 0.3896,
      "step": 5943
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.342539611611853,
      "learning_rate": 4.6678484797405e-08,
      "loss": 0.411,
      "step": 5944
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.0700500862174644,
      "learning_rate": 4.6323390673373434e-08,
      "loss": 0.4819,
      "step": 5945
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.407487929296645,
      "learning_rate": 4.5969646067400285e-08,
      "loss": 0.4312,
      "step": 5946
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.847861297694688,
      "learning_rate": 4.561725107585346e-08,
      "loss": 0.4124,
      "step": 5947
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.557253604952598,
      "learning_rate": 4.5266205794735617e-08,
      "loss": 0.4415,
      "step": 5948
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.622285848101991,
      "learning_rate": 4.491651031968136e-08,
      "loss": 0.4634,
      "step": 5949
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1098755126128153,
      "learning_rate": 4.456816474595782e-08,
      "loss": 0.4185,
      "step": 5950
    },
    {
      "epoch": 0.96,
      "grad_norm": 10.545771803147769,
      "learning_rate": 4.4221169168462975e-08,
      "loss": 0.3651,
      "step": 5951
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.722286213623457,
      "learning_rate": 4.387552368172898e-08,
      "loss": 0.4034,
      "step": 5952
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.306308663369756,
      "learning_rate": 4.35312283799183e-08,
      "loss": 0.4065,
      "step": 5953
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.959896793442997,
      "learning_rate": 4.318828335682701e-08,
      "loss": 0.4284,
      "step": 5954
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.700360230701288,
      "learning_rate": 4.28466887058826e-08,
      "loss": 0.4077,
      "step": 5955
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.503813471126561,
      "learning_rate": 4.250644452014507e-08,
      "loss": 0.4456,
      "step": 5956
    },
    {
      "epoch": 0.96,
      "grad_norm": 18.6928897347029,
      "learning_rate": 4.216755089230584e-08,
      "loss": 0.3787,
      "step": 5957
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.928056008004031,
      "learning_rate": 4.183000791468994e-08,
      "loss": 0.4804,
      "step": 5958
    },
    {
      "epoch": 0.96,
      "grad_norm": 12.208786307475588,
      "learning_rate": 4.149381567925215e-08,
      "loss": 0.4481,
      "step": 5959
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.361227644459601,
      "learning_rate": 4.1158974277580866e-08,
      "loss": 0.4201,
      "step": 5960
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.394638556460001,
      "learning_rate": 4.0825483800895905e-08,
      "loss": 0.4183,
      "step": 5961
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.6884245863734,
      "learning_rate": 4.04933443400507e-08,
      "loss": 0.354,
      "step": 5962
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.721707770805464,
      "learning_rate": 4.0162555985526766e-08,
      "loss": 0.3642,
      "step": 5963
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.619694059761135,
      "learning_rate": 3.9833118827442565e-08,
      "loss": 0.4028,
      "step": 5964
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.30243908453077,
      "learning_rate": 3.950503295554409e-08,
      "loss": 0.4487,
      "step": 5965
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0902467334362889,
      "learning_rate": 3.917829845921095e-08,
      "loss": 0.4376,
      "step": 5966
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.443638228152987,
      "learning_rate": 3.885291542745584e-08,
      "loss": 0.3356,
      "step": 5967
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.076784529531071,
      "learning_rate": 3.8528883948921183e-08,
      "loss": 0.3764,
      "step": 5968
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.819100128845204,
      "learning_rate": 3.8206204111882475e-08,
      "loss": 0.4556,
      "step": 5969
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0060458522836582,
      "learning_rate": 3.788487600424606e-08,
      "loss": 0.4349,
      "step": 5970
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.855538328597058,
      "learning_rate": 3.7564899713550815e-08,
      "loss": 0.4344,
      "step": 5971
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.305390947429676,
      "learning_rate": 3.724627532696812e-08,
      "loss": 0.3409,
      "step": 5972
    },
    {
      "epoch": 0.96,
      "grad_norm": 9.311060838478724,
      "learning_rate": 3.6929002931297975e-08,
      "loss": 0.4236,
      "step": 5973
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.741499464320747,
      "learning_rate": 3.661308261297625e-08,
      "loss": 0.348,
      "step": 5974
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.928163664086753,
      "learning_rate": 3.629851445806687e-08,
      "loss": 0.4815,
      "step": 5975
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.632046612815195,
      "learning_rate": 3.59852985522674e-08,
      "loss": 0.4048,
      "step": 5976
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.010620011046495,
      "learning_rate": 3.5673434980906806e-08,
      "loss": 0.4678,
      "step": 5977
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.382062508792096,
      "learning_rate": 3.536292382894435e-08,
      "loss": 0.4098,
      "step": 5978
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3178493772666569,
      "learning_rate": 3.505376518097292e-08,
      "loss": 0.4851,
      "step": 5979
    },
    {
      "epoch": 0.96,
      "grad_norm": 10.621455062820207,
      "learning_rate": 3.474595912121514e-08,
      "loss": 0.4085,
      "step": 5980
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.280743721388792,
      "learning_rate": 3.443950573352672e-08,
      "loss": 0.4556,
      "step": 5981
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.774075484445219,
      "learning_rate": 3.41344051013931e-08,
      "loss": 0.3951,
      "step": 5982
    },
    {
      "epoch": 0.96,
      "grad_norm": 13.129593829413892,
      "learning_rate": 3.3830657307932224e-08,
      "loss": 0.3486,
      "step": 5983
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.145772291670653,
      "learning_rate": 3.352826243589346e-08,
      "loss": 0.4234,
      "step": 5984
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.821949491111831,
      "learning_rate": 3.322722056765759e-08,
      "loss": 0.416,
      "step": 5985
    },
    {
      "epoch": 0.96,
      "grad_norm": 12.153632784286645,
      "learning_rate": 3.292753178523733e-08,
      "loss": 0.3008,
      "step": 5986
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.116915657369894,
      "learning_rate": 3.262919617027516e-08,
      "loss": 0.3465,
      "step": 5987
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.121280067440487,
      "learning_rate": 3.233221380404605e-08,
      "loss": 0.472,
      "step": 5988
    },
    {
      "epoch": 0.96,
      "grad_norm": 12.925447474057027,
      "learning_rate": 3.2036584767456965e-08,
      "loss": 0.4176,
      "step": 5989
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.974661526885103,
      "learning_rate": 3.1742309141044594e-08,
      "loss": 0.4365,
      "step": 5990
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.200548717387913,
      "learning_rate": 3.1449387004978125e-08,
      "loss": 0.4252,
      "step": 5991
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.323820782915598,
      "learning_rate": 3.115781843905763e-08,
      "loss": 0.44,
      "step": 5992
    },
    {
      "epoch": 0.97,
      "grad_norm": 11.325014056647124,
      "learning_rate": 3.086760352271401e-08,
      "loss": 0.3958,
      "step": 5993
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.1471808269529475,
      "learning_rate": 3.057874233501068e-08,
      "loss": 0.3872,
      "step": 5994
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.138364169444989,
      "learning_rate": 3.0291234954640256e-08,
      "loss": 0.4126,
      "step": 5995
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.070428328212516,
      "learning_rate": 3.0005081459928976e-08,
      "loss": 0.4044,
      "step": 5996
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.027306126888241,
      "learning_rate": 2.9720281928831694e-08,
      "loss": 0.3495,
      "step": 5997
    },
    {
      "epoch": 0.97,
      "grad_norm": 15.484251643676506,
      "learning_rate": 2.9436836438936356e-08,
      "loss": 0.3925,
      "step": 5998
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.981061043060834,
      "learning_rate": 2.9154745067460632e-08,
      "loss": 0.387,
      "step": 5999
    },
    {
      "epoch": 0.97,
      "grad_norm": 8.013327818517764,
      "learning_rate": 2.8874007891255273e-08,
      "loss": 0.4435,
      "step": 6000
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.845035399263454,
      "learning_rate": 2.859462498679966e-08,
      "loss": 0.4426,
      "step": 6001
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.067345512731572,
      "learning_rate": 2.831659643020568e-08,
      "loss": 0.4305,
      "step": 6002
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.050833092650646,
      "learning_rate": 2.8039922297216638e-08,
      "loss": 0.4216,
      "step": 6003
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.716401291648597,
      "learning_rate": 2.776460266320502e-08,
      "loss": 0.4236,
      "step": 6004
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.350410826299381,
      "learning_rate": 2.7490637603176385e-08,
      "loss": 0.3445,
      "step": 6005
    },
    {
      "epoch": 0.97,
      "grad_norm": 8.35827400590957,
      "learning_rate": 2.7218027191766027e-08,
      "loss": 0.3516,
      "step": 6006
    },
    {
      "epoch": 0.97,
      "grad_norm": 9.334357280179175,
      "learning_rate": 2.6946771503240653e-08,
      "loss": 0.41,
      "step": 6007
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.768189884383511,
      "learning_rate": 2.6676870611497817e-08,
      "loss": 0.3524,
      "step": 6008
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.548425267408917,
      "learning_rate": 2.6408324590065926e-08,
      "loss": 0.4556,
      "step": 6009
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.4917488053342565,
      "learning_rate": 2.6141133512103677e-08,
      "loss": 0.3823,
      "step": 6010
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.272435594183172,
      "learning_rate": 2.5875297450402848e-08,
      "loss": 0.462,
      "step": 6011
    },
    {
      "epoch": 0.97,
      "grad_norm": 35.1506432110317,
      "learning_rate": 2.5610816477382728e-08,
      "loss": 0.4226,
      "step": 6012
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.088717262821666,
      "learning_rate": 2.5347690665096236e-08,
      "loss": 0.428,
      "step": 6013
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.822472235741214,
      "learning_rate": 2.5085920085226035e-08,
      "loss": 0.3644,
      "step": 6014
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.80356632167269,
      "learning_rate": 2.482550480908563e-08,
      "loss": 0.4111,
      "step": 6015
    },
    {
      "epoch": 0.97,
      "grad_norm": 10.576413389135011,
      "learning_rate": 2.456644490761939e-08,
      "loss": 0.4035,
      "step": 6016
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.073589034326763,
      "learning_rate": 2.4308740451401413e-08,
      "loss": 0.4074,
      "step": 6017
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.101722832469393,
      "learning_rate": 2.4052391510638873e-08,
      "loss": 0.456,
      "step": 6018
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.188207502027041,
      "learning_rate": 2.379739815516757e-08,
      "loss": 0.4172,
      "step": 6019
    },
    {
      "epoch": 0.97,
      "grad_norm": 9.640262788199468,
      "learning_rate": 2.354376045445472e-08,
      "loss": 0.3838,
      "step": 6020
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.859405950750676,
      "learning_rate": 2.3291478477598383e-08,
      "loss": 0.4327,
      "step": 6021
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.160002489421595,
      "learning_rate": 2.304055229332691e-08,
      "loss": 0.3565,
      "step": 6022
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.910831820358226,
      "learning_rate": 2.279098197000007e-08,
      "loss": 0.4103,
      "step": 6023
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.3546984441467425,
      "learning_rate": 2.254276757560736e-08,
      "loss": 0.4485,
      "step": 6024
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.163151368607638,
      "learning_rate": 2.2295909177769138e-08,
      "loss": 0.4853,
      "step": 6025
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.7349808329194625,
      "learning_rate": 2.205040684373605e-08,
      "loss": 0.3533,
      "step": 6026
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.222925335470593,
      "learning_rate": 2.1806260640390155e-08,
      "loss": 0.432,
      "step": 6027
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.4717750983998545,
      "learning_rate": 2.15634706342438e-08,
      "loss": 0.353,
      "step": 6028
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.799953132698664,
      "learning_rate": 2.1322036891439634e-08,
      "loss": 0.4373,
      "step": 6029
    },
    {
      "epoch": 0.97,
      "grad_norm": 10.125779045981059,
      "learning_rate": 2.1081959477750604e-08,
      "loss": 0.5125,
      "step": 6030
    },
    {
      "epoch": 0.97,
      "grad_norm": 8.440219872737547,
      "learning_rate": 2.0843238458580494e-08,
      "loss": 0.4668,
      "step": 6031
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.5206951620778995,
      "learning_rate": 2.060587389896285e-08,
      "loss": 0.4048,
      "step": 6032
    },
    {
      "epoch": 0.97,
      "grad_norm": 9.641644872522267,
      "learning_rate": 2.0369865863563708e-08,
      "loss": 0.4048,
      "step": 6033
    },
    {
      "epoch": 0.97,
      "grad_norm": 17.07710476526659,
      "learning_rate": 2.0135214416677205e-08,
      "loss": 0.3758,
      "step": 6034
    },
    {
      "epoch": 0.97,
      "grad_norm": 8.072103970624342,
      "learning_rate": 1.990191962222887e-08,
      "loss": 0.4563,
      "step": 6035
    },
    {
      "epoch": 0.97,
      "grad_norm": 9.036322343789793,
      "learning_rate": 1.9669981543775085e-08,
      "loss": 0.4108,
      "step": 6036
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.468016271365962,
      "learning_rate": 1.943940024450197e-08,
      "loss": 0.4287,
      "step": 6037
    },
    {
      "epoch": 0.97,
      "grad_norm": 8.982370095216224,
      "learning_rate": 1.9210175787226503e-08,
      "loss": 0.4124,
      "step": 6038
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.207761329786003,
      "learning_rate": 1.898230823439484e-08,
      "loss": 0.3655,
      "step": 6039
    },
    {
      "epoch": 0.97,
      "grad_norm": 8.36000357932267,
      "learning_rate": 1.8755797648085105e-08,
      "loss": 0.414,
      "step": 6040
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.971061568619134,
      "learning_rate": 1.8530644090005156e-08,
      "loss": 0.4566,
      "step": 6041
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.227771808491149,
      "learning_rate": 1.8306847621492042e-08,
      "loss": 0.3761,
      "step": 6042
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.503577117007659,
      "learning_rate": 1.808440830351532e-08,
      "loss": 0.3914,
      "step": 6043
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.608909397934487,
      "learning_rate": 1.7863326196673193e-08,
      "loss": 0.3955,
      "step": 6044
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.921338206461803,
      "learning_rate": 1.764360136119414e-08,
      "loss": 0.3871,
      "step": 6045
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.3376490927286895,
      "learning_rate": 1.742523385693806e-08,
      "loss": 0.4085,
      "step": 6046
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.271484616459547,
      "learning_rate": 1.7208223743392927e-08,
      "loss": 0.3339,
      "step": 6047
    },
    {
      "epoch": 0.97,
      "grad_norm": 8.089954834932664,
      "learning_rate": 1.6992571079679775e-08,
      "loss": 0.3662,
      "step": 6048
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.419170742417445,
      "learning_rate": 1.6778275924547726e-08,
      "loss": 0.3948,
      "step": 6049
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.533755278309194,
      "learning_rate": 1.656533833637619e-08,
      "loss": 0.4893,
      "step": 6050
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.1627924485816115,
      "learning_rate": 1.6353758373175986e-08,
      "loss": 0.455,
      "step": 6051
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.453473858991174,
      "learning_rate": 1.6143536092586564e-08,
      "loss": 0.4363,
      "step": 6052
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.438991704220839,
      "learning_rate": 1.593467155187933e-08,
      "loss": 0.449,
      "step": 6053
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.113592727555225,
      "learning_rate": 1.5727164807953777e-08,
      "loss": 0.3305,
      "step": 6054
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.532545517923866,
      "learning_rate": 1.5521015917340787e-08,
      "loss": 0.3631,
      "step": 6055
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.815106777587831,
      "learning_rate": 1.5316224936200997e-08,
      "loss": 0.4427,
      "step": 6056
    },
    {
      "epoch": 0.98,
      "grad_norm": 15.847134939799764,
      "learning_rate": 1.5112791920325332e-08,
      "loss": 0.3917,
      "step": 6057
    },
    {
      "epoch": 0.98,
      "grad_norm": 16.843670174152184,
      "learning_rate": 1.491071692513446e-08,
      "loss": 0.4279,
      "step": 6058
    },
    {
      "epoch": 0.98,
      "grad_norm": 33.4052993289339,
      "learning_rate": 1.4710000005678792e-08,
      "loss": 0.3912,
      "step": 6059
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.31791918515353,
      "learning_rate": 1.4510641216639587e-08,
      "loss": 0.3296,
      "step": 6060
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.7520733528678925,
      "learning_rate": 1.4312640612327289e-08,
      "loss": 0.4864,
      "step": 6061
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.861304287285569,
      "learning_rate": 1.4115998246683193e-08,
      "loss": 0.4014,
      "step": 6062
    },
    {
      "epoch": 0.98,
      "grad_norm": 9.461470870694528,
      "learning_rate": 1.3920714173278338e-08,
      "loss": 0.4587,
      "step": 6063
    },
    {
      "epoch": 0.98,
      "grad_norm": 11.489053227249872,
      "learning_rate": 1.372678844531239e-08,
      "loss": 0.3613,
      "step": 6064
    },
    {
      "epoch": 0.98,
      "grad_norm": 14.099669619567528,
      "learning_rate": 1.3534221115616975e-08,
      "loss": 0.3677,
      "step": 6065
    },
    {
      "epoch": 0.98,
      "grad_norm": 10.391123156960205,
      "learning_rate": 1.3343012236652353e-08,
      "loss": 0.3675,
      "step": 6066
    },
    {
      "epoch": 0.98,
      "grad_norm": 9.674963939351567,
      "learning_rate": 1.3153161860509078e-08,
      "loss": 0.4745,
      "step": 6067
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.065294697914851,
      "learning_rate": 1.2964670038908e-08,
      "loss": 0.3566,
      "step": 6068
    },
    {
      "epoch": 0.98,
      "grad_norm": 27.473325332228793,
      "learning_rate": 1.2777536823199155e-08,
      "loss": 0.3834,
      "step": 6069
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.6655308473195625,
      "learning_rate": 1.2591762264362872e-08,
      "loss": 0.3946,
      "step": 6070
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.020211802786527,
      "learning_rate": 1.2407346413009224e-08,
      "loss": 0.4304,
      "step": 6071
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.808388710844722,
      "learning_rate": 1.2224289319378025e-08,
      "loss": 0.3905,
      "step": 6072
    },
    {
      "epoch": 0.98,
      "grad_norm": 12.23554763369316,
      "learning_rate": 1.204259103333938e-08,
      "loss": 0.4683,
      "step": 6073
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.370507947373715,
      "learning_rate": 1.1862251604393138e-08,
      "loss": 0.3781,
      "step": 6074
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.93501585259362,
      "learning_rate": 1.1683271081668334e-08,
      "loss": 0.3382,
      "step": 6075
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.830234682021126,
      "learning_rate": 1.1505649513923744e-08,
      "loss": 0.3947,
      "step": 6076
    },
    {
      "epoch": 0.98,
      "grad_norm": 16.45824783868109,
      "learning_rate": 1.1329386949548993e-08,
      "loss": 0.418,
      "step": 6077
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.570556827407191,
      "learning_rate": 1.115448343656289e-08,
      "loss": 0.4123,
      "step": 6078
    },
    {
      "epoch": 0.98,
      "grad_norm": 9.89208313669047,
      "learning_rate": 1.098093902261399e-08,
      "loss": 0.3959,
      "step": 6079
    },
    {
      "epoch": 0.98,
      "grad_norm": 9.190697344643453,
      "learning_rate": 1.0808753754980029e-08,
      "loss": 0.3755,
      "step": 6080
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.440818441358894,
      "learning_rate": 1.0637927680570149e-08,
      "loss": 0.3819,
      "step": 6081
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.07350933645867,
      "learning_rate": 1.0468460845921014e-08,
      "loss": 0.3578,
      "step": 6082
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.766170133810189,
      "learning_rate": 1.0300353297200139e-08,
      "loss": 0.4433,
      "step": 6083
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.717861764776151,
      "learning_rate": 1.0133605080204779e-08,
      "loss": 0.4559,
      "step": 6084
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.357904689740886,
      "learning_rate": 9.968216240361927e-09,
      "loss": 0.4218,
      "step": 6085
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.304135888341806,
      "learning_rate": 9.804186822728324e-09,
      "loss": 0.4177,
      "step": 6086
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.5765103713169974,
      "learning_rate": 9.641516871989331e-09,
      "loss": 0.3729,
      "step": 6087
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.899938727792872,
      "learning_rate": 9.480206432461725e-09,
      "loss": 0.424,
      "step": 6088
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.651215525487839,
      "learning_rate": 9.32025554809035e-09,
      "loss": 0.3798,
      "step": 6089
    },
    {
      "epoch": 0.98,
      "grad_norm": 10.4178826421065,
      "learning_rate": 9.161664262449799e-09,
      "loss": 0.4189,
      "step": 6090
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.350177918918874,
      "learning_rate": 9.004432618745507e-09,
      "loss": 0.4377,
      "step": 6091
    },
    {
      "epoch": 0.98,
      "grad_norm": 10.52618921866084,
      "learning_rate": 8.848560659810989e-09,
      "loss": 0.4496,
      "step": 6092
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.796982675187409,
      "learning_rate": 8.694048428110614e-09,
      "loss": 0.4422,
      "step": 6093
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.264471589571443,
      "learning_rate": 8.540895965737928e-09,
      "loss": 0.3813,
      "step": 6094
    },
    {
      "epoch": 0.98,
      "grad_norm": 18.847214823380735,
      "learning_rate": 8.38910331441567e-09,
      "loss": 0.4607,
      "step": 6095
    },
    {
      "epoch": 0.98,
      "grad_norm": 29.551052843500198,
      "learning_rate": 8.23867051549576e-09,
      "loss": 0.4111,
      "step": 6096
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.473369761896919,
      "learning_rate": 8.089597609960976e-09,
      "loss": 0.493,
      "step": 6097
    },
    {
      "epoch": 0.98,
      "grad_norm": 11.709032964177176,
      "learning_rate": 7.94188463842216e-09,
      "loss": 0.3716,
      "step": 6098
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.8048252363823805,
      "learning_rate": 7.795531641121013e-09,
      "loss": 0.442,
      "step": 6099
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.4448797375536495,
      "learning_rate": 7.65053865792842e-09,
      "loss": 0.3508,
      "step": 6100
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.985813284660222,
      "learning_rate": 7.506905728343893e-09,
      "loss": 0.37,
      "step": 6101
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.069218528610057,
      "learning_rate": 7.364632891496692e-09,
      "loss": 0.3951,
      "step": 6102
    },
    {
      "epoch": 0.98,
      "grad_norm": 9.04308757565744,
      "learning_rate": 7.223720186146366e-09,
      "loss": 0.4016,
      "step": 6103
    },
    {
      "epoch": 0.98,
      "grad_norm": 10.737650633776012,
      "learning_rate": 7.08416765068165e-09,
      "loss": 0.4,
      "step": 6104
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.138620122231401,
      "learning_rate": 6.945975323119913e-09,
      "loss": 0.3915,
      "step": 6105
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.593341527781504,
      "learning_rate": 6.809143241109373e-09,
      "loss": 0.4294,
      "step": 6106
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.669509005373945,
      "learning_rate": 6.673671441925766e-09,
      "loss": 0.4769,
      "step": 6107
    },
    {
      "epoch": 0.98,
      "grad_norm": 9.09312722026993,
      "learning_rate": 6.539559962476238e-09,
      "loss": 0.4178,
      "step": 6108
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.5198948654101425,
      "learning_rate": 6.4068088392960084e-09,
      "loss": 0.3464,
      "step": 6109
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.073676465949116,
      "learning_rate": 6.275418108550591e-09,
      "loss": 0.3973,
      "step": 6110
    },
    {
      "epoch": 0.98,
      "grad_norm": 9.220166490288445,
      "learning_rate": 6.1453878060335755e-09,
      "loss": 0.3546,
      "step": 6111
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.142059881878264,
      "learning_rate": 6.0167179671694055e-09,
      "loss": 0.414,
      "step": 6112
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0801394384599652,
      "learning_rate": 5.8894086270111505e-09,
      "loss": 0.449,
      "step": 6113
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.202552803901914,
      "learning_rate": 5.7634598202416235e-09,
      "loss": 0.4114,
      "step": 6114
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1227019272606347,
      "learning_rate": 5.638871581172822e-09,
      "loss": 0.4401,
      "step": 6115
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.907913478093451,
      "learning_rate": 5.515643943745375e-09,
      "loss": 0.4063,
      "step": 6116
    },
    {
      "epoch": 0.99,
      "grad_norm": 14.254601387068977,
      "learning_rate": 5.393776941530759e-09,
      "loss": 0.4105,
      "step": 6117
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.081722083976935,
      "learning_rate": 5.273270607727976e-09,
      "loss": 0.388,
      "step": 6118
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.368456224003473,
      "learning_rate": 5.1541249751668745e-09,
      "loss": 0.3865,
      "step": 6119
    },
    {
      "epoch": 0.99,
      "grad_norm": 8.331727097624388,
      "learning_rate": 5.0363400763059346e-09,
      "loss": 0.4022,
      "step": 6120
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3116252920895939,
      "learning_rate": 4.919915943232822e-09,
      "loss": 0.4794,
      "step": 6121
    },
    {
      "epoch": 0.99,
      "grad_norm": 21.66870043846836,
      "learning_rate": 4.80485260766439e-09,
      "loss": 0.4469,
      "step": 6122
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.859805987432393,
      "learning_rate": 4.691150100948338e-09,
      "loss": 0.4298,
      "step": 6123
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.216466742032423,
      "learning_rate": 4.578808454058781e-09,
      "loss": 0.5016,
      "step": 6124
    },
    {
      "epoch": 0.99,
      "grad_norm": 23.178312008392336,
      "learning_rate": 4.4678276976017895e-09,
      "loss": 0.4078,
      "step": 6125
    },
    {
      "epoch": 0.99,
      "grad_norm": 10.946148994979515,
      "learning_rate": 4.358207861810959e-09,
      "loss": 0.4062,
      "step": 6126
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.917774853260408,
      "learning_rate": 4.249948976550178e-09,
      "loss": 0.355,
      "step": 6127
    },
    {
      "epoch": 0.99,
      "grad_norm": 12.102880410817235,
      "learning_rate": 4.143051071311965e-09,
      "loss": 0.4048,
      "step": 6128
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.663957136127433,
      "learning_rate": 4.037514175218027e-09,
      "loss": 0.4194,
      "step": 6129
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.394378401234621,
      "learning_rate": 3.933338317019808e-09,
      "loss": 0.4383,
      "step": 6130
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.892768247197947,
      "learning_rate": 3.83052352509794e-09,
      "loss": 0.4829,
      "step": 6131
    },
    {
      "epoch": 0.99,
      "grad_norm": 9.647023320495624,
      "learning_rate": 3.729069827461685e-09,
      "loss": 0.3949,
      "step": 6132
    },
    {
      "epoch": 0.99,
      "grad_norm": 13.554356542917054,
      "learning_rate": 3.628977251749488e-09,
      "loss": 0.4906,
      "step": 6133
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0527465350347402,
      "learning_rate": 3.530245825229539e-09,
      "loss": 0.4369,
      "step": 6134
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.878133802501732,
      "learning_rate": 3.4328755747992103e-09,
      "loss": 0.4155,
      "step": 6135
    },
    {
      "epoch": 0.99,
      "grad_norm": 8.015421641386913,
      "learning_rate": 3.336866526985061e-09,
      "loss": 0.4232,
      "step": 6136
    },
    {
      "epoch": 0.99,
      "grad_norm": 10.898180899022222,
      "learning_rate": 3.2422187079417242e-09,
      "loss": 0.4286,
      "step": 6137
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.110808241441318,
      "learning_rate": 3.148932143455241e-09,
      "loss": 0.4444,
      "step": 6138
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.346760391923627,
      "learning_rate": 3.0570068589380606e-09,
      "loss": 0.3917,
      "step": 6139
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.020984681275212,
      "learning_rate": 2.9664428794340393e-09,
      "loss": 0.4735,
      "step": 6140
    },
    {
      "epoch": 0.99,
      "grad_norm": 9.43866629901102,
      "learning_rate": 2.877240229614553e-09,
      "loss": 0.358,
      "step": 6141
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.72577302897655,
      "learning_rate": 2.7893989337818283e-09,
      "loss": 0.379,
      "step": 6142
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.987038193873231,
      "learning_rate": 2.7029190158656125e-09,
      "loss": 0.417,
      "step": 6143
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.232591282472939,
      "learning_rate": 2.6178004994253936e-09,
      "loss": 0.4136,
      "step": 6144
    },
    {
      "epoch": 0.99,
      "grad_norm": 10.179370395692516,
      "learning_rate": 2.5340434076503994e-09,
      "loss": 0.4398,
      "step": 6145
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1745821739528415,
      "learning_rate": 2.4516477633579338e-09,
      "loss": 0.4839,
      "step": 6146
    },
    {
      "epoch": 0.99,
      "grad_norm": 8.313312570973585,
      "learning_rate": 2.370613588994486e-09,
      "loss": 0.3689,
      "step": 6147
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.362322241481521,
      "learning_rate": 2.2909409066362854e-09,
      "loss": 0.4822,
      "step": 6148
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.397634502037918,
      "learning_rate": 2.2126297379887473e-09,
      "loss": 0.4366,
      "step": 6149
    },
    {
      "epoch": 0.99,
      "grad_norm": 10.84490569463106,
      "learning_rate": 2.1356801043853624e-09,
      "loss": 0.3666,
      "step": 6150
    },
    {
      "epoch": 0.99,
      "grad_norm": 9.728637687901125,
      "learning_rate": 2.060092026789917e-09,
      "loss": 0.4045,
      "step": 6151
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.316861827030633,
      "learning_rate": 1.9858655257942726e-09,
      "loss": 0.4193,
      "step": 6152
    },
    {
      "epoch": 0.99,
      "grad_norm": 10.141829091733241,
      "learning_rate": 1.9130006216200314e-09,
      "loss": 0.467,
      "step": 6153
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.727186055319543,
      "learning_rate": 1.841497334117426e-09,
      "loss": 0.4014,
      "step": 6154
    },
    {
      "epoch": 0.99,
      "grad_norm": 8.021105975067629,
      "learning_rate": 1.771355682765874e-09,
      "loss": 0.4793,
      "step": 6155
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.364216295047285,
      "learning_rate": 1.7025756866739795e-09,
      "loss": 0.3746,
      "step": 6156
    },
    {
      "epoch": 0.99,
      "grad_norm": 23.229018878342714,
      "learning_rate": 1.6351573645795316e-09,
      "loss": 0.4345,
      "step": 6157
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.1957081400054985,
      "learning_rate": 1.5691007348489495e-09,
      "loss": 0.3539,
      "step": 6158
    },
    {
      "epoch": 0.99,
      "grad_norm": 26.040534784527363,
      "learning_rate": 1.5044058154778385e-09,
      "loss": 0.4139,
      "step": 6159
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.224006707459801,
      "learning_rate": 1.441072624090989e-09,
      "loss": 0.3572,
      "step": 6160
    },
    {
      "epoch": 0.99,
      "grad_norm": 14.733831569741248,
      "learning_rate": 1.3791011779423769e-09,
      "loss": 0.4434,
      "step": 6161
    },
    {
      "epoch": 0.99,
      "grad_norm": 10.213729027054061,
      "learning_rate": 1.3184914939140537e-09,
      "loss": 0.4228,
      "step": 6162
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.48150459004017,
      "learning_rate": 1.2592435885178112e-09,
      "loss": 0.3246,
      "step": 6163
    },
    {
      "epoch": 0.99,
      "grad_norm": 12.120620765998419,
      "learning_rate": 1.2013574778951819e-09,
      "loss": 0.4133,
      "step": 6164
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.467612450847542,
      "learning_rate": 1.1448331778152189e-09,
      "loss": 0.3639,
      "step": 6165
    },
    {
      "epoch": 0.99,
      "grad_norm": 9.181595434787408,
      "learning_rate": 1.0896707036772703e-09,
      "loss": 0.4424,
      "step": 6166
    },
    {
      "epoch": 0.99,
      "grad_norm": 14.609234451513899,
      "learning_rate": 1.0358700705082048e-09,
      "loss": 0.4195,
      "step": 6167
    },
    {
      "epoch": 0.99,
      "grad_norm": 12.651273337820202,
      "learning_rate": 9.834312929657419e-10,
      "loss": 0.4055,
      "step": 6168
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.954822724360298,
      "learning_rate": 9.323543853351212e-10,
      "loss": 0.4844,
      "step": 6169
    },
    {
      "epoch": 0.99,
      "grad_norm": 10.223454056212427,
      "learning_rate": 8.826393615318784e-10,
      "loss": 0.4072,
      "step": 6170
    },
    {
      "epoch": 0.99,
      "grad_norm": 8.021186471007278,
      "learning_rate": 8.342862350985137e-10,
      "loss": 0.3348,
      "step": 6171
    },
    {
      "epoch": 0.99,
      "grad_norm": 8.915500727027704,
      "learning_rate": 7.872950192083783e-10,
      "loss": 0.3985,
      "step": 6172
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.061696961393825,
      "learning_rate": 7.416657266634542e-10,
      "loss": 0.4145,
      "step": 6173
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.597530538370256,
      "learning_rate": 6.973983698943532e-10,
      "loss": 0.3481,
      "step": 6174
    },
    {
      "epoch": 0.99,
      "grad_norm": 8.335415548569067,
      "learning_rate": 6.544929609597628e-10,
      "loss": 0.4649,
      "step": 6175
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.227930452211167,
      "learning_rate": 6.129495115497764e-10,
      "loss": 0.4048,
      "step": 6176
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.029770965236446,
      "learning_rate": 5.727680329808971e-10,
      "loss": 0.4794,
      "step": 6177
    },
    {
      "epoch": 1.0,
      "grad_norm": 9.520516198579397,
      "learning_rate": 5.339485362004793e-10,
      "loss": 0.3492,
      "step": 6178
    },
    {
      "epoch": 1.0,
      "grad_norm": 9.116180764939346,
      "learning_rate": 4.96491031782842e-10,
      "loss": 0.4271,
      "step": 6179
    },
    {
      "epoch": 1.0,
      "grad_norm": 8.921383798167614,
      "learning_rate": 4.603955299337104e-10,
      "loss": 0.4318,
      "step": 6180
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0746654954229136,
      "learning_rate": 4.2566204048577473e-10,
      "loss": 0.4271,
      "step": 6181
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.969422497493513,
      "learning_rate": 3.9229057290146587e-10,
      "loss": 0.3772,
      "step": 6182
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1897269848044565,
      "learning_rate": 3.6028113627240015e-10,
      "loss": 0.4551,
      "step": 6183
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.554933802975019,
      "learning_rate": 3.2963373931882425e-10,
      "loss": 0.4483,
      "step": 6184
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.019045594479564,
      "learning_rate": 3.003483903890603e-10,
      "loss": 0.3606,
      "step": 6185
    },
    {
      "epoch": 1.0,
      "grad_norm": 10.078260501888897,
      "learning_rate": 2.724250974628362e-10,
      "loss": 0.3736,
      "step": 6186
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.738962420198265,
      "learning_rate": 2.458638681457348e-10,
      "loss": 0.4026,
      "step": 6187
    },
    {
      "epoch": 1.0,
      "grad_norm": 10.910467437794013,
      "learning_rate": 2.2066470967418985e-10,
      "loss": 0.338,
      "step": 6188
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.522931697443439,
      "learning_rate": 1.9682762891382045e-10,
      "loss": 0.4746,
      "step": 6189
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.459417452162901,
      "learning_rate": 1.74352632357766e-10,
      "loss": 0.3763,
      "step": 6190
    },
    {
      "epoch": 1.0,
      "grad_norm": 11.202578332491813,
      "learning_rate": 1.5323972612890647e-10,
      "loss": 0.4162,
      "step": 6191
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.634305084844753,
      "learning_rate": 1.3348891597930734e-10,
      "loss": 0.4169,
      "step": 6192
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.422162267430779,
      "learning_rate": 1.1510020728910943e-10,
      "loss": 0.4444,
      "step": 6193
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.87311337589163,
      "learning_rate": 9.807360506874919e-11,
      "loss": 0.3867,
      "step": 6194
    },
    {
      "epoch": 1.0,
      "grad_norm": 9.567841007067475,
      "learning_rate": 8.240911395562823e-11,
      "loss": 0.4014,
      "step": 6195
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.59661168114939,
      "learning_rate": 6.810673821855407e-11,
      "loss": 0.409,
      "step": 6196
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.727390032355771,
      "learning_rate": 5.516648175274419e-11,
      "loss": 0.3805,
      "step": 6197
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.456222955044605,
      "learning_rate": 4.358834808371182e-11,
      "loss": 0.3949,
      "step": 6198
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.997693936588838,
      "learning_rate": 3.3372340366155663e-11,
      "loss": 0.3969,
      "step": 6199
    },
    {
      "epoch": 1.0,
      "grad_norm": 9.391928347366203,
      "learning_rate": 2.4518461382849745e-11,
      "loss": 0.4083,
      "step": 6200
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.8280251099775136,
      "learning_rate": 1.702671354575358e-11,
      "loss": 0.4475,
      "step": 6201
    },
    {
      "epoch": 1.0,
      "grad_norm": 18.325389450510528,
      "learning_rate": 1.0897098895457092e-11,
      "loss": 0.4361,
      "step": 6202
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.00372671048808,
      "learning_rate": 6.129619102845929e-12,
      "loss": 0.4431,
      "step": 6203
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.627549124373033,
      "learning_rate": 2.7242754663259207e-12,
      "loss": 0.4502,
      "step": 6204
    },
    {
      "epoch": 1.0,
      "grad_norm": 13.06124087765242,
      "learning_rate": 6.810689129332915e-13,
      "loss": 0.4299,
      "step": 6205
    },
    {
      "epoch": 1.0,
      "grad_norm": 19.593519222619403,
      "learning_rate": 0.0,
      "loss": 0.4042,
      "step": 6206
    },
    {
      "epoch": 1.0,
      "step": 6206,
      "total_flos": 5571263301926912.0,
      "train_loss": 0.45520276639429247,
      "train_runtime": 233098.4711,
      "train_samples_per_second": 3.408,
      "train_steps_per_second": 0.027
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 6206,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 3000,
  "total_flos": 5571263301926912.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}